CUDA vs OpenCL needs a reality check...

[Madpacket]

OpenCL has already won as it's open. Unlike DirectX, nVidia does not have the marketshare to push CUDA likle Microsoft did with direct x.

Game over.

 

[cmdrdredd]

No offense, but this needs a new thread. It has nothing to do with CUDA vs. OpenCL. It is important IMO though.

It does relate to OpenCL. Something makes it not catch on, not just money. I'm not a programmer, but someone familiar with it might know what I'm asking about it.

 

[3DVagabond]

It does relate to OpenCL. Something makes it not catch on, not just money. I'm not a programmer, but someone familiar with it might know what I'm asking about it.

Well, I'd say M$ is why Linux and OpenGL haven't taken off. That's not going to effect OpenCL though.

 

[Olikan]

Well, I'd say M$ is why Linux and OpenGL haven't taken off. That's not going to effect OpenCL though.

if M$ decides to push direct compute?

 

[AnandThenMan]

Really though...what I wanna understand is why people claim open source stuff is the best thing ever when it doesn't catch on that much? Linux, OpenGL, OpenCL all open source and yet the market for them is comparatively small.

You need to look at the overall picture. Look a the Internet, the vast majority is open source/open consortium. Imagine if the Internet was run on a closed, proprietary backbone. Flash is a notable exception, but it will eventually be replaced. As for Linux, it has a very small install base on the desktop, but dominates in servers and super computer applications. Think about that for a second, if closed systems are so great, why don't we see them more in the most critical applications?

Basically we see closed systems become popular more because of the money thrown behind them than the actual merits of the product or code. Intel is by far #1 in graphics but by far the worst at it.

 

[SlowSpyder]

if M$ decides to push direct compute?

Then Intel and AMD are every bit as happy probably.

 

[Ken g6]

I made CUDA and Open CL apps three years ago. OpenCL documentation was harder to find and the compiler was buggy. Although I worked around the bugs the OpenCL app was never as fast as CUDA.

I haven't looked at either lately though.

 

[NTMBK]

I've written apps in both. CUDA just flat out has nicer syntax, and is a more pleasant language to work with. But to my mind, OpenCL is still the horse to back in this race- certainly in the general consumer space. It's just everywhere. Every mainstream APU has a built in OpenCL-capable accelerator. If I were to focus my efforts on one over the other, it'd be the one which lets me address all of my users, not just 15% of them.

In academia, and HPC? Entirely different story. You can dictate what hardware you buy, and have full control over the platform you run it on. CUDA all the way.

 

[-Slacker-]

Lonbjerg, judging by your use of emoticons when discussing the possible shortcomings of opencl and, well, also by your conduct in this thread, it looks like you are happy opencl may not be up to par. Don't you want opencl to succeed?

 

[blastingcap]

Lonbjerg, judging by your use of emoticons when discussing the possible shortcomings of opencl and, well, also by your conduct in this thread, it looks like you are happy opencl may not be up to par. Don't you want opencl to succeed?

Is that a serious question?

 

[BallaTheFeared]

OpenCL has already won as it's open. Unlike DirectX, nVidia does not have the marketshare to push CUDA likle Microsoft did with direct x.

Game over.

Seems like an odd thing to say, considering Nvidia created the market and has way more market presence than AMD.

Truth of the matter is OpenCL and AMD aren't really factors right now, and Knights Corner is actually the biggest threat to Nvidia.

 

[ShintaiDK]

DitectCompute is the only winner. Just like DirectX.

 

[zlatan]

DitectCompute is the only winner. Just like DirectX.

DirectCompute is a very limited API, mostly designed to extract the DX graphics pipeline. C++ AMP is a better solution for the problem.

 

[NTMBK]

Truth of the matter is OpenCL and AMD aren't really factors right now, and Knights Corner is actually the biggest threat to Nvidia.

Depends on what end you're looking at. Knights Corner is certainly the biggest threat in the high end, HPC market, no doubt about that. But OpenCL+AMD+Intel+NVidia is the sensible option for getting GPGPU into consumer-level software. (Or DirectCompute, or C++AMP. Anything portable.)

 

[Kippa]

Does anyone here even use Premiere Pro? :ninja:

I have Premiere Pro CS 5.5. I don't use it too heavily, just for putting together 3d animations into an mp4 video.

 

[wand3r3r]

Who uses CUDA? (What for?)

A real question.

 

[Jaydip]

Who uses CUDA? (What for?)

A real question.

I do use it everyday for medical imaging purposes.

 

[Kippa]

Many years ago I did programming at college and uni doing Cobol, Pascal, Assembly and Modula 2. The last time I did any coding was back in 1997, a very long long time ago. I am curious though, for programmers out there that actually program in CUDA and OpenCl, what is the difference like programming wise?

 

[Dribble]

Let me just grab some key-words for you

Now its 2013, and apparently they are "statisfied" with OpenCL (since their makeing a OpenCL version).

Last I read on this the reason Adobe introduced OpenCL was because apple went AMD, that I think was the only platform it was available for (no OpenCL on PC's) and it was used in less places then CUDA too. They are hardly the shining light of OpenCL adoption.

OpenCL did and still has the problem of no one investing serious time and money into it - sure there are *supporters* and companies that say they like it, but none put the manpower and money that nvidia put into CUDA, which is why CUDA is still light years ahead of OpenCL. That's always the problem with open standards - its often not really in one companies interest to push them hard, so they don't get the development effort, and due to the need to agree everything in with cross company committees change is slow.

So who is going to be the company to really push OpenCL:
AMD - no money to invest.
Microsoft - behind their own standard of direct compute.
Intel - too caught up in x86 and adding extensions to that which can be used to do OpenCL but aren't really pushing it.

Apple might be the best bet but they are hardly champions of all things open - if they do support OpenCL it'll be towards their own ends and might quickly end up as ClosedAppleCL.

 

[Jaydip]

Many years ago I did programming at college and uni doing Cobol, Pascal, Assembly and Modula 2. The last time I did any coding was back in 1997, a very long long time ago. I am curious though, for programmers out there that actually program in CUDA and OpenCl, what is the difference like programming wise?

OpenCL I'm not aware about but CUDA is very much like C.In fact if you know C you can leverage most of your knowledge using CUDA.You only need to learn about the programming differences between GPU and CPU.

 

[Ibra]

Who uses CUDA? (What for?)

A real question.

Nvidia's blog have plenty of examples. Or you can't visit there because they are banned you for comments section trolling? :sneaky:

 

[Arkadrel]

Or you can't visit there because they are banned you for comments section trolling? :sneaky:

Why the negative tone, the implications, and name calling?
Why do people feel the need to mock others on this forum?

Im gonna go out on a limb and guess, hes innocent of what you accuse him off, because most people usually are (and its the right thing to do, until proved otherwise).

Im guessing wand3r3r doesnt even have a account there.
Neither do I for that matter, before you jump on me.

 

[wand3r3r]

^ No I don't have an account with NV.

Nvidia's blog have plenty of examples. Or you can't visit there because they are banned you for comments section trolling? :sneaky:

You seem well versed on NV marketing, is it a coincidence?
Now go back into your cave.

I am curious about developers actually dealing with the topic, not shills opinions.

 

[NTMBK]

^ No I don't have an account with NV.



You seem well versed on NV marketing, is it a coincidence?
Now go back into your cave.

I am curious about developers actually dealing with the topic, not shills opinions.

You tell him to "go back into [his] cave"... and then make a completely baseless accusation about him being a shill. There's more than one troll in this conversation.

 

[wand3r3r]

With his post history unfortunately it might not far fetched. He's either the most deranged fan of NV, or ... well I guess I shouldn't go there.

However, if you read my post, I said not shills opinions i.e. NV's blog.

(I did tell him to go back in his cave, which is pretty mild compared to what he keeps saying around here)