Cuda, le fameux langage de programmation GPGPU de nVidia débarque dans une nouvelle version 4.1. Voici l'ensemble des modifications annoncées sur le site de nVidia pour les développeurs :
Citation:
Try the new compiler!
New LLVM-based compiler delivers up to 10% faster performance for many applications
New & Improved “drop-in” acceleration with GPU-Accelerated Libraries
Over 1000 new image processing functions in the NPP library
New cuSPARSE tri-diagonal solver up to 10x faster than MKL on a 6 core CPU
New support in cuRAND for MRG32k3a and Mersenne Twister (MTGP11213) RNG algorithms
Bessel functions now supported in the CUDA standard Math library
Up to 2x faster sparse matrix vector multiply using ELL hybrid format
Learn more about all the great GPU-Accelerated Libraries
Enhanced & Redesigned Developer Tools
Redesigned Visual Profiler with automated performance analysis and expert guidance
CUDA-GDB support for debugging MPI applications, multi-context debugging, and assert() in device code
CUDA-MEMCHECK now detects out of bounds access for memory allocated in device code
Parallel Nsight 2.1 CUDA warp watch visualizes variables and expressions across an entire CUDA warp
Parallel Nsight 2.1 CUDA profiler now analyzes kernel memory activities, execution stalls and instruction throughput
Learn more about debugging and performance analysis tools for GPU developers on our CUDA Tools and Ecosystem Summary Page
Advanced Programming Features
Access to 3D surfaces and cube maps from device code
Enhanced no-copy pinning of system memory, cudaHostRegister() alignment and size restrictions removed
Peer-to-peer communication between processes
Support for resetting a GPU without rebooting the system in nvidia-smi
New & Improved SDK Code Samples
simpleP2P sample now supports peer-to-peer communication with any Fermi GPU
New grabcutNPP sample demonstrates interactive foreground extraction using iterated graph cuts
New samples showing how to implement the Horn-Schunck Method for optical flow, perform volume filtering, and read cube map texture
New LLVM-based compiler delivers up to 10% faster performance for many applications
New & Improved “drop-in” acceleration with GPU-Accelerated Libraries
Over 1000 new image processing functions in the NPP library
New cuSPARSE tri-diagonal solver up to 10x faster than MKL on a 6 core CPU
New support in cuRAND for MRG32k3a and Mersenne Twister (MTGP11213) RNG algorithms
Bessel functions now supported in the CUDA standard Math library
Up to 2x faster sparse matrix vector multiply using ELL hybrid format
Learn more about all the great GPU-Accelerated Libraries
Enhanced & Redesigned Developer Tools
Redesigned Visual Profiler with automated performance analysis and expert guidance
CUDA-GDB support for debugging MPI applications, multi-context debugging, and assert() in device code
CUDA-MEMCHECK now detects out of bounds access for memory allocated in device code
Parallel Nsight 2.1 CUDA warp watch visualizes variables and expressions across an entire CUDA warp
Parallel Nsight 2.1 CUDA profiler now analyzes kernel memory activities, execution stalls and instruction throughput
Learn more about debugging and performance analysis tools for GPU developers on our CUDA Tools and Ecosystem Summary Page
Advanced Programming Features
Access to 3D surfaces and cube maps from device code
Enhanced no-copy pinning of system memory, cudaHostRegister() alignment and size restrictions removed
Peer-to-peer communication between processes
Support for resetting a GPU without rebooting the system in nvidia-smi
New & Improved SDK Code Samples
simpleP2P sample now supports peer-to-peer communication with any Fermi GPU
New grabcutNPP sample demonstrates interactive foreground extraction using iterated graph cuts
New samples showing how to implement the Horn-Schunck Method for optical flow, perform volume filtering, and read cube map texture
Pour résumer :
- Un nouveau compilateur qui offrira un gain de performances de 10% sur une grande partie des applications juste en recompilant les applications existantes.
- Des améliorations / simplifications de fonctions mathématiques et de traitement d'image.
- Des améliorations pour l'intégration de Cuda aux outils de développement. De nouveaux outils d'aide au débogage et optimisation.
- Quelques nouvelles fonctionnalités dans l'API, la capacité de faire communiquer les GPU en mode "peer to peer"
- De nouveaux exemples de code fournis pour implémenter les nouveautés.
Nous autres plieurs allons surtout retenir le potentiel gain de 10% si Stanford met à jour ses cores avec cette version et espérons que les nouvelles fonctionnalités pourront aider à l’amélioration des cores à l'avenir, même si pourtant Cuda devait progressivement être remplacé par OpenCL...
Source : PCInpact
Source complémentaire : Site des développeurs nVidia
Adanorm
Le: 10/11/11
