Could try Quake 2 (not finished yet)
http://virtuallyfun.superglobalmegacorp.com/2 … -to-ms-dos-pt3/

DosFreak wrote:

Could try Quake 2 (not finished yet)
http://virtuallyfun.superglobalmegacorp.com/2 … -to-ms-dos-pt3/

The creator of the port says it runs about Q1 speed in DOS.

kithylin wrote:

The only game it does poorly in is Quake I though, I've always thought it was just because it's an AMD system and Quake I was optimized with pentium instructions and thus leaned towards Intel CPU's only.

Not Pentium instructions, but FPU-optimized for Pentium. This hurt AMD K5/K6 and Cyrix CPUs, but PPro/P2/P3 are fine, Athlon should be faster clock-by-clock than P3. Your GF2 should have VESA 2.0, and I suppose the MTRRs are already configured correctly since you get good performance with Descent II.
Barton should have 512kB L2 cache + the L1 caches, so better than any P3.

Could you compare your values against http://www.vgamuseum.info/index.php/benchmarks ?
That also uses a KT133A mainboard, and an Athlon XP 2200+ so not too different from yours. (This scores quite a bit higher than Maulwurf's/Phils P3 at least when using fastvid - 104fps vs. 60fps at 640x480.)

idspispopd wrote:

Not Pentium instructions, but FPU-optimized for Pentium. This hurt AMD K5/K6 and Cyrix CPUs, but PPro/P2/P3 are fine, Athlon sho […]
Show full quote

kithylin wrote:

The only game it does poorly in is Quake I though, I've always thought it was just because it's an AMD system and Quake I was optimized with pentium instructions and thus leaned towards Intel CPU's only.

Not Pentium instructions, but FPU-optimized for Pentium. This hurt AMD K5/K6 and Cyrix CPUs, but PPro/P2/P3 are fine, Athlon should be faster clock-by-clock than P3. Your GF2 should have VESA 2.0, and I suppose the MTRRs are already configured correctly since you get good performance with Descent II.
Barton should have 512kB L2 cache + the L1 caches, so better than any P3.

Could you compare your values against http://www.vgamuseum.info/index.php/benchmarks ?
That also uses a KT133A mainboard, and an Athlon XP 2200+ so not too different from yours. (This scores quite a bit higher than Maulwurf's/Phils P3 at least when using fastvid - 104fps vs. 60fps at 640x480.)

I can give a try tomorrow some time I suppose. With certain systems, the MTRR's and FastVid have no effect what so ever on AthlonXP core chips. In all the testing I've done on my AMD "super-dos" system, I can not see any noticeable difference with or without fastvid/MTRR-enabler, it's the exact same FPS either way.

however it should be noted I'm using the mobile AthlonXP-M chip in there, not the desktop version, so it only runs at 266 Mhz Bus, I think 300 with the small overclock I can get out of it.

It's this chip: http://www.cpu-world.com/CPUs/K7/AMD-Mobile%2 … H2000FQQ3C.html with only 256 KB cache, and after some checking I've found it's Throughbred core not barton. Sorry.. I mis-spoke on that one. It's the fastest I can get to work in that KT-133A motherboard (the A version of the chipset that is) And even then it comes in as "AMD Unknown" during POST, the bios doesn't officially recognize it, but it does run fine anyway.

I tested games on P4 3ghz on via chipset motheroard, and then on intel motherboard (Intel 875P). Sb live + geforce 5600. Games in dos and win ran noticeably slower on via chipset, didnt run benchmarks though.

kithylin wrote:

I can give a try tomorrow some time I suppose. With certain systems, the MTRR's and FastVid have no effect what so ever on AthlonXP core chips. In all the testing I've done on my AMD "super-dos" system, I can not see any noticeable difference with or without fastvid/MTRR-enabler, it's the exact same FPS either way.

however it should be noted I'm using the mobile AthlonXP-M chip in there, not the desktop version, so it only runs at 266 Mhz Bus, I think 300 with the small overclock I can get out of it.

It's this chip: http://www.cpu-world.com/CPUs/K7/AMD-Mobile%2 … H2000FQQ3C.html with only 256 KB cache, and after some checking I've found it's Throughbred core not barton. Sorry.. I mis-spoke on that one. It's the fastest I can get to work in that KT-133A motherboard (the A version of the chipset that is) And even then it comes in as "AMD Unknown" during POST, the bios doesn't officially recognize it, but it does run fine anyway.

The MTRRs should always matter. Maybe they are already set for some reason? I don't know if the System or Video BIOS can set those, but Windows 98 is able to set them (depending on the video driver). I don't know if any of the DOS MTRR tools can dump/display the current setting.

About your CPU: I don't think the precise model matters very much. FSB266 is still faster than any P3, and if the frame buffer doesn't fit in the cache anymore the exact size of the cache is probably also somewhat irrelevant. (800x600 would just fit into 512kB.)

idspispopd wrote:

The MTRRs should always matter. Maybe they are already set for some reason? I don't know if the System or Video BIOS can set those, but Windows 98 is able to set them (depending on the video driver). I don't know if any of the DOS MTRR tools can dump/display the current setting.

About your CPU: I don't think the precise model matters very much. FSB266 is still faster than any P3, and if the frame buffer doesn't fit in the cache anymore the exact size of the cache is probably also somewhat irrelevant. (800x600 would just fit into 512kB.)

I've tried the fastvid and like I said above, the exact same results in FPS with fastvid on and fastvid off, and even the other one to set MTRR does not increase speed any, and the athlon one doesn't either for K7.

I'm fairly sure that the "extra instructions" that get "turned on" in older pentium class systems are just automatically always-on in the newer chips.

And I'm using KT-133A with the athlonXP chip because it's the fastest / most advanced CPU I can get to work with a native ISA slot for ms-dos native-mode ISA sound. And get to use a AWE64 with it.

Fastvid made a huge difference for me. I loaded Quake (shareware 1.06) for the first time and was disappointed by the low frame rates. At 800x600, I was getting between 9 and 10 FPS.

I have a 700MHz P3 (100 FSB), 128MB RAM (unknown specs), Geforce FX5500 PCI (no AGP on MB), and CT1330A.

I did not enable Write Posting because Fastvid tells me I have an 82440 MB.

Here are the results after running Fastvid (demo1.dem with sound on)

640x480 : 44.3
800x600 : 31.0

I finally got my hands on a 1GHz 100 FSB Slot 1 PIII, so I will retest soon.

What I would really like to know is how much effect memory (timing and amount) has on the result.

Its not about extra instructions, but changing memory access settings. Some BIOSes and video drivers do that automatically. Some don't, even with later chips.