Run it in a MS-DOS session with a GPU clock logger in the background?

It is probably running at its normal speed. Without the drivers loaded there is nothing handling the power-saving features of the card.

The same goes to processors. They run at their maximum non-turbo speed, unless you have a fancy BIOS (I think my Z99-A has that) that allows you to choose to use the minimum possible multiplier until the OS ACPI takes over.

philscomputerlab wrote:

Run it in a MS-DOS session with a GPU clock logger in the background?

That would be in windows with the drivers loaded, if I understand what you're saying. That's not what I'm trying to determine. I'm trying to figure out what it's clock speeds are while it's -actually in dos-, as in boot to and running in ms-dos 6.22.

alexanrs wrote:

It is probably running at its normal speed. Without the drivers loaded there is nothing handling the power-saving features of the card.

The same goes to processors. They run at their maximum non-turbo speed, unless you have a fancy BIOS (I think my Z99-A has that) that allows you to choose to use the minimum possible multiplier until the OS ACPI takes over.

I've been working with a geforce2 ultra I got for $2 1.5 years ago and flashed a custom bios to it, hoping to try and put it in 3D clocks while in DOS or normally not 3D. I've done this with some other cards (GT-240, GT-640) and it's worked. But I've never worked with cards this old.

I was wondering if anyone has written any way to monitor or even check and report gpu speeds for dos.. no one has heard of such a thing?

I don't think anyone ever wrote such na utility. Your best bet might be some sort of benchmark. Run some benchmark that might be affected by the clockspeeds (you're not using the accelerated functions, so I'm not sure it would have a big impact on anything, really) a few times, then boot into Windows, force the clocks to whatever you want to test and run the benchmarks again on a full screen DOS session, then compare the results.

EDIT: Do clockspeed monitoring utilities work on Windows without the drivers loaded? If so you might be able to monitor it in Windows 9x safe mode.

kithylin wrote:

That would be in windows with the drivers loaded, if I understand what you're saying. That's not what I'm trying to determine. I'm trying to figure out what it's clock speeds are while it's -actually in dos-, as in boot to and running in ms-dos 6.22.

In full-screen DOS session, does the Windows driver even matter?

Windows enables Linear Frame Buffers, so if you don't enable them through DOS tools, your results will be off anyway.

You can try a few full-screen DOS sessions and change the clock speed and see if it actually makes a difference. Because what the VGA Benchmark Database showed is that the graphics card doesn't matter than much in DOS, it's really dependent on the processor and Linear Frame Buffers being enabled.

Can you explain a bit more what this is about?

For DOS performance on fast PCs you shouldn't be using Nvidia cards. They have a performance limit built in. AMD doesn't. This has nothing to do with clock speeds however.

philscomputerlab wrote:

In full-screen DOS session, does the Windows driver even matter? […]
Show full quote

kithylin wrote:

That would be in windows with the drivers loaded, if I understand what you're saying. That's not what I'm trying to determine. I'm trying to figure out what it's clock speeds are while it's -actually in dos-, as in boot to and running in ms-dos 6.22.

In full-screen DOS session, does the Windows driver even matter?

Windows enables Linear Frame Buffers, so if you don't enable them through DOS tools, your results will be off anyway.

You can try a few full-screen DOS sessions and change the clock speed and see if it actually makes a difference. Because what the VGA Benchmark Database showed is that the graphics card doesn't matter than much in DOS, it's really dependent on the processor and Linear Frame Buffers being enabled.

Can you explain a bit more what this is about?

For DOS performance on fast PCs you shouldn't be using Nvidia cards. They have a performance limit built in. AMD doesn't. This has nothing to do with clock speeds however.

Actually (thanks to your benchmark suite you gave us) I was able to find out that in fact the video card does make quite a big difference in "true mode dos", that is booting to ms-dos without loading windows. I have my computer in the other room, a AthlonXP barton core chip @ 1.66 Ghz with a minor overclock to 1.81 ghz. I ran your benchmark suite in true mode dos and tried different video cards. Nothing else what so ever changed in the system other than pulling one card and putting in another. I tried a Kyro II AGP card, and a geforce2 ultra and the kyro returned significantly different results in the dos benchmarks, from some of them being almost -200 FPS less, and the fastest being my geforce2 ultra card. So I was thinking since I now know card performance makes the difference, then to try and flash a custom bios to it so it runs 3D clocks in DOS. Because I think most of those cards run "boot up" or 2D clocks when they're not "in windows with drivers loaded".

Basically I'm just trying to figure out how to get the best dos performance.

I was just trying to see if I could find a way to see if my bios flashing had any effect, to try and discern what speed the card is running at while in dos, see if my bios flashing even did anything. It doesn't seem to effect the results so.. I'm not sure if it worked the way I thought it did.

My results from yesterday's testing:
Geforce 2 Ultra
=============
672.2 FPS - 3DBench
504.3 FPS - PCP Bench
577.0 TICS - Doom
280.0 FPS - QUAKE

Kyro II
========
462.8 - 3DBench
360.9 - PCP Bench
568 TICS - Doom
233.0 FPS - Quake

ET6000 (PCI)
========
601.5 FPS - 3DBench
505.1 - PCP Bench
708 TICS - Doom
264.3 FPS - Quake

#9 S3 (PCI)
========
497.9 FPS - 3DBench
338.0 - PCP Bench
770 TICS - Doom
210.4 FPS - Quake

Also different video cards have a rather drastic effect on VESA modes in certain games. Like for example Descent II run in 1280x1024 in real mode dos. When I tried that with the #9 S3-based PCI card, it was nearly a slideshow and looked to be below 10 FPS. But swapping it for the geforce2 ultra card and I can run it smooth and fast with no visible delay at all.

So... I'm just trying to fiddle with different things and figure out what I can do to make stuff faster in that machine. Also so far.. no other card I've tested from any vendor or format is any faster in the dos benchmarks than my geforce2 ultra in the AGP system. I even have an AMD X1300 here which is a quite a bit slower. I haven't tried the tests in windows but I might today and see what the difference is.

EDIT: Also I can enable LFB in DOS with the test suite, but it literally has no effect what so ever on the scores, they don't change either + or - a single FPS, on that system.

Non Intel processor sometimes behave different with LFB. Special tools are often required to unlock graphics performance. Do you have an Intel system to compare with?

My best benchmark score is on a Core i7 with an AMD graphics card. It got almost 2200 fps in PCPBENCH and almost 1000 fos in Quake 🤣

philscomputerlab wrote:

Non Intel processor sometimes behave different with LFB. Special tools are often required to unlock graphics performance. Do you have an Intel system to compare with?

My best benchmark score is on a Core i7 with an AMD graphics card. It got almost 2200 fps in PCPBENCH and almost 1000 fos in Quake 🤣

The only problem with those systems, is they're not actually usable as dos gaming machines, not fully with sound. I'm concerned with speed and performance in that system over there in the other room I own because I can actually use it.. it has a fully functional ISA slot for MS-DOS SOUND compatibility, and full DOS wavetable support. So I'm trying to figure out the fastest thing possible out of every tiny little aspect for that machine. I've already fine-tuned the CPU speed, ram speed, and ram timings, and now the next part is to focus on video card performance and figure out what I can do there.

And no I don't have an intel machine that's good with DOS that's even close to that machine's performance. I have a Dell here that can do Pentium-III/533, but that's not much point in that thing either when I have this 1.8 ghz amd system.

EDIT: I think I can get a gateway machine's motherboard that could run a 1.1 ghz tulatin coppermine chip.. but it has no ISA slot either so I'm not terrible interested in even trying that thing.

Cool.

You can click on View > List and then sort by processor type.

Here a list of Athlon XP system. Maybe this is helpful...

https://docs.google.com/spreadsheet/lv?key=0A … rowsperpage=250

Not sure if the link opens for you. If not, open the database, go view > list and filter for Athlon XP 😀

But it looks like your system is performing very well!