Interesting, I suspect it'll be slightly faster than the am5x86 133mhz

It might match up well to the Am5x86 at 160. The Cx chip has a better FPU too and this might even get close to the POD83 in that way unlike a Am @ 160.

swaaye wrote:

It might match up well to the Am5x86 at 160. The Cx chip has a better FPU too and this might even get close to the POD83 in that way unlike a Am @ 160.

At stock clocks it might be better, but how much overclocking potential does it have? I would guess it's probably near it's stability limits at 133mhz or else they wouldn't have had to bump the voltage part way through the production run. The AMD or POD might be better after overclocking is taken into consideration. Still, since these appear so infrequently, it would be nice to see some numbers especially on a UMC motherboard. The fact that it has 'cooling fan required' printed on the top tells me that it is going to be HOT even at 133mhz which is not promising for overclocking.

Yeah I'm not sure I'd expect the 133 to overclock much if at all. The Am 133s usually will do 160 on the other hand and that's probably what it'll take to compete with this Cx 133.

My POD will do 100 on some boards but I'm not sure how common that is. The POD is a picky thing though and doesn't perform well in very many boards due to hardware issues.

I've just finished a series of testing with the Cyrix 5x86-133 and compiled results with 9 different benchmark programs. These include,
Norton Sysinfo v8.0, PC-Config v9.33, PcpBench v1.97, Speedsys v4.78, Cachechk v4.0, Roy Longbottom's Dhrystone, Linpack, and Whetstone.

I also cross-compared the results to 13 other socket 3 cpus. The purpose of this is to show a performance comparison between cpus, not necessarily the fastest possible configuration. I used a PC Chips M919 for testing, however the Biostar MB-8433UUD v3.0 generally showed slightly improved performance over the M919. Also, switching to a VLB graphics card tends to increase 3D performance from that of the PCI ones. For example, pcpbench results increase from 8.2 to 9.2 with a Diamond Stealth 64 VRAM (4MB).

The cpus compared were,
Cyrix 5x86-133, 120, 100, and 100 with branch prediction on
AMD x5-160, 133, and AMD DX4-100
Intel Pentium Overdrive P24T-100 and 83.
Intel DX4-100, DX-33, SX2-50.

Unfortunately, the DX2-66 I had in the collection no longer boots.

Refer to the attached PDF for the comparison chart.

wow, thats really nicely setup little pdf there!

I really like how @133 the cyrix is still whipping the amd@160 like a red headed stepchild in a Kentucky walmart. I wish I had one. 🙁

awesome! as I suspected the cx5x86 would beat the other 5x86 chips clock by clock 😀
Now where do I sign up to get one? 😉

Now that's significantly faster than the equivalent AMD chip. Impressive!

If you examine the values of a pair of related columns, eg Cyrix 133 & 120, or Cyrix 120 & 100, or AMD 160 & 133, some of these related values seem a little bit unexpected.

rg100, are you refering to the counter-intuitive trend with VESA and System Memory bandwidths? The higher system memory bandwidth reported by Speedsys for C120, A160, P100 units are because those cpus are the only ones with a 40 Mhz front-side bus, so it makes sense that their memory bandwidth would be higher than their 33 Mhz counterparts, even at a lower external clock rate (where C=Cyrix, A=AMD, P=Pentium).

For the same pairs of cpus, the lower VESA memory bandwidth for the 40 Mhz FSB units is somewhat confusing. I think it is related to the motherboard automatically adding a HOST-to-PCI clock divisor for PCI slots when the FSB is set to 40 Mhz.

My favourite dummy benchmark row is the PC-Config, which relates the tested cpu to that of a typical Intel Pentium 100. I thought it quite agreeable that the AMD X5-133 came in at a Pentium 76, right where it should be, and likewise for the AMD X5-160 at a P90. The Cyrix 5x86-120 clocked in at about a fictitious Pentium 103, where the Cyrix 5x86-133 was a Pentium 110. PC-Config noted that the results of the Cyrix 5x86-133 were equivalent to a Cyrix 6x86-120.

I still have a boat load of DOS benchmark programs that I haven't had time to test which I might add to the next revision of this comparison sheet, that is if they are any good (including Quake1). I am also considering adding an AMD X5-200 column, a Cyrix 5x86-133 w/Enhancements OFF column, and a Cyrix 5x86-133 w/Biostar MB-8433UUD v3.0 motherboard column. I'd also like to establish stability of the Cyrix 5x86-133 and POD100 in Windows. Let me know if there is any other test which might add to the conclusivity of these results.

feipoa wrote:

I was able to obtain a Cyrix 5x86-133, 3.7V

Ah good, now I can rest in peace 😜

I wonder if different versions of the Cyrix 5x86 have actual different performance per Mhz (meaning they are really different chips, instead of coming from a single product line with no changes except for scaling)?

Thanks a lot for the explanation. I think your Biostar rev 3.0 mobo has some useful BIOS options for the Host to PCI clock divisor, including 1:1. I had no problems with it set to 1:1 @ 50 MHz FSB running Windows 98, and so I think it would be a better board for testing, because you'll know that this divisor value remains constant throughout all of your tests.

Also, you're more likely to get the "X5-200" working on the Biostar than on the PC Chips board. Quake1 is a good test. I would be interested to see if your overclocked POD @ 100 MHz and X5-200 can handle its timedemo test.

For a Windows stability test, recently I ran these 3 apps at the same time on a 486 mobo I was testing: SuperPi.exe, Prime95.exe, Sandra 2002 Pro (memory or cache tests), and then I ran 3DMark 99 Max while those other 3 tests were in the background. I picked those tests at random. You might like to pick others, but running a bunch of tests at the same time seemed like a good stress test. There's also WinTune 97 and 98 you could try...

BTW, good work and explanation! 😀 PS - do you know if there are any more Cyrix 133 CPUs floating about waiting for a buyer? 😉

Ah good, now I can rest in peace

I wonder if different versions of the Cyrix 5x86 have actual different performance per Mhz (meaning they are really different chips, instead of coming from a single product line with no changes except for scaling)?

@Tetrium
On the contrary -- this is where you come in! In time, you will have an early 1996 model Cyrix 5x86-133 that will require testing, particularly if I cannot get mine working in Windows. I've been having trouble getting anyone from that 24-unit eBay auction to test their chips. In the least, I'd settle down if I could get stepping/revison information. Cpushack had, at one point, offered to let me test out his varieties, but he's been too busy to send anything and I am doubtful that he will.

@rg100
Unfortunately, the Biostar MB is in regular use and as a server. I'm okay for doing quick no-risk tests, and I might make an exception for the Cyrix 5x86-133. As for the AMD X5-200, I'll have to get it working on the PC Chips M919 some how. I've been waiting for another v3.0 or 3.1 Biostar to show up on eBay, but I haven't seen one. Know of any secret stashes?

Those are probably good starter tests for Windows-based benchmarks. Sandra2002, Passmark v4, 3DMark 99Max, SuperPi, Prime. The last 3 I haven't been able to find online though. I think I have an old shareware version of the WinTunes.

Extra Cyrix 5x86-133's!? I'm not sure, I'll ask my source again. He only had the one, and I wanted more. On the more pricey side of things, you can start throwing numbers in the range of $2000 USD in cpu-world to see if someone bites. There's 24 happy new owneres of these processors out there as of recently. It would be interesting to know how many of these are really left in the world.

My ski weekend starts now, so I'll be holding off on any extra tests until returning.

Here yer go 😀

Link to SuperPi
Link to Prime95
Link to 3DMark 99 Max

You're going to need some good fortune with the X5-200 on the PC Chips board. I hope it works, but IMHO it's not going to be stable. Secret stashes of 3.x Biostar mobos? If only I knew. In fact, it's only recently that I realised it was a good board. I just dug it out of my retro junk pile, and decided on a whim to overclock its brains out.

Please note that SuperPi crashes when you run it for the first time. Actually, the version above is for w9x. I think Tetrium will know where to get the NT based version! 😀

Tetrium, how much did you pay for your Cyrix 133 CPU? Pity I missed that opportunity! So, the seller had a huge pile of them, and decided to sell them all in one sale? Strange!

feipoa wrote:

@Tetrium On the contrary -- this is where you come in! In time, you will have an early 1996 model Cyrix 5x86-133 that will requ […]
Show full quote

Ah good, now I can rest in peace

I wonder if different versions of the Cyrix 5x86 have actual different performance per Mhz (meaning they are really different chips, instead of coming from a single product line with no changes except for scaling)?

@Tetrium
On the contrary -- this is where you come in! In time, you will have an early 1996 model Cyrix 5x86-133 that will require testing, particularly if I cannot get mine working in Windows. I've been having trouble getting anyone from that 24-unit eBay auction to test their chips. In the least, I'd settle down if I could get stepping/revison information. Cpushack had, at one point, offered to let me test out his varieties, but he's been too busy to send anything and I am doubtful that he will.

@rg100
Unfortunately, the Biostar MB is in regular use and as a server. I'm okay for doing quick no-risk tests, and I might make an exception for the Cyrix 5x86-133. As for the AMD X5-200, I'll have to get it working on the PC Chips M919 some how. I've been waiting for another v3.0 or 3.1 Biostar to show up on eBay, but I haven't seen one. Know of any secret stashes?

Those are probably good starter tests for Windows-based benchmarks. Sandra2002, Passmark v4, 3DMark 99Max, SuperPi, Prime. The last 3 I haven't been able to find online though. I think I have an old shareware version of the WinTunes.

Extra Cyrix 5x86-133's!? I'm not sure, I'll ask my source again. He only had the one, and I wanted more. On the more pricey side of things, you can start throwing numbers in the range of $2000 USD in cpu-world to see if someone bites. There's 24 happy new owneres of these processors out there as of recently. It would be interesting to know how many of these are really left in the world.

My ski weekend starts now, so I'll be holding off on any extra tests until returning.

$2000 is WAY overboard, no need to overreact.
I really don't know the little details about group buys. I only check from time to time if theres something interesting for me there. The Cyrix 133 was just one of the group buys I was interested in. Got a couple K6-2+'s from there also (I got plenty of K6-3+'s already and wanted some of the 2's just so I have them available).

If you want further opportunities then I'd suggest what I've suggested you weeks ago, theres nothing more I can add about it. It's not like I never missed a group buy of interesting chips before(some Cyrix M2's went for mighty cheap), it's just life.

And if your source has a couple more for you, then good luck!

I know the chip is interesting, it's just a shame they never were available in similar quantities to the AMD chip.

Really? Well it was a stab in the dark. How much do you think people would sell their Cyrix 5x86-133's for?

K6-III+ 450 is a great chip! I ran mine in a Super7 w/AGP for sometime (FIC 503A motherboard). I think I had overclocked it to 500 or 550 without issue. I was also able to use the Cyrix MII 400GP in the 503A.

@rg100
Do you happen to have a finalised step-by-step guide for how you ran the AMD-X5 ADZ or ADW at 200 Mhz for the AMI and AWARD BIOS's? If so, could you send or copy/paste it?

Step 1 could be use an ADW, not an ADZ. 😀 I only made some basic summary comments on Vogons, like a checklist of hardware to try -

* Good PSU, I use an ATX with an ATX-to-AT adapter.
* Try a CF rather than HDD.
* Use a CF reader device mounted on to the IDE pins, not one on the end of a cable.
* Try an IO card, rather than the mobo's integrated IDE.
* Experiment with longer IO recovery times in the BIOS, if that option exists.
* Copper cooler with paste. For "mobo on a test bench" testing purposes, put something heavy on it, to weigh it down on top of the CPU.
* If the mobo doesn't have a 5V jumper setting, cut pin S4 on the CPU.
* Have a bunch of PCI VGAs handy. Good ones to try are Diamond Viper, some of the Virge 325s. Dell Nitro won't work @ 50 FSB. Good test to see which ones are likely to cope @ 200 is to set the AMD-X5's multi to 3, and the FSB to 50.

I think that's everything. Let me know how you get on! I want to see other people succeed with the "486 200 MHz" experiment. You've got a good board (Biostar). It would be great to see it stable in Windows. I understand you use NT. The challenge would be to get that stable, because DOS is easier. BTW, I hear what you say about it being used for a server. However, I think you're going to struggle getting your other PC Chips board running at 200 MHz.

retro games 100 wrote:

I want to see other people succeed with the "486 200 MHz" experiment.

It's just one of many things that are on my "to do" list 😵
I can't really pick 🤣!
I guess I'll continue posting pics of my retro rigs for the time being.
Also, I still have a lot of tidying to do in the attic. Got a stack of about 1 meter high of motherboards, ram, graphics cards and trays of CPU's all waiting to be tested.

Oh, and a pile...or should I say, mountain, of packaging material to sort out. I want to keep the handy stuff for several reasons.

I don't know how much the Cyrix 133 would sell for, but consider this:
The Cyrix 133 was said to be considered mythical by another Vogons user I can't remember the name of. But it's quite possible that there are still more stocks of these chips waiting to be found. It seems to always go that way.
The Cyrix 133 seems a bit too rare for how rare it really should be. Consider feipoa wanted a Cyrix 133 and couldn't get one of the known owners to sell him one. Low and behold, quickly after feipoa found another source and now has this chip also!

Personally I wouldn't want to spend too much on really rare chips, and some chips I really want are still missing in my collection also (Cyrix MII PR433, AMD Thunderbird 1400B...I'd even take a 1300B with both hands 😜).
I know I have a lot, but I barely use what I already have. It's one of the reasons I haven't been buying like MAD like I did last year 🤣 😜

I can actually say that most of what I'd want, I already have 😁
But ofcourse theres always more/new stuff appearing I'd want hehe!! 😜