Re: Problem with hitting breakpoints
Hi Sten, Sorry for the long time in replying.We have finally managed to reproduce the issue in our labs.We are currently looking into its cause and seeing if we can fix it or find a workaround. Thanks...
View Articleиспользование инструкций интел
при использовании SSE1,2,3,4, и другихинструкций интел происходит замедлениеисполнения нужно использовать дешифраторинструкций с заменой на свои для AMD Use instructions intel using SSE1, 2,3,4, and...
View ArticleRe: Managing two r290 - ubuntu 12.04 - driver problem?
maybe relevant? http://devgurus.amd.com/thread/168198
View ArticleGrafikkarten Probem Mit dem MSI GX70 3be Gaminglaptop
Hallo zusammen ich habe mir den MSI GX70 Laptop geholt mit der AMD Radeon HD 8970M Grafikkarte.Ich habe mir die Windows 8 Testversion installiert und auch alle Treiber installiert, doch ich habe bei...
View ArticleRe: Re: [Bug Report] GLSL shader source not entirely displayed
I use this code now, and it is still running perfectly with Catalyst 13.4. But I have still the same problem with CodeXL. Both provided strings are null terminated. GLchar* profileLookupTable[] = {...
View ArticleGPU driver startup overhead
I'm working on a software stack for fast and portable (both in terms of platform support and ease of copying around) software that is partially based on hardware-accelerated 3D APIs. Today I decided to...
View ArticleClik here to view.
Re: Managing two r290 - ubuntu 12.04 - driver problem?
Not for me, I have no onboard graphics. My motherboard is a Gigabyte 990FXA-UD3. From memory, I'm using PCI-E slots 1 and 4, both are 16x slots.
View ArticleRe: ADL_Workstation_GLSyncSupportedTopology_Get, missing documentation
Ok, Do you have an example?
View ArticleRe: Re: Re: [Bug Report] GLSL shader source not entirely displayed
Hi Ziple,Well, trying out the code you pasted, it seems that there's a bug in CodeXL in addition to what I mentioned. This bug will be fixed in the next version of CodeXL. In the meantime, you can work...
View ArticleRe: Problem with bit operations with openCL
Exactly. it was never fixed. I'm glad that after more than half a year, we are back to where we started. This sounds like progress, AMD style.
View ArticleRe: Managing two r290 - ubuntu 12.04 - driver problem?
I don't know if is connected, any way I add another detail.When I use the gpu for computational purpose the temperature raise 93° as expected but the fan speed does not grow (I tried uber and quite...
View ArticleRe: Performance penalty of mixing vector and scalar computation on GCN
On GCN the physical vector type is not int4, it's int64 Scalar instructions aren't working on individual vector elements, they have a separate 64 bit register space on which they work separated from...
View ArticleRe: Kaveri APUs and fp64 GPU support?
moozoo wrote: AnandTech Portal | Floating point peak performance of Kaveri and other recent AMD and Intel chipsI would like to point out that the author of article, Rahul Garg, is also the original...
View ArticleRe: Suggest Feature you want in AMD APP
Meteorhead wrote: CUDA is dying, but all too slowly. Hoping they get their act together on the OpenCL front however.CUDA won't die just like Fortran won't die. Some organizations refuse to rewrite...
View ArticleRe: CodeXL 1.3 installation on Ubuntu 12.04 LTS
Can you also look at the /tmp directory and send attach the log file if one was created. it will be named "CodeXL-username.log", Anyways Valentines Day is Coming I wish All Very Happy Valentines Day 2014.
View ArticleClik here to view.
CL_DEVICE_ADDRESS_BITS returns 32 for a 6GB Radeon 7970
I have a Radeon 7970 and need more than 4GB. Currently, querying the device for adddress bits returns 32. I assume this is why I can't fully use my 6GB . Is it possible to use more than 4GB like CUDA?...
View ArticleRe: Suggest Feature you want in AMD APP
I think it is much more difficult to support so many different types of hardware and perhaps that is why nvidia is able to produce a more consistent feature set. Maybe OpenCL is not 100% open etc....
View ArticleRe: Performance penalty of mixing vector and scalar computation on GCN
"There are instructions to extracts a specific element from a vector register into a scalar reg: v_readlane_b32, v_readfirstlane_b32. They eat 1 cycle." That's exactly what I've been looking for, thank...
View ArticleRe: R290 linux - Two stage reduction - Low performance
Hi, What happens when you keep the vector size 10000000? Do you get similar performance as mentioned in the article?What if you change the work group size from 64 to 256? You may need to take care of...
View Article