vetz wrote:

Two more boards added and front post updated. I could not replicate RacoonRider's result on the ASUS P/I-P55T2P4 board, though mine is a rev 3.0 board and his is a 3.1. Main difference is that I have 2x256kb cache, while his has 512kb integrated.

The final difference is 0.25 points, so close enough 😀 Perhaps it's the BIOS settings? There's an interesting thing about this board's BIOS confirmed by tom's hardware: the 60ns EDO RAM setting is faster than any manual DRAM setting.

RacoonRider wrote:

vetz wrote:

Two more boards added and front post updated. I could not replicate RacoonRider's result on the ASUS P/I-P55T2P4 board, though mine is a rev 3.0 board and his is a 3.1. Main difference is that I have 2x256kb cache, while his has 512kb integrated.

The final difference is 0.25 points, so close enough 😀 Perhaps it's the BIOS settings? There's an interesting thing about this board's BIOS confirmed by tom's hardware: the 60ns EDO RAM setting is faster than any manual DRAM setting.

That could be it 😜 I had it set on manual settings with x222/x222 DRAM Read/Write timings. If the 60ns EDO RAM setting is indeed faster than that might be true on the other ASUS 430HX boards as well.

Rebenched the board. The 60ns DRAM settings are not the fastest in the latest BIOS. I got worse results than with manual settings.

Ewww... I tested with 60ns. Perhaps THG was not right after all (or is that right only for some revisions?). I might rebench the board with x222/x222, but it's so deep in the shed that I'm not going to do it any time soon.

And it turns out, V4.51PG is not BIOS version 🙁

Hi, vetz!

How about some more tests 😀 ?

I've got 3 new boards: Intel ALTserver\CS Dual Socket 5 board with 430NX and asynchronous cache and 2 pretty generic ones from Tekram(TX) and Lucky Star(VX) ready to go 😀

RacoonRider wrote:

Hi, vetz!

How about some more tests 😀 ?

I've got 3 new boards: Intel ALTserver\CS Dual Socket 5 board with 430NX and asynchronous cache and 2 pretty generic ones from Tekram(TX) and Lucky Star(VX) ready to go 😀

Go for it 😀

I have a support ticket in on Dropbox to get the results back online, it's not traffic related. If not, I'll find a new solution 😀

vetz wrote:

Two new boards benched: […]
Show full quote

Two new boards benched:

Abit PE-5 (SiS 50X) Socket 5
PC Chips M519 (Opti Viper) Socket 7

The PC Chips board is the worst performing Socket 7 board I've ever seen

Phil: The SP97-XV was never shipped, the seller refunded me 🙁

I can confirm very similar performance from the Opti Viper M. My board is an A-Trend ATC-1545. Using 256Kb cache and and S3 trio64V2/Dx reults are

3d bench 77.8
PCP bench 25
Doom realticks 1535
Quake 22.7

Slow chipset but interesting and I still love the board

s.mouse wrote:

Slow chipset but interesting and I still love the board

I got my hands on one this week too and I can say the same, it's so slow that sometimes even a POD-83 in a UMC888x 486 board could outperform the PCChips Viper + P100 😁 A POD@100 is often (or usually) faster. I'm not sure if I should keep it, it's a very interesting chipset but I'm quite sure I wouldn't use the board when I have the complete Intel chipset series from FX to TX.

Shuttle HOT-591P (MVP3 512k) was a disappointment too. It works and performs better than the old OPTi but does nothing special and lags behind the Intel pack and Aladdin V. I don't need them except for some chipset comparison but this is what's already going on here. Also I may not even have a PCI Matrox, I usually test with Riva 128 which has PCI and AGP versions too.

Hello,
I want to update the HOT-591P bios.
From 591PWIQB to 591PS025. Unfortunately, when trying to update in the awdflash program pops up: "Source File Not Found!"
What is the problem with?

I have uploaded three Socket 5 boards (all are VLB/PCI), and nine Socket 7 boards. All I have tested with G200 PCI and fastest bios settings.

Gona wrote on 2021-01-13, 19:35:

I have uploaded three Socket 5 boards (all are VLB/PCI), and nine Socket 7 boards. All I have tested with G200 PCI and fastest bios settings.

Nice work!

I do have a question regarding the Quake results which seems too good to be true. Which version and screensetting did you use? I believe the settings in Phil's collection is a bit different.

vetz wrote on 2021-01-14, 15:00:

I do have a question regarding the Quake results which seems too good to be true. Which version and screensetting did you use? I believe the settings in Phil's collection is a bit different.

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch menu,
"c) Quake timedemo"
It is v1.06.
"quake.exe +timedemo demo1 +exec mode0.cfg -nosound -nocdaudio"
in the mode0.cfg "vid_mode 0".
In my own benchmarks I'm using v1.08 and "fullscreen" without status bar but here it is important to use same method this is why I have used Phil's pack without any mod.

I've uploaded another Pentium VLB motherboard scores and I also got 21.9 in Quake. Are you guys running 3Dbench and PCP in VGA (not 640x480)? But the Doom is run in high settings, right?

It's actually quite a bit better with VLB — I'm going to do a detailed post soon.

Gona wrote on 2021-01-14, 18:55:

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch m […]
Show full quote

vetz wrote on 2021-01-14, 15:00:

I do have a question regarding the Quake results which seems too good to be true. Which version and screensetting did you use? I believe the settings in Phil's collection is a bit different.

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch menu,
"c) Quake timedemo"
It is v1.06.
"quake.exe +timedemo demo1 +exec mode0.cfg -nosound -nocdaudio"
in the mode0.cfg "vid_mode 0".
I my own benchmarks I'm using v1.08 and "fullscreen" without status bar but here it is important to use same method this is why I have used Phil's pack without any mod.

From my experience, Phil's settings differ from the default v1.06 settings and yield different results. Also, Quake v1.08 with default settings tends to yield a higher benchmark value compared to default v1.06. This is thus 3 different results you can receive for what users may classify as "default Quake". As such, I have always just used v1.06 (demo) at default without sound.

jheronimus wrote on 2021-01-14, 20:25:

I've uploaded another Pentium VLB motherboard scores and I also got 21.9 in Quake. Are you guys running 3Dbench and PCP in VGA (not 640x480)? But the Doom is run in high settings, right?

It's actually quite a bit better with VLB — I'm going to do a detailed post soon.

Yes, PCP not in 640x480 and Doom in high settings. OPTi Pentium VLB motherboards perform better with VLB videocards (because the PCI bridged from the VLB), and with a fast VLB videocard (like ARK1000) OPTi beat the VIA Pentium VLB (but still behind the Pentium PCI motherboards).

Gona wrote on 2021-01-14, 18:55:

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch m […]
Show full quote

vetz wrote on 2021-01-14, 15:00:

I do have a question regarding the Quake results which seems too good to be true. Which version and screensetting did you use? I believe the settings in Phil's collection is a bit different.

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch menu,
"c) Quake timedemo"
It is v1.06.
"quake.exe +timedemo demo1 +exec mode0.cfg -nosound -nocdaudio"
in the mode0.cfg "vid_mode 0".
In my own benchmarks I'm using v1.08 and "fullscreen" without status bar but here it is important to use same method this is why I have used Phil's pack without any mod.

Settings might have changed in his benchmark suite from 2014 to 2017 (if it has it must been by mistake). I still have the harddrive where I ran all the tests from, so I can go and check the settings to confirm.

vetz wrote on 2021-01-15, 09:54:

Gona wrote on 2021-01-14, 18:55:

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch m […]
Show full quote

vetz wrote on 2021-01-14, 15:00:

I do have a question regarding the Quake results which seems too good to be true. Which version and screensetting did you use? I believe the settings in Phil's collection is a bit different.

I have downloaded and used Phil's dosbench_v_1.4_jan_2017.zip (the latest now) without any touch, and I have used Phil's batch menu,
"c) Quake timedemo"
It is v1.06.
"quake.exe +timedemo demo1 +exec mode0.cfg -nosound -nocdaudio"
in the mode0.cfg "vid_mode 0".
In my own benchmarks I'm using v1.08 and "fullscreen" without status bar but here it is important to use same method this is why I have used Phil's pack without any mod.

Settings might have changed in his benchmark suite from 2014 to 2017 (if it has it must been by mistake). I still have the harddrive where I ran all the tests from, so I can go and check the settings to confirm.

Take a "compare by content" with the 2014 and the 2017 config files.

I own a matrox g450 PCI, is it equivalent for the benchmark?

CachoAlpuy wrote on 2022-11-03, 13:07:

I own a matrox g450 PCI, is it equivalent for the benchmark?

I'm going to make a Matrox PCI VGA benchmarks on P100 and my Soltek SL-54A5-s motherboard and we will see differences. This process will take a few days.

CachoAlpuy wrote on 2022-11-03, 13:07:

I own a matrox g450 PCI, is it equivalent for the benchmark?

Matrox Millennium, Mystique, Mystique 220, Millennium II and G200 PCI cards are produce very similar benchmark results. G450 PCI is a littlebit faster but this is still a less then 1% difference, so I say Matrox G450 PCI results are also acceptable. Note that G450 PCI will not boot in all motherboards (as I remember none of my 486 boards can boot with the G450 PCI).

Cards before Matrox Millennium are designed for Windows only and has really weak DOS performance. The G550 LowProfile PCI is a really different desing as G550 AGP (or G450 PCI) and has also a really weak DOS performance.