NVIDIA CUDA @ Hardware.fr

Friday 10th August 2007, 04:04:00 PM, written by Rys

Hardware.fr have published an analysis of NVIDIA CUDA, looking at the architecture of the first hardware, the software side, performance versus x86 processors running the same application on GPU and CPU, the role AMD plays in the GPU computing space and more.

In terms of the CUDA programming model, Damien talks about how kernels are structured for and executed on the GPU, explaining the threads/warps/grids/blocks terminology, before talking about that how that maps to current GPUs and their SM arrangement.

PTX is touched on, Tesla's role in NVIDIA pushing into the HPC market is refreshed, and then current CUDA applications are talked about before VMD is benchmarked. VMD is an application partly developed by John Stone at UIUC, which has a CUDA path, and the article analyses performance on supporting CUDA GPUs (including a 2x 8800 Ultra + 1x 8800 GTX configuration, showing 3-way G80 scaling).

So if you're interested in the performance traits and programming of CUDA and how the programming model maps to the GPU, a highly-threaded device with lots of local memory bandwidth, we very much suggest you read. Non-French speakers might want to take a translator!

nvidia ± cuda, hardware, fr, VMD, perf, watt

CUDA 4.0 and Parallel Nsight 2.0 released
NVIDIA Fermi GPU and Architecture Analysis
NVIDIA's Parallel Nsight finally released
NVIDIA GeForce GTX 460 - GF104 breaks cover
PhysX87, ancient tragedy in 5 acts by RWT
So long, Chris, and thanks for all the fish
NVIDIA GF100 graphics architecture details
NVIDIA Fermi: new GPU architecture, starting with GF100
NVIDIA release OpenCL GPU drivers for Linux and Windows
NVIDIA GeForce GTX 275 at $250 to fight HD 4890