shadermark 2.1 released

Lezmaka said:
It's be interesting to see if the 6800's gain much speed with the 66.31 recommended drivers on dave's system, but I think he's got something pretty big coming up this week and may not have the time :)

I agree since the 66.31 are now WHQL. If you look at ellingsen1's results you should get a general idea since he used the 66.31's ... :)
 
trinibwoy said:
Dave, I'm sure you've noticed but people are getting much higher nvidia scores with the 66.31 driver that is recommended for use on 2.1. Why didn't you use that one?

Generally I don't follow leaks and Beta drivers, we stick to using WHQL drivers, but in this case I thought I'd use the latest drivers that I had from NV for the 6600 review. Didn't spot the recommendations for Shadermark, but the 66.31's don't appear to be available from NV's site yet. Plus, I've not actually seen results in this thread running at reference speeds.
 
I think I'll go along with the 66.31 results as they are WHQL drivers available though not yet released on Nvidia's site. It does seem the NV40 results are better than the x800xt pe ...

The reference clock that people have posted here are valid default clocks with some Ultra cards you can buy. What would be interesting is to see an NV40 UE score ...
 
Here are some results with the card clocked at reference speeds added to the stock speed. I think both are needed as all the Ultra's I have seen on the shelves so far have came clocked at a 425 core.
Edit: Tested at driver defaults (Quality)
Code:
XFX Ultra 66.31 Drivers
     425/1.15          400/1.1
   PS2.0 PS3.0 PS2.0  PS3.0
2   962   962    902    902
3   781   781    721    721
4   781   781    721    721
5   712   709    661    661
6   721   721    699    699
7   672   672    601    601
8   409   385    385    361
9  1077  1076    960    960
10  841   841    815    815
11  721   721    692    688
12  481   481    451    451
13  418   401    391    361 
14  481   480    454    454
15  277   337    261    317
16  330   330    312    312
17  450   450    420    420     
18   55    57     52     53
19  165   179    154    172      
20   55    55     52     51
21   88    88     83     83  
22  119   120    112    114
23  132   134    120    120
24   80    80     75     75 
25   94    95     88     89
26   90    90     84     85
 
pharma said:
Lezmaka said:
It's be interesting to see if the 6800's gain much speed with the 66.31 recommended drivers on dave's system, but I think he's got something pretty big coming up this week and may not have the time :)

I agree since the 66.31 are now WHQL. If you look at ellingsen1's results you should get a general idea since he used the 66.31's ... :)

The beta 65.xx seems to work. Only the 61.xx (didn't try the 62.xx) have problems with the 3_0 shadow mapping shader.


Thomas
 
After seeing some results there´s something curious... someone notice that in the HDR tests the X800XT-PE is so much faster against the 6800Ultra? There is some reason for this?

It remembers some of the firsts results in the rthdribl demo between the R420 and NV40.
 
There are a lot of "interesting" things to note about the results.

It's really hard to tell without knowing what a particular test does, but I think the X800 shows an advantage when alpha fillrate is a primary factor. This is probably due to the fact that the 6800 can only fill 8 pixels with alphablending enabled.

There are other interesting things, like why is the 6800 faster in the 2a tests than the 2/2b/3 tests? I'd guess that the 2a backend is optimised for the 5xxx series and the reduced register pressure happens to be a help on the 6xxx series on these longer shaders.
 
kemosabe said:
Forgive my newbieness to this particular topic, but how are we to expect any significant performance gains on the NV4X series in games that support SM3.0 in light of these results? :?

recompiling a 2.0 shader to a 3.0 target does not create a different shader, that's why you won't expect gain or loss on the same hardware: it executes at the same speed. minus micro variations of specs.

The difference is that 3.0 allows you to write more shaders in different and subtile ways. For example there is no low precision color interpolator anymore, there is a generic texture sampler in the vertex shading unit, there are vface and screen coordinate input registers, you can ask for a specific LOD in a dependant texture read. There are no limit on those dependant texture read. You can early exit from a shader based on dynamic conditions, etc.. etc..

There are numerous things that are not possible in plain 2.0 (or so complicated that you want to save the trouble), that are possible either in 2 extended or only in 3.0. And for the shaders that relies on those features, it's hard to tell how they would compare versus 2.0 simply because you can't compare without introducing arbitrary bias (oversimplification in one case, or change of paradigm in another case). Of course there is also no telling how these things will behave in the future, because you can't predict the future based on the comparisons you see right now.
 
LeGreg said:
recompiling a 2.0 shader to a 3.0 target does not create a different shader

It may do. If you have some HLSL with a branch it will compile to a different set of instructions depending on the target. One will implememt the branch in a ps2 style the other with true dynamic branching.
 
Thanks for the insight. :) Doesn't that suggest that the real benefit will come with games that are developed from the ground up using SM3.0?
 
kemosabe said:
Thanks for the insight. :) Doesn't that suggest that the real benefit will come with games that are developed from the ground up using SM3.0?

That would appear to be the case. However, there are some who subscribe to a philosophy that developers should not provide effects that are only viable on one set of hardware.
 
Anybody run it on a FX? Mostly for fun I fired it up on a 5800: While stepping through shaders 1-19 seemed to work just fine (albeit slow, as expected) the benchmark mode just briefly flashed 'Shader not supported, press <H>' before quitting to desktop.
 
Zaphod said:
Anybody run it on a FX? Mostly for fun I fired it up on a 5800: While stepping through shaders 1-19 seemed to work just fine (albeit slow, as expected) the benchmark mode just briefly flashed 'Shader not supported, press <H>' before quitting to desktop.

I'am currently investigating this issue, found a MRT detection bug, but forwarded it to nvidia, could be driver issue....

Thomas
 
pixel shader version: Pixel Shader 3_0

shader 2 1017 fps
shader 3 826 fps
shader 4 828 fps
shader 5 756 fps
shader 6 765 fps
shader 7 680 fps
shader 8 404 fps
shader 9 1103 fps
shader 10 895 fps
shader 11 765 fps
shader 12 508 fps
shader 13 423 fps
shader 14 510 fps
shader 15 340 fps
shader 16 340 fps
shader 17 469 fps
shader 18 60 fps
shader 19 192 fps
shader 20 57 fps
shader 21 93 fps
shader 22 127 fps
shader 23 142 fps
shader 24 84 fps
shader 25 100 fps
shader 26 96 fps

Gainward 6800U (450/1200) - P4 3.2@3.6 - 66.31 - quality
 
crusher_pt said:
After seeing some results there´s something curious... someone notice that in the HDR tests the X800XT-PE is so much faster against the 6800Ultra? There is some reason for this?

It remembers some of the firsts results in the rthdribl demo between the R420 and NV40.

Is that demo/shader demo thinger using MRT? If it is, NV40's performance drops off after 3 render targets.
 
trinibwoy said:
That would appear to be the case. However, there are some who subscribe to a philosophy that developers should not provide effects that are only viable on one set of hardware.

Yes, like 99.9% of all game developers, no doubt...;)
 
Zaphod said:
Anybody run it on a FX? Mostly for fun I fired it up on a 5800: While stepping through shaders 1-19 seemed to work just fine (albeit slow, as expected) the benchmark mode just briefly flashed 'Shader not supported, press <H>' before quitting to desktop.

I'm getting this same problem with my FX 5900. I'm going to bench now with my XT PE. I'll post the result soon.
 
performance:

X800 Pro (Catalyst 4.9 WHQL, just ignore shader 19 and 20, because they have problems)

Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 122683392
HAL (pure hw vp): RADEON X800 PRO (Anti-Detect-Mode: off, gamma correction: DAC)



options

pixel shader version: Pixel Shader 2_b

results:
shader  1 (                                           Pixel Shader Precision Test): pixel shader 1.1 full precision: s16e7
          (                                           Pixel Shader Precision Test): pixel shader 1.4 full precision: s16e7
          (                                           Pixel Shader Precision Test): pixel shader 2.0 full precision: s16e7
          (                                           Pixel Shader Precision Test): pixel shader 2.0 partial precision: s16e7
          (                                           Pixel Shader Precision Test): pixel shader 3.0 full precision: s0e0
          (                                           Pixel Shader Precision Test): pixel shader 3.0 partial precision: s0e0

shader  2 (                                            Per Pixel Diffuse Lighting -   ps_2_b):	 692 fps	  1.4442 mspf	    3463 rendered frames
shader  3 (                            Per Pixel Directional Light Shader (Phong) -   ps_2_b):	 502 fps	  1.9906 mspf	    2512 rendered frames
shader  4 (                                  Per Pixel Point Light Shader (Phong) -   ps_2_b):	 503 fps	  1.9899 mspf	    2513 rendered frames
shader  5 (                                   Per Pixel Spot Light Shader (Phong) -   ps_2_b):	 416 fps	  2.4038 mspf	    2081 rendered frames
shader  6 (                                        Per Pixel Anisotropic Lighting -   ps_2_b):	 501 fps	  1.9941 mspf	    2508 rendered frames
shader  7 (                                         Per Pixel Fresnel Reflections -   ps_2_b):	 461 fps	  2.1713 mspf	    2303 rendered frames
shader  8 (                                          Per Pixel Car Surface Shader -   ps_2_b):	 281 fps	  3.5605 mspf	    1405 rendered frames
shader  9 (                                         Per Pixel Environment Mapping -   ps_2_b):	 936 fps	  1.0684 mspf	    4680 rendered frames
shader 10 (                                    Per Pixel Environment Bump Mapping -   ps_2_b):	 604 fps	  1.6553 mspf	    3021 rendered frames
shader 11 (                                                Per Pixel Bump Mapping -   ps_2_b):	 484 fps	  2.0682 mspf	    2418 rendered frames
shader 12 (                                       Per Pixel Shadowed Bump Mapping -   ps_2_b):	 160 fps	  6.2600 mspf	     799 rendered frames
shader 13 (                                        Per Pixel Veined Marble Shader -   ps_2_b):	 240 fps	  4.1733 mspf	    1199 rendered frames
shader 14 (                                                 Per Pixel Wood Shader -   ps_2_b):	 256 fps	  3.9124 mspf	    1278 rendered frames
shader 15 (                                                 Per Pixel Tile Shader -   ps_2_b):	 189 fps	  5.3046 mspf	     943 rendered frames
shader 16 (        Per Pixel Refraction and Reflection Shader with Phong Lighting -   ps_2_b):	 233 fps	  4.3002 mspf	    1163 rendered frames
shader 17 (                             Per Pixel BRDF-Phong/Anisotropic Lighting -   ps_2_b):	 295 fps	  3.3888 mspf	    1476 rendered frames

performance impact of heavy alpha blending

shader 18 (                                  Fur Shader With Anisotropic Lighting -   ps_2_b):	  28 fps	 35.2134 mspf	     142 rendered frames

performance impact of multiple shaders switches

shader 19 (                                                    Combination Effect -   ps_2_b):	  91 fps	 10.9832 mspf	     456 rendered frames

performance impact of dynamic flow control

shader 20 (                Dual Layer 8x8 PCF Shadow Mapping without Flow Control -   ps_2_b):	  31 fps	 32.2707 mspf	     155 rendered frames
shader 21 (                                                         not supported)

performance impact of floating point filtering

shader 22 (     High Dynamic Range Shader - low quality version without filtering -   ps_2_b):	 142 fps	  7.0567 mspf	     709 rendered frames
shader 23 (                                                         not supported)
shader 24 (         High Dynamic Range Shader - high quality without fp filtering -   ps_2_b):	  99 fps	 10.0702 mspf	     497 rendered frames

performance impact of multiple render targets

shader 25 (  Per Pixel Edge Detection And Hatching Shader using 1 RT and 2 Passes -   ps_2_b):	  82 fps	 12.2603 mspf	     408 rendered frames
shader 26 (   Per Pixel Edge Detection And Hatching Shader using 2 RTs and 1 Pass -   ps_2_b):	  81 fps	 12.3647 mspf	     405 rendered frames

6800 GT (66.31 WHQL, High Quality)

Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 135266304
HAL (pure hw vp): NVIDIA GeForce 6800 GT (Anti-Detect-Mode: off, gamma correction: DAC)



options

pixel shader version: Pixel Shader 3_0

results:
shader  1 (                                           Pixel Shader Precision Test): pixel shader 1.1 full precision: s10e8
          (                                           Pixel Shader Precision Test): pixel shader 1.4 full precision: s10e8
          (                                           Pixel Shader Precision Test): pixel shader 2.0 full precision: s23e8
          (                                           Pixel Shader Precision Test): pixel shader 2.0 partial precision: s10e5
          (                                           Pixel Shader Precision Test): pixel shader 3.0 full precision: s23e8
          (                                           Pixel Shader Precision Test): pixel shader 3.0 partial precision: s10e5

shader  2 (                                            Per Pixel Diffuse Lighting -   ps_3_0):	 766 fps	  1.3062 mspf	    3829 rendered frames
shader  3 (                            Per Pixel Directional Light Shader (Phong) -   ps_3_0):	 632 fps	  1.5820 mspf	    3161 rendered frames
shader  4 (                                  Per Pixel Point Light Shader (Phong) -   ps_3_0):	 637 fps	  1.5703 mspf	    3185 rendered frames
shader  5 (                                   Per Pixel Spot Light Shader (Phong) -   ps_3_0):	 578 fps	  1.7306 mspf	    2890 rendered frames
shader  6 (                                        Per Pixel Anisotropic Lighting -   ps_3_0):	 595 fps	  1.6820 mspf	    2973 rendered frames
shader  7 (                                         Per Pixel Fresnel Reflections -   ps_3_0):	 541 fps	  1.8497 mspf	    2704 rendered frames
shader  8 (                                          Per Pixel Car Surface Shader -   ps_3_0):	 309 fps	  3.2313 mspf	    1548 rendered frames
shader  9 (                                         Per Pixel Environment Mapping -   ps_3_0):	 917 fps	  1.0908 mspf	    4584 rendered frames
shader 10 (                                    Per Pixel Environment Bump Mapping -   ps_3_0):	 680 fps	  1.4701 mspf	    3402 rendered frames
shader 11 (                                                Per Pixel Bump Mapping -   ps_3_0):	 595 fps	  1.6816 mspf	    2974 rendered frames
shader 12 (                                       Per Pixel Shadowed Bump Mapping -   ps_3_0):	 381 fps	  2.6227 mspf	    1907 rendered frames
shader 13 (                                        Per Pixel Veined Marble Shader -   ps_3_0):	 323 fps	  3.0975 mspf	    1615 rendered frames
shader 14 (                                                 Per Pixel Wood Shader -   ps_3_0):	 389 fps	  2.5732 mspf	    1944 rendered frames
shader 15 (                                                 Per Pixel Tile Shader -   ps_3_0):	 283 fps	  3.5274 mspf	    1418 rendered frames
shader 16 (        Per Pixel Refraction and Reflection Shader with Phong Lighting -   ps_3_0):	 269 fps	  3.7170 mspf	    1346 rendered frames
shader 17 (                             Per Pixel BRDF-Phong/Anisotropic Lighting -   ps_3_0):	 375 fps	  2.6658 mspf	    1876 rendered frames

performance impact of heavy alpha blending

shader 18 (                                  Fur Shader With Anisotropic Lighting -   ps_3_0):	  44 fps	 22.6305 mspf	     221 rendered frames

performance impact of multiple shaders switches

shader 19 (                                                    Combination Effect -   ps_3_0):	 148 fps	  6.7739 mspf	     739 rendered frames

performance impact of dynamic flow control

shader 20 (                Dual Layer 8x8 PCF Shadow Mapping without Flow Control -   ps_3_0):	  47 fps	 21.3272 mspf	     235 rendered frames
shader 21 (                   Dual Layer 8x8 PCF Shadow Mapping with Flow Control -   ps_3_0):	  75 fps	 13.3708 mspf	     374 rendered frames

performance impact of floating point filtering

shader 22 (     High Dynamic Range Shader - low quality version without filtering -   ps_3_0):	 106 fps	  9.4411 mspf	     530 rendered frames
shader 23 (            High Dynamic Range Shader - high quality with fp filtering -   ps_3_0):	 119 fps	  8.4145 mspf	     595 rendered frames
shader 24 (         High Dynamic Range Shader - high quality without fp filtering -   ps_3_0):	  71 fps	 14.0551 mspf	     356 rendered frames

performance impact of multiple render targets

shader 25 (  Per Pixel Edge Detection And Hatching Shader using 1 RT and 2 Passes -   ps_3_0):	  82 fps	 12.1969 mspf	     410 rendered frames
shader 26 (   Per Pixel Edge Detection And Hatching Shader using 2 RTs and 1 Pass -   ps_3_0):	  78 fps	 12.8313 mspf	     390 rendered frames

Volari V8 (r1.06.52.rar from volarigamers.com)

Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 68157440
HAL (pure hw vp): LianGuan V3XT/V5/V8 Series (Anti-Detect-Mode: off, gamma correction: DAC)



options

pixel shader version: Pixel Shader 2_0

results:
shader  2 (                                            Per Pixel Diffuse Lighting -   ps_2_0):	  48 fps	 20.6470 mspf	     243 rendered frames
shader  3 (                            Per Pixel Directional Light Shader (Phong) -   ps_2_0):	  27 fps	 36.6576 mspf	     137 rendered frames
shader  4 (                                  Per Pixel Point Light Shader (Phong) -   ps_2_0):	  29 fps	 34.6633 mspf	     145 rendered frames
shader  5 (                                   Per Pixel Spot Light Shader (Phong) -   ps_2_0):	  21 fps	 47.3903 mspf	     106 rendered frames
shader  6 (                                        Per Pixel Anisotropic Lighting -   ps_2_0):	  25 fps	 40.7903 mspf	     123 rendered frames
shader  7 (                                         Per Pixel Fresnel Reflections -   ps_2_0):	  26 fps	 37.7849 mspf	     133 rendered frames
shader  8 (                                          Per Pixel Car Surface Shader -   ps_2_0):	  17 fps	 59.5323 mspf	      85 rendered frames
shader  9 (                                         Per Pixel Environment Mapping -   ps_2_0):	  71 fps	 14.1198 mspf	     355 rendered frames
shader 10 (                                    Per Pixel Environment Bump Mapping -   ps_2_0):	  56 fps	 17.8297 mspf	     281 rendered frames
shader 11 (                                                Per Pixel Bump Mapping -   ps_2_0):	  22 fps	 44.7142 mspf	     112 rendered frames
shader 12 (                                       Per Pixel Shadowed Bump Mapping -   ps_2_0):	  10 fps	 97.3405 mspf	      52 rendered frames
shader 13 (                                        Per Pixel Veined Marble Shader -   ps_2_0):	  12 fps	 83.3608 mspf	      60 rendered frames
shader 14 (                                                 Per Pixel Wood Shader -   ps_2_0):	  17 fps	 60.2483 mspf	      83 rendered frames
shader 15 (                                                 Per Pixel Tile Shader -   ps_2_0):	  11 fps	 91.9855 mspf	      55 rendered frames
shader 16 (        Per Pixel Refraction and Reflection Shader with Phong Lighting -   ps_2_0):	  11 fps	 90.6699 mspf	      56 rendered frames
shader 17 (                             Per Pixel BRDF-Phong/Anisotropic Lighting -   ps_2_0):	  13 fps	 74.1757 mspf	      68 rendered frames

performance impact of heavy alpha blending

shader 18 (                                  Fur Shader With Anisotropic Lighting -   ps_2_0):	   2 fps	643.5165 mspf	       8 rendered frames

performance impact of multiple shaders switches

shader 19 (                                                    Combination Effect -   ps_2_0):	   3 fps	359.1570 mspf	      14 rendered frames

performance impact of dynamic flow control

shader 20 (                                                         not supported)
shader 21 (                                                         not supported)

performance impact of floating point filtering

shader 22 (                                                         not supported)
shader 23 (                                                         not supported)
shader 24 (                                                         not supported)

performance impact of multiple render targets

shader 25 (                                                         not supported)
shader 26 (                                                         not supported)

quality comparison:

X800 Pro (ignore 19 and 20)
Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 122683392
HAL (pure hw vp): RADEON X800 PRO (Anti-Detect-Mode: off, gamma correction: FBW)



options

pixel shader version: Pixel Shader 2_b
shader    2   mse: 0.88548978
shader    3   mse: 0.87381872
shader    4   mse: 0.93206660
shader    5   mse: 0.99856695
shader    6   mse: 0.86144098
shader    7   mse: 0.67858760
shader    8   mse: 1.13319461
shader    9   mse: 0.38536231
shader   10   mse: 1.85095596
shader   11   mse: 1.07751083
shader   12   mse: 1.10503546
shader   13   mse: 0.64727815
shader   14   mse: 0.73469766
shader   15   mse: 1.18856907
shader   16   mse: 1.27490107
shader   17   mse: 1.45594374
shader   19   mse: 1.80908807
shader   20   mse: 3.69012801
error reading image files...
shader not supported
shader   22   mse: 8.33308570
error reading image files...
shader not supported
shader   24   mse: 8.56113847
shader   25   mse: 0.80111027
shader   26   mse: 0.80111027

6800 GT
Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 135266304
HAL (pure hw vp): NVIDIA GeForce 6800 GT (Anti-Detect-Mode: off, gamma correction: FBW)



options

pixel shader version: Pixel Shader 3_0
shader    2   mse: 0.56019783
shader    3   mse: 0.54107634
shader    4   mse: 0.52548536
shader    5   mse: 0.52206866
shader    6   mse: 0.53719330
shader    7   mse: 0.43692334
shader    8   mse: 0.71981208
shader    9   mse: 0.40015380
shader   10   mse: 1.38709100
shader   11   mse: 0.64171155
shader   12   mse: 0.60582415
shader   13   mse: 0.66471100
shader   14   mse: 0.33196863
shader   15   mse: 1.29484272
shader   16   mse: 0.89861298
shader   17   mse: 0.30872790
shader   19   mse: 0.85641321
shader   20   mse: 0.56255150
shader   21   mse: 0.56256898
shader   22   mse: 7.91011810
shader   23   mse: 8.11475976
shader   24   mse: 8.11018531
shader   25   mse: 0.53175004
shader   26   mse: 0.53177166

V8
Code:
ShaderMark v2.1 - DirectX 9 HLSL Pixel Shader Benchmark - ToMMTi-Systems (http://www.tommti-systems.com)

video mode / device info
(1024x768) X8R8G8B8 (D24X8) vram used 68157440
HAL (pure hw vp): LianGuan V3XT/V5/V8 Series (Anti-Detect-Mode: off, gamma correction: FBW)



options

pixel shader version: Pixel Shader 2_0
shader    2   mse: 0.98031743
shader    3   mse: 1.08029334
shader    4   mse: 1.04153983
shader    5   mse: 1.01915328
shader    6   mse: 0.96252314
shader    7   mse: 0.95538425
shader    8   mse: 2.08199787
shader    9   mse: 0.73123391
shader   10   mse: 1.91277218
shader   11   mse: 1.23583603
shader   12   mse: 1.21484502
shader   13   mse: 1.98894437
shader   14   mse: 1.10344315
shader   15   mse: 2.22119427
shader   16   mse: 1.52149582
shader   17   mse: 1.21849759
shader   19   mse: 1.61868954
error reading image files...
shader not supported
error reading image files...
shader not supported
error reading image files...
shader not supported
error reading image files...
shader not supported
error reading image files...
shader not supported
error reading image files...
shader not supported
error reading image files...
shader not supported

Thomas
 
Back
Top