In some cases yes!

Here is what I got for all my slower machines, benched them all but the Soyo P2 300.

1Pentium-120 - S3 VIRGE DX - PCI
23DBENCH = 108fps
3PCPBENCH = 27.2
4DOOM = 52.97
5Quake = 25fps
6
7Pentium 233MMX - RIVA TNT NV4 - PCI
8
93DBENCH = 181.1
10PCPBENCH = 59.1
11DOOM = 103.31
12Quake = 46.5
13	
14Cyrix 6x86MX-PR-200 - Voodoo Banshee - PCI
153DBENCH = 230.2
16PCPBENCH = 47.6
17DOOM = 79.63
18Quake = 28.1
19
20AMD-K6-2-500 - Voodoo 3 2000 - PCI
213DBENCH = 317.2
22PCPBENCH = 120.5
23DOOM = 97.13
24Quake = 70.3
25
26AMD-K6-2+-500 Geforce FX 5500 - PCI
273DBENCH = 226
28PCPBENCH = 141.2
29DOOM = 67.29
30Quake = 76.2
31
32Intel Celeron-333 - Voodoo 3 3000 - AGP
333DBENCH = 509.5
34PCPBENCH = 92.2
35DOOM = 67.23
36Quake = 60.4

Got really odd results with the K6-2+ especially compared to the normal K6.
The Cyrix did really well with 3dbench and doom, and tanked on quake 😵 .

Edit: Just benched my PII 450 for comparison since it also has a PCI 3dfx card:

1Pentium II 450 - Voodoo 3 3000 PCI
23DBENCH = 598.5
3PCPBENCH = 141.3
4DOOM = 100.52
5Quake = 97.6

I wonder why the P-233 MMX getting such a good fps rate in doom?, even the PII-450 did not beat it.
Also I added all these to the database.

Edit2: Adding my Soyo board, fixed the cpu fan on it. Doing alot tonight. My as well bench all my machines 500mhz and slower.

1Pentium II 300 - Riva TNT NV4 - AGP
23DBENCH = 439.5
3PCPBENCH = 96.8
4DOOM = 67.53
5Quake = 65.1

Much like the Celerons scores. I was thinking maybe Doom liked the TNT in the P-233mmx, but this is the same 16mb NV4 GPU, but its AGP. What is with that P-233mmx doom score 🤣 .

Edit3: Just re ran the P-233MMX, and yes it got the same high doom fps score. Does doom just run really fast on the P55c? or is it something to do with the SIS chipset its using?

AlphaWing wrote:

I wonder why the P-233 MMX getting such a good fps rate in doom?, even the PII-450 did not beat it.
Also I added all these to the database.

Did you use FASTVID for PII-450?

Got some more results from the Russian guys I know from a different forum.

Phil, could you please sort the list?

P.S. philscomputerlab, is that you as well?

RacoonRider wrote:

Phil, could you please sort the list?

P.S. philscomputerlab, is that you as well?

List sorted again 😀

Yes philscomputerlab is my new forum name / website / YouTube channel!

RacoonRider wrote:

AlphaWing wrote:

I wonder why the P-233 MMX getting such a good fps rate in doom?, even the PII-450 did not beat it.
Also I added all these to the database.

Did you use FASTVID for PII-450?

Yea I did. That sis board still won.

I just got an S3 Virge GX/2 board and tried it out, and reported my results. To my own surprise, it performed quite similar to my GeForce FX 5500. Some tests showed that it was faster, while others reported the same speed. It also responded to the VGA boost differently, often being slower than my 5500.

This tests is more lemitrd by cpu, chipset and bus speed on PCI systems and more modern.
Video card and his bus is not a bottleneck from about 486 systems.

I've started my test on 486 DX2-66 with bios default.
Tested few videocards and found similar benchmarks.
Then, I've statret bios optimisation, and on absolutely same system got 10 and 20% higher bench: (two bottom lines)

So, cache and ram timings, bus speed - is highly sensivity for dos gaming system.
I'm upload full results later, after finish my tests.

So, cache and ram timings, bus speed - is highly sensitivity for dos gaming system.

This is not really something new and unusual. Tight timings and high bus speed goes hand in hand with higher performance in every way - not just games.

When doing benchmarks it is always good to use the best possible BIOS optimizations. 10-20% gain is not uncommon with good optimizations.

BTW:
Your last and best results are quite fast for 486DX/2-66.

kixs wrote:

BTW:
Your last and best results are quite fast for 486DX/2-66.

I know. Moreover, later I've got 3D Bench:
51,3 - Am486 DX2-66 (33*2)
68,6 - Am486 DX4-100 (33*3) WB L1
81,8 - Am486 DX4-120 (40*3) WB L1
71,4 - i486 DX4-100 (33*3)
85,8 - i486 DX4-100 overclock to 120 (40*3)
so looks like this is one of top configuration, even faster, than average AMD 5x-133

RacoonRider wrote:

Made a long video card benchmark run with my new build-in-progress, AMD 5x86 -133 on Acer AP43 (SIS 496/497). […]
Show full quote

Made a long video card benchmark run with my new build-in-progress, AMD 5x86 -133 on Acer AP43 (SIS 496/497).

Matrox Athena showed the expected poor DOS performance; however I expected better from V2200. S3 videocards show impressive scores yet I was confused when 64V+ turned out to be slower than 64, given that 64 has 60ns EDO ram and 64V+ - 45ns. S3Trio 3D/2X has the best score of all S3s, yet I expected it to slightly outperform Millenium II and ET6000. Well, it didn't, despite SDRAM and being manufactured in 2000.

The top scores are achieved with Matrox Millenium II, S3 Trio 3D/2X, ET6000, Matrox Mystique, but ET6000 did not work in DOS text mode, so Doom score is unknown. We see almost equal scores: at this point, the CPU is no longer bottlenecked by the videocard.

P.S. I'm adding Millenium II score into the list. Are the other scores worth adding? It would create confusion if I added V2200 and Athena scores there; S3 scores are more or less identical; ET6000 is inconsistent.

Rendition cards are well known for having really slow VGA performance (mode 13h or Mode-X/Y) while having good VESA performance. You can use renutil to remap mode 13h to a VESA mode, but this is not compatible with Doom (or other games using Mode-X/Y).
Regarding the difference between 64V+ and 64: Faster RAM does not automatically make a difference, the timings actually used are set by the video cards BIOS. I you are interested you can use MCLK to check the current RAM timings and set faster timings. S3 cards usually have a big margin here, I suppose manufacturers sourced different RAM chips and used conservative timings in the BIOS.

AlphaWing wrote:

RacoonRider wrote:

AlphaWing wrote:

I wonder why the P-233 MMX getting such a good fps rate in doom?, even the PII-450 did not beat it.
Also I added all these to the database.

Did you use FASTVID for PII-450?

Yea I did. That sis board still won.

I thought FASTVID doesn't do anything for Mode-X/Y, but according to http://web.archive.org/web/20020408044840/htt … orials.php?id=4 it can double performance in Doom 2. Does it make a difference in other games?
Classic Pentium (MMX) has a faster bus transfer than PPro/PII when not using FASTVID or similar tools.

Doom 2 is not mode X.

Uploaded few results to Stock and Overclocked pages. Please validate and sort.

PS Why you not summ 3D Bench in to "SUMM of FPS"?
3D Bench is highly sensitivity to ram and cache speed, so it better show any difference in tuning.

easy_john wrote:

PS Why you not summ 3D Bench in to "SUMM of FPS"?
3D Bench is highly sensitivity to ram and cache speed, so it better show any difference in tuning.

3dbench craps out on fast machines.

easy_john wrote:

Uploaded few results to Stock and Overclocked pages. Please validate and sort.

Done!

I can some Cyrix chips and also a IntelDX4 Overdrive. Nice.

@ GeorgeMan

Do you have any more information about your 386 motherboard? Maybe a photo because I couldn't find anything online 😀

@ Elinda

There is an entry with onboard graphics. Do you think it is connected via AGP internally?

The entire sheet should be validated now!

easy_john wrote:

I know. Moreover, later I've got 3D Bench: 51,3 - Am486 DX2-66 (33*2) 68,6 - Am486 DX4-100 (33*3) WB L1 81,8 - Am486 DX4-120 (40 […]
Show full quote

kixs wrote:

BTW:
Your last and best results are quite fast for 486DX/2-66.

I know. Moreover, later I've got 3D Bench:
51,3 - Am486 DX2-66 (33*2)
68,6 - Am486 DX4-100 (33*3) WB L1
81,8 - Am486 DX4-120 (40*3) WB L1
71,4 - i486 DX4-100 (33*3)
85,8 - i486 DX4-100 overclock to 120 (40*3)
so looks like this is one of top configuration, even faster, than average AMD 5x-133

Can you also post a picture of Speedsys and Cachechk with Am486 DX4-100 (33*3) WB L1 and i486 DX4-100 (33*3). What motherboard are you running this on? Thanks!

philscomputerlab wrote:

I can some Cyrix chips and also a IntelDX4 Overdrive. Nice.

IntelDX2/4 Overdrive (DX2odpr66, DX4odpr100) is absolutely same as intel DX2/4, but work on 5v mobo.
I can't find any difference in features or speed between odpr and ordinary DX2/4, so I stop tests 486 Overdrive and use only iDX4-100.
Pentium overdrive (PDPV) is more interesting, but mine is still not arrived.

kixs wrote:

Can you also post a picture of Speedsys and Cachechk with Am486 DX4-100 (33*3) WB L1 and i486 DX4-100 (33*3). What motherboard are you running this on? Thanks!

Amd DX4-100 (33*3) WB
.
i486 DX4-100 (33*3) WT
.
Mobo: A-trend Atc-1415 (UM8886AF/UM8881F), bios: minimal timings, cache: 256-20ns, ram: 32mb simm

Unfortunately I don't have WB intel "&EW" right now for side by side comparison.
But I do test on Amd, and there is not to much increasing on WB cache:
.

And here you can see difference between 128 and 256 kb cache (line 3 and 4)
.
(this tests on different mobo - 4dps)

I can't post cachechk, since I save only speedsys and benchmark results and don't run any other utilities. But I can do it later, if it's important.
PS sorry about watermarks, you can open picture link and there pictures w/o it.

Wow... this was fast 😁

Thanks for the Speedsys. I usually run Cachechk too. I kinda think its shows a little better how fast the memory is then Speedsys.

kixs wrote:

I usually run Cachechk too. I kinda think its shows a little better how fast the memory is then Speedsys.

Done for amd dx4-100 for this time:
cachechk:

1CACHECHK V7 11/23/98  Copyright (c) 1995-98 by Ray Van Tassle. (-h for help)
2 CMOS reports: conv_mem= 640K, ext_mem= 31,744K, Total RAM= 32,384K
3 "AuthenticAMD"  486 Clocked at 100.0 MHz
4 Reading from memory.  
5 MegaByte#:      --------- Memory Access Block sizes (KB)-----
6       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
7 0:   11   11   11   11   11   20   20   20   20   30   --   --   --    us/KB
8 1:   11   11   11   11   11   20   20   20   20   30   30   30   30    us/KB
9 2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18  <--- same as above.
1019 20 21 22 23 24 25 26 27 28 29 30 31  <--- same as above.
11
12 Extra tests----
13 Wrt   16   16   16   16   16   16   16   16   16   16   16   16   16<-Writing
14 This machine seems to have both L1 and L2 cache. [reading]
15    L1 cache is   16KB-- 103.6 MB/s  10.1 ns/byte (279%) (185%)      3.9 clks
16    L2 cache is  256KB--  55.9 MB/s  18.8 ns/byte (150%) (100%)      7.2 clks
17    Main memory speed --  37.1 MB/s  28.2 ns/byte (100%) [reading]  10.8 clks
18    Effective RAM access time (read ) is 112ns (a RAM bank is 4 bytes wide). 
19    Effective RAM access time (write) is  60ns (a RAM bank is 4 bytes wide). 
20    "AuthenticAMD"  486 Clocked at 100.0 MHz.  Cache ENABLED.
21 Options: -t0

ctcm:

1XMM not installed     !
2CTCM uses Memory from address   00120000
3Int15-Memory=32505856
4
5CTCM, translated by Thomas Pabst, copyright ct-Mag. V1.5b/t2
6Processor-Timing   : i486DX,i487DX,i486DX2,am486DX,am486DX2,am486DX4,am5x86
7Processor CPUID    : am486DX4-WB/ Typ:00 Fam:04 Mod:09 Stp:04
8Clck               : 99.8 MHz
9internal Bus       : 32 Bit between CPU and primary Cache or Memory          
10FPU                : i487-Typ
11L1 Cache           : 16 KByte,4-way associative
12L2 Cache           : 256 KByte, direct mapped
13Write Strategy  L1 : Write Back
14Write Strategy  L2 : Write Back
15Dirty Tag L2       : ok        
16
17         Through Put & Bus Performance: Main Memory from  00120000
18
19Best Time for  16K MOVSD (Cache /Page Hits) :   125 mcs - 130.9 MByte/s
20average t. for 16K MOVSD (Miss + Hit)       :   461 mcs -  35.5 MByte/s
21average t. for 16K MOVSD (if  clean)        :   645 mcs -  25.4 MByte/s
22average t. for 16K MOVSD (if  dirty)        :  1012 mcs -  16.2 MByte/s
23Main Memory    16K MOVSD (Cache misses)     :  1077 mcs -  15.2 MByte/s
24
25average with  256 KB L2-Cache /DOS (640K)  :   519 mcs -  31.5 MByte/s
26average with  256 KB L2-Cache /WIN (4M  )  :   633 mcs -  25.9 MByte/s

(Bios options may be not 100% identical, as in previous test, i've test to many boards and cpu, so may be change something)
I'm don't use this utils before. What is more important in this reports?