Redneck Rampage @ 1600x1200.

retro games 100 wrote:

I can think of one, I'm sure others can think of better examples. System Shock 1, in SVGA mode. […]
Show full quote

I can think of one, I'm sure others can think of better examples. System Shock 1, in SVGA mode.

At what point does it become pointless to upgrade your PC, in an attempt to play the most demanding DOS game in it's best quality settings, with the smoothest frame rate possible? For example, is there no point in swapping out a P2 CPU for a P3 CPU? Is there no point in swapping out an old graphics card released in the mid '90s, for something more powerful that was released say 3-5 years later? (I'm just talking about SVGA DOS games here.)

One last thing - how can I determine what frame rate my 3D DOS & Win9x games are running at?

Thanks a lot for any comments people. 😀

Try the pcplayer benchmark to see what happens on different hardware

Thanks - I DL'd pcplayer benchmark & speedsys. (I found them via google -> which pointed straight back to a vogons thread.) 😀

Regarding frame rates, at what point does it become pointless to attempt to increase this value? How many frame rates can the human eye / brain see, before any further increase in this value results in no discernable improvement?

I would like to get the "redneck rampage" demo, and crank this game up to the max (I may need to wait until I get a 1600 x 1200 monitor), and see just what hardware is required to achieve a 'maximal' (maximum discernable quality) frame rate.

Has anyone forgotten Quake? 😕

Malik wrote:

Has anyone forgotten Quake? 😕

nah, I blame temporary alzheimers disease

I just tried the Quake 1 demo (1.06), and inside the video options, the highest resolution available was 800x600. I tried choosing that setting (using a P2 @350mhz, 100fsb CPU), and the 5 second test mode looked very choppy. In fact, this video test mode reverted back to 320x200 mode, presumably because the 800x600 video mode test was not successful.

Another thing I've noticed with Quake 1 (the demo) is that when I walk about in that initial section, right at the very beginning, I appear to suddenly lurch forward every 5 seconds or so. Walking about is not always 100% smooth. Perhaps I need to shut down to "pure DOS" and run Quake, rather than use a Windows 98 "dos box" to run it.

Edit: I tried "pure DOS" for Quake 1 demo, and I now have extra video resolution modes to choose from, the maximum being 1280x1024, which is the native resolution of my 19" flat screen monitor. It runs very choppy. The best hardware I have that I can try is a P3-based board, slot 1, with a P3 800mhz. Other than that, I could try a P4-based board, but I would have no ISA slots. But Quake doesn't have any funky MIDI music, so perhaps it's OK to run it on a non ISA-based board? I guess I just need something for the sound effects. Would this game be happy with a PCI-based sound card for that?

I ran pcpbench on 2 machines:

1) Pentium IV, 2.4ghz, Windows 98 (I booted using a dos disk), Diamond Viper AGP 32mb graphics. Using pcpbench.exe without any command line parameters gave me a "score" of 68.1, and visually speaking, it looked quite good.

I also ran the Quake demo ver. 1.06 at 1280x1024 video mode, and it was horribly choppy. (Does this game look good on any PC, when run at very high resolutions?)

2) Pentium III, 800mhz (the best I can muster), Windows 98 (booted up using a dos disk), ATI Radeon 9250 AGP 256mb graphics. Using pcpbench.exe without any command line params gave me a significantly lower score of 44.9, and visually speaking, it looked ok.

Question to the people on this website who have built very fast P3-based systems: Have you done this in order to get very good frame rates in your DOS games, running at resolutions beyond say 800x600? I thought (and it now appears incorrectly), that you have built these fast P3 machines in order to run DOS games really well, and also to run win9x games well too. (I was thinking of building 2 seperate machines for this purpose - any old P3 for all DOS games, and a P4 for win9x games.)

I've tried both Pcpbench.exe and speedsys.exe, and they both look like great utilities!

Is there also a utility that I can run, to determine the framerate of a specific DOS and/or win9x game? Say for instance, I want to know what frame rate Doom or Quake runs on my rig. Can I run a framerate utility, then launch a specific game, to find out its framerate?

Thanks a lot people. 😀

Armored Fist 2
Extreme Assault
Battlespire

Cybermage. It took a P3-600 Coppermine to smooth that out in SVGA. I was running a P3-450 and it would get choppy when I got close to the walls. 😀

And yeah, System Shock in SVGA is serious stuff.

Quake is extremely optimized for P5/P6 CPUs so it doesn't need as much CPU power IMO. Unless you're not on a P5/P6....

Yes! I have been waiting for a chance to see my rig flex its muscles in pure DOS. I'll was going to mention that Janes ATF is likely a good benchmark however I only have the gold edition which works through Windows and DirectX (more properly DirectDraw), so I'm not going to use that as a comparison. I do remember with the dos version, a voodoo5 and my Athlon XP running at 11.5x100 I could get about 50fps at 1024x768. In windows I'm always hovering around 60, so I think there is some V-Sync involved.

Right now the system is currently running at a 2200+ rating; 1.67ghz (12.5x133) and a Radeon 9800xt (VBE 2.0 as reported by pcpbench). Epox 8KTA3 based of a Via KT133a (The fastest Socket A platform with an ISA slot I could find). When I run it in LFB mode I get some corruption at the bottom of screen. and at 640x480 I get intermittent banding in the main scene. It does however complete it successfully. I haven't bothered with any VBE utilities yet. I'm also capped at at 1280x1024 because of my monitor.

640x480x8bpp 191.0 LFB // 122.1 NOLFB
800x600x8bpp 134.7 LFB // 84.8 NOLFB
1024x768x8bpp 88.9 LFB // 52.6 NOLFB
1028x1024x8bpp 56.6 LFB // 32.9 NOLFB

800x600x16bpp 80.0 LFB // 45.4 NOLFB
1024x768x16bpp 54.6 LFB // 28.7 NOLFB

1280x1024x32bpp 9.6 (oddly enough the artifacts are gone, and got the same speed for LFB/NOLFB)

If there is a demand to see it, and if I get some time I might try the Voodoo5 to see the differences in the 2D core. I would think that Athlons are gonna be the best bet for high speed pure DOS enviroment. On your P4 based score above you didn't mention the Resolution or bit depth, so I assume it's the default of 640x400x8. The reason why my 1.67ghz part walked all over a P4 2.6 is because the P4's FPU is actually rather weak. The ALU is okay but I don't' think this bench uses it very much. P4's really shine when you are using SSE2 code, and that stuff simply wasn't around for back then. I believe that the fastest P4 you could get this going with is Nothwood "B" 3.06ghz. Anything faster is either based of a Prescott core, or needs a Intel 865 chipset for 800mhz FSB setting. Both of those are the point where DOS compatibility goes out the window.

retro games 100 wrote:

At what point does it become pointless to upgrade your PC, in an attempt to play the most demanding DOS game in it's best quality settings, with the smoothest frame rate possible?

I've heard differing opinions, usually 60 fps but others say 72 fps is the maximum the human eye can perceive a difference. Definitely if you never go under 72 there is no practical reason to try and squeeze out more.

Those big numbers! 😳 Much bigger than mine! 🙁 Damn, I'm doing this all wrong! 😢 (No wonder you call your rig a rocket! 😀 )

I'm going to try pcpbench again, on both the P3 and P4 machines, but with an nvidia 5200 and then with a voodoo3. Hopefully, I'll be able to claw out an extra 10% improvement? [Edit: From the games mentioned above, I've got Cybermage & Battlespire, although I can't find the Cybermage disc ATM, but I've managed to find the Battlespire disc. I'll try it out when I change over graphics cards...]

But, looking at your jaw-droppingly impressive figures 😎 , I now realise I must rethink my ill thought out retro plans. I guess I could keep my slot 1 boards for "basic" dos & win9x usage, but for anything really fast, I won't bother with the P4 mobo, but get something non Intel instead.

I think I'd like to mess about with 3 new types of board, categorized by the number of ISA slots they have: 2, just 1, and none at all.

For the 2 ISA slots board, I could get something Duron-based perhaps? (What's the fastest 2 ISA slot board out there? Perhaps it's Intel based, like a socket 370 tuatalin 1.4ghz?)

For the 1 ISA slot board, I could get something similar to what prophase_j is using.

For the no ISA slots board, what's the fastest possible machine I could run windows 98 on? (Must have win 98 mobo chipset drivers.)

Thanks. 😀

Hater Depot wrote:

retro games 100 wrote:

At what point does it become pointless to upgrade your PC, in an attempt to play the most demanding DOS game in it's best quality settings, with the smoothest frame rate possible?

I've heard differing opinions, usually 60 fps but others say 72 fps is the maximum the human eye can perceive a difference. Definitely if you never go under 72 there is no practical reason to try and squeeze out more.

That's kind of heart-warming news. So, in other words, it's not essential to go completely OTT on the speed factor. Get the frame rate anything above about 70, and you're OK.

But I wonder what rig you'd need to get 1280x1024x32bpp at 70 frames per second? 😉

retro games 100 wrote:

I think I'd like to mess about with 3 new types of board, categorized by the number of ISA slots they have: 2, just 1, and none at all.

How about 3? For 3 ISA slots board, you could get something like P4 3.06 GHz.

retro games 100 wrote:

That's kind of heart-warming news. So, in other words, it's not essential to go completely OTT on the speed factor. Get the frame rate anything above about 70, and you're OK.

But I wonder what rig you'd need to get 1280x1024x32bpp at 70 frames per second?

It strikes me as a bit odd to go for these resolutions+fps in Dos VESA modes. I am very happy some of my favorite dos games support 640x480, but for most games one will have to go back to 320x200 again, despite the available computing power.
At 640x480 I am confident they run smooth enough on my slot-1 system with 2 ISA slots. Even before I gave it a faster CPU. (Now it gives 3600 3Dmarks'01, but that benchmark is aimed at windows 3D gaming.)

I always notice it when a CRT screen is set to a 60 Hz refresh rate, and insist on 85 or 75 minimum. On the other hand when a game runs at 40 fps I am usually satisfied. I read some console games come preset to do just 25 fps. Not saying I would advocate that to be enough 😉.

Note that if you go for a system without ISA slots and want to run real dos, you will probably need NMI and/or DDMA supporting mainboards to get dos sound emulation working.

I've changed graphics cards, and re-run the tests.

Test #1. P3, slot 1, 800mhz. Now with voodoo3 SG-RAM 16mb card. (Got rid of AGP 2x ATI Radeon 9250, 256mb). Result was 44.9, now it's 41.2. 🙁

Test #2. P4, 2.4ghz. Now with AGP 4x Innovision 5200fx, 128mb). Result was 68.1, now it's 67.3. 🙁

🙁

Looks like I need to rethink all of this...I need some mobo purchasing advice! As in my previous post, I'd like to mess about with boards with 2, 1, and 0 ISA slots. (The "0 ISA slot" board won't be for DOS, but it won't be for XP either.)

Incidentally, I tried to test Battlespire (8 times in fact, with 4 different graphics cards, on 2 machines, in 2 video modes. All 8 attempts failed. The launcher runs for about 1 second, then returns to the windows 98 desktop. Running in DOS, it complains about not being able to detect vesa mode 2.)

@KAN,

Isn't that Soyo board one of those "industrial" things? (I think I'd be happier to go for something that the "masses" use.)

@gerwin,

Don't you want to run Redneck Rampage @ 1600x1200? 😉 🤣

About those "slow" benchmark results you got on the P3-800 and P4 machines:
Most likely your BIOS does not enable MTRRs for LFB to use under DOS. Utilities like FastVid or MTRRLFBE can fix this.

However, even with optimizing, VESA modes will be very slow in high resolutions. Faster CPUs or different video cards won't help much (for comparison, just run a non-accelerated video driver on your newest and fastest Windows/Linux machine and see how it crawls...).

I agree with gerwin on this one. Playing DOS games at more than 640x480 doesn't make much sense. Unaccelerated DOS games won't get much prettier in higher resolutions, and usually a frame rate above 30 FPS doesn't give you any gameplay advantages. Games that require fast reaction times (like multiplayer FPS) either are frame limited by their engine or come in an accelerated version.

Redneck rampage, no thanks. I am content with Jonof's/Proasm's Duke 3D and Shadow Warrior in OpenGl. That is cheating I know. : )

VastVid or MTRRLFBE, I will try these too later, But IIRC I read somewhere it was not needed for DOS 7.10?

I tried pbpbench at default settings without anything else loaded, and got the following results from three seperate systems:
1) Pentium-III-S 1000/105MHz with Geforce MX440 (with 128-bit memory): 50.5 fps
2) Pentium-III-EB 866/133MHz with Geforce MX440 (with 64-bit memory): 37.5 fps
3) AMD Sempron 3000+/333MHz with Geforce 7600 GS: 81.8 fps

For a gaming system it would be wise to stay clear of 64-bit graphics card memory. That may also be what causes the fps difference between the two pentium systems here?

Just comparing this to prophase_j's results... ouch. 😜