sliderider wrote:

One day I will build a ALR Revolution 6x6. […]
Show full quote

SquallStrife wrote:

kool kitty89 wrote:

There were some dual socket 7 board too though. (not sure how many CPUs supported multiprocessing, but most/all of the Pentiums seem to have done so)

Ooooh yeah!

The very good Socket 7 system! 😎

One day I will build a ALR Revolution 6x6.

http://www.vanvleet.net/ALR%20Revolution%206X6.htm

Booyah!

I recall a thread on here recently (maybe semi-recently) where a guy built one of those, complete with modified case.

Can't seem to find it, but it had loads of pictures in it. A really cool build.

sliderider wrote:

Drools. Does he have PPro's in it or PII Overdrives?

overdrive 333MHz, 6 of them. 😀

I used to be a hater. Me and my shiny Pentium.

I came across this while browsing another one of my old boot magazines. These chips did pique my interest at the time.

That was an interesting article. Thank you for sharing it. Some of the Cyrix Media GX systems also have ISA and PCI slots. I wonder if the PCI slots on them are only for audio cards, or if graphics cards also work. Has anyone tried this and how much performance will a PCI expansion card offer over the Cyrix integrated CPU-graphics solution? For the Cyrix integrated graphics controller, communicating with the CPU at full CPU clock frequency sure sounds tempting on paper.

The speed of CPU/graphics communication could be an advantage, but that would obviously be limited by the feature set of the graphics controller/accelerator. (if it's slow/weak/feature-poor, that speed isn't going to help much)

On the other side, there's also the performance disadvantage of sharing main RAM for graphics (a similar problem regardless of the controller being on-chip with the CPU -chipset embedded graphics have similar problems). One major advantage there would be for CPU based rendering, since you could write/copy/modify the framebuffer directly in main RAM without having to fiddle with copying to video memory over the ISA/VESA/PCI/AGP bus. However, most/all software assumes a conventional dedicated VRAM configuration and bypasses that advantage too.

Too bad the Media GX was limited to low FSB speeds and lacked external cache support. (either of which would have been useful for closer-to-mid-range systems, and better future proofing in general -as faster chipsets/RAM got cheap)
And as I mused on before, implementing the 5x86/GX core (sans embedded video/sound) in socket 7 form factor would have been really interesting too. (primarily as a low-end counterpart to the 6x86/MII, but potentially more than that too if yields ended up consistently higher, allowing higher clock speeds -and thus more competitive FPU/multimedia performance; maybe the 5x85 core would have ended up as the low-end multimedia/gaming option with the "full" but slower clocked 6x86 and MII catering more to the high-performance business/desktop market -assuming they didn't beef up FPU performance or manage to get specific game/driver support with more integer-centric implementations)

And, of course, it's unfortunate that Cyrix didn't manage to balance their system-on-a-chip and mainstream CPU design/development better. (hell, had they managed to stay competitive in mainstream CPUs, they probably would have had a much better chance to get established in embedded stuff too . . . as it is, they -or National Semiconductor at least- seemed to switch focus to embedded R&D too too soon and too heavily, and they ended up falling behind in the mainstream while also failing to establish a strong niche in the embedded market -NS's own financial problems certainly didn't help either)

The Red Hill museum has an interesting take on this subject:
http://redhill.net.au/c/c-8.html

A weird and innovative design, the MediaGX arrived in 1997. It was an all-in-one device combining CPU, memory controller, graphi […]
Show full quote

A weird and innovative design, the MediaGX arrived in 1997. It was an all-in-one device combining CPU, memory controller, graphics card and PCI controller on a single chip. In its success, it destroyed an entire company.

Because it seemed to have so much potential in the low-cost market, as a set-top box component in particular, it dragged Cyrix's attention away from the main market — orthodox high-performance desktop parts — and attracted the interest of other companies, notably National Semiconducter, which bought Cyrix largely on the strength of the MediaGX design, and over the next year or so proceeded to mismanage the company into oblivion.

→ A very unusual way of mounting a CPU. Yes, it's just as thin and flat as it looks in the picture. It's a MediaGX-166 from a Compaq Presario P2200.

The MediaGX was developed by Cyrix's second design team, the same team that had produced the 5x86, as a low-cost component for mass-market home systems. With a MediaGX-based system, the video card and sound card functions were performed on the CPU itself. This resulted in a cheap and reasonably well-performed system, but it was non-standard and rather restrictive.

The single-chip motherboard was unique to the MediaGX and couldn't be chip-upgraded to a Pentium or 6x86, and the built-in sound and graphics prevented these from being upgraded too. In short, the MediaGX was mainly of interest to brand-name manufacturers selling cheap and underpowered systems to first-time buyers through the supermarket outlets. Amstrad and Commodore were both defunct by this time, but they would have loved it.

Hah, the MediaGX is what killed Cyrix. It certainly looked good on paper though, and apparently National Semiconductor really thought so too. A problem with big companies like that (Nat. Semi.) is that there is too much red tape, progress is slow, and takes 3x more employees (and 5x more money) than a small company would need to do the same thing.

feipoa wrote:

Hah, the MediaGX is what killed Cyrix. It certainly looked good on paper though, and apparently National Semiconductor really thought so too. A problem with big companies like that (Nat. Semi.) is that there is too much red tape, progress is slow, and takes 3x more employees (and 5x more money) than a small company would need to do the same thing.

In hindsight, the GX probably wouldn't have hurt them at all had they kept it as a more moderate priority (with mainstream desktop parts maintained as the primary business model). They could have eased their way into the embedded market more gradually and still been at/near the head of the pack for x86 based SoC designs (since few others were pursuing that at the time).

It could have made even more sense if they specifically designed the embedded cores to also work as low-end standalone CPUs (for desktop or mobile applications), thus getting better value out of engineering resources spent on the embedded project.
Again, a socket 5/7 5x86 derivative could have been very interesting in such a role, especially if it had consistently better yields than the contemporary MI/MII parts. (meaning consistently higher clock speeds and potentially more Pentium-like performance attributes with weaker ALU but faster FPU making it an attractive low-end gaming alternative once Pentium-optimization became common ~1997 onward; unless Cyrix had beefed up the MII's FPU too, in which case the 5x86-derived core could still fit well in the low-cost/low-power niche for desktops and laptops -and probably would have been much better than the Winchip)

Plus, short of getting bought up and changing management (as with the NS merger), the foray into SoC designs could have meant more investment interest and better PR. (and that alone could have made it worthwhile in the short run)

feipoa wrote:

A problem with big companies like that (Nat. Semi.) is that there is too much red tape, progress is slow, and takes 3x more employees (and 5x more money) than a small company would need to do the same thing.

This is an excellent point too, and one of the big reasons the IBM+Cyrix partnership ended up working as well as it did was their separation preventing IBM's bureaucracy from meddling with Cyrix's productivity. (and as a bonus even compared to Nat. Semi., IBM had access to leading-edge semiconductor fabrication technology and was directly involved with x86 PCs in a fairly wide range of market sectors -meaning they'd have some interest in Cyrix's CPUs in their own x86 based machines too, somewhat like IBM had done with their own licensed x86 derivatives prior to the Cyrix partnership)

There's a ton of other examples in history where smaller companies ended up ruined by integration with corporate giants or by self-inflation to similar effect.

The IBM/Cyrix deal worked as well as it did because before then the only fabs that Cyrix had access to were older ones that didn't have the capacity or the quality that Cyrix really wanted but they had no other choice. IBM also needed 486/5x86 designs so they would't have to pay a huge fee to Intel for their chips so they benefited from a cross licensing agreement in exchange for the use of their more modern fabs that were capable of turning out the quantities Cyrix needed with a minimum of defective chips.

sliderider wrote:

The IBM/Cyrix deal worked as well as it did because before then the only fabs that Cyrix had access to were older ones that didn't have the capacity or the quality that Cyrix really wanted but they had no other choice. IBM also needed 486/5x86 designs so they would't have to pay a huge fee to Intel for their chips so they benefited from a cross licensing agreement in exchange for the use of their more modern fabs that were capable of turning out the quantities Cyrix needed with a minimum of defective chips.

This is all true, but it also doesn't contradict any of what I said.

Cyrix went from manufacturing chips through various 3rd parties (ST/TI/etc/etc) prior to the IBM deal and were usually limited to relatively low priority facilities with older process technology and/or lower yields. (and also limited time to market since Cyrix was at the mercy of those 3rd parties rather than in direct control over the plants -so first silicon could be delayed)
The partnership with IBM gave Cyrix relatively high priority access to modern, high-tech plants with relative secure and stable access to them.

However, my point IBM in my previous post was no in comparison to Cyrix's previous situation (with 486 SLC/DLC and SX/DX parts), but compared to what happened under National Semiconductor, where Cyrix was merged into NS's infrastructure and bureaucracy, causing problems that Cyrix had never dealt with prior to that. (though similar to those they'd have faced had they merged with IBM rather than forming the looser partnership they had)
On top of that, NS itself fell into financial difficulties and didn't have cutting-edge manufacturing capabilities on the level of IBM. (which, among other things, may have contributed to the very late introduction of 2.2V 250 nm Cyrix CPUs)

I remember me and my brother both having Cyrix pr 233 computers and I remember him having heating issues with his. The strange part about it I never had any heating issues with mine. Those computers were great for the time we had em.

From my recent tests of Cyrix 6x86 CPUs, I have found that even the smallest overclock can be enough to show instability. The instability may be temperature related, but may also be from the frequency element.

Examples,

Cyrix 6x86-90GP, this is an 80 MHz processor, typically at 2x40, although I've never seen a 40 MHz socket 5/7, but maybe they exist? I ran it beautifully at 1x75 MHz for my ensuite of benchmark programs, which takes about 3-4 hours at that speed. I did start to notice some instability at 1x83 Mhz (a 3 MHz overclock) once the processor had been powered up for some time. To complete the benchmark tests, I had to wait for it to cool down. To run a Cyrix 6x86 at 1x, it seems the 4x setting can be used.

Cyrix 6x86MX-233PR, this is a 188 Mhz processor, typically run at 2.5 x 75. I was able to run the ensuite of benchmarks at 133 and 150 Mhz. I was able to run all but 3 tests at 166 MHz as well, but 200 MHhz was out of the question.

Cyrix MII-366PR typically runs at 2.5 x 100, or 250 MHz. It ran fine at 233 MHz but I couldn't even run the DOS tests at 250 Mhz (83x3.0).

All the chips which struggled at their rated speed got very hot. I should probably plunk down a thermistor to quantify "very hot". Even after adding heatsink compound to the HS/FAN, the above examples still failed.

feipoa wrote:

From my recent tests of Cyrix 6x86 CPUs, I have found that even the smallest overclock can be enough to show instability. The instability may be temperature related, but may also be from the frequency element.

It's probably a mix of both issues (logic stability at high speeds and heat dissipation). The more inherently sensitive/near-limit a chip is, the more temp-sensitive they tend to be, and models that naturally run quite hot (like the faster ~.6 micron 3.5V 6x86 models or the faster 350 nm 2.9V MIIs) will be more problematic in general.

Cyrix chips were late to transition past 350 nm, so you see some very high-strung CPUs at relatively high clock speeds for that process. (AMD only went to 233 MHz with their 350 nm K6 -and that was at 3.2V- while Intel pushed the early 350 nm PII to 266 MHz -alongside the 233 -at least in AMD's case, they'd been struggling with good yields at 350 nm, but the switch to 250 nm changed that dramatically -beyond AMD's expectations, pushing higher yields much faster than anticipated and releasing the K6 300 early)

So the 2.9V MII 366 may very well be very cooling sensitive. (worth looking into) The one I got on ebay came with a fairly large (albeit passive) aluminum heat sink . . . though that would need a good case fan to work well. http://www.ebay.com/itm/280824575282?ssPageNa … 984.m1497.l2649

Do you have a 2.9V 75x3.5 PR333? (that's the fastest -core- clocked mass produced 2.9V model, Red Hill's site mentions it along with the 3x83 PR333 and mentions that it was both more stable and better performing -also the fastest MII they ever sold since the PR366 wasn't available to Australia wholesalers -let alone the PR400)

Cyrix 6x86-90GP, this is an 80 MHz processor, typically at 2x40, although I've never seen a 40 MHz socket 5/7, but maybe they exist? I ran it beautifully at 1x75 MHz for my ensuite of benchmark programs, which takes about 3-4 hours at that speed. I did start to notice some instability at 1x83 Mhz (a 3 MHz overclock) once the processor had been powered up for some time. To complete the benchmark tests, I had to wait for it to cool down. To run a Cyrix 6x86 at 1x, it seems the 4x setting can be used.

That's also running the FSB much higher than its rated for though. (like if you ran a 100~133 MHz rated part -PR120-166- at 1x100+)

Haha! Scored a 5x86-100GP off ebay. Had to buy a PowerMac DOS compatibility card to get it but I reckon it was worth it. 😁

Chewhacca wrote:

Haha! Scored a 5x86-100GP off ebay. Had to buy a PowerMac DOS compatibility card to get it but I reckon it was worth it. :D

I've never been able to get any of my Cyrix 5x86-100GP's to run at 133 MHz, however some can run at 120 MHz (2x60 or 3x40). I've had pretty good luck getting an IBM 5x86c-100HF and Cyrix 5x86-120GP to run at 133 MHz though.

Chewhacca wrote:

Haha! Scored a 5x86-100GP off ebay. Had to buy a PowerMac DOS compatibility card to get it but I reckon it was worth it. 😁

Lol! I got one of my Cyrix chips that way... I guess Mac serves a purpose for me after all. Honestly it's cheaper that way for some reason as a rule.

But as feipoa said, I'm not sure if you can do 133 on those... I'll have to do some benchmarks and see what comes up. But it's still faster at 120MHz than an AMD at 133 in my experiences.

Continued from: VIA C3/C7 Discussion

kool kitty89 wrote:

It was my impression that the Jalapeno chip was an SoC using a proprietary form factor (as with GX and MXi), but the Mojave/M3 w […]
Show full quote

swaaye wrote:

I did some digging... […]
Show full quote

I did some digging...

Cayenne - Core chip for Cyrix MXi and Gobi.
Jedi - Former name for Cyrix Joshua (before Gobi).
Gobi - Former name for Cyrix Joshua.
Joshua - Former core for VIA Cyrix III (never finished).
MXi - Cyrix (NSM) integrated chip [Proprietary]
Jalapeno - Former codename for Cyrix Mojave.
Mojave - Cyrix M3 [Socket 370]
Serrano - Cyrix M4 [ ? ]

Joshua = Cyrix's Cyrix III
Samuel = Centaur's Cyrix III

MIII---
Cyrix's next-generation Mojave chip is based on the 0.18-micron 'Jalapeno' core and will start at 600MHz. It's a SoC like MediaGX. It features 256KB of on-chip Level 2 cache, an 11-stage deep-pipeline, a new floating point unit, a 3D graphics engine, 3DNow! and MMX instructions, support for DVD playback and a 3.2GBps memory interface (RAMBUS).

It was my impression that the Jalapeno chip was an SoC using a proprietary form factor (as with GX and MXi), but the Mojave/M3 was the conventional desktop counterpart to that. (intended for Super Socket 7 and/or Socket 370)
And that, in turn, became the Joshua core used in the preproduction VIA Cyrix III.

I think the 11-stage pipeline comment is also a mistake as the M3 and other 6x86 based designs were still relying on the old 7-stage superpipelined architecture based on the original 6x86 -which would have been the main contributing factor to limited clock speed scalability, rather like the K6-2+/III+. (the Centaur design did use an 11 stage pipeline though, so that's probably where the confusion comes from)

Hmm, looking around more, there's mention of the 11 stage pipeline (and ground-up design nature) of the M3 (of various code-names) well before the VIA merger, so I was wrong.

http://www.eetimes.com/electronics-news/40382 … ix-is-uncertain
http://www.eetimes.com/electronics-news/40364 … s-Jalapeno-core
http://www.eetimes.com/electronics-news/41484 … -Processor-Core

So the MIII/Jalapeno/Mojave was the ground-up new 11-stage CPU design planned, but seems to have never been completed.

The Cayenne/MXi/Gobi core appears to be M2 based with added 3DNow and improved FPU along with additional graphics/memory interface in the MXi design and 256k L2 cache on the Gobi (which later became the VIA Joshua, reaching preproduction as a S370 part)

Edit: the Centaur Samuel Core actually uses a 12-stage pipeline, so I was wrong with that comment too. (on the 11-stage Cyrix core being confused with the Centaur one)

I know I mentioned the issue of Cyrix's lag in going beyond 350 nm parts before, but looking again, that period (when AMD and Intel went 250 nm) is even more significant than I'd considered before.

Prior to AMD and Intel's move to 250 nm parts in 1998, Cyrix's newest 6x86MX models had managed to keep up relatively close in clock speeds with mid 1997 featuring the 200/66 MHz PR233 alongside Intel's Pentium II 266 and 300 and AMD's K6 233 (albeit the latter was achieved at 3.2V, with the 200 being the fastest 2.9V K6). While Intel obviously had a lead in raw clockspeed at the time, the gap to AMD was relatively small (especially considering the "factory overclock" nature of the 350 nm K6 233).

However, in 1998, thing changed with AMD's massive jump in yields and clock speeds (and lower power/heat) with the 250 nm K6 and K6-2 as well as Intel's 250 nm PII (and the arrival of the Celeron -especially the Mendocino later on the year), while Cyrix/IBM were still stuck on older 350 nm chips and substantially lower clock rates and yields (and considerably larger dies -even compared to the Mendocino) while also not resorting to voltage boosts as AMD had done previously.

I'm not sure what led to the delay in moving beyond 350 micron chips (issues with the NS merger?), but whatever the case, that lag seems to be an even more definitive cause of Cyrix's decline than the management/bureaucratic problems tied to the NS buyout. (imagine if the K6 had stayed 350 nm through 1998)

That note of sticking to 2.9V is also interesting; I wonder why Cyrix didn't resort to AMD's menthods of "factory overclocking" with the likes of the 3.2V K6 233 (or the various 2.3/2.4V 250 nm parts later-on). Maybe the chips ran unacceptably hot at those voltages, or maybe the boost didn't help stability on the M2 as it did on the K6. (that would be an interesting thing to test)

I just got this IBM PR333 chip 😀

Upsidedown image? You just fixed it!

That baby should be able to handle 75x3.5 (262 MHz) at 2.9 V, and maybe even 83.3x3.5 (292 MHz) at 2.9-3.1 V if you are lucky. My IBM PR333 works well at 262 MHz.

If your motherboard supports Cyrix 4X CPUs, you should be able to run it at 4x66 (266 MHz)

Yeah sorry, it should be fixed now.

I also have a Cyrix MII-333GP (75MHz). Which I take is basically the same chip but 75MHz FSB and not 83?

Cyrix had so many different names for the same chip!