The CPU Shack Museum CPU History Museum for Intel CPUs, AMD Processor, Cyrix Microprocessors, Microcontrollers and more. Thu, 25 Feb 2021 19:10:19 +0000 en-US hourly 1 The 486 CPU Era – The Birth of Overclocking. – Part 2 Thu, 25 Feb 2021 19:07:50 +0000 In Part 1 of The 486 CPU Era – The Birth of Overclocking, we covered some of the basics of the 486 era and where it came from, as well as the various brands/types of 486s of the era (many of which we will test and attempt to overclock.  In Part 2 we will discuss […]

The post The 486 CPU Era – The Birth of Overclocking. – Part 2 first appeared on The CPU Shack Museum.]]>
In Part 1 of The 486 CPU Era – The Birth of Overclocking, we covered some of the basics of the 486 era and where it came from, as well as the various brands/types of 486s of the era (many of which we will test and attempt to overclock.  In Part 2 we will discuss the hardware selection and rational, testing environment and benchmarks! (and a healthy dose of Overclocking with some perhaps surprising results)

Choosing a Motherboard

Socket 5, GIGABYTE GA586AM, UM8891BF / UM8892BF chipset – Good but not good enough

Choosing a motherboard for the 80486 platform is not easy. There are several criteria or approaches for the implementation of such projects. 1. Consider whether you need PCI slots? 2. The need for VLB slot(s) 3. The need for everything on one board.

Since I set myself the task of assembling the most productive Socket 3 system, the presence of ISA and VLB slots was a secondary matter for me, PCI slots were a priority due to their speed characteristics. The fastest chipset was required from the motherboard – this is the UMC 8886/8881. Revisions of this chipset were later used in Socket 5 Pentium motherboards that supported FSB 60/66 MHz and higher. The board must have 4 slots for RAM with support for EDO RAM, the minimum total size is 128 MB (4x 32 MB).

The total size of the L2 cache should be equal to 1 MB, so the motherboard should contain 8 sockets for such microcircuits.

Due to the use of different processors with different input voltages, the board must support a choice of voltages from 3.3 V to 5 V in small steps, in order to be able to “smooth” overclocking. Accordingly, the overclocking capability on the bus from 33 to 50 MHz and higher should be implemented. So which board do we end up with?

Socket3, Shuttle HOT-433, UMC 8886AF / 8881F chipset

I was looking for such a board for a long time, I spent more than a year looking for it. There is a very small number of models of such motherboards that satisfy all the requirements I have described above and which, let’s say, are problematic to find, and the cost of such a motherboard is sometimes commensurate with the cost of an initial or average modern motherboard, if you look for it on the world flea market eBay. Although you may be lucky and find an exhibit for a penny, I was “unlucky” and I bought my copy for a substantial amount, although now it has grown in value due to the shortage. It was not easy to part with money, given the complete practical uselessness of such a product these days, but it was worth it!

As a result, I settled on a version of the motherboard manufactured by Shuttle, the HOT-433, based on the UMC 8886AF / 8881F chipset. For its age, this board looks gorgeous, befitting its name =)

Shuttle, or formerly Holco, is a major Taiwanese manufacturer of motherboards and PCs that has been operating since 1983. I personally associate Shuttle with the XPC Barebone mini PC or compact PCs in interesting multimedia cases, which flourished in the second half of the 2000s.

As you can see from the photo, we have an AT motherboard form factor, so you need to use an AT power supply, or use an AT->ATX adapter, which I did. I have connected a good 500W FSP PSU, which will be enough power for this assembly with more than five times the headroom or even more. (todays video cards require more power then an entire 486 system back in the day).

DIN-5 Keyboard

There is also a DIN-5 connector on the board for connecting an old-style keyboard. You can use an authentic retro keyboard, or connect your favorite modern keyboard through a DIN-PS/2 adapter (what? you want USB?). During testing, I used both options, both keyboards worked without problems.

One of the key advantages of Shuttle HOT-433 is support for 128 MB of ordinary EDO RAM, or 256 MB of registered EDO RAM! By modern standards, this is like 128 and 256 GB of DDR4 RAM for a modern performance system now. A very limited number of Socket 3 motherboards are capable of supporting this amount of RAM.

128MB od 50ns EDO

The motherboard also officially supports the installation of an external 1MB L2 cache, for which 8 tag ram slots are available, which will cache all 256MB of RAM. I used 4 modules of 32 MB 60ns EDO RAM manufactured by Kingston, which gave a total of 128 MB and a 1MB L2 cache (15ns). This volume is enough even for the comfortable work of Windows XP, and with 256 MB you can swing at Windows Vista =)

The motherboard VRM supports a wide variety of voltages. The user has access to values: 3.3, 3.45, 3.6, 4.0 and 5.0 volts, which allows you the flexibly set the desired values during overclocking.

1MB of 15ns Cache

Due to the documented capabilities of Shuttle HOT-433 it allows one to select the following system bus values: 25, 33, 40, 50 MHz. And thanks to unofficial features, the board allows setting sky-high values for Socket 3 CPUs – 60, 66 and even 83 MHz! (since this chipset was also used for Pentiums, such support was happened to stick with the 486 model as well)  Boards with similar capabilities, according to my information,  also include: Biostar MB8433-UUD of later revisions and PC-Chips M919 486 VIP.

Among other bonuses Shuttle HOT-433 has onboard IDE and FDD controllers. It should be noted that not all Socket 3 motherboards have discrete controllers, thus the user will lose an extra PCI / ISA slot for installing such a controller. And also the motherboard has a PS/2 mouse-port connector, which is not made in the usual round shape, but in the form of a special connector on the motherboard, where you need to connect a block with an external “round” PS / 2 connector. However, serial-port mice, familiar for those years, work perfectly in the ports of the same name.

VRM with a large ground plane (also helps with heat dissipation) and many many jumpers

If this is your first time with 486 motherboards, get ready to read the manual! (they originally came printed on a substance known as paper, as opposed to a PDF) So, as the configuration of the processor type, the amount of L2 cache memory, the voltage and frequency of the system bus, and much more are selected using jumpers. There are more than a dozen jumpers on the board, so it would be better if they were colored for various parameters, and of course you have to print a paper manual to set everything up correctly.


The motherboard BIOS is based on AMI Firmware that might even surprise the modern user accustomed to UEFI. Here are some photos of this interesting interface.


The main BIOS menu vaguely resembles modern UEFI, with different windows with different parameters and scroll bars.

Mouse Support!

But what’s most interesting is the mouse cursor! which you can select and change the options you want. And this function was available in the early 90s of the last century, simply by connecting an ordinary mouse with a serial interface. Everything worked without any drivers, why then they decided to abandon this technology and remember it again only after a couple of decades – is a good question. But the function is nice, you can, for example, quickly check the performance of all your tailed serial rodents.

Cache Settings

In the BIOS of the  Shuttle HOT-433 it is possible to select not numerous, by modern standards, but useful parameters: type of cache operation (WB / WT), system bus dividers, cache timings, memory delays. You can also configure the IRQ and select the necessary parameters of the integrated devices. When the system is configured, it’s time to move on to building the test bench and installing the Operating System.

The Test Stand

The Main Hardware Components of the System

• Intel Pentium Overdrive 63 MHz
• Intel Pentium Overdrive 83 MHz
• Intel 80486 DX4-100 100 MHz
• Intel 80486 DX2-66 66 MHz
• Intel 80486 SX 33 MHz
• Intel Overdrive 486SX-20 40 MHz
• AMD Am5x86-P75 133 MHz
• AMD Am486 DX4-100 100 MHz
• Cyrix 5×86-100GP 100 MHz
• Cyrix CX486-DX2-66GP 66 MHz
• Cyrix CX486S (FasCache) -40GP 40 MHz
• UMC U5S-SUPER40 40 MHz
• Shuttle HOT-433, UMC 8886AF / 8881F chipset
• Kingston KTC-2430/64-CE, 64MB Kit (2 X 32MB), EDO non-Parity, 60ns 5V (though the chips are 50ns)
Video card
• Creative Graphics Blaster RIVA TNT CT6700, PCI – 16 MB
Storage device
• Seagate Medalist 3210 3.2 GB ATA-33 256K Cache 5400rpm
Power Supply
• FSP 500-60GLN (3.3V – 30A, 5V – 30A, 12V – 2x 18A)

Testing was carried out in Windows XP and MS-DOS 8.0 using the following software

• Super Pi mod. 1.5XS (task 1M)
• SiSoftware Sandra 2002
• AIDA64 5.50.3600
• PC Player Benchmark
• Superscape Benchmark v.1.0c
• TOPBENCH v.3.8
• Speed Test v.2.1
• DooM v.1.09
• Quake v.1.06
• Speedsys v.4.70


The idea was to test everything in Windows XP, since, in my opinion, this is the most universal OS from Socket 3 to LGA 1200/2066. But I won’t be able to test everything, and I knew this in advance, but whoever wants to try it himself will be disappointed. All of the above processors, with the exception of the Intel Pentium Overdrive 63/83 MHz, will not work with Windows XP, although it would be more accurate to say that the OS itself does not support them, due to the lack of the necessary instructions (CPUID, and CMPXCHG8B) for the normal operation of the OS. Which ones you see below in the photo:

Therefore, at this stage, the older brothers will compete with Intel Pentium Overdrive.

Super Pi mod. 1.5XS (task 1M)
Minutes (less is better)

This test shows that the number of pi with a million decimal places, overclocked to 100 MHz, is calculated by a Socket 3 Pentium Overdrive in 33 minutes and 25.314 seconds. Its full-fledged Socket 7 Pentium 100 with SDRAM memory and a more modern platform performs the same task in 20 minutes 11.623 seconds, while a modern Core i9 9900K at 5 GHz takes only 7.859 seconds.

If we hypothetically imagine that we overclocked the Pentium Overdrive to 5 GHz, then the result would be 40.106 seconds, or a lag due to the difference in the architecture of our Overdrive by 5.1 times compared to the Coffee Lake representative, which is the eighth generation of Intel Core processors. But the difference between these two processors is 22 years. Whether it is a lot or a little, I don’t undertake to answer.

CPU-Z Vintage Edition 1.02

With the release of a special version of CPU-Z – Vintage Edition, it became possible to display large characteristics of the processor and components of the system under test, in relation to the standard version, as well as to evaluate the performance of processors. The result of a pair of Ovedrive’s in relation to each other.

SiSoftware Sandra 2002

Below on the screenshots you can see the performance of the Intel Pentium Overdrive overclocked to 100 MHz:



AIDA64 5.50.3600

To evaluate the performance of the memory subsystem of this platform, you can take a look at the screenshot of the Cache and Memory benchmark from the AIDA64 test package:

It’s time to switch to DOS and compare all the representatives of the 4th generation processors with each other.

PC Player Benchmark
Frames per second, FPS (higher is better)

Crazy 3D test in 320×200 resolution with rich 8-bit color. Although the Creative Graphics Blaster 3D accelerator based on the Nvidia RIVA TNT GPU with 16 MB of video memory was responsible for the graphics all the time, the whole burden of displaying the final image fell on the processor. The top three were the Pentium Overdrive overclocked to 100 MHz, the second place was taken by a representative from AMD Am5x86-P75, overclocked to 160 MHz and having a PR rating of Pentium 90, bronze went to the i486 DX4-100 overclocked to 120 MHz.

The outsiders turned out to be processors without an FPU unit, but the UMC GREEN, overclocked to 50 MHz, I must say, was a surprising performer.

Curiously the Cyrix saw little gain when overclocked from 100 to 120MHz while the AMD and Intel chips did.  Likely there is a flaw in the Cyrix preventing it from performing to its fullest potential.

the same 3D accelerator from Creative – Graphics Blaster RIVA TNT, CT6700

Superscape Benchmark v.1.0c
Frames per second, FPS (higher is better)

Another 3D test, the results of which surprised me a little. The overclocked Pentium Overdrive is still the leader, but the results of processors with a 40MHz FSB turned out to be lower than those with the default 33MHz FSB. With increasing FSB, we had to increase the L2 cache timings by one wait state, most likely this parameter affected the performance in this test. It would be curious if this could be ‘solved’ with faster L2 cache.

Score (more is better)

The processor benchmark, working in real-time mode, is more intended for testing older generations of processors. There is also the effect of switching to 40 MHz FSB and a decrease in performance. But the results of processors manufactured by UMC and the Cyrix 5×86-100GP 100 MHz performed even better than in the previous test, thanks to which the Cyrix 5×86 managed to take the 3rd place.

Speed Test v.2.1
CPU score (more is better)

A processor benchmark that measures the performance of integer and floating point operations. Unfortunately, this test did not work with processors without an FPU, as well as with Cyrix processors and its clones. But this test is interesting because only my AMD Am5x86-P75 overclocked to 200 MHz could pass it. I don’t know the algorithms for this test, but most likely it does not use the “newfangled” instructions from Intel Pentium Overdrive.


Quake v.1.06 (320х200) (perhaps the most important?)
Frames per second, FPS (higher is better)

To run Quake, FPU support is required by the processor, it is not available on SX versions of processors. It will be possible to run relatively well only on the Intel Pentium Overdrive overclocked to 100 MHz. From the launch nuances, the game for some reason flatly refused to work on the Cyrix 5×86 overclocked to 120 MHz. Such is the nature of overclocking.  Its not unlikely that some parts of the CPU core handle the overclock better then others, making some basic tasks work, while others will not.

Also of note the Pentium Overdrive, running at 100MHz, is more then twice as fast as a 100MHz DX4 (or 5×86 or AMD).  The Pentium has dual integer pipelines, it can issue 2 instructions per clock, and we clear see that benefit here.

Speedsys v.4.70
Final Score (more is – better)


Speedsys is a very popular and versatile performance test for this type of system. In addition to the final assessment, it will show the exchange rate with caches, the performance of the RAM and, if desired, the speed of the hard disk. I think its results should be a good guide.

This time the results turned out to be natural, no obvious performance drop was noticed when overclocking the FSB from 33 to 40 MHz. The leaders were both Intel Pentium Overdrive models, overclocked AMD Am5x86-P75 to 160 MHz and the 120 MHz Cyrix 5×86-100GP. The weakest link is of course the 33MHz Intel i486 SX. The UMC U5S-SUPER40 running at 50 MHz bypassed, as claimed in its advertising materials, the Intel i486 DX2-66 at 66 MHz, notably without an FPU).  Comparing the UMC @ 40MHz to the 40MHz SX Overdrive shows the Intel slightly outperforming it (likely due to the larger caches in the Overdrive)


This was my first acquaintance with the 4th generation of processors in this format and it did not go easy. More than one month has passed from idea to implementation, but this time, I hope, was not wasted. After Socket 7, there is a performance gap that separates all 486 and the next generation of processors. It’s scary to think about how people worked, played and created on the 386s and 286s, 8086, and few complained about the lack of performance, since the very opportunity to get acquainted with the PC was probably a miracle. Now everything is different, (un)natural selection has done its job …

So far, I have only managed to go up and down, leaving my deep performance research for the next time. Of course, you can get completely different numbers, in some cases it is even higher, but in this comparison, the dynamics and proportionality of the growth of results are important. Already at the end I noticed that the use of a more “modern” Riva TNT video card results in a decrease in performance compared to video cards of its time with 1 or 2 MB of video memory (the CPU is unable to keep the Video Cards RAM filled, slowing it down). But in this configuration, I got exactly such numbers, and they correlate with each other.

If suddenly someone wants to try to build their first 486 machine, then the best choice would be the Intel Pentium Overdrive, or as a cheaper alternative – AMD Am5x86-P75 overclocked to 160 MHz.

Editors note:

This leads to a bit of fun we like to call What If?  Earlier we mentioned Darwin, and the evolution of the processors post 486 period, ultimately resulting in AMD and Intel as sole contenders in the x86 realm. But what if this process was based entirely on the engineering prowess of the respective species and not the outside forces of the courtroom? What if UMC and Cyrix (whom Intel actually licensed technology for the P6 from) succeeded? Cyrix may have continued to be hampered by not having its own fab’s but even today AMD is technically fabless, and Intel is outsourcing fab work to TSMC and Samsung.  We see the performance of the UMC 486 architecture, where would this have led?  Having their own fabs made certain aspects much easier.  It would be curious to benchmark the U5D with a FPU.  And What If IBM has been allowed to market their own 486 design?  

The post The 486 CPU Era – The Birth of Overclocking. – Part 2 first appeared on The CPU Shack Museum.]]> 0
The 486 CPU Era – The Birth of Overclocking. – Part 1 Sun, 21 Feb 2021 22:40:40 +0000 Introduction 486 CPU Era – the birth of Overclocking – this is how I decided to call everything that was in the pre-Pentium era, which I did not find and become familiar with until a couple of months ago. (Another Article in cooperation with max1024 of Belarus – Edited/Expanded by Me) If we abstract from […]

The post The 486 CPU Era – The Birth of Overclocking. – Part 1 first appeared on The CPU Shack Museum.]]>

486 CPU Era – the birth of Overclocking – this is how I decided to call everything that was in the pre-Pentium era, which I did not find and become familiar with until a couple of months ago.

(Another Article in cooperation with max1024 of Belarus – Edited/Expanded by Me)

If we abstract from the very first Pentiums, which appeared using Socket 4 in two speeds of 60 and 66 MHz, then these processors won popular fame and love in motherboards based on Socket 5 and 7. Such machines could be seen in the early 90s on which while playing C&C, Warcraft and other RTS games. The Sega Mega Drive II and Super Nintendo game consoles competed with expensive computers. Moreover, the consoles were far ahead in popularity (and to be honest, the graphics and game play were better) and I got used to the joystick much earlier than to the mouse and keyboard.

The question arises, what was there before all these Pentiums? And the answer, if you dig deeper, can discourage or even confuse any inveterate computer enthusiast, since the cultural layer of “hardware” from the very first processor belonging to the x86 architecture to the first representatives of the superscalar architecture is much larger than from the Pentium 4 to the freshly released Intel Core i9-11900K, which belongs to the Rocket Lake family of 11th generation Intel Core processors. It is not so easy to digest this entire historical layer, so I have outlined the framework for myself.

To simplify the chosen concept, I decided that the platform should in any case support the PCI interface, since it is, firstly, relatively fashionable and “modern” and, secondly, gives more room for my experiments with the accumulated PCI expansion cards. I did not impose other, special requirements on the test platform, except that according to the established tradition, it should be the most powerful and fastest set that is possible to assemble.

Here I think some of the readers of this article the “True oldies” will say: “what is this nonsense, where is the ISA, VLB and 8-bit only?”, But everything has its time, we will gradually dive into the depths of the prehistoric hardware sea, otherwise decompression cannot be avoided. [Editor’s note, I grew up on an 8-bit 8088 and of course connected the PC Speaker to a 100 Watt Stereo Amp, the loudest 8-bit beeps ever]

typical VLB videocard – V7 Mirage P64 on S3 Vision 864, 2 Mb (before they hid all the good stuff with a heatsink)

So, let’s play from the presence of the PCI bus, which appeared just during the heyday of 4th generation processors, “fours” or simply – four hundred and eighty-sixths, which first appeared back in 1989 or today it is 32 years ago. “Almost like yesterday” the oldies will say, “We were not born yet,” the rest will answer, although this is not the point.

The previous generation of 386 processors was content to exchange data with peripheral devices more often at the “width” of 8 and 16 bits, although the entire generation of processors belongs to the first microprocessor architecture supporting 32 bits, but despite this, motherboards designed for them had no  32-bit PCI bus. Although this could not have happened historically, since the specification is new, in relation to the previous buses, it (PCI) was first implemented in 1992. This means that the whole choice comes down to the whole variety of 486 processors, and there was enough variety in those years, not that today there is a choice between “red” and “blue”.

Species diversity

Charles Darwin “Each species is fertile enough for the surviving offspring to multiply, to ensure the growth of the population” 

In his landmark and world-famous work “The Origin of Species by Natural Selection”, Charles Darwin devoted a special role to species diversity and natural selection. If his theory of the origin of species is empirically correlated with the struggle for survival in the processor market, then we can conclude that Evolution has already ended and only two subspecies remain on the highest pedestal of the food chain – “red” and “blue”. And we just have to be unwitting witnesses of the Revolution, when finally one species completely engulfs or annihilates the other and reigns … I think we don’t need such happiness =) However, Charles Darwin’s theory remained a theory.

Returning to the species diversity of all 486 processors, I want to note that they were produced at different times by as many as seven manufacturers who produced not one processor model, but even several parallel lines for various tasks with different processor models. Modern users only dream of such a variety, there is not much to choose, and this is precisely the result of evolution and natural [and unnatural selection in the case of Intel suing several out of existence] selection. Before getting to know the representatives of processors from different manufacturers, I will start with an advertising kit from Intel, designed specifically for their 486 processors.

Before you is a masterpiece of art, more correctly – an advertising product released in 1992, which tells about the outstanding microprocessors released by Intel.

The first exhibit is a processor from the 386 family – Intel 80386SL. It is a mobile microprocessor with a 386 core and power-saving features not found in the desktop 386. This processor has an unusual design for x86 microprocessors – the land grid array (LGA), most were in a Plastic QFP. Interestingly, the 80386SL had over 800,000 transistors – three times as many as the desktop 386SX / DX (integrating many chipset features onto the die).

The second processor belongs to the 80486 family – Intel 80486SX. The first 486 processor without FPU. The product ad claims the i486SX has 800,000 transistors, although this is contradicted by other Intel sources. According to them, the i486SX processors had about 1.2 million transistors for 1 micron chips and 0.9 million transistors for 0.8 micron chips. Conclusion – no one is immune from errors, even inside the CPU, though they may have deducted the transistors that were used by the disabled FPU.

80486-25 Pre-DX SX249

The third exhibit is the DX version of the i80486 processor. The difference between DX and SX processors is that DX chips have a working built-in module for calculating floating point operations, while SX does not (it was disabled in early chips, and removed in later versions completely). The die of this chip is the same size as the SX die- approximately 16 mm x 10 mm, that is, 160 mm2.

The Intel Pentium 60 microprocessor closes the evolutionary branch of development. This microprocessor is an early production version with the FDIV error, which was in the floating point unit in the original Pentium processors manufactured by Intel in 1994. The error was expressed in the fact that when performing division on floating point numbers using the FDIV processor instruction, in some cases the result could be incorrect.

Early 1993 Pentium 60 – With the FDIV bug (and overheating problems later fixed by moving to a huge heatspreader package)

Intel knew about the problem, but preferred not to expand on it. They believed that since this defect is significant only for a narrow circle of users (mathematicians and other scientists), then users who want to replace the processor should contact the company and prove that it is they who need this replacement. (Charles Darwin would say about this in our time that if you need a modern video card, then you must first prove that you cannot play modern games without ray tracing the best one with an acceptable FPS for you =))

Intel had had many bugs in their processors before (early 486s and early 386s were VERY buggy, some 386s so much so that they were limited to 16-bit code, the FDIV error though became a bit of a media sensation.  Perhaps a sign that computers had finally become mainstream enough that a bug could generate that level of attention.

And then I will briefly talk about those seven manufacturers of 486 processors, the models of which I will test below and even try to overclock them =)


Let’s start with Intel. In the fall of 1989, at the Comdex Fall computer show, held in Las Vegas, Nevada, USA, Intel presented its new fourth-generation 32-bit x86-compatible microprocessors built on a hybrid CISC-RISC core. They received the name Intel 486 or i486 for short. The frequency range of processors started at 16 MHz and ended at 100 MHz. The entire 486 CPU line was divided into two camps, with and without an integrated floating point unit (FPU). The first type of processors received the suffix “DX”, and the second “SX”. It should be noted that initially the FPU was embodied in silicon, but if a defect was found in this block during the production of processors, with the remaining processor blocks fully functional, the processor was simply turned into the “SX” version, or some kind of “Celeron” these days.  The very first production of 486s (in 1989) did not make this distinction, it was simply the 80486, no DX to be found, the DX was added at the turn of the decade, in 1990.

Early processor models had 8 KB of unified L1 cache for code and data. In later versions of the CPU, the cache size has been increased to 16 KB. The early i486 models worked with the cache on the Write Through principle (WT for short), but later learned to work with the cache using the Write Back (WB) function. When using this more productive principle of working with data, if there was a copy of the data in the cache, the information was written only to the cache memory, writing to the RAM was not performed, while with the pass-through write, the data was always written to the RAM, even if they were already present in cache.

classic 486 set – external L2 cache with a 1 MB capacity)

The i486 processors did not have a second-level cache; instead, an external cache was located on the motherboard (it worked at the CPU bus frequency), the volume of which could be increased with your own hands by adding more capacious memory chips to special sockets, or simply increasing their number, if, of course, the motherboard had this function.  Some motherboards has a special slot for a cache module called a COAST (Cache On A STick) to simplify this. The L2 cache size was measured in kilobytes and in the maximum configuration could reach 1024KB.

Speaking of the i486 processors on this overclocking resource, it is worth noting that it was this generation of processors that gave the whole world such a concept as – Overclocking, in its full understanding, since it was in the i486 that the first multipliers of the base frequency or simply multipliers appeared.  Previously overclocking required clock crystals on the motherboard to be replaced.  The i486DX2 processors had a multiplier of x2, and the lineup consisted of models with frequencies from 40 to 66 MHz. The manufacturer’s assortment included the only model i486SX2 processor with a frequency of 50 MHz. The i486DX4 processors had a x3 multiplier, although the name of the model itself hints at a x4 multiplier. These models included processors with a frequency of 75 and 100 MHz.

During the 486s, there was also a fashion for Upgrade processors, and even Intel itself encouraged this activity and even released a separate line of Intel 486 OverDrive processors. These processors were intended (initially) to be installed in a special Socket, which was designed to install the i487SX mathematical coprocessor, which, in fact, was an ordinary i486DX, but with a different sequence of contacts (CPU legs). Often on motherboards one could find a soldered i486SX and a second socket for installing a mathematical coprocessor i487SX or OverDrive, or even two sockets, one of which was made of blue plastic and had the inscription “Intel OverDrive ready” or just a blue socket without designations.

Ordinary 486 processors had 168 pins, special “Overdrives” designed to replace CPUs soldered on motherboards had one more pin – 169. An additional pin served as a kind of key for installing such a solution into the socket, and also had a shutdown function the processor soldered on the motherboard.

When an i487SX or OverDrive was installed in the second socket, the first soldered i486SX or any other processor installed in the socket was automatically completely disabled and the second processor intended for the upgrade was already engaged in all the calculations.

After Intel introduced the 486 OverDrive processors in 1992, they used a convoluted model numbering system. Each CPU was labeled the same as the CPU it was supposed to replace. If, for example, you had an i486SX-25 MHz, then you had to buy an OverDrive labeled ODP486SX-25. Such an upgrade system was misleading and later Intel abandoned it, starting to mark OverDrive processors with their real frequency. This is because the first OverDrive processors worked at double the clock speed with built-in coprocessors, that is, they were i486DX2 processors. So the ODP486SX-25 was actually an i486DX2-50.

The popularity of these i486 OverDrive processors was great. The main advantages of the Intel 486 OverDrive processors were as follows:
• OverDrive processors contained built-in voltage regulators
• Supported write-back (WB) cache
• had a non-removable radiator and were able to operate without forced air cooling

To extend the lifespan of the 486 platform, Intel in February 1995 releases a special version of the processor – Intel Pentium OverDrive with a frequency of 63 MHz. The 83 MHz version was released only 8 months later, when the age of 486 machines was rapidly approaching its end. The processor itself is a real Pentium core! Modified to interact with the i486 bus. Its bus was 32-bit, which is half the size of a full-fledged Pentium, but to compensate for the processor, the L1 cache size was doubled.

The processor is rated at 3.3 volts with an onboard voltage regulator providing 5 volts from the motherboard. The fan built into the CPU heatsink is powered directly from the CPU chassis.


The main competitor to Intel in those days was AMD. Processors of the Am486 family were functional analogs of the competitor’s processors and initially used the same microcode of the Intel 80386 processor and the Intel 80287 math coprocessor.

Later, some models used their own microcode. In April 1993, the first processors with the Am486DX and Am486SX markings were born, similar to Intel models. DX and SX processors worked at the system bus frequency, the frequencies corresponded to similar Intel models, but the cost was 20% or more percent less.  These original AMD processors also used Intel microcode, and are nearly impossible to distinguish from their Intel brethren.

A little later, a year later, there were models with multipliers, as well as with write-back cache. The frequencies of the Am486DX2 models were in the range from 50 to 80 MHz, and the Am486DX4 from 75 to 120 MHz, which by 20 MHz in clock frequency exceeded the TOP of the line i486DX4-100 from Intel.

But AMD did not stop there and in 1995 released the following processor models called Am5x86. The processor die was manufactured using a 350-nm process technology and had 1.6 million transistors. The L1 cache was doubled to 16 KB, and the multiplier was 4. The processor was running at 33 MHz FSB, and thanks to the x4 multiplier, the resulting core clock speed was an impressive 133 MHz. In terms of performance, the Am5x86 was comparable to a Pentium processor with a frequency of 75 MHz, which was clearly indicated on the processor marking in the form of the inscription “Am5x86-P75”, while the processor model itself looked like AMD Am486DX5-133ADW/Y/Z.

These processors overclocked well and easily turned into the Pentium 90 equivalent when running at 160 MHz (40 x4). Not a large percentage of processors could even work at 200 MHz (50 x4). It is believed that processors with a “Z” at the end are more successful overclocking, since according to the manufacturer’s specification they have an operating temperature of 85 C, versus 55 C with a marking ending in “W” or 75C for the “Y”. Another popular saying says that the later the processor is released, the higher the chance of it working at 200 MHz. I have personally met processors made in 1996, 1997 and 1998, but even now they are more often found from 1996.

Although the Am5x86-P75 processor is labeled with the number “5”, which should symbolize the 5th generation of processors, in fact, this processor belonged to the fourth, such a marketing move from AMD.


“Blue giant” or IBM stood at the origins of all computer History, it is not for nothing that the slogan “IBM compatible PC” already meant that buying such a computer you will receive a hardware platform compatible with a large number of software products. Until the mid-80s of the last century, IBM felt very confident in the PC market, but when a new market player appeared – Compaq, which in 1985 released its first 80386 Deskpro 386 computer, the situation for IBM changed dramatically. To increase the share of its PCs sold in the computer market, IBM entered into contracts for the supply of processors with Intel, and then with AMD. Additionally, IBM at its manufacturing facilities produced Intel processors, the 486 version of which looked like this:

The lineup of Intel clones consisted of 33, 50 and 66 MHz models. If the bare processor was a regular version of Intel i486DX and apart from the marking on the case “MFG (manufactured) BY IBM” did not stand out, then the version with a heatsink immediately catches the eye, thanks to its unusual blue color and the name “BLUE LIGHTNING”.  The BLUE LIGHTNING name was borrowed from another line of QFP only 486 processors that IBM had designed and made inhouse.  The PGA versions are NOT related to the earlier QFP BL, BL2 and BL3 x86 processors.

IBM Blue Lightning – Cyrix Based Version

When IBM’s contract with INTEL ended, the Blue Giant signed a new one with Cyrix and subsequent IBM processors were already clones of another American processor manufacturer (Cyrix).  IBM models from Cyrix ranged from 50 to 100 MHz. The difference between the same model from Cyrix and IBM was that the IBM manufacturing facility had stricter quality control as well as tolerances [for their own marked chips], resulting in better die quality and lower operating voltage. IBM (as well as ST and TI) manufactured dies for Cyrix as Cyrix was a fabless company.

IBM 50G6663 486SLC2-66 – IBMs Own design

So if you come across an ancient IBM processor, this does not mean that it is IBM, it could be an Intel, or a Cyrix…or an IBM.


Cyrix Corporation was an American microprocessor developer that was founded in 1988 and specialized in the supply of mathematical coprocessors for 286 and 386 processors. Cyrix was a CPU manufacturer with no manufacturing facilities of its own (fabless). For the manufacture of processors, they used the manufacturing facilities of SGS-Thomson (now ST Microelectronics), Texas Instruments and IBM (all three of which possessed x86 licenses from Intel to make such manufacturing ‘legal’) . Since 1993, Cyrix had launched its 486 processors, which include the Cx486S, Cx486DX, DX2 and DX4 models.

Cyrix’s first 486 processor was the Cx486S (codenamed M5). It was designed as an alternative to the Intel 486SX as it did not have an integrated floating point unit (FPU). However, the processor had a 2KB write-back cache and a special “Write-Burst” (WB) signal that gave a slight performance boost in some applications, provided the motherboard was able to use this feature. These processors were labeled “FasCache” to emphasize this feature, as most processors used slower write-through (WT) caches. Three models of such processors were released with frequencies of 25, 33 and 40 MHz.

A little later, the Cyrix Cx486DX models (codenamed M6) appeared, which in their essence were Cx486S plus an internal floating point unit (FPU) and a fourfold increase in L1 (8 KB) cache size. The operating voltage for the processors was 5 V, and the lineup consisted of models with 33, 40 and 50 MHz (50MHz models were very often faked from slower ones). Despite the impressive 50 MHz, such processors were not in demand, since at such a bus frequency the peripheral devices installed with them worked extremely unstable, as were the motherboards (the same was true of Intel’s DX-50 processors).

Five months, it took such a period of time to release models with doubled and tripled frequency, in relation to the release of the first 486 processor. All of these processors had 8KB internal L1 cache. Many models had writeback L1 cache (WB) and were available in 3.3V, 4V, and 5V versions (and several voltages in between, and multi voltages, the variety was mainly due to making them compatible with what motherboards could provide) . The processors were based on their own microcode and were about 5-10% slower than the real Intel 486. The processors were produced with or without proprietary green heatsinks. At this time, Cyrix processors gained popularity thanks to upgrades of old systems designed for a single 5 volt supply, since they could install the popular 66 and 80 MHz models, which were powered by Cyrix from 5 V, versus 3.3 V from competitors.

In 1995 Cyrix released its fastest and latest Socket 3 processor, the Cx5x86 (codenamed M1sc). It was no longer just a 486 processor, but something more, as it contained elements of the next fifth generation architecture (Pentium, P5). Cyrix 5×86 had a 64-bit internal and 32-bit external data bus, had a parallel operation system, branch prediction (which sometimes was enabled, and sometimes not) and optimization of instruction execution. The large 16K internal L1 cache could be configured for both write-through and write-back. Cyrix Cx5x86 processors have a voltage of 3.6 V and are designed to operate at 80, 100, 120 and 133 MHz.

However, the 133 MHz version is extremely rare and is in great demand among CPU collectors. Most CPUs can be configured to use x2 or x3 multipliers, but some models support x3 and x4 multipliers. Sometimes this can be seen in the markings on the surface of the processor or heatsink, and sometimes not. The processor also has the ability to programmatically disable multiplication, and special knowledge and utilities are required to properly configure it in different operating systems. Now everyone relies on Plug & Play, before it was much more complicated.

As a result, the Cx5x86 turned out to be a very productive processor and only had the Intel Pentium Overdrive 83 MHz and AMD Am5x86-P75 with a frequency of 133 MHz to truly compete with. But, despite the lower clock speed, the Intel Pentium Overdrive had a much more powerful coprocessor and the Cx5x86 processors lagged behind in the speed of floating point calculations (something that would ‘plague’ Cyrix CPUs for many years.

The decline of the Cyrix Corporation as an independent structure began in the fall of 1997, when Cyrix was bought by National Semiconductor. In 1999, the combined company Cyrix National Semiconductor was acquired by the Taiwanese company VIA Technologies.  National Semiconductor retained the MediaGX designs (based on the Cyrix 5×86) which then were later sold to AMD, who went on to continue making the GX line for several years.  The Cyrix Cx486DX core continued to be licensed and used by other companies for nearly 20 years, including by ST, and ZF Micro.

ST – IT’s ST and Texas Instruments

For Socket 3, you can still find processors whose name is not very familiar to anyone these days and these processors have a very unusual appearance. These are all Cyrix processors manufactured by two semiconductor giants ST Microelectronics and Texas Instruments under license.

ST Microelectronics (ST) is a multinational electronics and semiconductor manufacturer headquartered in Geneva, Switzerland, formed from the merger of two semiconductor companies in 1987: SGS Microelettronica from Italy and Thomson Semiconducteurs from France. In accordance with the agreement between SGS-Thomson and Cyrix, Cyrix has granted SGS-Thomson the right to use certain microprocessors developed by Cyrix as part of SGS-Thomson’s intellectual property. SGS-Thomson was authorized to manufacture and sell such processors under its own name in unlimited quantities, with Cyrix receiving royalties from the sale of such CPUs. In addition, Cyrix is authorized to sell SGS Thomson chips under its own name.

The lineup of ST processors consisted of different models from the usual DX version with a frequency of 40 MHz to the DX4-120 model operating at 40 MHz FSB and having a x3 multiplier, as well as a processor model based on the Cyrix Cx5x86 core with a clock frequency of 120 MHz. Such processes were configured as well as Cyrix processors.

TI TI486DX4-G100-GA – Cyrix Based

Texas Instruments Incorporated (TI) is an American technology company headquartered in Dallas, Texas that develops and manufactures semiconductors and a variety of integrated circuits that are used in virtually all areas of life. The field of activity and history of this company is so vast, and the number of scientific discoveries and achievements is so great that suffice it to say that TI produced the world’s first silicon transistor in 1954, and the people from this company were later involved in the emergence of such large IT companies, like Intel.

So processors marked ST – IT’s ST are not something unusual, but the same models Cyrix Cx486S, Cx486DX, DX2 and DX4, but with some minor modifications, about which little is known. The collaboration between Cyrix and TI was shorter, and despite the fact that the development of a modified processor based on the Cx5x86 was carried out by TI, the management of TI decided to leave the processor business and the processor was never released. TI made the usual Cyrix DX2 and DX4 clones, with very colorful markings.  These are identical to their Cyrix counterparts.  But TI didn’t just make Cyrix clones.  They had a 486 entirely of their own design as well.

TI 486SXL2-G66-GA – TI’s own design – no FPU – odd pinout

The SXL2 is a clock doubled 486 processor without an FPU.  They also had a non clock doubled SXL-40 (curiously these actually secretly support clock doubling).  These had a bit different pinout and are almost always found on upgrade adapters (or special motherboards that support them)



UMC is another very interesting manufacturer of the 80486 processor family. UMC or United Microelectronics Corporation is a Taiwanese microelectronics manufacturer founded back in 1980. Now UMC is one of the three world leaders in contract manufacturers of semiconductor microcircuits along with Taiwanese TSMC and American GlobalFoundries inc. I think these three manufacturers are familiar to you, because news about them appears on a regular basis in news feeds and all major manufacturers of central processors and graphics accelerators directly depend on their production capacity and the degree of mastering more subtle technical processes.

Let’s go back to 1993, when UMC presented its 486 processor called GREEN CPU. Unlike the clones from AMD that are practically indistinguishable from Intel i486 in terms of performance (and microcode), and the slower 486s from Cyrix, the UMC processor worked faster than the i486 from Intel at the same clock speed. UMC engineers revised the Intel product code and added their own developments that increased the processor’s performance. In addition to the processor, UMC also produced its own motherboard chipsets, I/O controllers, network and graphics chips, and other necessary chipsets needed to create a complete and unified computer ecosystem. And it should be noted that it was the system logic from UMC that was the fastest among all chipset manufacturers.

In 1994, Intel filed a lawsuit against UMC for infringement of its patents for their i486 processor, as a result of which the sale of UMC processors in the United States was prohibited. There is even a special inscription on UMC processors that reads: “Not for U.S. sale or import ”. As a result, the habitat of the processors fell on Eastern Europe and the countries of the former USSR.  You can learn more about Intel vs the World in a previous article.  In many ways it was less ‘survival of the fittest’ and more ‘survival of he who has more lawyers and money’ to the detriment of the consumer in many cases.

What were the UMC GREEN processors? These are processors with a frequency of 25-40 MHz, in a plastic or ceramic case. The first level cache was 8 KB, the standard voltage is 5 volts. Versions of the processor with the “SUPER” prefix worked at a reduced voltage equal to – 3.3 V. A distinctive feature of the processors was the absence of a floating point unit, but even despite this, the processor was very fast and in terms of speed at 40 MHz it could easily compete with other models manufacturers operating at 66 MHz.

Due to litigation with Intel, UMC released an analogue of the i486DX with an FPU unit, but without settling all the disputes and did not include it in the final versions of processors, which were extremely small in number. Engineering samples with the doubled frequency of the UMC486DX2 were released, but the matter did not go beyond samples. In the same 1994, Intel filed lawsuits against UMC and its distributors, UMC responded to the lawsuits with an antitrust lawsuit, and ultimately the case was settled out of court, as a result of which UMC withdrew its product and stopped production of its 80486 processors.

In Part 2 we will discuss choosing motherboards and the appropriate testing hardware, and then?! Benchmarks and some overclocking!


The post The 486 CPU Era – The Birth of Overclocking. – Part 1 first appeared on The CPU Shack Museum.]]> 0
The Story of the Soviet Z80 Processor Wed, 27 Jan 2021 00:23:53 +0000 Before we get into the fascinating story of the Soviet (specifically the Angstrem) Z80 clone it’s good to understand a bit about the IC industry in the USSR.  There were many state run institutions within the USSR that were tasked with making IC’s.  These included analogs of various western parts, some with additional enhancements, as […]

The post The Story of the Soviet Z80 Processor first appeared on The CPU Shack Museum.]]>
Before we get into the fascinating story of the Soviet (specifically the Angstrem) Z80 clone it’s good to understand a bit about the IC industry in the USSR.  There were many state run institutions within the USSR that were tasked with making IC’s.  These included analogs of various western parts, some with additional enhancements, as well as domestically designed parts.  In some ways these institutions competed, it was a matter of pride, and funding to come out with new and better designs, all within the confines of the Soviet system.  There were also the various Warsaw Pact countries (BulgariaCzechoslovakiaEast GermanyHungaryPoland and Romania), that were aligned with the USSR but not part of it.  These countries had their own IC production, outside of the auspices and direction of the USSR.  They mainly supplied their own local markets (or within other Warsaw Pact countries) but also on occasion provided ICs to the USSR proper, though one would assume an assortment of bureaucratic paperwork was needed for such transfers.

This resulted in some countries developing similar devices, at rather different times, or different countries focusing on different designs.  East Germany was all in on the Z80, Romania, Poland and Czechoslovakia made clones of the 8080, Bulgaria, the 6800 and 6502. They were though, seperate from the USSR’s own institutional system, so while East Germany had a working Z80 in the early 1980’s the USSR did not.  It is this distinction we will focus on today

This article is largely from guest author Vladimir Yakovlev, translated from Russian, and edited/expanded by me.

By the end of the 80s – beginning of the 90s, clones of the British Sinclair ZX Spectrum computer, a simple, cheap computer with a huge library of games originally released in 1982, were being distributed in the USSR. The “strapping” of the central processor instead of the original ULA microcircuit was done on small logic microcircuits of the 555 (74LS) series and the like, but the Z80 itself had to be bought from abroad. Naturally, the thought arose, to start making the processor yourself. After all, the processor itself, developed in 1976 for the microelectronic industry, was not too complicated.

In 1990, the development of an analogue of the Z80 was organized in Zelenograd near Moscow at the Scientific Research Institute of Precise Technology (NIITT) and the “Angstrem” plant. Initially, Zelenograd was conceived as a center of the textile industry, but was later reoriented to the development of electronics and microelectronics by Nikita Kruschev after he visited Silicon Valley (California, USA) in 1959. To this day, Zelenograd has retained the status of a scientific center and the informal name “Russian Silicon Valley”.

The chief designer was appointed Yuri Otrokhov, who had previously led similar developments. Otrokhov, who served as a tanker in his youth (military service being mandatory in the USSR), called the project the T34 microprocessor.

Otrokhov: “T-34VM1 is the internal designation of the KR1858VM1 processor, assigned by me at the stage of development and production in honor of my first tank, on which I learned to drive.”

Here is one of the versions of the creation of the clone, outlined by one of the employees of NIITT at that time, Boris Malashevich [1]:

“Otrokhov, like his colleagues in the department, knew how to develop original microprocessors, but they had not yet had to reproduce analogs. Therefore, the developers included specialists from NIITT divisions who are able to restore the electrical circuit of the IC according to its topology. For 9 months after four iterations, they managed to make an NMOS microprocessor T34VM1 (KM1858VM1, KR1858VM1) – a complete analogue of the Z80A microprocessor, to be made using a 2-micron technology” (The original Zilog version was on a 4 micron process).

While Otrokhov and his team worked at Angstrem to make a NMOS Z80, a similar team was working at ‘Transistor’ in Minsk Belarus to make a CMOS version, later known as the KR1858VM3.

Due to the incredible popularity and demand for the Z80, many analogue manufacturers worked without a license, so in total less than half of all Z-80 produced were licensed products from Zilog or its official partners (SGS, Mostek, etc).

From an interview with the creators of the Z80 [2]:

Faggin: Yes, we were concerned about others copying the Z80. So I was trying to figure what we could
do that that would be effective, and that’s when I came across an idea that if we use the depletion load
the mask that doesn’t leave any trace, then I could create depletion load devices that look like
enhancement mode devices. And by doing that we could trick the customer into believing that a certain
logic was implemented, when it was not. Then I told Shima, “Shima, this is the idea how to implement
traps. Put traps, you know, figure out how to do the worst possible traps that you can imagine,” and then
Shima with his mind, that was steel mind, was able to actually figure out a bunch of traps that he could
talk about.
Shima: I didn’t count [on] talking about that mostly. I placed six traps for stopping the copy of the layout
by the copy maker. And one transistor was added to existing enhancement transistors. And I added a
transistor looks like an enhancement transistor. But if transistors are set to be always on state by the ion
implantations, it has a drastic effect on very much. I heard from NEC later the copy maker delayed the
announcement of Z80 compatible product for about six months. That is what I got from NEC. And finally
a total transistor of Z80 became 8,200 while a total of transistor of 8080 was 4,800.

In the course of the design, due to the fact that the development team had specialists in both the creation of new ICs and the reproduction of analogs, Zilog’s tricks aimed at copy protection were identified and decrypted. For example, the topologist saw the 3-Input-NAND Gate element, but this element worked as 2-Input-NAND Gate. The topology and layout of the resulting clone was different, but the functionality did not differ from the original. At first, it was possible to identify such traps, making sure that the circuit was inoperable, only by examining the circuit elements inside the die using probe analyzers. But, having understood the principle of constructing traps, a mechanism for their detection was also developed. As a result, it was possible to make a full-fledged analog of the Z80, although the electrical circuit and topology of the T34MV1 had some differences.

The German Connection

It is known that the T34VM1 and subsequent ones, produced in the USSR, contain differences in undocumented commands, which exactly coincides with the logic of the U880 processor from East Germany. They do not set the CY flag when the OUTI command is executed (by the result of adding the number issued to the port and the value of the L register after the operation), and the hidden system bus register, the contents of which are available through the undocumented flags F3 and F5, have a different logic of operation.

Microprocessor from the German Democratic Republic.

MME 80A – Export Version – No Date Code

Мanufactured by VEB Mikroelektronik “Karl Marx” Erfurt (abbreviated as MME; part of Kombinat Mikroelektronik Erfurt) in the German Democratic Republic.

And here is the microprocessor manufactured by the Angstrem plant.

MME 80A CPU – Dated early 1991 – Soviet style package

Firstly, the case of the microcircuit is clearly Soviet, produced by the Semiconductor Devices Plant (Yoshkar-Ola) [3]. And secondly, the marking features are typical for the products of Zelenograd “Angstrem”.

After some time, the marking was changed to a more familiar one.

Angstrem T34VM1 – Sample Dated 9212

Ceramic package of the T34VM1 Z80-compatible processor, manufactured in the Soviet Union/Russia by Angstrem in 1992 and later years. The ‘ОП’ marking means “experimental batch”.

Sometimes there were more exotic markings.

Angstrem T34 008 – Dated 9332

The markings suggest an 8MHz version, well within what a 2 micron process is capable of

The Angstrem plant also produced many microprocessors in plastic.

Angstrem KR1858VM1 – Dated 9303 – Sample

Angstrem KR1858VM1 – Dated 9312 – Not marked Sample

KR1858VM1 manufactured by the Angstrem plant, produced up to 9303 inclusive, also is marked ‘ОП’. The example issued 9312, no longer has such a mark.

T34 die – MME U880 Rev 5

Looking at the die of an early T34 marked processor it is noted that there is an inscription “U880 / 5” on the T34VM1  [4]

KR1858VM1 die marked MME 1990 Rev 6 – A ~1.6x shrink of the previous version allowed speeds of up to 8MHz

The die of a later KR1858VM1 contains the inscriptions “U880 / 6” [5]

The topologists had to copy exactly the East German version one-to-one, except in places where it did not comply with the topological restrictions associated with the NMOS manufacturing technology available at the Angstrem plant. Which was eventually done. And the products themselves were made from crystals obtained from MME. Indeed, since Soviet times, there have been quite close relationships between MME and NIITT.

On October 3, 1990, the unification of Germany took place. Stocks of finished dies had been accumulated at MME. But the question of patent purity immediately appeared (now joining the West Zilog could have pressed claims of copyright infringement, they ended up not, and Germany continued to make the unlicensed U880 with Zilog even using Thesys as a distributor). Most likely the dies and photomasks were transferred to the Soviet Union even before this event.   This allowed Angstrem to use MME dies while they finished work on their own version.

Towards the end of 1993 Angstrem began making clones using its own photomasks. The inscription U880 disappeared from the die. An image of a heart appeared in the center.

Angstrem die – Later version with a heart in the middle, and limited other markings

Angstrem continued to make Z80s through the 1990’s.

Angstrem KR1858VM1 from 1995

Other former Soviet institutions also made clones of the Z80, including Kvazar (in Kiev Ukraine) and Electronica (now VSP-Mikron).  Details on these are lacking greatly, but looking at a die from an early 1993 version of a Electronica T34 we see a very similar die to that of the Angstrem, including the heart in the middle of it as well.  Dies of the Kvazar look similar to the early Angstrem versions, but lack the heart in the middle of the die.  It’s possible Angstrem supplied dies (or masks) to other institutions as well.

Unfortunately Angstrem is no more, being hit hard by US Sanctions it was taken over by the VEB bank in 2019.

If you have more info/details that can further fill in some of the blanks (especially about Electronica/Kvazar versions) drpop us a line.

[2]. Three founding members of Intel microprocessor spin-out Zilog Corporation, Federico Faggin, Masatoshi Shima and Ralph Ungermann describe the early days of the company and the development and marketing of the Z80 microprocessor that became one of the highest volume and longest-lived architectures in the industry.


The post The Story of the Soviet Z80 Processor first appeared on The CPU Shack Museum.]]> 0
Shanghai – World’s 1st 45nm Monolithic Quad Core x86 CPU – October, 2008 Fri, 08 Jan 2021 23:21:02 +0000 In sports, particularly Baseball, its often said that the longer a record is to say, they less impressive it is.  ‘Most Home Runs Ever’ is much more of an impressive record then ‘Most Home runs in the 7th inning against a left handed pitcher with a runner on 3rd’  Both are of course records, the […]

The post Shanghai – World’s 1st 45nm Monolithic Quad Core x86 CPU – October, 2008 first appeared on The CPU Shack Museum.]]>
In sports, particularly Baseball, its often said that the longer a record is to say, they less impressive it is.  ‘Most Home Runs Ever’ is much more of an impressive record then ‘Most Home runs in the 7th inning against a left handed pitcher with a runner on 3rd’  Both are of course records, the first, many may even know the answer (Barry Bonds), the second? I’m sure someone can look it up but I have no idea.

So when I got this interesting commemorative AMD Opteron Sample it seems fitting to break down the record engraved on it ‘Shanghai – World’s 1st 45nm Monolithic Quad Core x86 CPU – October, 2008’  That seems impressive, and the reality is that it was (and is) and its a testament to the very hard work the design team, whose names are engraved for perpetuity on the chip, put into it.  The Shanghai was a third gen Opteron that followed the very troubled Barcelona, so it was really a make or break design for AMD.

Intel Core 2 Quad Q9100 QAVK Engineering Sample – Dual 45nm dies – Mid 2008

The most impressive aspect of the record is ‘First monolithic quad core x86 CPU.’  This was putting 4 x86 cores on a single die. Now Shanghai wasn’t the first to do this, as Barcelona had done so previously, thus the addition of ’45nm’ to the record.  Barcelona was made on a 65nm process whereas Shanghai shrank that to 45nm.  At the time Intel had the Quad-Core Clovertown Xeons (65nm) and had (in 2007) just released the Harpertown/Yorkfield Quad-Cores made on a new 45nm process.  All of these used two dual core dies in a single package. Intel was able to catch up later with the Nehalem based processors in 2009.

Was there other single die Quad-cores at the time?  What if we look outside of the realm of x86?  In 2008 IBM released the z10 quadcore processor, it was a single die, running at up to 4.4GHz (!) but it was made on a 65nm process.  Likewise, the UltraSPARC T2 was a 8-core CPU from 2007 but again, only on a 65nm process.  Freescale released the 45nm quadcore, single die P3 series P2040 PowerPC processors, but in 2010.  MIPS had the quadcore 1004K in 2008 but only on 65nm. So it seems AMD may have had a better record then they thought.

What if we stretch what we call a processor? There were at the time some fairly simple large multicores like the Tilera TILE64 (64-basic 32-bit cores) made on 45nm process, but they are less of a general purpose CPU.  Perhaps the closest is the Sony CELL Processor in the Playstation 3, which IBM was moving to 45nm in 2008 and had 1x PowerPC core + 7 SPEs. Perhaps AMD could have made a claim to the first 45nm single die CPU ever, even including non-x86 chips.


The post Shanghai – World’s 1st 45nm Monolithic Quad Core x86 CPU – October, 2008 first appeared on The CPU Shack Museum.]]> 0
SEMICON WEST: A Blast from 1996 Fri, 20 Nov 2020 22:24:10 +0000 In 1970 an industry group was started called SEMI (Semiconductor Equipment and Materials International).  They were formed to represent, as the name implies, all the various people/companies involved in making semiconductors.  This wasn’t so much the Intel’s and AMD’s but the companies that made the equipment, chemicals, and even software they used to actually design, […]

The post SEMICON WEST: A Blast from 1996 first appeared on The CPU Shack Museum.]]>

SEMICON WEST 1996 PLCC68 Memorabilia

In 1970 an industry group was started called SEMI (Semiconductor Equipment and Materials International).  They were formed to represent, as the name implies, all the various people/companies involved in making semiconductors.  This wasn’t so much the Intel’s and AMD’s but the companies that made the equipment, chemicals, and even software they used to actually design, fab, package and test chips.

In 1971 they had their first tradeshow, SEMICON WEST, at San Mateo Fairground, California.  They continue to have events around the world, SEMICON WEST is now in San Francisco (and there was a corresponding SEMICON EAST that started in 1973 in New York, but no longer exists).

SEMI not only provides an avenue for vendors and technology to be showcased, but they also work to put forth standards in industry, as well as education.  It was SEMI in the 1970’s who worked to develop standard wafer sizes, can you imagine if there was no standard sizes for such a principal component? Madness!

Lack of molded markings (usually date/country/lot would be included) suggest this was made specific for the conference.

These conferences have seminars on such compelling topics as ‘Chemical Mechanical Polishing’ and ‘Photosensitive Benzocyclobutene for Stress-Buffer and Passivation Applications.’  Today they also include vendors and information on hiring, and personnel management in the semiconductor industry, as well as safety, environmental, and education.  Certainly not as flashy as CeBIT or COMDEX, but perhaps equally if not more important.

The pictured chip was given away as swag during SEMI/WEST 1996.  Its a pretty typical PLCC68 package with the logo from that years conference.  On the back there is a complete lack of markings (even in the mold) suggesting this may have been a run specifically made for the conference, probably by a packaging vendor.

The post SEMICON WEST: A Blast from 1996 first appeared on The CPU Shack Museum.]]> 2
SSQ22667-001: An 80C186 for the Space Station Thu, 22 Oct 2020 01:55:08 +0000 Recently some interesting CPUs showed up on eBay and other IC selling sites.  They were marked SSQ22667-001 and made by Intel.  Some were conveniently also labeled SQ80C186-12.  Packaged in a 68-pin CQFP package, they typically would be labeled as a MQ80C186 (Military CMOS 186 running at 12MHz) but these were as ‘SQ’ prefix, and had […]

The post SSQ22667-001: An 80C186 for the Space Station first appeared on The CPU Shack Museum.]]>

Intel SSQ22667-001 SQ80C186-12 – Space Rated CMOS 80186

Recently some interesting CPUs showed up on eBay and other IC selling sites.  They were marked SSQ22667-001 and made by Intel.  Some were conveniently also labeled SQ80C186-12.  Packaged in a 68-pin CQFP package, they typically would be labeled as a MQ80C186 (Military CMOS 186 running at 12MHz) but these were as ‘SQ’ prefix, and had the weird SSQ22667-001 part # on them as well. Others in the same package were marked SSQ22668 and 22669. So what was special about these CPUs? Was this some random House # for an OEM?  Nope, these were made for NASA, specifically to conform with MIL-STD-975.  To learn a bit more about how these MIL-STD’s work, lets take a journey back to the 1960’s (everyone knows hat was a fun time)

Back in the 1960’s integrated circuits were getting to be more standard, and more available. Many companies were making many different types (generally simple logic at the time, but that was changing fast).  The US Military was, of course, an early user of integrated circuits, as they could afford them, and IC’s allowed for some cutting edge technology.   To make purchasing and stocking such components easier, the military, as they usually do, decided there needed to be some standards, and ICs for the military, should be available in higher standards

Intel MC1702A/B – MIL-STD-883 Class B – 1976

then those destined for your microwave oven or digital alarm clock.  Thus in May of 1968 the MIL-STD-883 was released.  This was (and continues to be, its on Rev L now) a standard created on Test methods and procedures for ICs, any IC’s.  It provides such things as inspection methods, burn-in methods, lot sampling, and a whole host of other ways to test and inspect IC’s.  As the years went by, different Classes of testing were added.  A computer chip the captains coffee pot did not need the same testing as a computer chip destined for a nuclear submarine, or one for use in Space.  Several classes were then created for space, S, V, Q and B, varying in the degree of testing needed.  Obviously a vehicle designed to take people to space should use higher quality parts then one launching unmanned missions.

As IC’s continued to be developed, and many devices became ‘standard’ like various RTL/TTL devices and the like, the Military wanted to define those better for themselves as well.  Thus in 1969 MIL-M-38510 was released.  38510, often called JAN38510 (Joint Army Navy Standard Naming which was used through Rev J in 1991) was a General Specification for Microcircuits.  It provided fit, form and function standards for various devices.  They could be made by anyone, anyway they liked, but to be marked/used as a JAN38510 device they had to meet what it defined that device to do.  This was all

Zilog JAN MIL-M-38510 52002BQA Z8002 CPU – 1987

based on existing devices, it simply took a commercial device, such as a 74181 ALU, and gave it a 38510 description and part number.  This ensured that no matter where the Military got that 38510 standard 74181 ALU it would behave the exact same.  The 38510 standard refer’d back to the MIL-STD-883 testing procedures, it in itself did not define any testing.

As things progressed, MIL-STD-883 with the how, and MIL-M-38510 with the what, NASA decided they should have their own standard (American government agencies like to compete).  Based on the 38510 standard,and the 883 testing standards NASA created MIL-STD-975 in 1976.  This was essentially a list of products that met NASA’s standards for all electronic devices.  Everything from capacitors, diodes, cables, oscillators and even some processors. Ultimately this was a great idea at the time.  It provided designers with a list of parts they could use that NASA had already certified as acceptable, rather then having to test/certify every single piece.  The cost and time saving were immense once the initial certification was done.  The list of certified devices was updated every few years through 1994 when the standard was canceled, likely because there was just too much new devices becoming available to keep up with.  Three levels of quality are used in this standard. Grade 1 parts arc very low risk, higher quality and

Illustration of the Ørsted spacecraft in orbit (image credit: DRSI)

reliability parts intended for critical applications (such as man rated space applications). Grade 2 parts are low risk, high quality and reliability parts for usc in applications not requiring Grade 1 parts. Grade 3 parts are higher risk, good quality and reliability parts but are not recommended for applications requiring high product assurance levels.

These particular SQ80C186s are made by Intel and listed as Grade 3 devices.  This is mainly because Intel decided not to take part in the NASA certification process, so their grading is based on their MIL-STD-883 QML (Qualified Manufacturer List) testing.  These parts were used on many satellite designs (such as PoSAT-1, Portugal’s first satellite in 1993 and the 1999-2014 Danish Ørsted Geomagnetic Mission)  as well as the International Space Station.  Its possible on the ISS they were used in a non-mission critical area where Grade 3 is acceptable.   Even as a Grade 3 device the replacement cost (in 2003) was $2,266.  Today they are a mere collectors item, as parts like these need to have a certified traceability with them, knowing where they have been and how/where they were stored is important to them being certified for use. These particular chips were made in 1993, a lot can happen in 27 years of storage and transport around the world.

In some cases a lower grade device still can be used in a critical area, as long as there is redundancy built in.  The Space Shuttle, for example, used milspec devices rather then NASA spec devices for its main computer.  This was allowable as that computer was 6-way redundant, (3 ways voting with an A side and B-side).  It can save some $$ up front but long term it caused maintenance and mission delay issues as components needed replaced more.  Another good example was the American Space Station Skylab back in 1973.  Its mission computer was designed with military spec devices, saving around $300,000.  It turned out to be very unreliable due to poorer quality devices (solder balls floating inside of device cavities) and had to be replaced with a space rated computer, at a cost of $3.3 million.  That computer continued to work without issue until Skylab’s 1979 deorbiting.

There are, of course, many other military/gov’t electronics standards.  In the 1990’s MIL-PRF-38535 was released which is a General Performance and Verification Requirements standard, a cross over between 883 and the 38510 standard.  This is still in use today.  Other countries/regions (such as the EU)  have their own standards as well, though much of the West followed American standards because it was simpler and made working together on joint projects easier.

At some point we all learn that buying higher quality usually saves time and money in the long run, its no different with ICs, and even more so when what they are in, is in orbit, or on its way to some distant moon or planet.

The post SSQ22667-001: An 80C186 for the Space Station first appeared on The CPU Shack Museum.]]> 1
Aircraft Instrumentation, Bitchin’ Betty and an 80C86 CPU Wed, 30 Sep 2020 04:46:53 +0000 Quite the combination I know, but of course all related.  Last week I got some boards in that were quite interesting.  They were all fairly early serial numbered, from the 1980s and military in design.  Now one thing about anything military is identifying it is pretty hard to do, especially when it hails from an […]

The post Aircraft Instrumentation, Bitchin’ Betty and an 80C86 CPU first appeared on The CPU Shack Museum.]]>

F-15 with P4 Instrumentation Pod – Looks like a missile under the wing, with blue and red stripe.

Quite the combination I know, but of course all related.  Last week I got some boards in that were quite interesting.  They were all fairly early serial numbered, from the 1980s and military in design.  Now one thing about anything military is identifying it is pretty hard to do, especially when it hails from an era before the Internet.  Many records from the 1980s have made it online, but OCR and transcription errors abound, a single wrong digit can turn an item made for a A-4 Skyhawk into a new blade from a lawnmower or a shiny new Navy mess tray.

Thankfully these boards all had a CAGE code which the US uses to identify each and every supplier.  In this case that code was 94987 which is Cubic Defense.  Cubic didn’t make lawnmower blades or mess trays but they did make a lot of instrumentation systems for aircraft (and they continue to do so).

F-16 with blue training pod under its left wing)

It turns out that training fighter pilots is best done without having to use live weapons, for obvious reasons, but in all other aspects should remain as true to lifer as possible, and then be able to be analyzed after that fact in order to learn from mistakes, and see who gets bragging rights for pulling the most G’s.  This means that the aircraft has to send and receive data as it would in combat, threat warnings have to go off when targeted, missiles have to be ‘launched (while being captive) at the appropriate times, and every aspect of the flight must be recorded, speed, roll rates, altitude, etc.

Cubic made pods, that attached to one of a fighters weapon hardpoints (typically the outermost) that did exactly that.  These pods interface with the aircraft’s flight systems (using the standard 1553 bus) as well as with ground based systems on the training range, forming a complete picture of what is going on between all the aircraft taking part.  These particular boards are from Cubic’s second generation digital pods, the P4 series (the first gen was, the P3). Specifically the P4A series.  Each pod contained a vast amount of sensors, antennas and instrumentation to monitor and record what was happening, as well determine if a missile as ‘launched’ to or from the fighter.

Cubic 185200-1 with Harris ID80C86 – The brains of the AN/ASQ-T25 P4AM Training Pod

At their heart was a Harris or Intel 80C86 processor, (Harris actually did the CMOS conversion on the 8086).  This is one of the earliest applications of the CMOS 8086.  In this case the 80C86 is running off of the normal 8284A clock generator and a 13.5MHz crystal. This results in a processor frequency of 4.5MHz, a bit under its 5MHz rating.  This is pretty typical of military applications, it generates less heat, draws less power, and gives more margins.  This particular board has a industrial spec CPU, later production versions had a full military qualified part (this board was a prototype).

1553 Bus Controller Board made by Cubic

A separate board handled 1553 bus communications (for talking to the rest of the aircraft).  Likely many other support boards were also included, but one of the more interesting ones contains a National Semi MM54104 DIGITALKER chip.  The DIGITALKER is a integrated circuit digital speech synthesizer designed by National based on the work of Forrest S. Mozer, thus his name being on the chip

Cubic 188135-1 with very early National Semiconductor DIGITALKER – This is from a AN/ASQ-T20 P4AX Training Pod (A P4A without Altimeter support)

This is a very early DIGITALKER chip, note the patent number on it: 4124125, which is a misprint, it should be 4214125 (the other one is for a heat exchanger), which appears on later examples of the DIGITALKER.  Using compressed voice samples int he two MD2764 EPROMs on the board the DIGITALKER synthesizes a voice, which is often called ‘Bitchin’ Betty.’  This voice can warn the pilot of various things such as missile locks, targeting, or in some cases, what maneuvers to perform to not become nose art on your enemies plane.  The pod has its own as it is simulating warning that would come if there was in fact an actual missile launch.  The pod can receive telemetry from the ground, and other pods that a simulated missile launch has occurred, and provides the needed inputs to simulate what happens.

Cubic P4 Pod Memory – 1 Mbit in a Intel 7110-4 module. This is a very early prototype of the board, hand made and hand wired.

The pods also record the likelihood of the missile attack being successful or not, useful for scoring the mock dogfighting. This and other data was stored on 1 Mbit Intel 7110-4 Bubble memory modules,  These were the hot new thing in the late 1970’s, providing a fairly fast form of nonvolatile memory. Today FLASH memory does the same ting, in s much more easy to use format.

Backside of the mission memory board. All handwired. Debugging had to be fun.


All of this is powered by the lowly 80C86 processor.  These P4A pods were all retired in 2006-2010 and replaced by much more advanced P5 instrumentation pods.  The P5 has longer range (both air to air and air to ground) much better encryption support (always a bummer when your dogfight gets hacked) as well as live monitoring (no more having to download all the data after the fact). Perhaps someday I’ll find some boards from a P5…..give it 40 years or so.



The post Aircraft Instrumentation, Bitchin’ Betty and an 80C86 CPU first appeared on The CPU Shack Museum.]]> 1
Finding the Limits of the Socket 8 Thu, 10 Sep 2020 02:38:13 +0000 Socket 8 processors have something magical and I really enjoy working with them. Earlier I wrote about them more than once and it would seem that everything has already been said, but in this article you will find out which PC configuration is truly the fastest on Socket 8, although it never existed in reality. […]

The post Finding the Limits of the Socket 8 first appeared on The CPU Shack Museum.]]>
Socket 8 processors have something magical and I really enjoy working with them. Earlier I wrote about them more than once and it would seem that everything has already been said, but in this article you will find out which PC configuration is truly the fastest on Socket 8, although it never existed in reality. I just gave this platform what it never had, it’s like giving the first representatives of the Skylake processor architecture, which was released back in 2015, DDR5 and PCI-Express 4.0 today.

Before starting another fascinating story about Socket 8 and the processors that were installed there, I will give links to my previous experiments:

Chapter 2: Mini-Mainframe at Home: The Story of a 6-CPU Server from 1997
and what got us started…
Part 1: Mini-Mainframe at Home: The Story of a 6-CPU Server from 1997

As you can see, my close acquaintance with this socket has existed for a long time and over the past few years we have clearly managed to make friends. It would seem that all Socket 8 processors have been studied and tested in various configurations, including an insane configuration of six processors in such a monster as the ALR Revolution 6×6. But quite recently I got my hands on a motherboard made by ASUS, which gave me the opportunity to take a fresh look at the use of processors and the performance they are able to give in a newer platform.

What is this board and what chipset is it based on? To name the heroine of today’s article, I will first dwell on the main chipsets for Socket 8 processors. The first chipsets for Intel Pentium Pro processors appeared in November of 1995, 25 years ago. Already at that time, they understood that the future was behind the parallel execution of various tasks. The Intel 450KX chipset, codenamed “Mars”, was introduced for workstations, and the Intel 450GX “Orion” for servers. Mars allowed for dual-processor configurations, and the Orion officially supported up to four physical processors. Although on the example of the super-server ALR Revolution 6×6, which is based on Intel 450GX, the number of processors could have been much larger and could easily double the official figure.

Nowadays the term chipset is often associated with a single chip located on the motherboard, but when applied to the first chipsets for Intel Pentium Pro processors, we are dealing with the physical seven chips that made up the “number of special chips” or “chipset.” These chipsets supported slow FPM DRAM standard RAM, the server GX chipset could operate with 4 GB of such memory, while the KX “was content” with 1 GB support (Intel figuring a workstations needed less RAM then a server). By the standards of the second half of the 90s, these were immense volumes of RAM

In May 1996, a more progressive chipset appeared – Intel 440FX “Natoma”, which quickly began to replace older system logic sets. Intel 440FX itself already consisted of a pair of microcircuits, support for SMP, faster EDO / BEDO DRAM memory types along with the outdated FPM DRAM (though limited to 1GB max of RAM), a new version (2.1) of the PCI bus standard, as well as support for Intel Pentium-II processors were announced.

Most motherboards based on the Intel 440FX “Natoma” chipset have a physical design in the form of a Socket, where the processor was installed, but there were exceptions with a few using the new Slot 1 slot, where the first Pentium-II and Pentium Pro were installed through special slot adapters. A good example is the ASUS KN97-X motherboard with the included Socket 8->Slot 1 adapter called the ASUS C-P6S1.

ASUS KN97-X motherboard with ASUS C-P6S1 slocket adapter

Each manufacturer of such slot motherboards produced their own slot adapters, but due to their small circulation, finding them is now problematic. Socket 8 processors feel good in such adapters and the presence of a more modern infrastructure of such motherboards obviously contributes to an increase in performance. But Intel, having released the Intel 440FX chipset, decided to stop further support for its Socket 8 processors, although it could really have extended their life cycle.  Why just sell people a new motherboard chipset, when you cold ALSO force them to buy a new CPU to go in it?

Slocket 8 – Socket 8 to Slot 1 Adapters

Getting the Pentium Pro to work on the 440LX based Slot 1 ASUS KN97-X is not easy, you need a special BIOS version for the Pentium Pro, since the motherboards were designed mainly for Intel Pentium II processors. Therefore, each time you want to change the generation of processors, the user had to install a new BIOS chip, otherwise, beyond the POST screen, on a BIOS designed for Pentium II, you will not see anything.

Intel AL440LX motherboard by Intel

In 2016, I was able to first “make friends” with Pentium Pro and the next set of logic from Intel for Pentium-II processors – Intel 440LX “Balboa”, which was presented in August 1997. The Intel 440LX “Balboa” was already fundamentally different from all previous Intel chipsets. There was support for high-speed SDRAM memory, an integrated IDE controller with support for Ultra DMA-33, and, most importantly, there was support for a high-speed graphical interface – AGP, which worked in 2x mode.

Dual lLot 1s on a 440BX

Later, compatibility was mastered with the popular Intel 440BX “Seattle” chipset, which appeared in April 1998, which for a long time was used in computers by hundreds of millions of users. The main feature of this chipset is support for 100 MHz system bus and the ability to comfortably overclock Slot processors. The 440BX still supported the Ultra DMA-33 standard , but the speed of the disk system has increased at least twice as compared to the old PIO mode on the first chipsets for Pentium Pro. Motherboard manufacturers went further and in a number of advanced products it became possible to set the FSB frequency up to 155 MHz and in exceptional cases even up to 200 MHz (ABIT BE6-II).  The 440BX was also the first chipset where a heatsink on the chipset (North Bridge ) became common.

The FSB frequency of all Pentium Pros and the Pentium II Overdrive 333 MHz is 66 MHz. The L2 cache operates at the core frequency, in contrast to the Pentium II, where the external L2 cache operates at only half the speed of the CPU core. The L2 cache size of the Pentium Pro is also available in capacities up to 1M making it oh so delicious =). The very idea that a retro dream system with a processor the size of a cigarette pack can be run even faster makes me apply all possible efforts to implement it. The unlocked multiplier of Intel Pentium Pro processors allows using not only the standard configuration (system bus x processor multiplier) 66 MHz x3, but also 100 MHz x2, thus increasing the bandwidth of all PC components, and if you add a high-speed SSD to the Ultra DMA-33 controller , then the responsiveness of the retro system increases significantly!

The presence of an AGP video card and modern discrete controllers using PCI expansion cards generally make such a system unattainable in terms of performance in relation to the reference from the mid-90s.

It would seem then that it is impossible to get much faster than Intel 440BX. But no, there is an alternative from VIA Technologies and its chipset for SLOT 1 motherboards for Intel processors –  the VIA Apollo Pro133A. This chipset appeared much later (late 1999) than the Intel 440BX in the computer market, so it immediately gave a new user the following advantages over motherboards based on the 440BX chipset. The disk subsystem has doubled in performance due to the native support of the Ultra DMA-66 protocol, and the exchange rate with the graphics accelerator has also doubled due to the ability of the AGP interface to work in 4x mode. It is also worth noting that VIA-based motherboards support the correct AGP bus frequency divider when using the system bus of 133 MHz and higher.  It should be noted that the Pro133 and Pro133A were the first chipsets for Slot 1 to officially support 133FSB (there were some overclock BX133 boards previously but the BX never officially supported it)

In numerous comparisons and reviews of those years, these two chipsets, despite the fact that they appeared on the market more than a year apart, showed almost identical performance, but in some tests they left each other behind by up to 5%. There was more technological superiority behind the VIA chipset however. In addition to all of the above, the VIA Apollo Pro133A chipset supported up to 1.5 GB of RAM, versus only 1GB  for Intel 440BX, and this memory functioned asynchronously with respect to the system bus. Besides, the RAM frequency could be 33 MHz higher than the FSB frequency. In subsequent VIA chipsets for the Socket 462 platform for AMD Athlon processors, this solution with asynchronous bus operation continued to be used with good success.

In addition to these main differences, there was also support for hardware monitoring, and a doubled number of supported USB ports, as well as the presence of an AC’97 audio codec. What happens if you put all these technologies into an ASUS motherboard? The answer is ASUS P3V4X.

ASUS P3V4X – VIA Apollo Pro 133A

So, what is the ASUS P3V4X motherboard? The ASUS P3V4X motherboard supports Intel Pentium II / III processors, four RAM slots make it possible to use 2GB of PC133 SDRAM, as well as Virtual Channel Memory. The motherboard has a universal AGP 4x slot, thanks to which you can install the fastest video accelerator from Nvidia – GeForce 7 series. 6 PCI slots and 1 ISA slot make it possible to connect any external expansion cards, including a tandem of a pair of 3Dfx Voodoo2 operating in SLI mode.

ASUS decided to use the older 596B southbridge instead of 686A on its board, as a result of which the P3V4X lacks the AC’97 sound controller and AMR slot, but that’s for the best in this case, you can add your own sound from Creative to your taste, and with a lone ISA slot, why not stick in a AWE64 Gold.

P3V4X with C-P6S1 Slocket and Pentium Pro

From an overclocking point of view, this is also an excellent board that allows you to increase the FSB up to 166 MHz, either using dip-switches or directly from the BIOS. The ability to choose timings for RAM and increase the voltage on the processor are also available.

The choice of the asynchronous mode of operation of the RAM is clearly demonstrated in the photo of this option in the BIOS of the ASUS P3V4X motherboard.

FSB/Memory Rations

I wonder what happens when the 1995 Pentium Pro and the Pentium II Overdrive are installed in a 2000 vintage motherboard that supports Intel Pentium III Coppermine processors? You will receive all the answers to these two questions and more by reading this article to the end.

Test bench and test results

The test bench will include the following Processors:
• Pentium Pro 200 MHz – 1024 Kb L2
• Pentium II Overdrive 333 MHz – 512K L2
• Intel Pentium II 333 MHz Deschutes 512K L2 (half core speed)
• Intel Pentium IIIE 667 MHz Coppermine 133FSB – 256K L2
• Slocket 8 slocket adapter ASUS C-P6S1.

• ASUS P3V4X, chipset VIA Apollo Pro133A – Slot 1
• 256 МB SDRAM PC133 (CL2);
Graphics card:
• Gainward Bliss 7800GS 512 МB, AGP@4x (Forceware 169.21)
• Kingston SSDNow V300 (60 Gb)

There is a bit of irony in using a video card with more RAM 512MB) then the system (256MB) its in.  Its also GDDR3 RAM running at 600MHz with a core clock of 425MHz.  Clearly the GPU will not be a bottleneck in this system (though it could be limited by the AGP 4X slot, as this card is a AGP 8X design)

Testing was conducted on Windows XP, Service Pack 3 using the following software:
• Super Pi mod. 1.5XS (task 1M)
• PiFast v.4.1
• wPrime v.1.43
• WinRAR x86 v. 5.40
• AIDA64 5.50.3600
• Cinebench 2003
• 3DMark 2000
• 3DMark 2001 SE
• PCMark 2004.

And now a little about what, how and with what will be compared. First, let’s see what results will be obtained on the ‘modern’ VIA Apollo Pro133A chipset with the processor/RAM mode in the 66/66, 66/100, 100/133 MHz ratio and further overclocking. That will give the asynchronous mechanism of operation of RAM in relation to the system bus for Intel Pentium Pro and Pentium II Overdrive processors.

Some Software Engineering was needed to make the board work…

Then the obtained results will be compared with reference platforms of that time, and also the classic Intel Pentium II operating at the same clock speed (but with its half speed L2 cache) aswell as  the Pentium II Overdrive will act as the competition. I also added an Intel Pentium IIIE 667 MHz with the “Coppermine” core, which has the same multiplier (x5) as the Intel Pentium II Overdrive, in order to compare both processors at the same clock speed and the same FSB frequency. It’s time to see what happened in the end and what the history of these processors could be if Intel gave the user a little more freedom.


Super Pi mod. 1.5XS (task 1M)
Minutes (less is better)

The first thing that can be stated is the difference in the operation of RAM with an increased clock frequency in relation to the system bus! This can be seen in the example of the Pentium Pro. The difference from the transition of “old” Intel 440GX chipsets to new ones is also clearly visible. In a test that calculates the value of pi with millions of decimal places and lasts 9 minutes and 3 seconds on the classic platform, on the VIA platform the same processor saves more than one minute of time, performing the same task in 8 minutes and 2 seconds, using the frequency FSB/DRAM formula – 66/100 MHz.

By moving or overclocking the FSB frequency of the processor to 100 MHz, the performance increases even more. It should be noted that the capacious 1 MB L2 cache functions perfectly at this frequency, and its stability limit is in the 110-112 MHz FSB limit.

A similar picture emerges with the Intel Pentium II Overdrive. From the transition from Intel 440GX to VIA we have a whole minute of time savings in calculations. At a frequency of 500 MHz the Intel Pentium II Overdrive the advantage of this processor is undeniable, only the Intel Pentium III with the “Coppermine” core is able to surpass the achieved result of the leader. But there is already a technological gap between these two processors, with a faster core design and instruction set.

PiFast v.4.1
Seconds (less is better)

In this test, the same dynamics of performance growth remains, although the test focuses more on the processor capabilities of the system and less on the RAM. Hence the interesting result of the Pentium Pro overclocked to 220 MHz, which outperforms the Pentium II 266 MHz on the i440LX chipset. Against the background of all this, the result of an ordinary Intel Pentium MMX 200 MHz on Socket 7 looks ridiculous, and the Socket 8 platform is just a real modern analogue of HEDT (High End DeskTop) from Intel.

wPrime v.1.43
Seconds (less is better)

wPrime once again shows that the situation is not changing and the performance gain of the VIA platform is quite significant. In this test, the Intel Pentium II Overdrive overclocked to 500 MHz managed to outperform the equal-frequency Intel Pentium III. The 512K L2 cache (twice that of the PIII)  played a role here.

WinRAR x86 v. 5.40
Kb/sec (more is better)

WinRAR, like any archiver, loves very much that the communication channel between the processor and RAM is as wide as possible, hence the higher the FSB frequency, the better the result. Of course, the frequency of the RAM and the processor itself plays an important role. Of all the obtained values, the result of the 333MHz Intel Pentium III stands out well, at which the system bus frequency was equal to a modest 66 megahertz, and the RAM worked at a frequency of 100 MHz. And under such conditions, the Pentium III bypassed even the Pentium Pro, operating at 220 MHz with a 110FSB and 146 MHz for RAM. We can state that Intel has designed the Pentium III core very well in comparison with previous generations of processors and the performance gain is quite impressive.

Also impressive is the Pentium Pro 200 running at 100/133 just beating a Pentium II Overdrive 333 at 66/66.  That shows very clearly the FSB/Mem clock bias of this test.

Against the background of all this, four Pentium II OverDrive 333 MHz in a system based on the i450GX chipset deliver speeds slightly higher than one Intel Pentium III with a frequency of 500 MHz.

AIDA64 5.50.3600
This is how Overclocked processors are identified by AIDA64 and CPU-Z

For greater confidence, I will also present the CPU-Z validation of the Pentium II OverDrive processor overclocked to 525MHz. At this frequency, the processor was not 100% completely stable, but it still has some margin of safety, this is primarily the merit of the motherboard – ASUS P3V4X.

AIDA64 5.50.3600
CPU Queen, score (more is better)

(click for Larger)

AIDA64 5.50.3600
FPU Julia, score (more is better)

(Click for larger)

But, probably, the most interesting thing is to look at the results of the Cache & Memory Benchmark, where at one frequency you can see how much the VIA Apollo Pro133A has gone ahead of the Intel 440FX, as well as a radical revision of the algorithm of the core and memory cache of Pentium II and Pentium III.


Cinebench 2003
score (more is better)

There is no difference in rendering on a single core with an equal processor frequency and various combinations of memory subsystem operation. The only remark is related to the Pentium II 333 MHz, which is inferior to other competitors with the same clock speed. And the blame for the defeat is the L2 cache, which operates at half the clock speed of its competitors. Pentium III, as always, won the victory among all test participants.

3DMark 2000, 3D score
score (more is better)

The lack of results for Intel Pentium Pro processors is explained by the absence of MMX instructions, without which this benchmark cannot work. But in the “3D parrot meter” of 2001 there is no such limitation, but even in spite of this it is clear that there is still a performance gain from switching to a faster operating memory.

At 333 MHz, the Pentium II lost out to the Pentium II Overdrive, as it should be (due to the ODs faster L2 Cache), but the Pentium III went into an uncatchable lead.

3DMark 2001 SE
score (more is better)

Here all the processors are already assembled, it’s a pity it is impossible to install a powerful AGP accelerator in a motherboard based on the Intel 440FX chipset, but in any case, such a combination would still be a loser.

PCMark 2004
score (more is better)

This is the last test within the framework of this article and it once again shows that a Socket 8 system in conjunction with the VIA Apollo Pro133A chipset is an uncompromising solution in terms of building a retro system and the possibility of its further expansion.


This experiment showed that the impossible is sometimes possible and often manufacturers of processors or motherboards artificially do not allow further operation of their components for a long period of time for economic reasons. Such artificial technical barriers are eliminated either by updating the BIOS of the motherboard, or by minor intervention in the circuitry, or by replacing the cooling system altogether, which we all recently observed in the example of the running on motherboards based on the very first chipset for processors with the Skylake microarchitecture – Z170 and processors that this chipset does not officially support. And such examples are not isolated, so such tests as we did here, perhaps will happen again with today’s processors 20 years from now.

It is good to do such experiments as long as the platform is available, but even after a couple of decades, one can understand why this happened “then”. In the configuration we tested, the Socket 8 system in conjunction with the VIA Apollo Pro133A chipset provides the highest performance ever obtained for Socket 8 processors and the ability to use the asynchronous mechanism of RAM operation where its frequency is higher than the system bus frequency is a positive decision.

It would seem that the end of this story has been set, but as always there is one (yes, one, there is always one) exception – the existence of a motherboard manufactured by ABIT, the SH6 with an i815E chipset and a Slot 1 interface. I have been looking for this board for many years and so far I have not been able to find it, but it would be very interesting to try to launch Socket 8 processors on it and compare the results obtained on the VIA platform.

Abit SH6 -Intel i815 Based Slot 1

Summing up today’s experiment, I want to note that I have not yet put an end to experiments with Socket 8, I’ll see you here.

Another article by max1024 – Edited by CPUShack

The post Finding the Limits of the Socket 8 first appeared on The CPU Shack Museum.]]> 4
HP NanoProcessor Mask Set Thu, 20 Aug 2020 22:10:29 +0000 Since we have a complete, and very early mask set of the HP NanoProcessor (donated by Mr Bower, thank you) it seemed fitting to scan them in (tricky at 600 dpi and 6 scans each (they are around 40x60cm) then I sent them over (500MB) my friend Antoine Bercovici in France to stitch and clean, […]

The post HP NanoProcessor Mask Set first appeared on The CPU Shack Museum.]]>
Since we have a complete, and very early mask set of the HP NanoProcessor (donated by Mr Bower, thank you) it seemed fitting to scan them in (tricky at 600 dpi and 6 scans each (they are around 40x60cm) then I sent them over (500MB) my friend Antoine Bercovici in France to stitch and clean, as well as remove the background.  THat allowed this cool animation of the mask being built.
These are made from a set of 100X Mylar masks

Here you can see how the 6 different mask layers are built up.  The last mask layer (black) is the bonding pads
Each individual layer is also shown, some are very simple, while others contain a lot more.

In the lower left corner of the masks you can see their layer number 1B 2A 3A…etc

You can see the original HP part number on the mask 9-4332A as well as ‘GLB’  GLB is a composition of the initials of the two designers of the chip: George Latham and Larry Bower.

Here is a larger version as well: HP NanoProcessor Mask Set



The post HP NanoProcessor Mask Set first appeared on The CPU Shack Museum.]]> 3
The Forgotten Ones: HP Nanoprocessor Sun, 09 Aug 2020 22:18:31 +0000 Back in the 1970’s the Loveland Instrument Division (LID) of HP in Colorado, USA was the forefront of much of HP’s computing innovation.  HP was a leader, and often THE leader in computerized instrumentation in the early 1970’s.  From things like calculators, to O-scopes to desktop computers like the 9825 and 9845 series.  HP made […]

The post The Forgotten Ones: HP Nanoprocessor first appeared on The CPU Shack Museum.]]>

Original Nanoprocessor prototypes from 1974-75. Note hand written wafer number, open die cover and early part number (94332)

Back in the 1970’s the Loveland Instrument Division (LID) of HP in Colorado, USA was the forefront of much of HP’s computing innovation.  HP was a leader, and often THE leader in computerized instrumentation in the early 1970’s.  From things like calculators, to O-scopes to desktop computers like the 9825 and 9845 series.  HP made their own processors for most all of these products.  The early computers were based on the 16-bit Hybrid processor we talked about before.  At around the same time, in 1974, the HP LID realized they needed another processor, a control oriented processor that was programmable, and could be used to control the various hardware systems they were building.  This didn’t need to be a beast like the 16-bit Hybrids, but something simpler, inexpensive, and very fast, it would interface and control things like HPIB cards, printers, and the like.  The task of designing such a processor fell to Larry Bower.

The result was a Control Oriented Processor called the HP nanoprocessor.  Internally it was given the identifier 94332 (or 9-4332), not the most elegant name, but its what was on the original prototypes and die.   The goal was to use HP’s original 7-micron NMOS process (rather then the new 5-micron NMOS-II process) to help save costs and get it into production quickly.

Nanoprocessor Features – Note the speed has been ‘adjusted’


The original design goal was a 5MHz clock rate and instructions that would execute in 2 cycles (400ns).  The early datasheets have this crossed out and replaced with 4MHz and 500ns, yields at 5MHz must not have been high enough, and 4MHz was plenty.

Handwritten Block diagram


The Nanoprocessor is interesting as it is specifically NOT an arithmetic oriented processor, in fact, it doesn’t even support arithmetic.  It has 42 8-bit instructions, centered around control logic.  These are supported by 16 8-bit registers, an 8-bit Accumulator and an 11-bit Program Counter.  Interface to the external world is via an 11-bit address bus, 8-bit Data bus and a 7-bit ‘Direct Control’ bus which functions as an I/O bus.  The nanoprocessor supports both external vectored interrupts and subroutines.  The instructions support the ability to test, set and clear each bit in the accumulator, as well as comparisons, increments/decrements (both binary and BCD), and complements.

Here is one mask (Mask 5 of 6) for the prototype Nanoprocessor. You can see its simplicity.  On the bottom of the mask you can see the 11-bit address buffers and Program Counter

2.66MHz 1820-1691 – note the -5V Bias Voltage marked on it

The Nanoprocessor required a simple TTL clock, and 3 power supplies, a +12 and +5VDC for the logic and a -2VDC to -5VDC back gate bias voltage.  This bias voltage was dependent on manufacturing variables so was not always the same chip to chip (the goal would be -5VDC).  Each chip was tested the and voltage was hand written on the chip.  The voltage was then set by a single resistor on the PCB.  Swapping out a Nanoprocessor meant you needed to make sure this bias voltage was set correctly.

If you needed support for an ALU you could add one externally (likely with a pair of ‘181 series TTL).  Even with an external ALU the Nanoprocessor was very fast.   The projected cost of a Nanoprocessor in 1974 was $15 (or $22 with an ALU),  In late 1975 this was $18 for the 4MHz version  (1820-1692) and $13 for the slower 2.66MHz version (1820-1691).

At the time of its development in 1974-1975 the Motorola 6800 had just been announced. The 6800 was an 8-bit processor as well, made on a NMOS process, and had a maximum clock rate of 1MHz.  The initial cost of the 6800 was $360, dropping to $175, then $69 with the release of the 6502 from MOS.  By 1976 the 6800 was only $36, but this is still double what a Nanoprocessor cost


An early ‘slide deck’ (the paper version equivalent) from December 1974 sets out the What Why and How of the Nanoprocessor.  The total cost of its development was projected to be only $250,000 (around $1 million in 2020 USD).  The paper compares the performance of the Nanoprocessor to that of the 6800.  The comparisons are pretty amazing.

Interrupted Count Benchmark

For control processing interrupt response time is very important, the Nanoprocessor can handle interrupts in a max of 715ns, compare that to 12usec for the 6800.   The clock rate of the Nanoprocessor is 4 times faster but the efficiency of its interrupts and instructions are what really provides the difference here.

The clock rate difference (1MHz vs 4) really shows here, but the Nanoprocessor is also executed 3 times the instructions to do the same task, and still is faster.

Even using an external ALU compared to the Motorola’s internal ALU, the nanoprocessor is better then twice as fast (thanks here to its much higher clock frequency)

Full Handshake Data Transfer. Interfacing to the outside world was the main driver of the Nanoprocessor. Here we see that it can ‘talk’ to other devices much faster then the 6800

All instructions on the Nanoprocessor take 500ns to execute compared to the 1-10u for the 6800.

Today we do benchmarks based on framerates in games, or render times, but you can see that benchmarks were even important back then.  How fast a processor could handle things determined how fast the printer could be, or how fast it could handle external data coming in.  It’s no wonder that the Nanoprocessor continued to be made into the late 1980’s and many of them are still in use today running various HP equipment.

Nanoprocessor User Manual – October 1974

A big thank you to Larry Bower, the project lead and designer of the Nanoprocessor, who donated several prototypes, a complete mask set, and very early documentation on the Nanoprocessor (amongst some other goodies)

Documentation so ealy it has many hand written parts, and some corrections.  This had to be a very annoying oops if it wasn’t caught early on.  Even Engineers get their left and right mixed up on occasion


The post The Forgotten Ones: HP Nanoprocessor first appeared on The CPU Shack Museum.]]> 1