I want to accelerate my programme. I have Radion RX 560 device. And see https://gpuopen.com/compute-product/opencl-sdk/ , this page showes there is a git hub:(https://github.com/GPUOpen-LibrariesAndSDKs/OCL-SDK),
but open that git hub URL there is no SDK. Only a md file.
By searching in Internet guys says developers should instal SDK like "AMDAPPSDK3.0" but I cannot find it in amd website. does it oboseleted? but what is the new toolkit to replace that ??
thanks.
Read this thread
Click on "Releases"...
then you can download the files...
Look at the release date... 27 June 2017.
I don't use OCL_SDK_Light. I use the openCL development libraries from the package manager under OpenSUSE Leap 15.1. No need to install the AMD APP SDK 3.0 or OCL_SDK_Light.
Okay the release is in the year 2017, but this doesn't mean that it's bad.
Hi
In my opinion It is dead and unsupported and left to "open source" i.e. people working for free in their spare time.
If you really want a hard time you could try using ROCm on Ubuntu Linux but that is not working properly either yet.
AMD compute platform for new users is a nightmare.
You will spend more time dealing with driver bugs, install issues, compiling your own GUI and crashes than actually producing any code.
Forget it if you are new to parallel compute and want to learn something that is supported properly then buy an Nvidia GPU and learn CUDA.
There are pleanty of free and supported resources on the Nvidia Developer Zone.
Bye.
If you do want to use OpenCL a point to note is newer Nvidia GPUs outperform pretty much all AMD cards in OpenCL benchmarks now. OpenCL support has been improved. They also have a website for people learning to program:
OpenCL | NVIDIA Developer
I stay at AMD. No good experience with Nvidia cards.
I guess you haven't used them then.
Not for open computing. That's right. But for graphics.
Currently I have one Notebook (Lenovo ThinkPad T460p) with a NVIDIA GPU. -> Graphical artefacts in some programs. I have also replaced the Mainboard. -> Same issue.
The programs where the artefacts exists with the NVIDIA GPU (940MX):
- Visual Studio
- Android Studio
I think I schould post this issue in the NVIDA forum. Already posted this in the Lenovo and Visual Studio MSDN forum.
But we are here in the AMD forum...
Hi,
I own an ASUS laptop ROG G751JL | ROG - Republic Of Gamers | ASUS USA with an Nvidia 965M GPU.
The GPU on the laptop died after two days of gaming with it.
I RMA'ed the laptop and the replacement has been fine for years so far.
Looking at the levono ThinkPad T440p it seems the CPU and GPU share the same cooling fan:
What are the CPU and GPU temps like when you see artifacting?
Are you sure you are not overclocking the GPU VRAM memory?
It may just be an Nvidia Driver issue though.
In any case - yes you should raise the problem on the Nvidia / Levono forum.
Why don't you post your question at a better AMD Forum for your type of question. AMD has its own OpenCL Forum from here: OpenCL.
Moderator dipak can answer your question plus White list you if applicable to post at OpenCL forum.
I have read that AMD SDK 3.0 has been replaced by that Github version.
This thread was moved to OpenCL Forum by Moderator Dipak. It has the links for AMD APP SDK 3.0 in it:where can i find standalone version of OpenCL AMD APP SDK 3.0 for windows
i found the offline standalone installers for APP SDK 3
Full Installers:
Web Installers:
The AMD APP SDK was dropped and the AMD OpenCL Compute pages were all taken down.
You have to use things like the the Internet WayBack Machine to even get any documentation.
All info needed is in this thread: https://community.amd.com/thread/228114
This is where I downloaded it from a year ago: Download AMD Accelerated Parallel Processing SDK 3.0.130.135
Norton Security still says the download is OK, but if you do trust that link to download it run your own antivirus checker on it.
Good luck.
Hi all:
So kind all of you, your suggestion are very useful for me. Now I downloaded sdk and CodeXL and learning them.
hope I can do it.
Thanks again!
I have used OpenMP and leveraged CPU cores efficiently, GPU cores are harder to use.