September 14th, 2022 ~ by admin

The History of the SUPER HEDT x86 PC


Last time we talked about the history of the development of high-end computers or High-End Desktop PC (HEDT). In the discussion of the article, some readers remembered motherboards that belonged to HEDT, but were one step higher, both in terms of their technical characteristics and in terms of the cost of ownership of the entire platform based on them. It is possible to name such very high-performance systems – Super HEDT, which often also had their own personal names, and we will talk about them this time.

Last time, the date of the appearance of Intel processors for LGA1366 and the announcement of the first motherboards for them was taken as the beginning of the countdown for the announcement of the HEDT platform. This event happened in the fall of 2008. The reason for the appearance of such platforms was dictated by the decision of the main chipmaker to create two platforms: one for all consumers, and the second specifically for enthusiasts who are ready to get additional performance and new features for extra money that were not available (or needed) for owners of conventional desktop platforms. The logic of making such a decision is simple and logical, but with Super HEDT platforms things are somewhat different.

There may be several reasons for the emergence of such platforms. One of which is to show a competitor that we can do even it better and even faster, to make performance extremes, at the cost of incredible efforts of engineers and, as a result, an incredibly high final price for the end user. Although, for such systems, the price, although it plays an important role, is not a deterrent, because there will always be enthusiasts who are willing to pay as much as the manufacturer dictates to have the very best.

By announcing such a platform, the manufacturer can be sure that the fame and success of his Super HEDT platform will smoothly spill over to his other products, even in the budget segment, because if he is capable of producing a Super “miracle” of engineering, then all of his products, will also have the same properties as the miracle ones, only they will be slower performance and a reasonable price. Super HEDT aura and advertising will do their job and competitors will be forced to return from heaven to earth, because they will have nothing to answer.

The second reason for the appearance of such systems may be for a diametrically opposite reason –  when the manufacturer fails to defeat a competitor on his football field and the only way out is to continue the game somewhere in the orbit of the planet, where the opponent will definitely not fly. Your own personal field – your own rules, even if the price of such a decision will also be cosmically high and beyond the reach of most earthlings. But they will talk about such a decision, though not for long, but long enough to lift you up onto the podium of media fame for a brief time..

And oddly enough, the first Super HEDT system recalled in this article will be the system that was born according to the second, more pessimistic scenario. Has anyone already guessed it?

To find the answer to this question, we will have to go back in time to the very end of November 2006. At this time, the golden years of AMD had already passed, the dominance of its extremely successful Socket 939 and the fastest single-core processors, which included the various Athlon 64 FX’s, had ended. With the introduction of new revolutionary Intel Core processors to the market, AMD had been cornered with its Socket AM2. The market turned towards multi-core, and AMD had nothing to boast of in performance per core. The appearance of the first quad-core Intel Core 2 Quad processors completely deprived AMD of any chance for worthy competition.

If you look at the range of processors available on the market in the fall of 2006, AMD had different Athlon 64 X2’s, manufactured according on a rather outdated 90 nm process technology, the fastest model was the Athlon 64 X2, 5200+ with a real clock speed of 2.6 GHz and a processor for ” green” enthusiasts, the Athlon 64 FX-62 with a clock frequency of 2.8 GHz and a recommended price of $1031. At this time, Intel was selling the Core 2 Duo E6700 with a clock frequency of 2.66 GHz and a recommended price of $530 and a dual-core flagship Core 2 Extreme X6800 at 2.93 GHz, 1066 MHz FSB and 4MB of L2 cache. The cost of all Extreme Editions then was $999. But in November, Intel got its first 4-core processors based on the “Kentsfield” core. The Intel Core 2 Extreme QX6700 was clocked at 2.66 GHz and became the first halo chip, and then more “popular” models followed to conquer the market. The “Kentsfield” core itself was not an honest “quad-core”, it consisted of two Core 2 Duo cores placed on the same substrate. (Hello to all chiplets, and a separate AMD Ryzen) Well, AMD did not go well with its Phenoms, but something had to be done, at least for the sake of media noise.

Core 2 Extreme X6800

The recipe for such an answer is always the same: let’s take our server platform, embellish it a little, adapt the BIOS for enthusiasts, and give the server processor a free multiplier. We will make the number of sockets more than one and the answer is ready. But server systems have a lot of limitations, besides, some of them are hardware-based and cannot be adapted to desktop standards. But in such cases, there is no time for compromises, it is better to sacrifice functionality, performance will not suffer much from this.


AMD Quadfather

Read More »

Posted in:
Boards and Systems

August 24th, 2022 ~ by admin

The Soviet CMOS 8085 CPU: 1821VM85A

Omitting the history of the creation of the first microprocessors (such as the 4004) , let’s turn to the moment when 8-bit microprocessors (Intel 8080 and  8085, Motorola 6800 and Zilog Z80 ) firmly conquered the market. It was the time of the second half of the seventies in the last century. It became obvious to specialists that the future belongs to microprocessors, and if you do not invest in these technologies, you will simply fall out of the number of the developed countries. This was well understood by the advisers to the leaders of the USSR at that time. But they also understood that, since the countries of the Eastern Bloc were somewhat late in this work from US, it would be wise to copy the microprocessors already developed overseas. After all, these microprocessors have already solved many of those problems that would take months and years to solve on their own, not to mention huge monetary costs. (It should be noted that the Soviet Union had its own original developments. For example, the 587 series microprocessor kit, this included three microcircuits.)

587IK1 587IK2 587IK3

At that time, it was not clear which chip needed to be copied – Intel, Motorola or Zilog. Each of them was good in its own way, and it was impossible to predict the outcome of the competition between them. In the end, it was decided to copy all microprocessors. And in order not to scatter forces, the enterprises of the USSR were entrusted with copying Intel products. Little Bulgaria got Motorola and Zilog chips were copied in East Germany. This is how Intel, without investing a single dollar, conquered the Soviet market. Microprocessors and microcontrollers under the names 580IK80A (8080A), 1821VM85A (80C85A), 1816VE48 (8048), 1816VE51 (8051)  became native to Soviet electronic engineers.


The hero of this article is the 1821VM85A-8 bit microprocessor, a functional analogue of Intel 8085A (but in CMOS). It has been developed since the beginning of the 80s at the Novosibirsk plant of semiconductor devices. Production began in 1985. According to Wikipedia, the manufacturing technology is CMOS, 3 microns. According to other sources 0.7 microns, silicone-on-sapphire. Clock frequency – 5 MHz. Theoretically, it can work at a higher frequency.

The die contains about 6500 transistors. When copying the 80C85, several schematic and topological errors were corrected. As a result, the analogue saves stored data without a minimum clock, but the original does not. This manifests itself when the clock frequency changes. When it is reduced to zero, the microprocessor falls asleep, but the contents of all registers remain unchanged. When clocking resumes, the microprocessor continues to execute the program from where it left off.

It was produced in both ceramic and plastic 40-pin DIP packages.

IM1821VM85A – 1990

IKM1821VM85 (tin pins) – 2012

military grade M1821VM85A – 1993

KM1821VM85 – 1991

IKR1821VM85A – 1994

KR1821VM85A – 1996

At the beginning of the marking of microprocessors, the letters I (И), K, M could be used. I did not succeed in finding out the meaning of the letter I (И) in the marking. Sometimes the letter A is missing from the microprocessor name. The letter I (И) is often present in the marking of the chips of the Novosibirsk plant of semiconductor devices. Here is an example of a microcontroller from this manufacturer.

IKM1850VE39 – 80C39 MCU from 1991

Marking the production date on chips was done in two ways. The first is the usual one, consisting of four digits. The first two digits are the year of manufacture, the second two digits are the week of that year. The second marking method corresponds to this table.

Read More »

Posted in:
CPU of the Day

July 14th, 2022 ~ by admin

The History of the HEDT x86 PC – Part 2 – AMD

AMD’s debut in the HEDT market

Part 2 of the History of the HEDT – See Part 1 (Intro + Intel)


The debut of the AMD High-End Desktop platform happened in the fall of 2017. For his main competitor, he was unexpected and even shrouded in some mystery. In March 2017, the first models of the AMD Zen microarchitecture processors appeared on store shelves, they were the first Ryzen, where the flagship was the 8-core AMD Ryzen 7 1800X, priced at a modest $499. At that time, Intel’s flagship in the desktop segment was the Core i7-7700K, which belongs to the Kaby Lake microarchitecture and had only 4 cores capable of processing 8 threads. The cost of the Core i7-7700K was then $350. The new six-core new flagship Coffee Lake microarchitecture Intel Core i7-8700K would only appear six months later at a price of $370.

AMD planned to release processors with the ZEN microarchitecture for the desktop, mobile and server markets, it did not originally plan to release HEDT processors.

The appearance on the market of AMD Ryzen Threadripper processors has been shrouded in mystery. According to one version, the merit of their appearance lies with the enthusiasm of a small group of AMD engineers who, in their spare time from their main work, experimented with creating a high-performance processor that could be even more productive than the desktop Ryzen. A group of enthusiasts have been working on this project for about a year, and at the end of the work they presented their developments to the company’s management, which, in turn, considered them promising and allocated the necessary funding for commercial development.

Socket TR4

Until May 2017, no one knew about the work on these processors, which would then be called the Ryzen Threadripper. In August 2017, the long-awaited announcement of the first HEDT platform from AMD with three models of AMD Ryzen Threadripper processors happened. The processors were installed in the new TR4 socket with a crazy number of contacts at that time, 4094 contacts. The most basic Ryzen Threadripper 1900X ran at 3.8 GHz, had 16 MB of L3 cache, and had 8 cores capable of processing 16 threads. Such a processor cost $549. The average was the Ryzen Threadripper 1920X with 12 cores and 32MB of L3 cache. Such a processor cost already $799. Twelve-core HEDT competitor from Intel with thermal paste under the cover – Core i9-7920X cost $1199. The flagship of the entire HEDT line from AMD was the 16-core Ryzen Threadripper 1950X, which was estimated at $999. Intel’s 16-core counterpart, the Core i9-7960X, was offered to enthusiasts at a price one and a half times more expensive, for $1,699.

What did AMD’s first HEDT platform offer to wealthy enthusiasts?

Read More »

Posted in:
Boards and Systems

July 11th, 2022 ~ by admin

The History of the HEDT x86 PC – Part 1


In this article, I would like to recall how the history of high-end computers or High-End Desktop PC (HEDT) began, what we now have in this segment, and what awaits us in the near future.

Until a certain point in time, the computer market for personal computers was not divided into subcategories. There was the concept of a personal computer, where its main criteria were: performance and cost. The higher the performance, the more expensive such a PC was. There was and still are ‘Workstation’ class PC’s but these are really more of a Business class (think CAD or Video editing) then what you would buy for your house. The problem of insufficient performance was solved for a very long time with the help of overclocking and it would seem that this order of things suited everyone.

If you want a faster video card or processor, buy the Top model. Until a certain point in time, everything was like that, but at the beginning of the 2000s, processor manufacturers realized that there was a certain group of buyers who were willing to pay more. Then they were called “enthusiasts”, and now they have been renamed “gamers”. Since 2003, Intel, in unison with AMD, has been releasing their processors for wealthy enthusiasts. The first processor model from “blue” was the – Pentium 4 Extreme Edition with a clock frequency of 3.2 GHz and a very nice price of $999 (A Celeron of that era was around $100 for 2.5-2.8GHz).

Thus, in this processor model, a beautiful and memorable cost, a defiant name and technological sophistication, the roots of which go deep into the server segment, are combined. The owners of the “extreme desktop processor” already considered themselves a completely different caste, and no overclocking of the older processor model could give an ordinary user the performance that an enthusiast had, and after all, extreme processors were also overclocked.

At first, AMD generally went the other way, a special separate platform was created for enthusiasts, where processors with their own separate socket were installed. Thus, the segmentation of the personal PC class took place at the physical level. We are talking about AMD Athlon 64 FX processors, and to be more precise, about the first model of this family – AMD Athlon 64 FX-51. I would call them timeless classics, still using a ceramic package, a separate socket, and special registered DDR-SRAM memory.

The release of these desktop processors for enthusiasts also marked the beginning of a new 64-bit era and changed the leaders of the processor industry. The yellow jersey of the leader shone on a green “background”, and Intel moved into the camp of catching up. As revenues and the image component of users and enthusiasts grew, marketers and simple engineers did not sit idly by. Performance is never enough (although it seems to me that for the last 5 years it has definitely been enough in any products of the middle-end segment) and something had to be offered to enthusiasts who were willing to pay even more. I suppose that such wealthy enthusiasts are now called creators or a close meaning of this term.

HEDT from Intel

Read More »

Posted in:
Boards and Systems

June 30th, 2022 ~ by admin

Chip of the Day: Soviet 573RF10 – a CMOS 8755A

Intel released the i8755 in 1976, the i8755A in 1977 (with better compatibility with the 8085A and 8086/8). The Intel 8755 is an UV- erasable and electrically reprogrammable ROM (UV-EPROM) and I/O chip. The EPROM portion has 16 384 bits, organized as 2048 words by 8 bits. The I/O portion has two general purpose I/O ports, each I/O port is individually programmable as input or output.  These were essentially a combination of the 8255 PIO and the 2716 EPROM on a single die/package. These were made on a NMOS process.

Intel C8755-8 – 1977

Intel C8755A – 1979

NEC D8755AD -1981

Toshiba TMP8755AC ’83

NEC and Toshiba released similar microcircuits behind Intel. Basically, the microcircuit was intended to work together with the 8085A microprocessor. It differs from its predecessor i8080A in that it has a multiplexed data and lower address bus. The standard three-bus architecture of the microprocessor system is obtained by multiplexing with the help of an additional external register. In this register, the low byte of the address is fixed by the special output signal of the microprocessor.

Intel 87C75PF Engineering Sample – 1988

By 1988, the 8755A was obsolete and Intel released the 87C75 instead (see article on the CMOS 87C75).

Novosibirsk IM1821VM85A – 1989

Around this time, the production of an analogue of the i8755A, the 573RF10 microcircuit, began in the Soviet Union. Why start producing a microchip that the world electronics leader is changing to a more advanced one? The fact is that at the beginning of 1988, the production of IM1821VM85A began in the USSR. This was a radiation hard analogue of the CMOS i80C85A. It was with it that the 573RF10 was supposed to work.

K573RF10E (gold pins) 1990

KM573RF10 – Gold ’92 / tin pins ’93

The chip is made in a 40-pin side-brazed ceramic DIP. Supply voltage +5 V. Programming voltage +21 V. It was produced at the Vostok fab in Novosibirsk on a CMOS process (to match the 80C85A).

Unmarked 573RF10

The 573RF10 is the only CMOS chip in the 573 series.

573RF10 die – single memory cell –

Intel 8755A die – 2 memory cells –

It is noticeable to the naked eye that the 573RF10 is own Soviet development. The 573RF10 and i8755A dies are completely different. The i8755 has two memory arrays clearly visible, while the 573RF10 has only one.
It must be said that the application of the 573RF10 chip was not wide enough. And in general, the idea did not take root. The next obvious step in evolution was the combination of a microprocessor, ROM and RAM, input-output ports in one chip which was frequently done on the MCS-48 and MCS-51 series MCU’s which were also being produced in the Soviet Union at the time.

Written by guest author Vladimir Yakovlev
Edited/Formatted by John Culver – The CPU Shack Museum
Pictures – The CPU Shack Museum and others

Posted in:
CPU of the Day

June 13th, 2022 ~ by admin

The History of Angstrem Memory IC’s in the USSR

This article is about memory chips manufactured by one of the entities – the leader of the electronic industry of the USSR – Angstrem. As you know, the Soviet Union ceased to exist in December 1991. We restrict ourselves to the development period of the considered memory chips produced at Angstrem, the end of 1991. Let’s make an attempt to track how the capacity of memory chips grew, how technologies were improved that allowed the Soviet Union not to let the world leaders in electronics go far from itself at that time. A small example: Angstrem’s Dynamic RAM 4K went into mass production in mid-1975, Intel introduced its own in 1974. Intel launched a 16K DRAM in 1977, and Angstrem released its counterpart in 1978.

Angstrem Headquarters

Angstrem was established in June 1963 in Zelenograd (outside of Moscow) as a pilot plant in conjunction with the Scientific Research Institute of Precision Technology. At Angstrem, new technologies for the production of microelectronics were developed, and pilot batches of new microcircuits were also produced. The debugged production technology was then transferred to other enterprises of the USSR and countries of Eastern Europe.
The development and manufacture of memory chips was one of the main activities of Angstrem. It was on them that new semiconductor structures and production technologies were more effectively worked out, and the stability of obtaining finished products is considered in world electronics as a sign of technology ownership. It’s relatively easy to make a small batch of good chips, it’s hard to make a process whereby a large amount of chips can be made and be reliable. It was the very low chip yield percentage that played a cruel joke on Angstrem when mastering the production process of the DRAM 565RU7 chip.


In 1966, Angstrem created the first MOSFET in the USSR, which was the first step towards the strict goal of creating CMOS integrated circuits. The first CMOS microcircuit, created in the Soviet Union in 1971, was the 16-bit Angstrom matrix of memory cells 1YaM881.The supply voltage is 6 volts instead of 5 volts, like the rest of the chips in this series.

1YaM881 – 1972

The next in a series of static RAM chips was the CMOS K561RU2 (K564RU2), released in 1976. 564 series of chips is a “military” analogue of the 561 series. In these series, there are several dozen microcircuits. The chip has an organization of 256 words by 1 bit.

561RU2 die – 16×16 256bit matrix clearly visible – The image is taken from the site with the permission of the author.

It contains 2067 integral elements. Supply voltage is 3-15 volts. It’s an analogue of CD4061A.  It should be noted that in most cases ‘analogue’ means similar to, not an exact copy or exactly compatible.  The USSR did make some compatible IC’s, but they mostly made stuff that was similar, but built to their own specifications/needs.

K564RU2A -1978

K561RU2 -1979

The package of the K561RU2 chip is wider than the standard packages of this series.

K565RU2 -1979

The K565RU2 static RAM chip was manufactured using NMOS technology. Chip capacity was 1024 bits (1024×1). Contains 7142 integral elements. An analogue of Intel 2102A, developed in 1974. K565RU2 appeared in 1977. It was originally designed to be placed in a ceramic package, but later, in order to reduce the cost of production, the dies began to be packed in plastic packages.

Read More »

Posted in:
CPU of the Day

June 5th, 2022 ~ by admin

CPU of the Day: P.A. Semi PA6T PowerPC

When Apple bought P.A. Semi back in 2008 it was the beginning of the era of the iPhone, and their was much speculation as to why Apple was buying a company that made low power high performance PowerPC processors.  Especially since the iPhone ran on ARM and the Mac had moved from PowerPC to x86.

P.A. Semi PA6T-1682M

P.A. Semi was started in 2003 by Daniel Dobberpuhl (who passed away in 2019).  Dobberpuhl was one of the truly greats of microprocessor design, with a career starting at DEC on the T-11 and MicroVAX, before helping DEC transition to the Alpha RISC design (21064).  It was Dobberpuhl who started the design center in Pal Alto (where P.A. Semi would later take its name from) that designed the DEC StrongARM processor.  A processor that was later purchased by Intel and became the XScale line of ARM processors.

After Intel bought the StrongARM line, he then helped start SiByte, making MIPS based RISC CPUs, and continued to do so when SiByte was purchased by Broadcom. So when he started P.A. Semi it was less about PowerPC and more about RISC, PowerPC just happened to be the architecture they chose to use.  The design team had extensive experience on a variety of CPU architectures, including SPARC, Itanium, and the early Opterons.  You can see why this acquisition was so attractive to Apple.

PA6T block diagram

In the few years (2003-2008) from when P.A. was founded to when Apple took them over, they did design, market, and sell a PowerPC processor line called PWRficient based on what they called the PA6T core.  The PA6T-1682M was a Dual core PowerPC processor (the 13xxM was the single core version) with each core running at up to 2GHz with 64K of L1 Instruction cache and 64K of L1 Data cache.  They were fab’d on a 65nm process by TI and ran at 1.1V.  The L2 cache was scalable and shared amongst the cores.  In the 1682M this was a 2M 8-way cache with ECC.  One of the most useful features was their clock stepping.  They could drop to 500MHz at only a few watts per core, and then back up to the full 2GHz in 25us.

AmigaOne X1000 (made by Aeon) PA6T-1682M

The PA6T was only on the marked for a few months (from the end of 2007 to April 2008) before Apple bought them for $300 million, but in this time P.A. Semi had numerous design wins.  Amiga selected it for use in the AmigaOne X1000 computer.  The AmigaOne did not hit market until 2011, which means that while P.A. Semi was bought and completely under control of Apple, they still continued to make, support, and supply their previous customers with the 1682M CPU.  Certainly Amiga wouldn’t be big enough to push Apple to continue making a chip?

They were not, but others were, and the PA6T was such a great processor that it had been selected and designed in to many computer system used by US Defense contractors, and if anyone doesn’t like change, its Defense contractors, so with some prodding by the US Dept of Defense Apple continued to make (or rather have TI make) the PA6T processors.  Curtis-Wright had designed the PA6T into their new CHAMP-AV5 DSP VME64 board, which was used for signals processing across numerous military applications.  They also also used the PA6T (at 1.5GHz) in the VPX3-125 SBC. Themis computers, NEC, Mercury and others designed in the PA6T. Extreme Engineering, another maker of PA6T based boards, referred to the design as ‘ground breaking.’

Extreme Engineering XPedite8070 SBC

It would have been interesting to see what P.A. Semi could have achieved had they not been gobbled up by Apple.  Clearly we see the results of the talent of the P.A. team in what Apple was able to accomplish with their A-series processors, but clearly P.A. had something special for the PowerPC architecture as well.

, ,

Posted in:
CPU of the Day

March 26th, 2022 ~ by admin

The DEC/Compaq Turbo Laser 6 AlphaServer KN7CH Processor

AlphaServer GS60 and GS140

The DEC TurboLaser 8200/8400 was a series of high end Windows NT compatible servers/workstations introduced in 1995.  After DEC was sold to Compaq (in 1998) the 8200/8400 were upgraded from the EV5/EV56 (21164/21164A) to the 21264/21264A (EV6/EV67).  Compaq wasn’t as bold with code names it seems so instead of being referred to as the TurboLaser they were simply called the TL6.  The machines themselves were also renamed from the 8200 to the GS60 and the higher end 8400 to the GS140.  GS referring to ‘Global Solution’ to reflect Compaq’s international marketing of the computers.  The GS60 was the lower end rackmount model supporting up to 6 CPUs and 12GB of RAM and the GS140 full cabinet model supporting up to 14 CPU and 28GB of RAM.  Both could be configured with either 21264 525MHz CPUs with 4MB of B-cache each or 700MHz 21264A CPUs with 8MB of B-cache each.  The 21264A added support for writeback cache, as well as its faster speeds and some new instruction set extensions.  Initially availability of these systems was in late November of 1999, coinciding with the release of the 21264A CPUs.  By the time of their release Alpha support for Windows NT was lagging, so most if not all systems were sold with Tru64 UNIX or OpenVMS OS.

The GS60/140 were large cases similar to a rackmount system but self contained.  The processor modules for them contained a pair of CPUs, the cache for the CPUs and the entire chipset.  They connected to the main computer with a very large connector that provided power (48VDC) as well as all the Memory/IO and clock signaling.  This was referred to as the TLSB (TurboLaser System Bus).   The fastest of these was the KN7CH (also known as the E2067-DA) which had dual 700MHz 21264A processors with 8MB of Cache each.

DEC KN7CH 6/700 Processor Board

This processor board is quite interesting, its a rather early board (PLDs are dated March of 2000) and the pair of Samsung 21264A processors are dated 9944, these are some of the very first production 21264As.  Also of interest is that these Samsung CPUs are 733MHz models (KP21264A-733UCN).  The 21264A was to be made in 600, 650, 667, 700, 733, and 750MHz versions, though I have only actually seen 667 and 733MHz versions.  Making only 2 speed grades of the processor would greatly simplify testing and logistics, and with a rather limited customer base, there wasn’t a clear marketing need to make so many different speeds, these were not CPUs that were generally available outside of OEM use.  These servers were also designed to be high reliability systems, running a 733MHz rated CPU at 700MHz would increase reliability by decreasing heat related wear and tear.

Build Sheet for a 8-Node GS140 with Eight 6 CPU GS140 6/700 Systems. Each with 12GB of RAM. A nice $9 million system

The entry price for the AlphaServer GS60 with 4 GB of memory was $199,990 ($340,000 in 2022). The AlphaServer GS140 system price started at $399,400 ($680,000 in 2022). These were very expensive systems.  One look at the processor board shows what that kind of expense gets you, a whole lot of gold.  Its hard to find another computer system built in 2000 that has 9 gold/ceramic chips on each processor board.  A single dual processor board was $45,000 ($76,000 in 2022USD), and each 4GB of RAM was another $49,000.  One can easily see how such a system could quickly cost several million dollars.  Each of these boards cost as much as a really nice car!  Lets look at what that $45,000 gets you

Top Row (L->R) SWI, Alpha 21264A, SWI – Bottom Row: TDI, TDI, TCC, TDI, TDI

2x KP21264A-733UCN. Each 21264A chip has a separate address and data bus for the B-cache and system operations. The 21264A chip has a 64-Kbyte instruction cache and a 64-Kbyte data cache.  These are made by Samsung on a 0.25u process and dissipate 85Watts at 2.0V.

20x IBM SRAM Cache Memory: 8-Mbyte ECC L2 cache per CPU made using 16x IBM 0418A81QLAA-4 512Kx18 8Mb ECC SRAM chips and 2x 128Kx36 / 2x 256×18 for the TAG RAM

2x DEC 21-47306-01 SWI: Two swizzle (SWI) chips receive data from the 256-bit wide DLSB (the DEC Local Bus) and pass it to one of the CPU chips over the 64-bit wide data interface bus.  These are located on either side of the pair of CPUs.

4x DEC 21-47307-01 TDI: Four TurboLaser Data Interface (TDI) chips receive data from the TLSB (the main system bus that connects all cards in the system) and pass the data over the DLSB to the two SWI chips.  These are the outer 4 chips on each end of the row of 5 gold chips on the bottom.  Each one handles 64-bits of the 256-bit TLSB.

1x DEC 21-47315-01 TCC: The TurboLaser control chip (TCC) takes commands from both CPUs and issues them to the TLSB. It also controls all data movements through the TDI and SWI chips. This is the center chip between the pairs of TDI chips.

2x AMD AM29F080DB-90EC: 5V 8Mbit Flash for the system firmware

4x Galaxy Power DC-DC Converters.  These regulate down the 48VDC supplied by the systems redundant power supplies to the voltages needed for the board.  There is a pair of 2.2V 7A converters for the CPUs, and a 7A 3.3V converter for all the I/O.  There also is a smaller 2A converter of unknown voltage (likely 5V).

Pair of Samsung KP21264A-733 Processors surrounded by cache chips

The TurboLaser line was replaced in 2002 by the WildFire servers (GS80, GS160 and GS320) which upgraded the CPU support to 32 21264Cs with 256GB of RAM.  Unfortunately by this time Compaq had merged with HP and the combined server line was a bit cluttered, having Alpha, PA-RISC, Itanium and Xeon based systems.  The Wildfire and its Marvel follow on were the end of the road for the Alpha.  Unfortunately the same thing happened with the PA-RISC and Itanium (ok maybe not so unfortunately with Itanic) as well.  The days of boards full of golden RISC are past, replaced by BGAs with enormous heatsinks.

Posted in:
Boards and Systems

February 11th, 2022 ~ by admin

How do you test a S3 GPU? With an HP 93000

GammaChrome XM18 – Engineering Sample

Recently I got in some very nice S3 GammaChrome GPUs.  The GammaChrome was S3 (owned by VIAs) follow on to the DeltaChrome and included support for such things at PCI-E.  The S18 (Code name Brooklyn) supported speeds of up to 500MHz and was made on a 130nm process by TSMC.  S3 also made a mobile version of the S18 called the XM18 (Code name Metro MPM) in 64MB and 32MB versions.  Clock speed on these was around 350MHz (memory on the samples I have is 350 so core should be similar).  The XM18 was packaged on a MPM (Multi Package Module) with 2 RAM chips and the GPU mounted on a small chip size BGA with around 800 balls.  This is very similar to how ATI packaged some of their mobile GPUs (like the Mobility Radeon 7500 and 9600).

HP 93000 (from HP Brochure)

So how do you test one of the XM18 Engineering Samples? Or any large scale chip for

86C813 ES Gamma Chrome XM18 ULP MPM64

that matter?  With Automated Test Equipment.  ATE systems are designed to rapidly test various chips to verify their design/performance before they go into full production (or to test samples of production ones).  The HP/Agilent 93000 (spunoff as Verigy in 2007 and acquired by Advantest Corporation in 2011) was introduced in 1999 to handle such testing, and at the time was rather revolutionary.  Previously most test systems used a simple test head that would mount the chip to be tested, with all the processing and customizations being contained in the main test machine.  This worked fine for a single design, but to test multiple chips got pretty expensive.  HP moved the testing to the test head directly, interfacing to the target chip via a large PCB.  This way changing chips only required updating the test program, and changing out the PCB.  Design changes required reworking a single PCB, rather then the entire test machine.

HP 93000 Test Head – Notice the 16 groups of pins (some covers and some mangled in this old sale photo)

The 93000 was the first ATE that achieved (on its low end (200Mbps) a cost of $1000/pin tested, and on the high end, test speeds of up to 1250Mbps (for the P1000 version, at a cost of $6-7000 per pin).  The XM18 has around 800 pins, half are probably power/ground so 400 some odd testable pins, in a mid range HP 93000 and you see these systems were not inexpensive. Well over a million dollars for a midrange system.

GammaChrome XM18 – Metro MPM Test Board

To use such a system the chip to be tested would be mounted on the test board, usually with a BGA socket.  This board breaks out all the various connections of the chip to 16 sets of contacts, which the probe head of the HP 93000 made contact with using spring loaded contacts.  The board is then clamped down and tests are run.

Connection List

These boards are very very large, each one is 17x23inches (43x58cm) and 5mm thick.  They weigh about 7lbs (3.1kg) as well.  They got used a lot and need to be rather robust and durable.  You can see the boards are marked with tables of all the connections, and where they are brought out to.  Useful information about what supporting equipment is need (sockets and stiffeners etc) is marked on the board as well.

Back of board. Notice all the capacitors, a crystal, and a series of 5VDC reed relays (the red devices)

These boards appear to be a ‘static’ type item, but they do require adjustment, notice the markings that say not to use this board, it needs recalibrated.  Looking closely at the board you can see capacitors have been removed/replaced, and many of the capacitors have felt tip marker markings on them.  Keeping the capacitance and inductances at their proper values 9and matched, considering the long trace lengths) would be a very important thing.

S3/VIA Matrix Test Board. The Matrix was the code name for the GammaChrome S14/S19

These test boards are from 2006, the 93000 systems are still being used today in upgraded form (now called the V93000) to test SoCs and other chips.  As chips have gotten more and more complex, faster, and with larger pin outs, test equipment continues to grow ins peed, and cost as well, but is an essential part to the process of designing, producing and supporting a successful GPU or CPU.

, , ,

Posted in:
Boards and Systems

January 7th, 2022 ~ by admin

The Many Sockets of VIA CPU’s

C5M – Ezra-T Prototype Pathfinder – PGA370

Most are familiar with the history of VIA so we won’t dive extensively into that but a quick summary is in order.  VIA was founded in California in 1987 before moving to Taiwan, and previous to 1999 was well known for making chipsets and other support chips for computers.  In 1999 VIA bought both Cyrix (from National Semiconductor) and Centaur Technologies (from IDT, who made the Winchip series of processors.

These purchases did two main things for VIA, it first gave them access to the x86 architecture, and it gave them legal leverage to continue down the x86 road.  Cyrix possessed a license to the P6 processor bus (through a cross licensing with Intel) that was good until 2006.  This allowed VIA to make what became the Centaur based CyrixIII/C3 processor on the P6 based Socket 370 platform. These are the processors and socket we are most familiar with for VIA CPUs.  With clock speeds of 466-1.2GHz and eventual support for the Tualatin based boards these chips were the most ‘public’ facing CPUs.  VIA also of course made many BGA versions, used in ITX form factor, and other mini type systems.

CNA – Isaiah – Interestingly using the old Pentium III-M pin out

The VIA designs, despite originally being called ‘CyrixIII’ were all based on the Centaur designed core.  Intel, as was its custom, sued VIA in 2001 asserting patent infringement, which it is likely VIA was expecting.  As with the case of Intel and Cyrix, VIA countersued, asserting Intel was infringing on patents VIA had acquired with the Centaur deal.  In 2003 a settlement was reached that included a 10 year patent cross license between Intel and VIA and allowed VIA to continue to make x86 compatible processors (extended in 2013 by 5 years until 2018(.  The deal also granted VIA a 4 year (with an extra optional year) license to continue to make chipsets compatible with Intel processors (they had originally signed a deal in 1998 to allow VIA to do so. This is how we continued to get VIA chipset based motherboards for Intel processors.  The deal also added a small detail that leads to todays discussion, it granted VIA a 3 year grace period to continue making bus and pin compatible processors up through 2006.

C5J (Left) and C5R (Right) – Banias Compatible Pentium M pin out

This last part is interesting, the fact that it was a grace period means it reflected what VIA was currently doing, not what they were planning to do in the future.  The obvious example here is the C3 line on Socket 370 using the P6/Tualatin bus, but that was pretty old news in 2003 so what was VIA working on?  CPU’s on more modern sockets of course, namely Socket 479 (mPGA479M) used by the Pentium-III-M (Tualatin) and Pentium M (Banias/Dothan).  These use the same physical socket on a motherboard, but the keying pins are different on the CPUs themselves.  These are all mobile designs which lend themselves well to VIAs low power designs.  VIA did also make several reference boards for these CPU’s so its clear that there was plans for releasing them to the broader market, and likely with additional motherboard support.

C5J (Left) and C5R (Right) – C5R is a 110nm part with a slightly larger die the the C5J

Another socket was just being developed at the time of this agreement, and that is perhaps the most interesting.  Intel LGA775 chips began sampling in late 2003, which is after the grace period of 3 years had begun so it would make sense for VIA to not develop CPU’s using a socket they were going to lose access to in a few years.  The package likely was in development for a couple years prior which is likely why VIA made a few (likely VERY few) samples for it.  The samples are marked C5R which is a C7 Esther core, if VIA’s naming is consistent, this would be the TSMC 110nm version of the 90nm C5J.

C5R With heatspreader and with heatspreader removed.

The Esther core code names are a bit confusing because of how some CPUID programs identify them. has a quite nice ID guide that goes into some great detail on them.  In summary there was a 90nm Rev A C5J made by IBM, and later a 90nm C5J (called Rev D) made by Fujitsu with some additional features.  This Rev D part often gets identified as a C5R, or a C5J shrink, neither of which is correct.  The actual C5R (and related C5Q) were what appear to be backup plans for the IBM produced parts, using a larger 130/110nm process at TSMC. Looking at the mPGA479 unfinished packages (labeled C5J and C5R) the die attach area on the C5R is actually slightly LARGER then the C5J (~35mm2 compared to 28mm2 of the C5J)

C5R Esther – 110nm TSMC in LGA775

Most VIA samples are labeled with the code name in Cxx format and not the marketing code name (Esther Isaiah etc) as each of the Marketing code names (for lack of a better term) consisted of many actual sub-cores.

Code Code Name Process Die Size
(sq. mm)
C5A Samuel TSMC 180nm 75
C5B Samuel 2 TSMC 150nm 52
C5C Ezra TSMC 150/130nm 52
C5M Ezra-T TSMC 130nm Proto Only (Pathfinder)
C5N Ezra-T TSMC 130nm 56 Cu Interconnects – Low-k – Tualatin Bus
C5X Nehemiah TSMC 130nm 78 10% Faster then C5XL – Higher power
C5XL Nehemiah TSMC 130nm 52 133FSB
C5XP Nehemiah Low Power C5XL – Not released
C5P Nehemiah TSMC 130nm 47 200FSB – DP Support
C5Y Nehemiah Unreleased – Adds SSE2
C5Z Nehemiah Unreleased – VIA V4 System Bus
C5I Esther 90nm Initial Esther – Almost Taped out
C5J Esther Rev A 90nm IBM
Rev D 90nm Fujitsu
C5Q Esther TSMC 130nm Unreleased
C5R Esther TSMC 110nm Unreleased – Samples Made
C5W Esther IBM 90nm SOI Canceled early
CNA Isaiah Fujitsu 65nm VIA Nano 1000/2000
CNB Isaiah Fujitsu 65nm VIA Nano 3000
CNQ Isaiah TSMC 40nm VIA Nano X2 4000 VIA Eden X2 4000, VIA QuadCore E U4000 / L4000 – (two die VIA Nano X2 or VIA Eden X2)
CNR Isaiah TSMC 28nm VIA QuadCore E – C4000, VIA Eden X4 C4000

Looking at the table above we can see VIA took many roads in the development of their CPUs, with many that went nowhere.  Some may see this as a lack of direction or focus, but in a lot of ways VIA seemed to be trying to figure out the best CPU for the market at the same time they were trying to make the best CPU from an engineering standpoint.  Where these two paths converged you had a marketable CPU that made it into mass production, and where they didn’t, or where legal road blocks arose, the design was canceled.  VIA’s CPU development is even more obscure now, though they have made a few other designs we will cover in a later article, as well as the return of Intel to the VIA party.




Posted in:
CPU of the Day