Yes they should be the very same, but if you are willing test them both, even cpuchk could show if there is any difference.

Nemo1985 wrote on 2020-02-11, 21:11:

Yes they should be the very same, but if you are willing test them both, even cpuchk could show if there is any difference.

I ran some quick test on both Ti 486 DX2-80 and It's ST 486DX2-80 and compared them against a Intel DX2-66. It's a VLB system with BIOS timings left on default. Both DX2-80s gave the same results.

Hello,

I am ran some dos benchmarks and cache check utilities on my Compaq Prosignia Server.
It has a Pentium Overdrive 83MHz, 112MB RAM and a Compaq EISA Qvision 1024/E graphics card:

Let's start with the xvesa102 tool, it also measures VRAM speed:

╔══ÁVESA video mode speed resultã════════════════════════════════════════════════════════╗
║ ║
║ VESA Version: 1.2 ║
║ OEM name: Compaq QVision 1024 (1 Mbyte) ║
║ Capabilities: 00000001 ║
║ Video memory: 1,024 Kbytes - Detected: 1,024 Kbytes ║
║ Video mode: 0101 - (640x480x8) ║
║ ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Read 8 bit: ████ 1,170 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Read 16 bit: ████████ 2,340 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Read 32 bit: █████████████████ 4,690 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Read 64 bit: ████████████████ 4,570 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Read 80 bit: ████████ 2,390 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Write 8 bit: █████████████ 3,720 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Write 16 bit: ███████████████████████████ 7,440 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Write 32 bit: ███████████████████████████████████████████████████████ 14,800 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Write 64 bit: ███████████████████████████████████████████████████████ 14,800 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
║ Write 80 bit: ██████████████████████████████████████████████ 12,400 Kb/s ║
║ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ║
╚══Á[ESC]=Close window,[F2]=Save result to TXT fileã═════════════════════════════════════╝

OK, so it's a already proven, that this graphics card is awesome under DOS 😀

Text output from the "Speedtst" tool:

The Speed Test, version 1.14 1996-Apr-11
ARA Copyright (C) 1994,95,96 (2:5020/234@FidoNet)
Agababyan Robert Assotiation

Used TMi0SDGL(tm) Version 1.18

Use SPEEDTST /? for help

þ Detected Long mode
þ Detected Pentium iP5 CPU (Intel Pentium-83MHz with Internal FPU)

Pentium iP54-75(1.5-120), Intel Triton 801762
Pentium iP54-75(1.5-120), VIA 82C875/876 801762
Pentium iP54-75(2-120), Intel Triton 794760
Pentium iP54-75(2-120), SiS 501/503 794760
Cx5x86-M1sc-100(3-150,Opt) 755187
Cx5x86-M1sc-100(3-150,Opt) 742857
Pentium iP5-100, ALR Revolution 666667
Pentium iP54-75(100), Intel Triton 664234
Cx5x86-M1sc-100(2-120,Opt) 602649
Cx5x86-M1sc-100(3-120,Opt) 602649
Pentium iP54-75(90), Intel Triton, ASUSTeK P54-TP4 600660
Am5x86-133-X5-P75(4-160), SiS 471, BTC 4SLD5.1 556575
Cx5x86-M1sc-120, ASUS PVI-486SP3 rev. 1.21 529070
Am5x86-133-X5-P75(3-150), SiS 471, BTC 4SLD5.1 521490
Cx5x86-M1sc-100(Opt) 504155
û This computer 495913 parrots, 183.9 XT
Am5x86-133-X5-P75 463104
Am5x86-133-X5-P75, SiS 471, BTC 4SLD5.1 461929
Nx586-90, NxVL System Logic, Alaris 449383
Pentium iP5-60(66), PCI58PL 446078
Pentium iP5-60(66), SiS 501/502/503, ASUS P5-SP 443902
i80486DX2-66(4-100), PC Chips 18 438554
.... skipped.

And then:
Output from the cachemess tool from the german computer magazin ct:

Dhrystone1:

Dhrystone2:

Then switching to "philcomputerlabs" dos benchmarks:
Superscape benchmark 1.0c ("for faster PCs"):

Then "chrisbenchmark". Unfortunately that does not work in 640x480 on that PC (vesa driver was loaded though), so just VGA:

More benchmarks of the same computer:

PC Player Benchmark score 7.2, that ran in 640x480 with the compaq vesa driver.

Doom:

and score:

then quake also chosen 640x480.
Not sure if it was really 640x480, looks a bit strange, but I never played quake honestly spoken.

Quake Score:

16.5 FPS is OK for an EISA computer, isn't it?

More Screenshots:

System Information 8.0

Landmark 6.0

speedsys 478:

and

and finally cachechk

The same benchmarks (some slight differences):
from my second 486 computer: mainboard micronics asic eisa 256kb cache wt, Am5x86-133 with adapter in wt-mode, 64MB RAM fast page mode 30 pin, graphics ATI ultra pro eisa 2MB VRAM, adaptec 2742W, IBM 9GB wide harddisks.

I post everything with pictures, sorry that they are bit blurry, but I think still more entertaining than ascii text 😀

So the first 5 pictures:
1)
First the "cachemess" utility ctcm from Heise Ct. The values are not bad for a mainboard from 1992, most important: The board's L2 cache covers the maximum possible RAM = 64MB:

2)
xvesa102 tool: VRAM speed for Video mode: 0101 - (640x480x8)
Wow, so the high end EISA ATI card seems to be really slower than the Compaq Qvision 1024/E under dos, interesting

3)
speedtst

4)
dhrystone1:

5)
dhrystone2:

The next 5 pictures:
1)
Superscape benchmark (for fast PCs) 😁

2)
Chris' benchmark, on this computer it worked in 640x480 8 bit, so not comparable with the number from the Compaq Prosiginia Server

3)
PC Player Benchmark 640x480

4)
Doom 640x480, result only:

1)
Quake 640x480: 10.9FPS, not too bad for a 486cpu in my opinion

2)
SI 8.0

3)
Landmark 6.0

4)
cachechk
So memory access is a bit slower than the Compaq Prosignia Server, also confirmed by this tool.

5)
sysinfo 478
I am little bit irritated that the harddisk benchmark gave a slower score for this machine (Wide SCSI hdd has slower write speed than the narrow SCSI?), but I won't investigate further.

And finally speedsys 478 cache speed:

More benchmarks for the last 2 machines:
Winstone 95 running Windows 98SE:

Compaq Prosignia Server:

Custom PC with mainboard Micronics ASIC EISA:

I have an M919 with 256k cache Stick

I have try 4x40MHz and 3x50MHz both running without problems at 3.3V
have try multiple benches but for me the doom ist most important, and DX4-150 gets 1825. as DX5-160 i only get 2072 (less iss better)

Would be intresting to know if ther is a FSB60 option on this Bord. I heard about it but dont find closer details.

Hm it really seems like the AMD 5x86 133 was almost strictly a better choice than the POD 83, because almost all chips/boards could do 160 MHz vs. only a small handful of POD 83. So the comparison is effectively POD 83 vs AMD X5-160.

And AMD X5-160 outperforms the POD 83 WB slightly overall, it's ~20% faster in ALU, and ~20% slower in FPU, not a huge delta either way. Plus a 486-based system at the time would have done more ALU stuff. And if you didn't have WB mode... even more an advantage for the AMD which just brute forces with higher clock speed. Not to mention the AMD was more than 3 times cheaper when it launched. It seems like it was an ultra-cheap option for people who didn't want to spend money on a whole system upgrade and just wanted to run Win9x, browse the web at somewhat normal speeds and bridge the gap to a Pentium 3 or K6/K7. The POD meanwhile, unless you had one of the last 486 boards and/or upgraded storage and memory, didn't offer enough improvement -- and if you were gonna spend money, you might as well have spend it on a new platform.

So... AMD with the crazy value option, some things never change... And at a time when computers were very expensive, so value went a long way. I wish I'd known about these options a child, it would have made a huge difference upgrading my 486DX-33.

The AMD Am5x86 wasn't just an upgrade, but some people bought a full fledged Am5x86-133 system with a new system. I did because it was cheap.

Qbcd wrote on 2020-06-09, 03:46:

Hm it really seems like the AMD 5x86 133 was almost strictly a better choice than the POD 83, because almost all chips/boards could do 160 MHz vs. only a small handful of POD 83. So the comparison is effectively POD 83 vs AMD X5-160.

And AMD X5-160 outperforms the POD 83 WB slightly overall, it's ~20% faster in ALU, and ~20% slower in FPU, not a huge delta either way. Plus a 486-based system at the time would have done more ALU stuff. And if you didn't have WB mode... even more an advantage for the AMD which just brute forces with higher clock speed. Not to mention the AMD was more than 3 times cheaper when it launched. It seems like it was an ultra-cheap option for people who didn't want to spend money on a whole system upgrade and just wanted to run Win9x, browse the web at somewhat normal speeds and bridge the gap to a Pentium 3 or K6/K7. The POD meanwhile, unless you had one of the last 486 boards and/or upgraded storage and memory, didn't offer enough improvement -- and if you were gonna spend money, you might as well have spend it on a new platform.

So... AMD with the crazy value option, some things never change... And at a time when computers were very expensive, so value went a long way. I wish I'd known about these options a child, it would have made a huge difference upgrading my 486DX-33.

I was never able to achieve realtime MP3 playback on my AMD 160Mhz 586 based SIS496 setup. The closest I got was with Winplay3 . I even tried DOS-based MP3 players. That was the only thing that was really disappointing to me with that system

Hm I see. I guess benchmarks don't really tell the whole story, in some tasks the Pentium outperformed the 486 several-fold, not just 2x clock for clock.

An Am5x86-160 shouldn't have any problem playing mp3s w/Winamp 2.05. It will help if you use wavout instead of directsound out, also disable the scrolling filename and spectrum analyser. I started a thread on benchmarking 486 chips with Winamp, but it never really caught on: Post your 486 Winamp benchmarks I was using 128 kbps files though.

feipoa wrote on 2020-06-14, 09:33:

An Am5x86-160 shouldn't have any problem playing mp3s w/Winamp 2.05. It will help if you use wavout instead of directsound out, also disable the scrolling filename and spectrum analyser. I started a thread on benchmarking 486 chips with Winamp, but it never really caught on: Post your 486 Winamp benchmarks I was using 128 kbps files though.

When I had one of those back in its heyday (AM5x86 133 overclocked to 160MHz on a SIS496 based board and 16MB of RAM ), I was never able to play MP3 files in realtime . The closest I got was with Winplay3 . I would love to still have that system and try again .

Yeah as cool as the POD83 is or even the Cyrix 5x86-120 the overclocked AMD really is the best compramise, faster than the Cyrix in Quake a bit slower vs POD83 but would do better for most games of the era.

Kind of sucks I really liked the Cyrix chip but AMD @ 160mhz is better.

You need to get a S1R3 Cyrix 5x86-120 chip so that you can have branch prediction enabled in Windows. This should compensate.

I've had the ESA-486 industrial motherboard for a few years now and don't think I've posted Speedsys for it.
It uses an SoC with a Cyrix 486DX4 core.
As you can see, there is no L2 cache. It does have 128MB of onboard PC100 SDRAM.
These hard drive numbers look pretty bad, don't they? Not sure why because I have UltraDMA enabled in the BIOS.