Do you have any benchmarks for dual Pentium pro CPU's ? Am running dual 200 at the moment and thinking of getting two Overdrives and was curious of the performance difference. Thanks

I benchmarked a tonne of 3D games on my PPRO 233-512K build here: Graphics card for Pentium Pro build

If I'm not mistaken, this is the thread that I then replaced the PPRO cpu with a PIIOD and the performance gain was extremely disappointing. This is in stark contrast to when user luckybob ran PIIOD's on his board when providing data for 686 benchmark comparison. I haven't yet had time to determine why the gain was negligible during my tests, but as I now have a dual PIIOD setup with an Intel PR440FX, it is on my list to reinvestigate.

I think I mentioned it earlier. But the motherboard i was using had USWC enabled. ( by default, btw) it made a 50% performance difference in doom/quake.

Also, none of these benchmarks can even see the 2nd cpu. You'd have to be running some flavor of NT/2000/Linux AND a program that was multi threaded. When it comes to gaming on old systems, multi cpu is just for dick waving.

Also, there are major differences between chipsets and manufacturers. My ASUS P65UP5/8 motherboard FLYS with overdrive chips. But my Tyan boards are as buggy as a discount hooker.

Thanks for the info, I had skimmed through this post looking and read a bit on what you had posted.

Will explore options as I am still deciding what price I want to pay for the pair and I noticed someone else here had issues with the W6-Li board and the OD CPUs

luckybob wrote on 2021-11-29, 00:49:

I think I mentioned it earlier. But the motherboard i was using had USWC enabled. ( by default, btw) it made a 50% performance difference in doom/quake.

You may have, but I don't recall specifically. That is an interesting observation. I thought USWC was an AGP feature? Were there some socket 8 boards which have this feature in the BIOS, or were you using a slot 1 AGP board with those downgrade adapters (slot1-to-socket8)?

luckybob wrote on 2021-11-29, 00:49:

Also, none of these benchmarks can even see the 2nd cpu. You'd have to be running some flavor of NT/2000/Linux AND a program that was multi threaded. When it comes to gaming on old systems, multi cpu is just for dick waving.

Indeed, but perhaps the PR440FX board's BIOS has some fix of sorts that speeds things up with a PIIOD compared to the VS440FX board I used in the above noted thread.

luckybob wrote on 2021-11-29, 00:49:

Also, there are major differences between chipsets and manufacturers. My ASUS P65UP5/8 motherboard FLYS with overdrive chips. But my Tyan boards are as buggy as a discount hooker.

I recall you had quite a bit of issues with those Tyans. Glad I switched gear and went with something else.

feipoa wrote on 2021-11-29, 01:18:

You may have, but I don't recall specifically. That is an interesting observation. I thought USWC was an AGP feature? Were there some socket 8 boards which have this feature in the BIOS, or were you using a slot 1 AGP board with those downgrade adapters (slot1-to-socket8)?

I'm not privy to the details of USWC, but I know the G250 Matrox loved it. And I used the P65UP5 for all of my original tests.

Indeed, but perhaps the PR440FX board's BIOS has some fix of sorts that speeds things up with a PIIOD compared to the VS440FX board I used in the above noted thread.

i recently got a 450GX board, overpaid quite a bit, but it had excellent memory throughput, but it has a crappy solder joint somewhere and doesn't post every time. I seriously have 100+ half-finished projects.

I recall you had quite a bit of issues with those Tyans. Glad I switched gear and went with something else.

it's not the fault of Tyan. The boards were fine new (i presume). They just didn't age well.

I think AMD K7 (Athlon) is a 786 class CPU.

I always thought that TYAN would be a good brand, but then I got hold of 3 used dual boards (one dual socket 370 and two dual socket 462)
Initially they worked good, but then they started to get flaky.
As I am not an electronics specialist, I gave it to one of our specialists in the DOSReloaded forum and he confirmed my suspicion: at least 2 of them used really cheap and bad caps.
After replacing them I got the Tiger 200T back and it is running stable as a rock. The two kings even compiled a current Gentoo linux in 3 days straight (ok, GUI/WM not yet installed).

So in principle they seem to be good boards, but not every component is up to their quality claims.

Skorbin wrote on 2024-04-19, 12:59:

...at least 2 of them used really cheap and bad caps.

So in principle they seem to be good boards, but not every component is up to their quality claims.

I don't think anyone that made these boards expected them to be still used today so while the components were top notch at the time, they were not made to last forever. This was par for the course back then (and still is to a certain extent). Luckily, today we do have access to long-life components, but in the era of planned obsolescence and 'price is the only metric', rarely are these type of components used unless it's a selling point.

Samir wrote on 2024-04-19, 18:00:

Skorbin wrote on 2024-04-19, 12:59:

...at least 2 of them used really cheap and bad caps.

So in principle they seem to be good boards, but not every component is up to their quality claims.

I don't think anyone that made these boards expected them to be still used today so while the components were top notch at the time, they were not made to last forever. This was par for the course back then (and still is to a certain extent). Luckily, today we do have access to long-life components, but in the era of planned obsolescence and 'price is the only metric', rarely are these type of components used unless it's a selling point.

Well only, other option is emulation, or some FCPGA boards.. and as im looking at 86Box code commits (scope is huge and there is a lot of exception, special situation to handle) is this a long way to make emulation to be less buggy than original HW..

ruthan wrote on 2024-11-16, 20:40:

Samir wrote on 2024-04-19, 18:00:

Skorbin wrote on 2024-04-19, 12:59:

...at least 2 of them used really cheap and bad caps.

So in principle they seem to be good boards, but not every component is up to their quality claims.

I don't think anyone that made these boards expected them to be still used today so while the components were top notch at the time, they were not made to last forever. This was par for the course back then (and still is to a certain extent). Luckily, today we do have access to long-life components, but in the era of planned obsolescence and 'price is the only metric', rarely are these type of components used unless it's a selling point.

Well only, other option is emulation, or some FCPGA boards.. and as im looking at 86Box code commits (scope is huge and there is a lot of exception, special situation to handle) is this a long way to make emulation to be less buggy than original HW..

Yep, emulation has come a long way and there are some devices that can even run the code better than the original hardware. Still for me, there's something to actually touching and feeling the hardware of the era while using the software from that era.

Cyrix 6x86mx, M II are the last "true" x86 processor. Since Pentium, K6 are really RISC core inside.......

If somebody is interested in some benchmark results of an OC-ed Cyrix 6x86L-PR200+ at 210 MHz (it is a M1-CPU, not M II using MX), they can be found here:
Cyrix appreciation thread

gonzo wrote on 2025-02-01, 09:47:

If somebody is interested in some benchmark results of an OC-ed Cyrix 6x86L-PR200+ at 210 MHz (it is a M1-CPU, not M II using MX), they can be found here:
Cyrix appreciation thread

I'm impressed you got it to 210MHz! Saying that, the 6x86L (non MX) is an awful CPU unless you are doing DOS or Windows Business Apps. It's FPU is like 2/3rds of a Pentium based on MHz. MX version seems to fix a few bugs too.

mwdmeyer wrote on 2025-02-05, 01:45:

gonzo wrote on 2025-02-01, 09:47:

If somebody is interested in some benchmark results of an OC-ed Cyrix 6x86L-PR200+ at 210 MHz (it is a M1-CPU, not M II using MX), they can be found here:
Cyrix appreciation thread

I'm impressed you got it to 210MHz! Saying that, the 6x86L (non MX) is an awful CPU unless you are doing DOS or Windows Business Apps. It's FPU is like 2/3rds of a Pentium based on MHz. MX version seems to fix a few bugs too.

Thanks!
The biggest advantage of the MX-version of the Cyrix-6x86 is the built-in recognition as a "pentium-class-CPU" for the most software - thit increases the useable software with it by far (compared to the 6x86L).
And, of course, the MMX-extention-set.