Warp3D SI : Tests thread

Just fired up OpenJK. My character seems to be wearing psychedelic pyjamas!

Never mind, I just checked ioUrbanTerror, and it works for me in 32-bit mode. It must be something different.

@ddni

Quote:
Please use version 1.8 of the driver (download the driver again). Version 1.9 was supposed to go to beta testers only.

Hans

@Hans
Going to 1.8 made no difference.
Changing texture detail in OpenJK from medium to high fixed the pyjamas.

@bszili

What is the latest version of openjk?
I am using one from may 2014.

@Breed

Just tried the new Wap3D-SI with RadeonHD v2.10 on my X1000 with HD7750 (passively cooled)

Cow3D

160-162 fps
171-175 fps (when 'b' pressed)

Not that I'm really into benchmarks.

TrevorD

Wrong thread

@amigakit

Quote:
i'm getting between 272 and 278 fps on Cow3D with the latest drivers and debug kernel on my X5K this morning. the card used is a passive 7750. nicely done, hans!

-- eliyahu

@ddni
That's probably the latest one. Both my website and os4depot has it.

Hello all

If you are tired of using my infamous Cow3D then you can also use/test my Microbe3D/demo-view-ppc with some of the animated babes joined in the package ;-P
https://www.youtube.com/user/wazp3d

http://aminet.net/package/util/libs/Microbe3D

Alain Thellier

Cow3D

160-165 fps

AmigaONE X1000 - Radeon HD 7950 and the last drivres.

So, in Cow3D
Sam460+HD7750 146fps (PCIe v2 x4)
X1000 +HD7750 175fps (PCIe v1 x16)
x5000 +HD7750 278fps (PCIe v2 x8)

@eliyahu

xkernel on X5000 or vanilla exec with debug?

@KimmoK

Quote:

I'm pretty sure that I read that only the xkernel is running on X5000.

I haven't seen ragemem results for the X5000 but I suspect that the video bus test results are much higher than the X1000. I'm not sure if the X1000 bus can be made to go faster by tweaking/updating the BIOS like with the Sam boards.

@Rob

I think on paper x1000 and x5000 should have same video bus speed.

X5000 + Cow3D is 1.5x faster than Cow3D on x1000.
X5000 MIPS rating is 1.5x higher than x1000. (per core)

So, to me it is because of the CPU. (unless the second core is used, but I think it can not be, as Cow3D + Warp3D are "legacy" SW...)

++
SAM460 results are superb good in cow3D, as it has 1/2 of the video bus speed and 2/3 of the performance of the PA6T core, but still it manages to perform 4/5 of x1000 speed.

(at the same time SAM460 reach about 1/2 of x5000 speed, more than MIPS would "hint")
+++

Now I realized that it could be that the SW is just better optimized for the core of a SAM460 by the compiler.

Look at performance difference between X1000 and Sam460 in the new QuakeII HD from HunoPPC to see a real life comparison.

Sam460ex Radeon HD7750
1024x768
map demo1.dm2 : 18.1 FPS

X1000 Radeon R9 270
1280x1024
map demo1.dm2 : 59.1 FPS

~3.3x faster

@ddni

ioquake3= 2.39x faster . so depends on application.

X1000 Radeon R9 270X
demo four 1280x1024 max settings 74fps

Sam460ex Radeon HD7750
demo four 1280x1024 max settings 31fps

@Spectre660

Yes everything depends on the application.
It is certainly inaccurate to look at the extremes and infer a difference.
What would be best is to look at a spread of applications, exclude the extremes and compare the mean.

I guess quake2 does more by the CPU than quake3. So quake2 might depend more on the CPU & bus.

And I imagine cow3D is far simpler for a CPU and quake gets more benefit from larger caches and faster memory.

I do not dare to suggests you to swap cards with SAM and x1000 + run same bench to see more .... ;)

@Kimmok
Yes Cow3D is extremely simple it almost not use the cpu (only to rotate the vertices) so you almost only benchmark the Warp3D-drawing-speed (and the graphic bus) with this program.

With Cow3D when you hit 'b' the full animation (the cow 360° turn) is bufferized so the cpu is no more used for the rotation
But for the faster cpus it dont make much differences anyway

Alain Thellier

@thellier

Thanks Alain.

What was making Cow3d seem so slow on X1000 before the recent upgrade?