@Seiya

Something still wrong with cow3d. I mean 37 and 4500 fps looks like again wrong fps calculation happens (too big difference between two), while gears results looks more or less real.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

i don't know..
maybe it could be depend by vmware version how to execute some 3D engines

Last edited by Seiya on 14-Apr-2015 at 03:26 PM.

_________________

as long as there is no common source that can be compiled for all platforms with no changes, there will be no remotely dependable results. the more people mess with the code and then arbitrarily decide if the benchmark result is sane or not, the less trustworthy it gets.

@wawa

however the source of cow3d should be the same for alls.

_________________

Just posted this to Amigans. Obviously you can't do the test on AROS or MorphOS but the thread was started for comparing test under OS4 anyway.

Universe Two by Crisot can be used for benchmarking pure Warp3D as can his Pxs-engine. It might also answer if there is something funny going on with Cow 3D.

Tested on X1000 witth R9 280 @850Mhz.

Universe Two: 72.893491 fps

Pxs-Engine has 2 executables. PXS_Ship renders 387 co-ordinates and 608 indics while PXS_Multiships renders 1935 co-ordinats and 3040 indices. I tested all resolutions available to me apart from 16-bit modes. I tested just once in each resolution rather than average multiple runs since the variation is so low there's no point.

I know some of you have 4x monitors so it would be interesting to see if the fps stay level all the way to 4x on your systems.

Pxs_Ship;

640x480x32: 72.823372 fps
800x600x32: 72.252017 fps
1024x768x32: 60.054924 fps
1152x864x32: 59.818174 fps
1280x720x32: 59.825574 fps
1280x800x32: 60.014453 fps
1280x960x32: 60.063571 fps
1280x1024x32: 60.063743 fps
1440x900x32: 60.160621 fps
1600x900x32: 59.867207 fps
1600x1200x32: 60.039655 fps
1680x1050x32: 59.884414 fps
1920x1200x32: 59.937315 fps

Pxs_Multi_Ships;

640x480x32: 75.015544 fps
800x600x32: 72.208206 fps
1024x768x32: 60.036888 fps
1152x864x32: 59.831259 fps
1280x720x32: 59.834494 fps
1280x800x32: 59.982838 fps
1280x960x32: 60.034092 fps
1280x1024x32: 60.064502 fps
1440x900x32: 60.138790 fps
1600x900x32: 59.867103 fps
1600x1200x32: 60.102927 fps
1600x1050x32: 59.902866 fps
1900x1200x32: 59.952533 fps

THIS is a nice example of HD7000/R7 level graphics:

https://www.youtube.com/watch?v=ZwWLns7-xN8

(Yes, this is actually Game Engine Footage!)



Now, let's go back to discuss using these cards to render Quake3 in "playable" 30FPS...

I just did quick test on my X5000 with MorphOS 3.10 (betatesting version) and Radeon X1950 and got over 164 FPS on 1280x1024 :)

@pampers

Is that Quake3 ?



Quote:

pampers wrote: I just did quick test on my X5000 with MorphOS 3.10 (betatesting version) and Radeon X1950 and got over 164 FPS on 1280x1024 :)

Last edited by Spectre660 on 04-Dec-2015 at 07:47 PM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@pampers

Quote:

1280x1024

1024x1024 is too low resolution for these cards and Quake 3. I mean, even with slow Warp3D difference between 640x480 and 1920x1080 is only few FPS.

@pampers

How's proceding development of radeon hd drivers under Mos?

@Spectre660

Yep

@kyle

I'm not correct person to ask.

Last edited by pampers on 04-Dec-2015 at 09:17 PM.

@pavlor

Quote:

I mean, even with slow Warp3D difference between 640x480 and 1920x1080 is only few FPS.

As I pointed out to you in post #74, it's very misleading to talk about differences as "only a few FPS" in contexts where the FPS count is so very low to begin with.

You should use percentage!

The difference between 1 and 2 FPS is only 1 FPS, yet the latter is 100% faster.

@TRIPOS

Well, X1000 scores cca 90 FPS with R9 280X:

IOQuake3 by HunoPPC
(Settings: everything maxed !!! / timedemo1 demo4)

1280 x 1024 = 94.7 fps
1600 x 1200 = 89.6 fps

Source: Amigans.net

Last edited by pavlor on 04-Dec-2015 at 09:38 PM.

@pavlor

This thread is full of several examples of Q3 benchmarks in various resolutions. Those rather close overall numbers (especially in FPS) is probably a result of ancient and trivial S/W such as Q3 running on modern top of the line GPU's, i.e. there is very little challenge for the H/W, no matter what, relatively speaking. The higher the GFX card, the less difference. The rather low numbers is definitely a result of poor drivers and/or poor concept/technology (and possibly poor implementation).

Come to think about it, the latter might affect/impact the first as well...

Anyway, pamper's MorphOS beta score is 73.18% (almost three quarters) faster than the OS4 high score you posted. H/W platform below the GFX card should be rather similar in overall performance comparison, from a GFX POV, right?

MorphOS score using a 2007 Radeon X1950.
OS4 score using a Radeon R9(!).

You do the math and fill in the gaps...

@TRIPOS

You wrote:

Quote:

talk about differences as "only a few FPS" in contexts where the FPS count is so very low to begin with.

90 FPS is very low?

Quote:

The difference between 1 and 2 FPS is only 1 FPS, yet the latter is 100% faster.

The difference between 89.6 and 94.7 FPS is only 5.1 FPS, yet the latter is 6 % faster...

Quote:

Anyway, pamper's MorphOS beta score is 73.18% (almost three quarters) faster than the OS4 high score you posted.

On first Radeon HD Warp3D driver release, difference was 3x in favor of MorphOS... as I see it, the gap is closing.

@pavlor

I've been told that there is a still some room for improvements on MorphOS side like support DMA on PCI-E Radeons so watch the gap ;)

@pampers

Quote:

I've been told that there is a still some room for improvements on MorphOS side

I´m sure there is much more room for improvements on OS4 side... well, its result is still laughable.

@pavlor

Well it gets harder and harder to make big improvements as more of graphic card features are supported. So going from bad to OK, is easy, going from OK to great is hard, great to excellent is near impossible to do.

When T&L (TCL) is supported in AmigaOS graphic drivers the morphOS and OS4 drivers should be a close match.

https://en.wikipedia.org/wiki/Transform,_clipping,_and_lighting

The next thing to support after TCL is OpenCL, I guess.

Last edited by NutsAboutAmiga on 05-Dec-2015 at 11:44 AM.
Last edited by NutsAboutAmiga on 05-Dec-2015 at 11:43 AM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

Quote:

I've been told that there is a still some room for improvements on MorphOS side like support DMA on PCI-E Radeons so watch the gap ;)

Update, 215 FPS with recent drivers on MorphOS 3.10 :)

@pampers

please "Timedemo demo2" in quake in software rendering ... no gl

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome