CUDA kompiler pro AMD GPU

[del42sa]

CUDA kompiler pro AMD GPU .

AMD support CUDA ( Jde vlastně o překladač umožňující AMD GPU pracovat s CUDA )

http://www.anandtech.com/show/9792/amd- ... r-amd-gpus

Today at Super Computing 15 (SC15) AMD is launching the Boltzmann Initiative and introducing a new HPC focused driver, a new compiler, and a runtime to translate CUDA code into something that can be compiled using AMD’s new compiler to run on AMD’s hardware. Those of you who are familiar with Ludwig Boltzmann may recall his contributions to Kinetic theory and development of Statistical Mechanics. AMD has named its new HPC initiative after him to highlight how his work is still being used in neural networking and Navier-Stokes algorithms on GPUs.

http://semiaccurate.com/2015/11/16/amd- ... on-in-hpc/

[Krteq]

Kurňa jsi mi to sebral. Tohle je asi nejzásadnější novinka v GPGPU odvětví za poslední léta!

Aneb další potvrzení ultimátní univerzálnosti GCN architektury

[PET5]

Teď ovšem půjde o to, jestli bude rychlejší výpočet za pomoci kompilátoru cuda, nebo přímo pod opencl.

[Krteq]

Pravda, ale až to příjde tak budou moci FirePro karty zpracovávat jak OpenCL kód, tak i kód psaný v CUDA

[ifkopifko]

del42sa> To čo cituješ (semiaccurate) znie síce fantasticky, ale na príslušnom slide tam je jedno zrejme dôležité slovíčko:

HiPify tools simplify porting from CUDA to HIP.

Takže by som krotil nadšenie, ono to tak ružové/jednoduché byť nemusí. Že existuje nejaký nástroj na uľahčenie portu ešte nemusí byť to isté, ako prostý automatický prevod.

Plus, keďže potrebuješ zdrojový kód, tak všetko stojí a padá na ochote developerov vydať aj AMD verziu svojho CUDA programu. Pokojne môžu mať aj nejakú tú motiváciu to nespraviť...

[Krteq]

To je ale zmínka jen o toolech (HiPify tools), celá infrastruktura by měla být dost robustní na to aby zvládala zpracovávat/překládat kód psaný v CUDA



BTW, nespekulovalo se o tomhle ve spojitosti s Huddym už někdy kolem roku 2008?

[ifkopifko]

Eh? To si si podľa mňa niečo zle vyložil. HiPify tools je jediná cesta ako preložiť zdrojový CUDA kód do HIP zdrojového kódu (okrem manuálneho prepísania samozrejme).

[Krteq]

Však já mluvil o celém HiP.

[oneb1t]

momentalne pisu openCL apku do skoly a jeslti bude HiP fungovat tak spatne jako openCL tak se na to muzou vyprdnout (je to probugovany a naladovy jak svina + malo funkcnich tutorialu)

[del42sa]

Now there are some unknowns here, including whether AMD can keep HIP up to date with CUDA feature additions, but more importantly there’s a question of just what NVIDIA’s reaction will be. CUDA is NVIDIA’s, through and through, and it does make one wonder whether NVIDIA would try to sue AMD for implementing the CUDA API without NVIDIA’s permission, particularly in light of the latest developments in the Oracle vs. Google case on the Java API. AMD for their part has had their legal team look at the issue extensively and doesn’t believe they’re at risk – pointing in part to Google’s own efforts to bring CUDA support to LLVM with GPUCC – though I suspect AMD’s efforts are a bit more inflammatory given the direct competition. Ultimately it’s a matter that will be handled by AMD and NVIDIA only if it comes to it, but it’s something that does need to be pointed out.

žaloba ? to snad ne ?

with the Boltzmann Initiative AMD is taking an important and very much necessary step to redefine themselves in the HPC space. By providing an improved driver layer for Linux supporting headless operation and a unified memory space, with a compiler for direct, single source C++ compilation on top of that, and a CUDA compatibility layer to reach the established CUDA user base, AMD Is finally getting far more aggressive on the HPC side of matters, and making the moves that many have argued they have needed to make for quite some time

[havli]

GPGPU obecne je tezka uzkoprofilovka, na normalni aplikace nepouzitelna vec.... AMD opet jen vyhodilo par slidu a tim to konci. Realna implementace bud nebude nikdy, nebo casem mozna jo, ale stejne nebude poradne fungovat. Vsak takovych "technologii" uz bylo.

Ted akorat zalezi, kolik penez do toho nasypou... Jiste by se nasla spousta jinych dulezitejsich oblasti, kam by se spis vyplatilo investovat.

[oneb1t]

ono je to nepouzitelny ale hlavne proto ze tomu AMD nedela skoro zadnou podporu...
zkus si rozbehat openCL hello world a hned pochopis proc to nikdo nepouziva i kdyby chtel

cuda je na tom trosku lip ale taky zadna slava

[ifkopifko]

To je ale zmínka jen o toolech (HiPify tools), celá infrastruktura by měla být dost robustní na to aby zvládala zpracovávat/překládat kód psaný v CUDA

Však já mluvil o celém HiP.

A čo presne si ty predstavuješ pod tým HIP-op? Aká "infraštruktúra" je zodpovedná za "spracovávanie/prekladanie CUDA kódu", okrem HiPify tools?

Mne to príde tak, že HIP je vlastne len programovací jazyk, ktorý má byť údajne kompatibilný s HCC a NVCC kompilátormi.

[blaho]

to uz by asi chcelo panove topic : CUDA kompiler pro AMD
aby to bolo nejak pokope
* toto tu je info topic

[del42sa]

upřesnění:

To bring applications written for CUDA onto AMD platforms, AMD's new HIP tool allows over 90 percent of CUDA code to be automatically converted into C++, the final 10 percent being manually converted in C++. This capability expands the installed hardware base available to run what were formerly exclusively CUDA-based applications.

http://www.eetimes.com/document.asp?doc_id=1328279&

We are not supporting CUDA. What we are providing is a standard C++ compiler for GPU/CPU plus tools that simplify porting CUDA apps to use it.

http://www.phoronix.com/forums/forum/ph ... gpus/page2

No change in direction other than adding some porting tools to make it easier for existing HPC users to run on our HW. This is the latest version of the HSA stack, and HCC is the latest version of the C++-over-HSA compiler. It just gets more interesting when you run it on a box of dGPUs instead of a single APU.

HSA isn't a programming language on its own (other than the HSAIL IR), it's a compute stack that supports a variety of standard languages and tools.

https://community.amd.com/community/amd ... -computing