For the oldies! What happened to Cyrix Processors

[Vince]

I'm not sure buying a firm would be seen as a transfer. I've no idea if VIA have the rights to produce X86 chips.

Don't quote me but I am pretty sure that the transfer of x86 and stipulations surrounding it cover the company being acquired for but not limited to its IP. Basically at its simplest, to produce x86 intel must say its ok. Given Intels hate for everything but money I find it difficult to believe they will allow any more competition that they need to.

To be honest we are lucky to have two players today given the history and I doubt Intel would let a third player in, as good as it would be to see a third player like Cyrix back producing cpu's. If anybody knows categorically all about x86 and the license I would be all ears.

 

[stewski]

I thought early Cyrix were fairly close to competition in everything except Floating Point?
I never had one but advised others to cheap out on them, something I was not thank for by those who played loads of quake...

 

[vanpeebles]

Haha I just found some old files with the spec on, it was a Cyrix M2 300 on a PC CHIPS M571, with a voodoo 1 card and a cirrus logic 2d card. The bus was normally set to 3.5 by 66mhz. Using the jumpers and delidded with Athlon cooler I had it running at 83mz. According to some test software at the time:

At 66mhz bus speed 300(Intel equiv 231) mhz it did 379 Million Instructions Per Second.
At 75mhz bus speed 333(Intel equiv 262) mhz it did 425 Million Instructions Per Second.
At 83mhz bus speed 350(Intel equiv 283) mhz it does! 479 Million Instructions Per Second.

 

[JasonM]

I thought early Cyrix were fairly close to competition in everything except Floating Point?

For regular instructions they were on average 25% faster then the Pentium 1 at same Mhz. However floating point was only around 2/3 of the Intel floating point.

I owned both the Cyrix 586 and the Cyrix M1, I benchmarked them at the time. I'm also a computer science graduate and in 1996 my final year degree project was a x86 super superscalar parallel simulator that was x86 clock cycle accurate running assember instructions over 2 x86 pipelines. I studied the Cyrix CPU's at the time and they were quite inspiring as there architecture was better then the Pentium.

 

[stewski]

For regular instructions they were on average 25% faster then the Pentium 1 at same Mhz. However floating point was only around 2/3 of the Intel floating point.

I owned both the Cyrix 586 and the Cyrix M1, I benchmarked them at the time. I'm also a computer science graduate and in 1996 my final year degree project was a x86 super superscalar parallel simulator that was x86 clock cycle accurate running assember instructions over 2 x86 pipelines. I studied the Cyrix CPU's at the time and they were quite inspiring as there architecture was better then the Pentium.

The reason I advised people to pick them was reviews had said similar I even mentioned that fpu had been noted as weaker, still the quakers didnt understand and blamed me.

Interesting project btw.

 

[JasonM]

The reason I advised people to pick them was reviews had said similar I even mentioned that fpu had been noted as weaker, still the quakers didnt understand and blamed me.

Interesting project btw.

At this time floating point was only used in things like AutoCad, so most people from 80386DX onwards had no use for the co-processor.

Intel had always been strong on floating point from their co-processors, and it was just bad timing that Quake arrived the same time Cyrix was trying to gain share. The fact that Cyrix was weak on floating point was not really their fault, as up until Quake no one was really using it!

Thanks for comment on project, somewhere I have it on a hard drive, and intend to make a YouTube video showing how it executes assembler instructions.

 

[stewski]

At this time floating point was only used in things like AutoCad, so most people from 80386DX onwards had no use for the co-processor.

Intel had always been strong on floating point from there co-processors, and it was just bad timing that Quake arrived the same time Cyrix was trying to gain share. The fact that Cyrix was weak on floating point was not really their fault, as up until Quake no one was really using it!

Thanks for comment on project, somewhere I have it on a hard drive, and intend to make a YouTube video showing how it executes assembler instructions.

Was in IT at exactly that period and droped off electro mech degree precisely due to a love of all things computing.

As CISC cpus went at the time I thought cyrix was a good value proposition, my family (sister) had one and it served their purpose.

Post it, I'd be very interested...

 

[hyperseven]

Had a choice between Cyrix and AMD for my first ever build, went with the K6-2

Not entirely sure but also had an Orchid righteous 3DFX paired with a Matrox Mystique in that same rig.

 

[Dalandius]

My understand was while the Cyrix M2 was a great processor but the guy who programmed Quake "used highly optimized assembly code designed almost entirely around the P5 Pentium's FPU." which ran slowly on the Cyrix M2.

 

[Huntyz]

The one single product that killed the Cyrix chips was id Software's Quake.

Quake used very cleverly optimised x86 code that interleaved FPU and integer instructions, as John Carmack had worked out that apart from instruction loading, which used the same registers, FPU and integer operations used different parts of the Pentium core and could effectively be overlapped. This nearly doubled the speed of FPU-intensive parts of the game's code.

The interleaving didn't work on Cyrix cores. It ran fine, but the operations did not overlap, so execution speed halved.

On every other benchmark and performance test we could devise, the 6x86 core was about 30-40% faster than the Intel Pentium core -- or the Pentium MMX, as nothing much used the extra instructions, so really only the additional L1 cache helped. (The Pentium 1 had 16 kB of L1; the Pentium MMX had 32 kB.)

But Quake was extremely popular, and everyone used it in their performance tests -- and thus hammered the Cyrix chips, even though the Cyrix was faster in ordinary use, in business/work/Windows operation, indeed in every other game except Quake.

 

[DarrenM343]

My understand was while the Cyrix M2 was a great processor but the guy who programmed Quake "used highly optimized assembly code designed almost entirely around the P5 Pentium's FPU." which ran slowly on the Cyrix M2.

I would think games were/are FPU intensive generally anyway?
I remember AMD around a similar time, maybe a few years later, also had a faster CPU but again suffered when it came to FPU. I think I had one for a year, an Athlon if I remember correctly
?

 

[mmj_uk]

The Cyrix M2 just had a really poor FPU, they were mainly designed as a cheap alternative for word processing and general office usage which used integer operations.

 

[malachi]

We all knew Cyrix were bad when Cex refused to take them for trade in due to the CPU's being so unreliable.

I had 3 and none of them worked with my PS2 keyboard or IE would randomly hard crash so I had to use Netscape Navigator for that period. Went Intel and AMD, all worked fine.

 

[HangTime]

They probably melted.

In all seriousness, I had a 6x86 PR200+ but the reality was it was a 150mhz cpu that performed about the same as a P133 in games. This was 20 years ago before I'd ever used the internet so doing research was much more difficult than it is now. I was able to overclock it in the BIOS to 180mhz by choosing 3*60 (kids, this was a big deal; in the 90s overclocking typically meant playing with physical jumpers on the motherboard) but my CPU fan conked and I didn't realise for ages so it got massively overheated (Cyrix were notorious for running hot even at stock). It then rapidly degraded over time so I had to keep dropping the clock speed to try and keep it stable before it finally packed up.

 

[jennidc]

I nearly went with a Cyril cpu for my first pc after using an amiga 1200 with a blizzard board, squirrel pcmcia scusi cdrom and 80mb hard drive for years. In the end I got a great deal on an Intel p150mhz chip that ran at 166 out of the box on the stock cooler.

 

[MagicBoy]

They probably melted.

Tru dat. I also had a PR200+, one of the early IBM badged 3.5V ones. Fan didn't fail but it got so hot it melted the ZIF socket. Was replaced with a 2.8V "L" version and a later TX board.

In general (integer) use the performance was stellar, but not much good once I got a 3Dfx card, as the FPU wasn't pipelined... so a P120 would outperform it. Quickly swapped to a P200MMX...

 

[mistersprinkles]

All I know is my buddy ended up with a Cyrix MII 300Mhz and compared to my Pentium II 300Mhz it was a joke for gaming. Starcraft was a slideshow on his machine literal slideshow it ran great on mine.

 

[Rroff]

I would think games were/are FPU intensive generally anyway?
I remember AMD around a similar time, maybe a few years later, also had a faster CPU but again suffered when it came to FPU. I think I had one for a year, an Athlon if I remember correctly
?

Yeah with a (advanced) 3D game engine a fair amount of floating point operation is pretty much inevitable. Carmack broke a lot of things back then - using multi-texturing for lightmaps for instance when everyone else was using single passes and many GPUs had poor multi stage texture support - it is another thing that finished some GPU brands back then that couldn't compete with those that could support stuff like that.

Indirectly Quake (1-3) killed off a few companies but largely I don't feel bad for them.

 

[MagicBoy]

All I know is my buddy ended up with a Cyrix MII 300Mhz and compared to my Pentium II 300Mhz it was a joke for gaming. Starcraft was a slideshow on his machine literal slideshow it ran great on mine.

To be fair to Cyrix, it ran at ~233Mhz and was sold as a Pentium 300Mhz equivalent, at 1/4 the price of the 300Mhz PII...

The mII was very little different to the 6x86Mx.

 

[ubersonic]

It was a shame the way Cyrix went down, I remember back when their last CPU came out reading the reviews on Tomshardware of the 1.4Ghz version, they said they were quite positive about it and it was a good rival for the Duron/Celeron and that they hoped the next CPU could take on the Pentium/Athlon.