Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-4041-8081-120121-160161-200201-

DXVA + VMR9 vs. MadVR vs. CUDA

Name: Anonymous 2011-04-25 10:39

Let's discuss the best method for GPU-accelerated H264 decoding

Name: Anonymous 2011-04-25 10:40

Sure. OpenCL. I'm afraid CUDA is obsolete from a software point of view. Both nVidia and AMD have exposed APIs for direct access to their video decoding hardware via OpenCL.

Name: Anonymous 2011-04-25 10:41

>>1,2
And for reference:

http://developer.amd.com/gpu/AMDAPPSDK/assets/OpenVideo_Decode_API.PDF

http://developer.download.nvidia.com/compute/cuda/sdk/website/C/src/cudaDecodeGL/doc/nvcuvid.pdf

Remember, OpenCL > DirectCompute > CUDA.

Name: Anonymous 2011-04-25 10:54

>>3
The other thing cool about OpenCL is that there is an OpenGL interop layer, so you can display the results of a decoded video frame directly to an output adapter, or use it as a texture to be rendered to a set of polygons, all on the GPU without having to copy anything to the system's main memory.

See: http://www.khronos.org/registry/cl/extensions/khr/cl_khr_gl_sharing.txt

Name: Anonymous 2011-04-25 19:02

Meanwhile, at JCT-VC:

Current indications are that the new standard could provide 2x better video compression performance (i.e. around half the bitrate for a similar quality level) at the expense of significantly higher computational complexity, compared with H.264/AVC.

http://www.vcodex.com/h265.html

I'm not sure how I feel about this. I mean, 2TB HDDs are around $70, but even mid-range CPUs and GPUs can cost hundreds of dollars each. Is this really moving in the right direction?

Name: Anonymous 2011-04-25 19:26

Yay, more patented shit. Fuck you, Americans.

Name: Anonymous 2011-04-25 21:40

>>3
Remember, OpenCL > DirectCompute > CUDA.

Why do you say that? I had the impression that CUDA was better able to take advantage of the hardware while OpenCL's only real advantage was portability across cards. Given that CUDA support has been extended to older cards and OpenCL apparently hasn't portability is hardly meaningful.

Ok, forget that nonsense. Is there anything about OpenCL that I'm missing? It seems like the technological chump here.

Name: Anonymous 2011-04-26 2:20

>>7
OpenCL works on DX10 hardware. OpenCL is an industry standard that works on AMD, Intel, nVidia, Imagination Tech (PowerVR), and various other GPUs, as well as CPUs due to it's heterogeneous nature. You can run OpenCL compute kernels easily on multiple GPU and CPU cores simultaneously on the same system. CUDA can't do this. I'd say OpenCL takes better advantage of hardware.

CUDA is to OpenCL as Glide was to OpenGL. And we all know what happened to Glide at the turn of the century. Proprietary standards never win out.

Name: Anonymous 2011-04-26 2:24

nVidia
PowerVR
Bring them closed source drivers.

Name: Anonymous 2011-04-26 2:30

>>7,9
>Imagination Tech (PowerVR)
PowerVR GPUs are the GPUs powering 90% of mobile devices. The OpenCL Embeddeed Profile capable PowerVR equipped mobile phones are coming out later this year, which includes the iPhone 5. OpenCL will pretty much become the defacto compute API.

Also, there's a big movement currently underway among game developers to move away from 3D-specific APIs like Direct3D and OpenGL and to build custom ``software'' renderers using compute languages, like OpenCL and DirectCompute. For example, DICE is using DirectCompute to implement Battlefield 3's tile-based fully-deferred shading renderer.

No one is bothering with CUDA becomes it's proprietary and only works on nVidia GPUS, and AMD's HD5000/HD6000 are the most popular DX11 GPUs currently (they have 75% of the DX11 market).

http://www.qotpa.com/2011/03/25/is-this-the-beginning-of-the-end-for-directx/

Name: Anonymous 2011-04-26 2:32

>>9
If you're interested AMD publishes all of the information for their newer GPUs including register mappings, shader instruction sets and so on. Unfortunately, the X.org developers are too slow, retarded and backwards to do anything with it.

Name: Anonymous 2011-04-26 2:44

>>11
I wasn't complaining about AMD, I was complaining about nVidia and PowerVR.

Name: Anonymous 2011-04-26 3:05

>>11
It ain't easy writing GFX drivers and microcode. Such effort extends over multiple years. I know nothing about this area so I can't really contribute effort here.

Name: Anonymous 2011-04-26 6:11

Anyone who thinks CUDA is better than OpenCL has fallen for nVidia's marketing propaganda and false advertisements.

Name: Anonymous 2011-04-26 8:06

Name: Anonymous 2011-04-26 8:11

Name: Anonymous 2011-04-26 8:17

Name: Anonymous 2011-04-26 8:22

Name: Anonymous 2011-04-26 8:27

Name: Anonymous 2011-04-26 8:33

Name: Anonymous 2011-04-26 8:38

Name: 7 2011-04-26 8:39

>>8,10
So... you're saying portability is the reason? That doesn't really work for me. Intel is not even worth considering, and as nVidia supports CUDA on much *older* cards at this time I wouldn't bet against CUDA being supported on more existing cards than OpenCL. Meanwhile my fairly recent AMD/ATI mobile card doesn't support OpenCL so the portability argument just doesn't register with me.

What I am concerned with is capability. CUDA seems to have lower level capabilities and in theory [C for CUDA/etc] could be optimized better as a result. The way OpenCL works... it just doesn't seem to have the ability to take full advantage of the hardware. It's been a while since I've read over the documentation, but I got the impression it was obviously suboptimal.

Anyway I am not personally concerned with how id Software optimizes Rage or whatever. It makes sense for them to use the common groung--people will buy AMD/ATI cards no matter what I think of their drivers. I'm looking at this for my own personal use, more something like FASTRA who is probably using nVidia primarily because they were the only sane choice at the time for GPGPU, but possibly they continue to go that way partly because CUDA has a greater performance potential than OpenCL. That's what I'm trying to investigate.

>>9,14
Please stop acting like a mentally challenged individual.

Name: Anonymous 2011-04-26 8:44

Name: Anonymous 2011-04-26 8:49

Name: Anonymous 2011-04-26 8:54

Name: Anonymous 2011-04-26 8:59

Name: Anonymous 2011-04-26 9:04

Name: Anonymous 2011-04-26 9:10

Name: Anonymous 2011-04-26 9:15

Name: Anonymous 2011-04-26 9:20

Name: Anonymous 2011-04-26 9:25

Name: Anonymous 2011-04-26 9:31

Name: Anonymous 2011-04-26 9:36

Name: Anonymous 2011-04-26 9:41

Name: Anonymous 2011-04-26 9:46

Name: Anonymous 2011-04-26 9:52

Name: Anonymous 2011-04-26 9:57

Name: Anonymous 2011-04-26 10:03

Name: Anonymous 2011-04-26 10:08

Name: Anonymous 2011-04-26 10:13

Name: Anonymous 2011-04-26 10:26

Name: Anonymous 2011-04-26 10:27

>>22
The oldest gpu that nvidia suppports CUDA on is the G80 series, which includes the Geforce 8000 series. This is also the oldest gpu upon which OpenCL is available, so your concerns here are moot.

Furthermore, how is OpenCL not low level? The OpenCL compute language is C99. In fact, both nVidia and AMD use the same underlying compiler for OpenCL--Clang/LLVM. This is also the same compiler nVidia has used for CUDA's C and C++ compilers. OpenCL produces just as optimized code on nVidia Gpus as CUDA. CUDA has no additional performance potential.

So what does CUDA have that OpenCL doesn't? C++ and Fortran language support for compute kernels. But it seems you are not interested in those anyway.

I posted this from my GNU/linux arm based tablet in the hottub, so I apologize for any grammatical or punctual errors.

Name: Anonymous 2011-04-26 10:32

Name: Anonymous 2011-04-26 10:36

Name: Anonymous 2011-04-26 10:41

Name: Anonymous 2011-04-26 10:46

Name: Anonymous 2011-04-26 10:50

Name: Anonymous 2011-04-26 10:55

Name: Anonymous 2011-04-26 11:00

Name: Anonymous 2011-04-26 11:05

Name: Anonymous 2011-04-26 11:09

Name: Anonymous 2011-04-26 11:14

Name: Anonymous 2011-04-26 11:19

Name: Anonymous 2011-04-26 11:23

Name: Anonymous 2011-04-26 11:28

Name: Anonymous 2011-04-26 11:33

Name: Anonymous 2011-04-26 11:38

Name: Anonymous 2011-04-26 11:42

Name: Anonymous 2011-04-26 11:47

Name: Anonymous 2011-04-26 11:52

Name: Anonymous 2011-04-26 11:57

Name: Anonymous 2011-04-26 12:01

Name: Anonymous 2011-04-26 12:06

Name: Anonymous 2011-04-26 12:11

Name: Anonymous 2011-04-26 12:15

Name: Anonymous 2011-04-26 12:20

Name: Anonymous 2011-04-26 12:25

Name: Anonymous 2011-04-26 12:30

Name: Anonymous 2011-04-26 12:34

Name: Anonymous 2011-04-26 12:39

Name: Anonymous 2011-04-26 12:44

Name: Anonymous 2011-04-26 12:48

Name: Anonymous 2011-04-26 12:53

Name: Anonymous 2011-04-26 12:58

Name: Anonymous 2011-04-26 13:02

Name: Anonymous 2011-04-26 13:07

Name: Anonymous 2011-04-26 13:12

Name: Anonymous 2011-04-26 13:14

>>42
Sorry, yes, nVidia supports OpenCL on their older cards too, but I meant the divide between CUDA cards and OpenCL-only cards. But it doesn't matter--I am not interested in portability even a little.

As far as the compiler goes, it doesn't matter what it is, it matters what the code it produces actually does. For example you could compile C to Java bytecode plus some runtime environment, but the system you're targeting is heavily abstracted. I don't think OpenCL is that bad, but if I recall correctly it is significantly more abstracted than the CUDA scenario. You don't seem to know so I'll stop asking and trust the conclusion from my memory.

Name: Anonymous 2011-04-26 13:17

Name: Anonymous 2011-04-26 13:21

Name: Anonymous 2011-04-26 13:26

Name: Anonymous 2011-04-26 13:31

Name: Anonymous 2011-04-26 13:32

>>78
significantly more abstracted than the CUDA scenario
No it is not. I have used both. They are pretty much isomorphic to one another in terms of feature support and abstraction.

You're starting to dwell into strawman territory with your ad hominem.

Name: Anonymous 2011-04-26 13:36

Name: Anonymous 2011-04-26 13:40

Name: Anonymous 2011-04-26 13:45

Name: Anonymous 2011-04-26 13:50

Name: Anonymous 2011-04-26 13:55

Name: Anonymous 2011-04-26 13:59

>>83
What ad hominem? If I called you a faggot that would be ad hominem. If I suggest that you don't you lack understanding of the question that's a completely valid point to make, and not fallacious in and of itself. (And if I explained the meaning of irony, that would merely be condescending.) Besides, this isn't a debate--I just wanted information I thought you had, but I am quickly losing faith that you'd understand it if you did have it. It's perfectly okay not to know something!

On that point, you haven't shown any understanding of the systems involved. Who cares what compiler is used? That's a complete red herring. There's nothing magical about the compiler, it's the way the resulting code is run--what exactly it targets--that makes all the difference. Even if you wrote the assembly by hand it wouldn't prove anything. All I was saying is: since your reasoning stops at the compiler used you don't have any useful information for me at all.

I'm just trying to explain myself but if you take that as insulting or possibly as an attempt to compromise your character: don't worry, I'm happy to let your character speak for itself without comment one way or another.

Name: Anonymous 2011-04-26 13:59

Name: Anonymous 2011-04-26 14:04

Name: Anonymous 2011-04-26 14:06

>>22
fuck you faggot, closed source drivers are bad and you know it

Name: Anonymous 2011-04-26 14:09

Name: Anonymous 2011-04-26 14:13

Name: Anonymous 2011-04-26 14:18

Name: Anonymous 2011-04-26 14:23

Name: Anonymous 2011-04-26 14:28

Name: Anonymous 2011-04-26 14:33

Name: Anonymous 2011-04-26 14:37

Name: Anonymous 2011-04-26 14:42

Name: Anonymous 2011-04-26 14:47

Name: Anonymous 2011-04-26 14:52

Name: Anonymous 2011-04-26 14:56

Name: Anonymous 2011-04-26 15:01

Name: Anonymous 2011-04-26 15:06

Name: Anonymous 2011-04-26 15:11

Name: Anonymous 2011-04-26 15:16

Name: Anonymous 2011-04-26 15:25

Name: Anonymous 2011-04-26 15:26

>>89
Oh boy, look at this:

http://www.lunarc.lu.se/Documents/nvidia-workshop/files/tutorial/CUDA_C_QuickRef.pdf

http://www.khronos.org/files/opencl-1-1-quick-reference-card.pdf

CUDA and OpenCL are isomorphic to one another, meaning, that can do the same fucking things.

Name: Anonymous 2011-04-26 15:30

Name: Anonymous 2011-04-26 15:31

Name: Anonymous 2011-04-26 15:36

Name: Anonymous 2011-04-26 15:40

Name: Anonymous 2011-04-26 15:45

Name: Anonymous 2011-04-26 15:50

Name: Anonymous 2011-04-26 15:55

Name: Anonymous 2011-04-26 15:59

Name: Anonymous 2011-04-26 16:04

Name: Anonymous 2011-04-26 16:09

Name: Anonymous 2011-04-26 16:14

Name: Anonymous 2011-04-26 16:18

Name: Anonymous 2011-04-26 16:23

Name: Anonymous 2011-04-26 16:28

Name: Anonymous 2011-04-26 16:33

Name: Anonymous 2011-04-26 16:38

Name: Anonymous 2011-04-26 16:43

Name: Anonymous 2011-04-26 16:47

Name: Anonymous 2011-04-26 16:52

Name: Anonymous 2011-04-26 16:57

Name: Anonymous 2011-04-26 17:02

Name: Anonymous 2011-04-26 17:06

Name: Anonymous 2011-04-26 17:11

Name: Anonymous 2011-04-26 17:16

Name: Anonymous 2011-04-26 17:21

Name: Anonymous 2011-04-26 17:26

Name: Anonymous 2011-04-26 17:27

http://dis.4chan.org/read/prog/1303840708

Name: Anonymous 2011-04-26 17:35

Name: Anonymous 2011-04-26 18:16

>>110-137
Why'd you have to spam this thread? I don't care if you spam the threads about autism or anuses, but this was a decent thread.

Name: Anonymous 2011-04-26 18:27

>>138

It wasn't particularly good, he spammed better threads. The OP is probably from the imageboards, all he says is ``discuss topic x'' whilst contributing absolutely nothing himself.

Name: Anonymous 2011-04-26 18:38

Very shitty thread. Deserve spam.

Name: Anonymous 2011-04-26 18:44

>>139
At the very least, it's better than the autism and /jp/ spam, which is what most recent threads have been.

Name: Anonymous 2011-04-26 18:55

>>141

So because most of the threads were shit you think posting more shit is okay?

That's poor logic.

Name: Anonymous 2011-04-26 19:02

HAX MY AUTISM

Name: Anonymous 2011-04-26 19:06

Name: Anonymous 2011-04-26 19:10

Name: Anonymous 2011-04-26 19:15

Name: Anonymous 2011-04-26 19:19

Name: Anonymous 2011-04-26 19:33

Name: Anonymous 2011-04-26 19:37

Name: Anonymous 2011-04-26 19:43

Name: Anonymous 2011-04-26 19:47

Name: Anonymous 2011-04-26 19:52

Name: Anonymous 2011-04-26 19:56

Name: Anonymous 2011-04-26 20:00

Name: Anonymous 2011-04-26 20:10

Name: Anonymous 2011-04-26 20:14

Name: Anonymous 2011-04-26 20:19

Name: Anonymous 2011-04-26 20:23

Name: Anonymous 2011-04-26 20:28

Name: Anonymous 2011-04-26 21:32

Name: Anonymous 2011-04-26 21:37

Name: Anonymous 2011-04-26 22:00

Name: Anonymous 2011-04-26 22:43

Name: Anonymous 2011-04-26 23:26

Name: Anonymous 2011-04-26 23:38

Name: Anonymous 2011-04-27 0:02

Name: Anonymous 2011-04-27 0:55

Name: Anonymous 2011-04-27 1:35

Name: Anonymous 2011-04-27 1:47

Name: Anonymous 2011-04-27 1:59

Name: Anonymous 2011-04-27 2:03

Name: Anonymous 2011-04-27 2:08

Name: Anonymous 2011-04-27 2:12

Name: Anonymous 2011-04-27 3:07

Name: Anonymous 2011-04-27 3:07

Name: Anonymous 2011-04-27 3:12

Name: Anonymous 2011-04-27 3:17

Name: Anonymous 2011-04-27 3:29

Name: Anonymous 2011-04-27 3:34

Name: Anonymous 2011-04-27 3:38

Name: Anonymous 2011-04-27 3:43

Name: Anonymous 2011-04-27 3:51

HI IM A PROGRAMMER IM 13 I CAN PROGRAM CALCULATOR SORRY FOR MY INGLISH IM NOT INGLISH I CREATED A RANDOM NUMBER GENERATOR HERE IT IS 0.11763595137311034

Name: Anonymous 2011-04-27 3:55

Name: Anonymous 2011-04-27 3:59

Name: Anonymous 2011-04-27 4:04

Name: Anonymous 2011-04-27 4:08

Name: Anonymous 2011-04-27 4:13

Name: Anonymous 2011-04-27 4:17

Name: Anonymous 2011-04-27 4:21

Name: Anonymous 2011-04-27 4:26

Name: Anonymous 2011-04-27 4:30

Name: Anonymous 2011-04-27 4:35

Name: Anonymous 2011-04-27 4:39

Name: Anonymous 2011-04-27 4:44

Name: Anonymous 2011-04-27 4:48

Name: Anonymous 2011-04-27 4:52

Name: Anonymous 2011-04-27 4:57

Name: Anonymous 2011-04-27 5:01

Name: Anonymous 2011-04-27 5:06

Name: Anonymous 2011-04-27 5:10

Name: Anonymous 2011-04-27 5:14

Name: Anonymous 2011-04-27 5:19

Name: Anonymous 2011-04-27 5:23

Name: Anonymous 2011-04-27 5:28

Name: Anonymous 2011-04-27 5:32

Name: Anonymous 2011-04-27 5:37

Name: Anonymous 2011-04-27 5:41

Name: Anonymous 2011-04-27 5:45

Name: Anonymous 2011-04-27 5:50

Name: Anonymous 2011-04-27 5:54

Name: Anonymous 2011-04-27 5:59

Name: Anonymous 2011-04-27 6:03

Name: Anonymous 2011-04-27 6:08

Name: Anonymous 2011-04-27 6:12

Name: Anonymous 2011-04-27 6:16

Name: Anonymous 2011-04-27 8:14

Name: Anonymous 2011-04-27 8:18

Name: Anonymous 2011-04-27 11:07

Name: Anonymous 2011-04-27 11:35

Name: Anonymous 2011-04-27 12:03

Name: Anonymous 2011-04-27 12:07

Name: Anonymous 2011-04-27 12:12

Name: Anonymous 2011-04-27 12:16

Name: Anonymous 2011-04-27 12:20

Name: Anonymous 2011-04-27 12:25

Name: Anonymous 2011-04-28 4:18

Don't change these.
Name: Email:
Entire Thread Thread List
All trademarks and copyrights on this site are owned by their respective parties. All comments and posts are the responsibility of their own posters.
- Tinychan + 2ch Mode -