GZDoom benchmarks - next round

[Enjay]

OK, well the only ATI machine that I have is my daughter's laptop so I'll do a test there as well.

56%


[edit]

Oh and something that might help people.

Code: Select all

alias bm "vid_vsync 0; vid_fps 1; stat rendertimes"

Type that in the console (or an autoexec.cfg) and optionally bind a key to it ( eg bind b bm) and it will make sure that vsync is off and the other stat reporting things are on without having to type them every time. access it by either typing bm at the console or hitting your bound key.

[/edit]

[Graf Zahl]

Thanks. These were interesting, especially having an Intel chipset among it. Amazing. They managed to make them as fast as a 5 year old Geforce. Some progress!

The ATI is also nowhere near up to the performance as NVidia, as expected.

But it's really amazing to see how much NVidia's drivers improved over the last year. If I remember your last measurements correctly the improvement is evens slightly than what I got.

[Enjay]

Glad they are of use. Yes, I was surprised by the intel chip too.

However, on UTNT - the version from Torm's site is tutnt-v106.zip

Code: Select all

 adding ./doom2.wad, 2951 lumps
 adding tutnt-v106.pk3, 16 lumps
 adding tutnt-v106.pk3:endmap01_gl.wad, 13 lumps
 adding tutnt-v106.pk3:gldefs.wad, 2 lumps
 adding tutnt-v106.pk3:intermap_gl.wad, 35 lumps
 adding tutnt-v106.pk3:titlemap_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt01_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt02_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt03a1_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt03a2_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt03b_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt04a_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt04b_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt04c_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tnt04cn_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tntle_gl.wad, 13 lumps
 adding tutnt-v106.pk3:tntres.wad, 6434 lumps
 adding tutnt-v106.pk3:voiceact.wad, 43 lumps
I_Init: Setting up machine state.

Code: Select all

This savegame needs these wads:
tutnt-v105.pk3:tnt04cn_gl.wad

It would seem that the test version is 105 but Torm now has 106 on his site.

[Gez]

Please do me a favor and don't use imageshack. My connection to them is lousy and I can't get any of the pictures downloaded without waiting several minutes each. Better use the DRDTeam filesharing server and zip them.

Okay.
http://files.drdteam.org/index.php/file ... 9600mgt.7z

Don't -file with full paths. They get stored in the savegame.

It's not just -file. "Send to" and "open with" do the same thing.

Also, if my memory isn't playing me tricks, I did not have to extract the wads from the zips the last time; only the pk3s like KDiZD. The archive manager back then was happy considering that par-lutz.zip:par.wad would do the job just fine for that "par.wad" that the savegame wants. Maybe it should ignore the path (folder and zip container) when comparing archive file names?

Oh and something that might help people.

Code: Select all

alias bm "vid_vsync 0; vid_fps 1; stat rendertimes"

I've got that in my ini:

Code: Select all

Name=fpsoff
Command=vid_fps 0; stat rendertimes; stat renderstats; vid_vsync 1;
Name=fps
Command=vid_fps 1; vid_vsync 1; stat rendertimes; stat renderstats; vid_vsync 0;

It would seem that the test version is 105 but Torm now has 106 on his site.

I just renamed tutnt.pk3 into tutnt-v105.pk3 and it worked and it liked it!

[Graf Zahl]

Also, if my memory isn't playing me tricks, I did not have to extract the wads from the zips the last time; only the pk3s like KDiZD. The archive manager back then was happy considering that par-lutz.zip:par.wad would do the job just fine for that "par.wad" that the savegame wants. Maybe it should ignore the path (folder and zip container) when comparing archive file names?

That may be a result of the resource manager rewrite a few months ago. I'll have to check the code.

[Enjay]

I just renamed tutnt.pk3 into tutnt-v105.pk3 and it worked and it liked it!

Yup, works for me too.

My machine:




However, my daughter's laptop (ATI) is significantly slower:




Full screenshots here:

http://files.drdteam.org/index.php/file ... hs/njt.zip

[Enjay]

Hmm, after getting such a low value on the ATI, I decided to look at the other machines too and they were equally unimpressive:

My old GeForce 6200


My wife's Intel

[Gez]

I'm not surprised. It's the slowest scene of the bench. I don't even get 35 FPS on it, despite a reasonable hardware. (Phobia is second-slowest.)

[Graf Zahl]

Yes, but >1.4 seconds per frame does not sound right.

[Gez]

Reminds me of the values I had on my now-defunct previous machine. The FPS counter reported values below 5 for about everything more intricate than the original IWADs. Yet, when playing, while it wasn't perfectly smooth, it felt more like 15-20 FPS than like 1-5 FPS: it wasn't smooth, but it wasn't a slideshow.

[Enjay]

However, slideshow would be an accurate description of how the game was in that test.

[Graf Zahl]

Reminds me of the values I had on my now-defunct previous machine. The FPS counter reported values below 5 for about everything more intricate than the original IWADs. Yet, when playing, while it wasn't perfectly smooth, it felt more like 15-20 FPS than like 1-5 FPS: it wasn't smooth, but it wasn't a slideshow.

That was different. Your old computer had problems with the CPU's Time Stamp Counter. Either it counted too slow or the CPU speed detection failed resulting in bogus values.

[Firebrand]

By looking at the screenshots I can spot a big number of linedefs displayed, it's likely to have bad FPS IMO. But I don't really know how the renderer works thought.

[Graf Zahl]

That large number is the time to render the linedefs in milliseconds. How this can reach this high is the question here...

[Rachael]

I was not able to set any of the shader options - the menu is disabled for me.