The benchmarks look brilliant!
Do you run the CPU @ 3 x 40 MHz?
With this chipset (mobo doesn´t matter too much) and an AMD 5x86 4 x 40 MHz you still stay quite a bit below these speeds (about 145/60/45).

Yes, the the CPU is running at 3x40 MHz. I gave it a try with 50 MHz, but didn't have any success. And for sure an AMD 5x86 at 4x40 MHz or Cyrix 5x86 at 3x40 MHz is a hard choice to make.

It'd be even better with a Cyrix 5x86-133 running at 4 x 40 😀 Might break the bank though

Is it Doom stable at 0WS? With a VLB CL-GD5428, it always crashed on me. I have a Trio64 VLB, but I haven't tested yet. What do the advanced settings in your BIOS look like? Do you have timings set to "fastest"?

Thanks

I don't own such a uniquehorn like a Cyrix 5x86 133 😁. But was there ever a Cyrix running stable at 160 MHz?

Doom runs perfectly fine and didn't notice any crash till now. Does not matter if the shareware Version out of Phil's Benchmark or the full versions of Doom and Doom II. I do not own any Cirrus Logic card, so I can't provide you any comparison. However I had some issues when I did run a Tseng4000.

I am using the 'fastest' settings and will provide a picture of the actual settings I am using later.

Here are the bios setting I am using.

Nussbips wrote on 2023-09-13, 14:33:

Hi guys, […]
Show full quote

Hi guys,

I started to follow vogons when I as trying to build a 486 for nostalgic reasons. These sweet memories when everything was easy and the biggest issue in life was how to finish LeChuck with a voodoo doll.

Soon a usual 486 was far to boring, so something more special was the new goal. I admit for me the VLB is the way to go. This lead me to the ASUS 486SV2GX4. PS/2 and no real time clock modul. Just pefect. I managed to install 1024 kb of cache, fixed the bios and got my hands on a Cyrix 120 GP. S3 Trio64 and ARK1000 VLB card. I did run in many issues, but thanks to Feipoa's and others endless efforts on this forum, I could solve most things and want to share the results.

I guess the results are quiet well for a non overclocked system. The ARK1000 really pushes the VGA speed, but a S3 Trio64 is playing in another league in SVGA tho. Both cards are running in transparent mode with 0 wait states. All stable Cyrix enhancements were active during benchmarking.

So I am curious to see, if the 486 topic gets some new life over here 😀.

Nice values.
But a little bit slow.
Both real benchmarks Quake and 3D Bench 1.0c are below the Am5x86 @ 160MHz.

Re: 3 (+3 more) retro battle stations

CoffeeOne wrote on 2023-09-22, 19:41:

Nice values. But a little bit slow. Both real benchmarks Quake and 3D Bench 1.0c are below the Am5x86 @ 160MHz. […]
Show full quote

Nussbips wrote on 2023-09-13, 14:33:

Hi guys, […]
Show full quote

Hi guys,

I started to follow vogons when I as trying to build a 486 for nostalgic reasons. These sweet memories when everything was easy and the biggest issue in life was how to finish LeChuck with a voodoo doll.

Soon a usual 486 was far to boring, so something more special was the new goal. I admit for me the VLB is the way to go. This lead me to the ASUS 486SV2GX4. PS/2 and no real time clock modul. Just pefect. I managed to install 1024 kb of cache, fixed the bios and got my hands on a Cyrix 120 GP. S3 Trio64 and ARK1000 VLB card. I did run in many issues, but thanks to Feipoa's and others endless efforts on this forum, I could solve most things and want to share the results.

I guess the results are quiet well for a non overclocked system. The ARK1000 really pushes the VGA speed, but a S3 Trio64 is playing in another league in SVGA tho. Both cards are running in transparent mode with 0 wait states. All stable Cyrix enhancements were active during benchmarking.

So I am curious to see, if the 486 topic gets some new life over here 😀.

Nice values.
But a little bit slow.
Both real benchmarks Quake and 3D Bench 1.0c are below the Am5x86 @ 160MHz.

Re: 3 (+3 more) retro battle stations

It's not a fair comparison to compare one CPU @120 with another @160.

debs3759 wrote on 2023-09-22, 20:35:

CoffeeOne wrote on 2023-09-22, 19:41:

Nice values. But a little bit slow. Both real benchmarks Quake and 3D Bench 1.0c are below the Am5x86 @ 160MHz. […]
Show full quote

Nussbips wrote on 2023-09-13, 14:33:

Hi guys, […]
Show full quote

Hi guys,

I started to follow vogons when I as trying to build a 486 for nostalgic reasons. These sweet memories when everything was easy and the biggest issue in life was how to finish LeChuck with a voodoo doll.

Soon a usual 486 was far to boring, so something more special was the new goal. I admit for me the VLB is the way to go. This lead me to the ASUS 486SV2GX4. PS/2 and no real time clock modul. Just pefect. I managed to install 1024 kb of cache, fixed the bios and got my hands on a Cyrix 120 GP. S3 Trio64 and ARK1000 VLB card. I did run in many issues, but thanks to Feipoa's and others endless efforts on this forum, I could solve most things and want to share the results.

I guess the results are quiet well for a non overclocked system. The ARK1000 really pushes the VGA speed, but a S3 Trio64 is playing in another league in SVGA tho. Both cards are running in transparent mode with 0 wait states. All stable Cyrix enhancements were active during benchmarking.

So I am curious to see, if the 486 topic gets some new life over here 😀.

Nice values.
But a little bit slow.
Both real benchmarks Quake and 3D Bench 1.0c are below the Am5x86 @ 160MHz.

Re: 3 (+3 more) retro battle stations

It's not a fair comparison to compare one CPU @120 with another @160.

Maybe. But it is interesting. In synthetic benchmarks the Cyrix (with all features enabled) is ahead the Am5x86, even with the lower clock. But in DOS games the Am5x86 is slightly faster.

Impressive values for an AMD @ 160 MHz as I never saw any reaching more fps in Quake than 17.7.

Nussbips wrote on 2023-09-24, 16:14:

Impressive values for an AMD @ 160 MHz as I never saw any reaching more fps in Quake than 17.7.

I got an IBM 5x86-100.

So I made some quick tests with my Asus VL and a S3 868 VLB graphics card, because I wanted to make a comparison between Cx5x86 @ 120 vs. Am5x86 @ 160.
Luckily the IBM chip seems to run fine at 120MHz, I will stability test it under Windows 98SE later on, then it will show, if it is really stable.

The good thing is I can run the IBM(Cyrix) 5x86 with the exact same jumper setting as the Am5x86 (except remove the jumper for x4 multiplier, so that the IBM runs with multiplier 3).
The Bios setting (Asus VL/I-SV2GX4 Rev.2.0) are all fastest except local bus settings T3 and Synchronize. The S3 868 VLB does not run with Transparent unfortunately.
Nevertheless it is a fair comparison, because both the AMD and the IBM run with the exact same settings.

The first result was very bad for the IBM, I expected that.
AMD: Quake low: 18.1fps, Doom fast: 1225 ticks
IBM: Quake low: 15.6fps, Doom fast: 1439 ticks

After comparing your speedsys screenshot with mine:

The CPU value is 56, while your figure shows 68, so I have to play around with the Cyrix enhancements, lets see.

CoffeeOne wrote on 2024-01-14, 13:50:

I got an IBM 5x86-100. […]
Show full quote

Nussbips wrote on 2023-09-24, 16:14:

Impressive values for an AMD @ 160 MHz as I never saw any reaching more fps in Quake than 17.7.

I got an IBM 5x86-100.

The attachment PXL_20240114_123530748.jpg is no longer available

So I made some quick tests with my Asus VL and a S3 868 VLB graphics card, because I wanted to make a comparison between Cx5x86 @ 120 vs. Am5x86 @ 160.
Luckily the IBM chip seems to run fine at 120MHz, I will stability test it under Windows 98SE later on, then it will show, if it is really stable.

The good thing is I can run the IBM(Cyrix) 5x86 with the exact same jumper setting as the Am5x86 (except remove the jumper for x4 multiplier, so that the IBM runs with multiplier 3).
The Bios setting (Asus VL/I-SV2GX4 Rev.2.0) are all fastest except local bus settings T3 and Synchronize. The S3 868 VLB does not run with Transparent unfortunately.
Nevertheless it is a fair comparison, because both the AMD and the IBM run with the exact same settings.

The first result was very bad for the IBM, I expected that.
AMD: Quake low: 18.1fps, Doom fast: 1225 ticks
IBM: Quake low: 15.6fps, Doom fast: 1439 ticks

After comparing your speedsys screenshot with mine:

The attachment cx586-120.jpg is no longer available

The CPU value is 56, while your figure shows 68, so I have to play around with the Cyrix enhancements, lets see.

Sorry to necro this post. I’m also getting 56 using the OP’s jumper and BIOs settings. Did you figure out how to get 68? I’m wondering if the OP used a command line tool (Peter Moss utility?) to get some extra speed.

I’m also only getting 14.8fps in Quake with a ET4000W32/VLB.

zuldan wrote on 2025-06-29, 07:56:

CoffeeOne wrote on 2024-01-14, 13:50:

I got an IBM 5x86-100. […]
Show full quote

Nussbips wrote on 2023-09-24, 16:14:

Impressive values for an AMD @ 160 MHz as I never saw any reaching more fps in Quake than 17.7.

I got an IBM 5x86-100.

The attachment PXL_20240114_123530748.jpg is no longer available

So I made some quick tests with my Asus VL and a S3 868 VLB graphics card, because I wanted to make a comparison between Cx5x86 @ 120 vs. Am5x86 @ 160.
Luckily the IBM chip seems to run fine at 120MHz, I will stability test it under Windows 98SE later on, then it will show, if it is really stable.

The good thing is I can run the IBM(Cyrix) 5x86 with the exact same jumper setting as the Am5x86 (except remove the jumper for x4 multiplier, so that the IBM runs with multiplier 3).
The Bios setting (Asus VL/I-SV2GX4 Rev.2.0) are all fastest except local bus settings T3 and Synchronize. The S3 868 VLB does not run with Transparent unfortunately.
Nevertheless it is a fair comparison, because both the AMD and the IBM run with the exact same settings.

The first result was very bad for the IBM, I expected that.
AMD: Quake low: 18.1fps, Doom fast: 1225 ticks
IBM: Quake low: 15.6fps, Doom fast: 1439 ticks

After comparing your speedsys screenshot with mine:

The attachment cx586-120.jpg is no longer available

The CPU value is 56, while your figure shows 68, so I have to play around with the Cyrix enhancements, lets see.

Sorry to necro this post. I’m also getting 56 using the OP’s jumper and BIOs settings. Did you figure out how to get 68? I’m wondering if the OP used a command line tool (Peter Moss utility?) to get some extra speed.

I’m also only getting 14.8fps in Quake with a ET4000W32/VLB.

Hi.
No, I did not figure it out.
But I lost interest in the CX586-100. While the systen seemed to be stable in DOS (most likely I would have found issues, too with more testing), there was no stability at all when running Windows 98.
The Am5x86 @ 160MHz works fine in Windows 98.
So I did not start playing around with the Cyrix special registers, because I would have needed to run the Cyrix at 100MHz in order to achieve stability, and then the Amd will always be faster, no matter what.

CoffeeOne wrote on 2025-06-29, 19:36:

Hi. No, I did not figure it out. But I lost interest in the CX586-100. While the systen seemed to be stable in DOS (most likely […]
Show full quote

zuldan wrote on 2025-06-29, 07:56:

CoffeeOne wrote on 2024-01-14, 13:50:

I got an IBM 5x86-100. […]
Show full quote

I got an IBM 5x86-100.

The attachment PXL_20240114_123530748.jpg is no longer available

So I made some quick tests with my Asus VL and a S3 868 VLB graphics card, because I wanted to make a comparison between Cx5x86 @ 120 vs. Am5x86 @ 160.
Luckily the IBM chip seems to run fine at 120MHz, I will stability test it under Windows 98SE later on, then it will show, if it is really stable.

The good thing is I can run the IBM(Cyrix) 5x86 with the exact same jumper setting as the Am5x86 (except remove the jumper for x4 multiplier, so that the IBM runs with multiplier 3).
The Bios setting (Asus VL/I-SV2GX4 Rev.2.0) are all fastest except local bus settings T3 and Synchronize. The S3 868 VLB does not run with Transparent unfortunately.
Nevertheless it is a fair comparison, because both the AMD and the IBM run with the exact same settings.

The first result was very bad for the IBM, I expected that.
AMD: Quake low: 18.1fps, Doom fast: 1225 ticks
IBM: Quake low: 15.6fps, Doom fast: 1439 ticks

After comparing your speedsys screenshot with mine:

The attachment cx586-120.jpg is no longer available

The CPU value is 56, while your figure shows 68, so I have to play around with the Cyrix enhancements, lets see.

Sorry to necro this post. I’m also getting 56 using the OP’s jumper and BIOs settings. Did you figure out how to get 68? I’m wondering if the OP used a command line tool (Peter Moss utility?) to get some extra speed.

I’m also only getting 14.8fps in Quake with a ET4000W32/VLB.

Hi.
No, I did not figure it out.
But I lost interest in the CX586-100. While the systen seemed to be stable in DOS (most likely I would have found issues, too with more testing), there was no stability at all when running Windows 98.
The Am5x86 @ 160MHz works fine in Windows 98.
So I did not start playing around with the Cyrix special registers, because I would have needed to run the Cyrix at 100MHz in order to achieve stability, and then the Amd will always be faster, no matter what.

Ok I enabled the Cyrix features and now I'm getting 16.1fps in Quake (v1.06). I have the same revision motherboard (but with only 256kb cache) / same CPU / same BIOS settings (BIOS with dirty bit enabled and confirmed working) and a ET4000W32P VLB but the OP is getting 17.6fps.

15x86.exe /LSSER=off /FP_FAST=on /RSTK_EN=on /LOOP_EN=on /BTB_EN=off

@feipoa any ideas?