VOGONS


First post, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Working on my massive benchmark database / project.

So I would like to use pcpbench

pcpbench/VGAMODE works great. I get results from slow to very fast machines.

However in default mode (mode 100) the scores seem a bit low.

E.g. on a PIII 1 GHz with a TNT2 M64 I get 58.2 This seems quite low doesn't it?

I also found that Superscape VGA Benchmark 1.0c doesn't work well on slow machines. Version 1.0 works as expected.

Reply 1 of 14, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie
Mau1wurf1977 wrote:

E.g. on a PIII 1 GHz with a TNT2 M64 I get 58.2 This seems quite low doesn't it?

Looks like the linear framebuffer is not cached. Try Fastvid (get it from here) or bench within Win9x DOS mode (Windows drivers turn LFB caching on by default).

Reply 2 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Damn I have allready ran a lot of tests...

Can I still "use" these benchmarks? In terms of comparing the numbers? the numbers do go higher with faster machines, but I thought a 1 GHz machine would get well over 100 frames?

Running pure dos, haven't got any windows on this machine.

I do remember UNIVBE from the old days. Do I still need it, even if the game benchmarks just fine?

Or do you recommend I get another video card, like a S3 Trio 64V+?

Thanks

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 3 of 14, by Dominus

User metadata
Rank DOSBox Moderator
Rank
DOSBox Moderator

Perhaps it just doesn't scale well when the cpu is well above their time of development.

Windows 3.1x guide for DOSBox
60 seconds guide to DOSBox

Reply 4 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Possible! I try running this UNIVBE and see what happens

EDIT: Well that didn't work.

It's telling me I have a not supported video chip.

Reply 5 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Ok vidspeed seems to have done something.

It does tell me my chipset isn't supported (It's a 440BX) but I went ahead, said yes to every question and ran the benchmark and the score improved.

There is slight flickering every now and then in the benchmark but apart from that it's smooth.

On a PIII 500 GHz the score improved from 41.2 to 68.3
Hmm still seems low to me...

Reply 6 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

I popped in another card, its a S3 Savage 4 but it gets me the exact same scores...

Which one should I stick with? TNT2 or S3 Savage 4?

TNT2 is passive, the Savage 4 has a little fan.

Reply 7 of 14, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie
Mau1wurf1977 wrote:

On a PIII 500 GHz the score improved from 41.2 to 68.3.
Hmm still seems low to me...

Sounds about right... Here are some comparison scores from my rig (K6-3+ 500 MHz, Voodoo5):

PCPBench 320x200x8: 165.3 FPS
PCPBench 640x400x8 uncached: 53.3 FPS
PCPBench 640x400x8 LFB: 80.8 FPS

Reply 8 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Cool!

The VGA score for this machine was 140.5.

So yea looks like tomorrow I have to rebench a few CPUs, but it shouldn't take me long at all...

Thanks for the help!

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 9 of 14, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

If it turns out that most of the cards still get the same scores, you're probably hitting some memory or bus bottleneck.
After all, VESA framebuffer graphics is very slow compared to proper 2D acceleration.

Reply 10 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Yea looks like that's what's happening...

I am very interested in seeing what the 1 GHz PIII with 133 FSB will score. Also I should have my Super Socket 7 board soon, then I can play with a ton of other chips.

Plan is to make a "Cache slowdown reference list" or something similar where you can go and see what CPU allows you to slow down to what machine... Still nutting out the details though...

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 12 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Hmm 5u3 I now found out that this tool also improves the scores in VGA mode.

3.g. 3DBench2 on a 500MHz PIII goes from ~300 to ~600!

5u3 wrote:

or bench within Win9x DOS mode (Windows drivers turn LFB caching on by default).

Hmm I am using a W98 Boot Disk. Wouldn't that be the same as W98 DOS mode? Or do you mean running cmd.exe from inside windows?

Reply 14 of 14, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

I see! I might just put it into my autoexec.bat when I bench on the Slot 1 machine.

The readme file that comes with fastvid mentiones that it's isn't needed / doesn't work on a Super Socket 7 system. I will also have to make sure I mention this on my "Disable Cache benchmark reference sheet" which I am working on.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel