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.

Matrox SX-900: Serial# 266 – Nov 1984

In today’s age of GPU’s the GPU is often used to offload the x86 processor.  Many tasks are well suited for the thousands of GPU cores on modern graphics cards, tasks that would be a large burden on an x86 processor.  In 1984 though, Matrox took a different approach to high-end GPU design.  Matrox was founded in Canada in 1976, and has been making graphics cards since they first released the S-100 bus ALT-256 in 1978.  Matrox kept up with the hardware changes of the time, released MULTIBUS boards, Q-Bus boards, and eventually PC compatible cards.

The SX-900 was the value (around $2000) version of their 2 board GXB-1000 (that was $3000-4000).  The Matrox SX-900 was a standard MULTIBUS card with support for 640x480x8bit graphics.  It supported a fill rate of 20 MPixels/sec which was very impressive in 1984.  By comparison, the Nvidia NV1 (STG-2000) released in 1995, was only capable of a 12MPixel/sec fill rate, albeit at a richer color depth.  So how did Matrox, in 1984, achieve such performance?

Matrox SX-900: Powered by a 80286-4 Processor and upD7220 Graphics Primitives Processor

Matrox used an Intel 80286 processor, running at 4MHz (the slowest 286 made) as a Display List Processor.  It handles all high level commands (256+) and then controls the rest of the cards hardware, including the NEC uPD7220 Graphics primitive processor and a advanced pixel processor (implemented in PALs/TTL).  Together they bring rather impressive performance.  The board supports up to 4096 colors (in a Lookup Table) but can only display 16 at a time. Interestingly the board has 512K of 150ns DRAM for use as video memory, more than enough for 640×480 graphics.  Also included is 640 bytes of 25ns ECL SRAM (5x AM9122-25PC), and 16K of 120ns CMOS SRAM implemented with 2 HM6264s.  Firmware (the same firmware used for the GXB-1000) is held in 4 27128 EPROMs for simple updating as needed.

The SX-900 was used in CAD systems, industrial automation, processor control, and other applications where data needed to be shown the user graphically, rather then on a green glowing monochrome text display.  One of the more famous applications was the University of Milan (in Italy) where the SX-900 (supported by Intel iSBC286 computing boards) controlled the K800 Superconducting Cyclotron, a 100MeV particle accelerator.  THis cyclotron ended up being moved and completed at Catania, also in Italy.

Many of these boards are still in use, dutifully displaying graphics and providing user interfaces to thousands of processor control systems in factories and institutions around the world.

Anandtech has an excellent article on the new Apple A8X processor that powers the iPad Air 2.  This is an interesting processor for Apple, but perhaps more interesting is its use, and the reasoning for it.  Like the A5X and A6X before it (there was no A7X) it is an upgrade/enhancement from the A8 it is based on.  In the A5X the CPU was moved from a single core to a dual core and the GPU was increased from a dual core PowerVR SGX543MP2 to a quad-core PowerVR SGX543MP4.  The A6X kept the same dual core CPU design as the A6 but went from a tri-core SGX543MP3 to a quad core SGX554MP4.  Clock speeds were increased in the A5X and A6X over the A5 and A6 respectively.

The A8X continues on this track.  The A8X adds a third CPU core, and doubles the GX6450 GPU cores to 8.  This is interesting as Imagination Technologies (whom the GPUs are licensed from) doesn’t officially support or provide an octa-core GPU.  Apple;s license with Imagination clearly allows customization though.  This is similar to the ARM Architecture license that they have.  They are not restricted to off the shelf ARM, or Imagination cores, they have free reign to design/customize the CPU and GPU cores.  This type of licensing is more expensive, but it allows much greater flexibility.

This brings us to the why.  The A8X is the processor the the newly released iPad Air 2, the previous iPad air ran an A7, which wasn’t a particularly bad processor.  The iPad Air 2 has basically the same spec’s as the previous model, importantly the screen resolution is the same and no significantly processor intense features were added.

When Apple moved from the iPad 2 to the iPad (third gen) they doubled the pixel density, so it made sense for the A5X to have additional CPU and GPU cores to handle the significantly increased amount of processing for that screen. Moving from the A7 to the A8 in the iPad Air 2 would make clear sense from a battery life point of view as well, the new Air has a much smaller batter so battery life must be enhanced, which is something Apple worked very hard on with the A8.  Moving to the A8X, as well as doubling the RAM though doesn’t tell us that Apple was only concerned about battery life (though surely the A8X can turn on/off cores as needed).  Apple clearly felt that the iPad needed a significant performance boost as well, and by all reports the Air 2 is stunningly fast.

It does beg the question though? What else may Apple have in store for such a powerful SoC?