CUDA Developer Drivers 4.1.21 RC 2

Easier Application Porting<br /><br /> Share GPUs across multiple threads<br /> Use all GPUs in the system concurrently from a single host thread<br /> No-copy pinning of system memory, a faster alternative to cudaMallocHost()<br /> C++ new/delete and support for virtual functions<br /> Support for inline PTX assembly<br /> Thrust library of templated performance primitives such as sort, reduce, etc.<br /> NVIDIA Performance Primitives (NPP) library for image/video processing<br /> Layered Textures for working with same size/format textures at larger sizes and higher performance<br /><br />Faster Multi-GPU Programming<br /><br /> Unified Virtual Addressing<br /> GPUDirect v2.0 support for Peer-to-Peer Communication<br /><br />New & Improved Developer Tools<br /><br /> Automated Performance Analysis in Visual Profiler<br /> C++ debugging in CUDA-GDB for Linux and MacOS<br /> GPU binary disassembler for Fermi architecture (cuobjdump)<br /> Parallel Nsight 2.0 now available for Windows developers with new debugging and profiling features.<br />

Requirements:	No special requirements
Platforms:	Mac, Mac OS X v10.6
Keyword:	Addressing,  Automated,  Communication,  Developer,  Gpudirect,  Improved,  Peertopeer,  Tools,  Unified,  Virtual
Users rating:	0/10