If I remember correctly the games of choice were Starcraft 1 and Quake 3, and I did get a different motherboard with the Cyrix.

The Pentium 133 was on an older VX style motherboard, I don't remember what the Cyrix came on.

I agree it was a touch faster in Windows than the Pentium 133, but not enough to make up for the poor game performance.

I benchmarked it and I believe the FPU performance was around the Pentium 120 I also had.

Oh wow, I just went through my CPU collection and I still have the PR200 (and a PR233 MX)! Cool.

I also have a couple of IBM/Cyrix 486 chips 😀

This was back in 1999 or 2000 when I was a poor school student 😀

mwdmeyer wrote:

I also have a couple of IBM/Cyrix 486 chips :)

5x86's?

mwdmeyer wrote:

If I remember correctly the games of choice were Starcraft 1 and Quake 3, and I did get a different motherboard with the Cyrix. […]
Show full quote

If I remember correctly the games of choice were Starcraft 1 and Quake 3, and I did get a different motherboard with the Cyrix.

The Pentium 133 was on an older VX style motherboard, I don't remember what the Cyrix came on.

I agree it was a touch faster in Windows than the Pentium 133, but not enough to make up for the poor game performance.

I benchmarked it and I believe the FPU performance was around the Pentium 120 I also had.

Oh wow, I just went through my CPU collection and I still have the PR200 (and a PR233 MX)! Cool.

I also have a couple of IBM/Cyrix 486 chips 😀

This was back in 1999 or 2000 when I was a poor school student 😀

In that context, the K6 and K6-II definitely would have been available too, but you're talking about really low-budget parts for the time in general.

Were you buying parts new or used? (for that timeframe and on a tight budget, buying used should have been much preferred -unless you had no good local used parts dealers) By mid 1999, used K6, Pentium MMX, and 6x86 MX chips should have been very reasonable (at least for those in the 233 MHz range).
New PR 233 MX and K6 233 parts had already been at or below $100 and Pentium MMX 233 at about $140 in mid 1998 given this chart:
http://www.realworldtech.com/altcpu/chart.htm (so used parts a year later should have been substantially less than that)

Edit: But you were sticking with your old socket 7 VX board, right, so you may not have had BIOS support for K6 or 6x86 MX. (in which case, Pentium MMX would be the best option -or maybe a fast WinChip variant -most of which tended to work without explicit BIOS support iirc, though FPS was rather weak so a high clock speed would definitely be needed to be worth it -though models with 3DNow! would also help for compatible games)
-Though the cheapest option is obviously what you already ended up doing in the end. (just overclocking the 133 to 166)

Also, you're talking about a PR200 classic, right? (75 MHz bus 150 MHz core) So your VX board had a 75 MHz setting? (I'm pretty sure the 6x86 classic only supported a 2x multiplier, so if you were running a 66 MHz bus, it would have been only 133 MHz -and effectively a PR166)
That's opposed to the PR200 MX which had both 75x2 and 66x2.5 MHz models.

Is your PR233 the 3x66 MHz bus version?

I got a motherboard with the Cyrix (different from the Pentium), I don't remember what model it was though (probably a 430TX). I was 13 or 14 years old at the time, I got the motherboard + Cyrix CPU for $50 if I remember correctly. My school friend sold it to me, all second hand, it wasn't until a few years later could I afford a new system.

My Pentium 133 cost me $5 for a VX motherboard from a random computer shop, the time went backwards on it which is why it was so cheap. The CPU was free because all the pins were bent! I spent a few hours unbending them.

Yes the Cyrix PR200 was the 75x2 model. I need to check the PR233.

I've got a few Socket 3 Cyrix 486 chips. Not 5x86.

I lie. They are IBM Blue Lightning 486 DX2/66 chips.

Also in the photo is a 486 DX50 and an AMD 486 SX2/66.

mwdmeyer wrote:

I got a motherboard with the Cyrix (different from the Pentium), I don't remember what model it was though (probably a 430TX). I was 13 or 14 years old at the time, I got the motherboard + Cyrix CPU for $50 if I remember correctly. My school friend sold it to me, all second hand, it wasn't until a few years later could I afford a new system.

My Pentium 133 cost me $5 for a VX motherboard from a random computer shop, the time went backwards on it which is why it was so cheap. The CPU was free because all the pins were bent! I spent a few hours unbending them.

Wow, OK, so that's definitely different circumstances than I was imagining . . . very resourceful for a 13/14 year old. 😉

Do you remember how much the PR200 cost you at the time?

Yes the Cyrix PR200 was the 75x2 model. I need to check the PR233.

So you managed to get the VX board to work at 75 MHz?

On another note, I noticed this odd part on ebay:
http://www.ebay.com/itm/Cyrix-MII-M2-Vintage- … =item231af6b018

It's a PR 266, but 3x66 (like the 66 MHz PR233) rather than 2.5x83. It's also not on the list here:
http://www.x86-guide.com/en/cpu/Cyrix-6x86.html (maybe a fake?)

Ha thanks. Yeah I believe I built my first computer (well pulled apart and rebuilt) when I was 6 years old, an old 12MHz 286.

Most of my first computers came from my dad's work when they would throw them out. I loved those days. He would tell me the day before that he would get me a computer from the office. The best time was when he said they were throwing away some 386s (I was running a 386SX 33MHz with 4mb of Ram), The next morning I ran down stairs to see what he had gotten me, it was a 486 DX33 with 8mb Ram, I was so happy!

All my first computers were an odd configuration as I could never afford a full machine but I did purchase upgrades from time to time. For example my Pentium 133 had 3 video cards in its life time. A 2mb S3 Trio, 4mb SiS 6326 and then finally a TNT2M64 32mb (I played half life on that machine!).

I never did get the Cyrix running with a 75MHz FSB on the VX motherboard. I tried and I believe the system reported as a PR166, it was such a long time ago now!

The Cyrix + Motherboard was $50 I believe. So not a bad deal. Just wished it was a Pentium 200 😀

My first new computer was when I purchased a Motherboard + Case + CPU + Ram. It was a AMD Duron 800MHz, Gigabyte GA-7IXE4 (http://ee.gigabyte.com/products/page/mb/ga-7ixe4/) and 64mb Ram. I used the Sound Card, Hard Drive, Video Card and Network Card from the Pentium 133! In fact it was somewhat a step down as my Pentium 133 had something crazy like 90-128mb ram in it towards the end of its life (started with 24mb).

I enjoyed those days. My main computer now is a boring 27" i7 iMac 😀

The closest I can find from the Cyrix literature is:

6x86MX - PR233GP 3x66
6x86MX - PR266GP 3x75
6x86MX - PR266GP 3.5x66

MII - PR266 2.5x83

However, there seem to be plenty differant collectors who have an MII-266GP 3x66
http://www.chipdb.org/cat-266-173.htm

Cyrix has, at times, changed their own PR ratings, so the MII-266GP may be one such CPU.

feipoa wrote:

The closest I can find from the Cyrix literature is: […]
Show full quote

The closest I can find from the Cyrix literature is:

6x86MX - PR233GP 3x66
6x86MX - PR266GP 3x75
6x86MX - PR266GP 3.5x66

MII - PR266 2.5x83

However, there seem to be plenty differant collectors who have an MII-266GP 3x66
http://www.chipdb.org/cat-266-173.htm

Cyrix has, at times, changed their own PR ratings, so the MII-266GP may be one such CPU.

Yes, as with the 3x83 MHz 333 vs 350, and per the 3x75 and 3.5x66 PR300 (vs the 266s you listed -and 266 probably would have been a more realistic rating).

Apparently there were some actual architectural improvements that led to re-rating some speed grades, but there also seems to be a general trend of less stringent ratings for the late Cyrix parts. (especially compared to the early ones that benchmarked faster than the Pentium equivalent -ie PR166 being comfortably faster than a Pentium 166 in integer performance)

On another note: looking around at some comments and comparisons on the FPU performance of the 6x86 vs Intel and AMD parts: the clock per clock performance of the FPU seems to be roughly equal to the K6 and P5 Pentium, but also application-dependent (there's disproportional performance for different operations).

In particular, if wiki's cycle time charts are accurate: http://en.wikipedia.org/wiki/X87
It shows that the 6x86 FPU is considerably slower at FMUL and FADD than the P5 (4 vs 1 cycle minimum), but noticeably faster for FDIV (which is a much more intensive operation in general at 24 vs 39 cycles minimum) and the P6 FPU is actually slower than the P5 in FMUL (2 min) but much faster in FDIV (16/17 min). For many multimedia applications (and especially 3D games), multiplication and division performance are major factors, so the slower FMUL (and add) could be mitigated (or even surpassed) by the FDIV performance. (and in the case of the P6's generally higher FPU benchmarks vs P5, this definitely seems to hold true -albeit the advantage to the 6x86 is much more modest and it's obviously at a universal disadvantage to the P6 FPU)

Also, given that general performance for the 6x86 FPU, had Cyrix extended the architecture to a pipelined dual-issue FPU derived from the same FPU logic, the performance probably would have roughly matched the P6 FPU clock for clock (or slightly exceeded it for some applications). Granted, they apparently did just that with the MIII Jalapeno/Cayenne core -along with 3DNow! and 256k L2 cache, but that was obviously much too late. (and, of course, went unreleased by VIA in favor of the WinChip III)

mwdmeyer wrote:

Ha thanks. Yeah I believe I built my first computer (well pulled apart and rebuilt) when I was 6 years old, an old 12MHz 286.

Most of my first computers came from my dad's work when they would throw them out. I loved those days. He would tell me the day before that he would get me a computer from the office. The best time was when he said they were throwing away some 386s (I was running a 386SX 33MHz with 4mb of Ram), The next morning I ran down stairs to see what he had gotten me, it was a 486 DX33 with 8mb Ram, I was so happy!

We did some of that at home when I was growing up too, but I didn't build my firs system until I was 10 (on new year's eve 1999), and I didn't know that much about the in-depth technical side of things until much later. (I knew some basic stuff, and how the case/peripherals/components fit together, but I didn't know more detailed things about the CPU, board, RAM, etc, and I always had help from my dad)

I don't think any of our systems were really like yours, but we did get a fair amount of discarded company parts/machines and my dad did a lot of bargain hunting with used parts, sales and wholesale dealers.

Then again, one oddity was our first shared family multimedia computer from 1993/94 . . . it was pretty powerful inside (fast 486, CD-ROM, decent amount of RAM, good sound -I think a Pro Audio Spectrum, DOS+Win3.x), but it used a mixed of parts (including a late 80s vintage baby AT case) and a 14" grayscale VGA monitor with a crack in the vents on the top/back. (so a pretty nice multimedia/gaming PC using a grayscale monitor 😉 -and we didn't get a color monitor for that machine until late 1995)
That old AT box was used as the shared family PC until 1999 when my dad built a new one (ATX iirc) and we rebuilt the AT box into my first PC. (I believe it was a K6-2 300 on a FIC503A and a Rage Pro PCI card -I forget how much RAM or if it had DVD -I think it was just CD, though the family PC had DVD at that point) That later got upgraded to a K6-2 550 (or 2+) and it sits as such (still working) in the garage at home.

All my first computers were an odd configuration as I could never afford a full machine but I did purchase upgrades from time to time. For example my Pentium 133 had 3 video cards in its life time. A 2mb S3 Trio, 4mb SiS 6326 and then finally a TNT2M64 32mb (I played half life on that machine!).

My machines tended to be made from hand-me-downs from my dad and/or the shared family PC (with a few exceptions where my dad upgraded things with newly obtained -though not always new- parts; and much more recently when I started buying my own parts).

I never did get the Cyrix running with a 75MHz FSB on the VX motherboard. I tried and I believe the system reported as a PR166, it was such a long time ago now!

Running at 133 MHz (PR166) on top of being a VX board definitely would explain the worse-than-P133 gaming performance. (at 150/75, there probably would have been a noticeable improvement, even for FPU-intensive stuff)

The Cyrix + Motherboard was $50 I believe. So not a bad deal. Just wished it was a Pentium 200 😀

$50 including another motherboard (aside from the VX board)?

The PR200 probably wouldn't have been nearly as disappointing on a well-matched board and 75 MHz. (though, given the timing and existing VX board, it probably would have made more sense to look for a used Petium MMX 166 or 200 in the $50 range, or a WinChip 2)

My first new computer was when I purchased a Motherboard + Case + CPU + Ram. It was a AMD Duron 800MHz, Gigabyte GA-7IXE4 (http://ee.gigabyte.com/products/page/mb/ga-7ixe4/) and 64mb Ram. I used the Sound Card, Hard Drive, Video Card and Network Card from the Pentium 133! In fact it was somewhat a step down as my Pentium 133 had something crazy like 90-128mb ram in it towards the end of its life (started with 24mb).

Funny that you mention that. A couple weeks ago, I gave 4 of our old EDO SIMMs to a friend (actually Apolloboy on these forums) to use in his Pentium 133 retro gaming box, hoping to get up to 32 MB (I'd assumed they were 8 MB SIMMs). As it turned out, they were 32 MB SIMMs, so when he booted it up, we were pleasantly surprised to find 128 MB installed. 😀

Had Cyrix extended the architecture to a pipelined dual-issue FPU derived from the same FPU logic, the performance probably would have roughly matched the P6 FPU clock for clock... they apparently did just that with the MIII Jalapeno/Cayenne core -along with 3DNow! and 256k L2 cache... and went unreleased by VIA in favor of the WinChip III.

I wasn't aware that Cyrix was planning on creating a dual FPU pipeline and integrate 256 KB L2 cache; seems I'm not to up on my Cyrix trivia.

The 6x86 FPU is considerably slower at FMUL and FADD than the P5 (4 vs 1 cycle minimum), but noticeably faster for FDIV.

If you look at the PDF for the Ultimate 486 Benchmark Comparison and refer to Appendix 4 (Normalised benchmark data), you can compare the various ALU and FPU operations independently (addition, subtraction, multiplication, and division). This is shown for PassMark tests 75-82. Here you discover that a Cyrix 5x86 is clock-for-clock comparable to a P54C Pentium in Integer and Floating-point division operations, but significantly lagging with addition, subtraction, and multiplication. Interesting is that the AMD X5 is also clock-for-clock comparable to a P54C in integer addition only. At the time of testing, I figured this was just PassMark algorithm-specific and didn't think much more of it.

I now wonder how well the Cyrix 6x86 compares to the Pentium with MMX add/sub/mult/div. I think PassMark also tests for this, but for the 6x86 benchmark comparison, I grouped these all up in a single MMX mark. Perhaps I should record all 4 MMX operations independently.

feipoa wrote:

Had Cyrix extended the architecture to a pipelined dual-issue FPU derived from the same FPU logic, the performance probably would have roughly matched the P6 FPU clock for clock... they apparently did just that with the MIII Jalapeno/Cayenne core -along with 3DNow! and 256k L2 cache... and went unreleased by VIA in favor of the WinChip III.

I wasn't aware that Cyrix was planning on creating a dual FPU pipeline and integrate 256 KB L2 cache; seems I'm not to up on my Cyrix trivia.

I mentioned this a couple times in my first few posts in the thread. (though it's kind of buried in my other ramblings 😉)

The extended 6x86 architecture was being developed for both desktop and embedded (GX style) chips at National Semiconductor from what I understand, but emphasis was on the embedded design (iirc, no L2 cache, but embedded video with 2D/3D acceleration features).

When VIA bought Cyrix, they proceeded with development of the design as the Joshua core Cyrix III in 370 form rather than Super 7. (133 MHz FSB, 22 million transistors, dual pipelined FPU, 3DNow! improved MMX -though still not 100% Pentium compatibility) That reached proproduction on 180 nm in early 2000, but VIA ended up switching to the WinChip4-derived Samuel for the production CIII.

http://en.wikipedia.org/wiki/Cyrix_III
http://www.cpuscorecard.com/cpuprices/vc3.htm
http://www.chipdb.org/img6134.search.htm
http://www.chipdb.org/img6014.search.htm
http://www.chipdb.org/img6015.search.htm

I believe the Cyrix MXi was the embedded (GX-like) counterpart:
http://www.cpu-world.com/CPUs/MXi/index.html

I haven't seen any benchmarks of the Josha Cyrix III unfortunately.

I don't remember everything I read and probably have selective memory for certain generations of hardware. I never really got into systems past a socket 7. To me, a socket 370/slot 1 are still modern systems which should only be run in dual-CPU configuration environments. I have a dual slot1 and dual socket 370 I use here for the "hard stuff." From my readings, I don't think Cyrix s370's ever supported dual CPUs.

Unfortunately, more and more web browsing activities are transitioning into the "hard stuff" category. Up to about 2 weeks ago, I was able to view wiki articles on my Cyrix 5x86 w/IE6. Now the browser hangs after a minute with full CPU usage. Luckily I can still use vogons. I hope they don't upgrade their forum software.

I see there are a few engineering samples of the Joshua's. While it would be nice to add the results of this CPU to the ultimate 686 benchmark comparison, I will probably never own one. The integrated memory/AGP controller of the MXi sounded fascinating. Cyrix definately had innovative ideas.

$50 including another motherboard (aside from the VX board)?

Correct. An MMX would have been faster, but getting another motherboard meant I was able to build my younger brother a computer for nothing 😀

mwdmeyer wrote:

$50 including another motherboard (aside from the VX board)?

Correct. An MMX would have been faster, but getting another motherboard meant I was able to build my younger brother a computer for nothing 😀

So you already had a free case/PSU/monitor sitting around you could use for your brother's machine?

On another note, buying a new motherboard makes for the potential for a much more flexible/comprehensive upgrade than CPU alone. (a different board could have support for newer/different CPUs and/or better support for some CPU models/types -like K6, 6-2, MII; newer RAM options, faster bus, AGP, etc) I'm assuming you'd still have gone with a Socket 7 or Super 7 board though, since those were the lower-cost options for the time. (and should have been common on the used market as well) -Plus, Socket 7 had the advantage of being common in both AT and ATX forms.

feipoa wrote:

I don't remember everything I read and probably have selective memory for certain generations of hardware. I never really got into systems past a socket 7. To me, a socket 370/slot 1 are still modern systems which should only be run in dual-CPU configuration environments. I have a dual slot1 and dual socket 370 I use here for the "hard stuff." From my readings, I don't think Cyrix s370's ever supported dual CPUs.

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)

As to Socket 370/Slot1 vs S7/SS7, they really are about the age if you think about it (and with many similar performance aspects), with the major difference being that Socket 7 was a backwards compatible evolution of an older platform. (and still carried board-level cache -which provided some interesting performance in ways PGA370 lacked -particularly the interesting case of the K6-III's 3 level cache)

Socket 370 got some higher performance CPUs via the late-gen PIII/Celerons (vs the K6-III+ being the fastest SS7 chip), but that's not so much a limitation of the socket/board architecture as marketing/business models. (especially in the case of the 100 MHz FSB PIII/Celerons -though official 133 MHz on SS7 was probably possible too, had chipset/CPU companies pushed for it)

And on that note: both the Cyrix and Centaur based cores Cyrix used in their 370 parts were originally being designed for SS7, but ported to 370 under VIA. (which in itself was a bit questionable, since both parts were aiming at the budget/value end of the market, and SS7 boards catered better to that than 370 -offering parts in both sockets might have made the most sense though)

I see there are a few engineering samples of the Joshua's. While it would be nice to add the results of this CPU to the ultimate 686 benchmark comparison, I will probably never own one. The integrated memory/AGP controller of the MXi sounded fascinating. Cyrix definately had innovative ideas.

Ironic that it was that very innovation that led to the National Semiconductor merger and many of the subsequent management problems. (and some of the financial problems for that matter, since NS hit hard times soon after, mitigating one of the potential biggest advantages of the merger: financial stability)

Prior to the merger, Cyrix certainly seemed to be capable of managing strong R&D for their desktop CPU design while still allowing for expanded products (like the Media GX).

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! 😎

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! 😎

I was looking around, and it looks like the K5 and 6x86 classic supported the OpenPIC multiprocessing standard, but apparently no motherboards ever supported it. The feature seems to have been dropped for the K6 and 6x86MX.

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 had a friend that built one of those.

jaqie wrote:

I had a friend that built one of those. http://puresimplicity.net/~hemi/Pics/Comps/ALR/test-mess-2.jpg http://puresimplicity.net […]
Show full quote

I had a friend that built one of those.

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