The CUDA Toolkit and SDK v2.2 is now released and available to all developers.
A brief overview of features–there are a lot:
- Officially adds support for Windows 7, Server 2003, Server 2008, Ubuntu 8.10, RHEL 5.3, and Fedora 10
- Includes cuda-gdb (hardware GPU debugger) for RHEL5 32 and 64-bit (officially supported and tested), but it may work on more platforms than just those
- Exclusive device mode in Linux: set some GPUs as exclusive-compute (can only own a single CUDA context) and some as non-compute (no CUDA contexts allowed) for easier management of clusters/MPI applications. See the manpage for nvidia-smi for how to set this and cudaSetValidDevices in the reference manual on how to best use this from CUDART.
- Zero-copy support: transparently read from certain system memory from a kernel on GT200 or MCP79 systems. See this post for more information on how it works.
- Asynchronous memcpy support on Vista/Server 2008/Win7
- Texture from pitchlinear memory: use this to avoid an additional memcpy at times in some scenarios.
4GB of pinned memory in a single allocation on most OSes
- maximum pinned memory per allocation increased in Vista to ~1.5GB
- pinned memory can be shared between contexts
- Multi-device OpenGL interop performance between a Quadro display card and a separate compute card is dramatically improved.
- Visual Profiler works on Vista
- Visual Profiler supports additional counters for GT200 to measure number of memory transactions of a given size, instruction throughput, etc.
- Blocking sync support for all platforms: allows the host thread to sleep and be awoken by driver when the GPU operation the host thread is waiting on is completed.
- Quite a few additional math functions added due to forum requests (feel free to keep posting requests, we do pay attention)
- __threadfence(): ensure that a thread’s pending memory writes are visible to all threads before continuing. It is explicitly not a global sync, unlike how it appears to some.
- Lots of bugfixes, of course; most importantly, killing a CUDA app should behave much, much better than it ever has before, especially when you’re on a dedicated compute card
185.85 for WinXP 32 Desktops/Workstations
185.85 for WinXP 32 Notebooks
185.85 for WinXP 64 Desktops/Workstations
185.85 for WinXP 64 Notebooks
185.85 for Vista 32 Desktops/Workstations
185.85 for Vista 32 Notebooks
185.85 for Vista 64 Desktops/Workstations
185.85 for Vista 64 Notebooks
185.85 for Win7 32 Desktops/Workstations
185.85 for Win7 32 Notebooks
185.85 for Win7 64 Desktops/Workstations
185.85 for Win7 64 Notebooks
CUDA Toolkit 2.2 for Fedora 10 32-bit
CUDA Toolkit 2.2 for Fedora 9 32-bit
CUDA Toolkit 2.2 for RHEL4.7 32-bit
CUDA Toolkit 2.2 for RHEL 5.3 32-bit
CUDA Toolkit 2.2 for SLED 10 SP2 32-bit
CUDA Toolkit 2.2 for SUSE 11 32-bit
CUDA Toolkit 2.2 for SUSE 11.1 32-bit
CUDA Toolkit 2.2 for Ubuntu 8.04 32-bit
CUDA Toolkit 2.2 for 32-bit Ubuntu 8.10
CUDA Toolkit 2.2 for Fedora 10 64-bit
CUDA Toolkit 2.2 for Fedora 9 64-bit
CUDA Toolkit 2.2 for RHEL 4.7 64-bit
CUDA Toolkit 2.2 for RHEL 5.3 64-bit
CUDA Toolkit 2.2 for SLED 10 SP2 64-bit
CUDA Toolkit 2.2 for SUSE 11 64-bit
CUDA Toolkit 2.2 for SUSE 11.1 64-bit
CUDA Toolkit 2.2 for Ubuntu 8.04 64-bit
CUDA Toolkit 2.2 for Ubuntu 8.10 64-bit
Debugger and Profiler
(Note that the Visual Profiler is included with the toolkit on all platforms except Mac OS.)
CUDA-GDB for RHEL 5.3 32-bit
CUDA-GDB for RHEL 5.3 64-bit
CUDA Visual Profiler for Mac OS 10.5
CUDA-GDB 2.2 User’s Manual
CUDA 2.2: Getting Started in Linux
CUDA 2.2: Getting Started in Mac OS
CUDA 2.2: Getting Started in Windows
CUDA 2.2 API Reference Manual (now with links!)
CUDA 2.2 Programming Guide
CUDA Visual Profiler 1.2 Readme
CUDA Visual Profiler EULA
CUDA Toolkit 2.2 EULA
CUDA 2.2 Release Notes for Linux
CUDA 2.2 Release Notes for Mac OS
CUDA 2.2 Release Notes for Windows