From 9bef70349e8743af78ff74b5f10e2d05468320e4 Mon Sep 17 00:00:00 2001 From: Tim Besard Date: Wed, 18 Sep 2024 16:20:48 +0200 Subject: [PATCH 1/4] Remove unneeded external tags. --- post/2023-09-19-cuda_5.0.md | 1 - post/2023-11-07-cuda_5.1.md | 1 - post/2024-04-26-cuda_5.2_5.3.md | 1 - post/2024-05-24-oneapi_1.5.md | 1 - post/2024-05-28-cuda_5.4.md | 1 - 5 files changed, 5 deletions(-) diff --git a/post/2023-09-19-cuda_5.0.md b/post/2023-09-19-cuda_5.0.md index 0636293..7bf4787 100644 --- a/post/2023-09-19-cuda_5.0.md +++ b/post/2023-09-19-cuda_5.0.md @@ -1,7 +1,6 @@ +++ title = "CUDA.jl 5.0: Integrated profiler and task synchronization changes" author = "Tim Besard" -external = true abstract = """ CUDA.jl 5.0 is an major release that adds an integrated profiler to CUDA.jl, and reworks how tasks are synchronized. The release is slightly breaking, as it changes how local diff --git a/post/2023-11-07-cuda_5.1.md b/post/2023-11-07-cuda_5.1.md index 8df48de..8ee789f 100644 --- a/post/2023-11-07-cuda_5.1.md +++ b/post/2023-11-07-cuda_5.1.md @@ -1,7 +1,6 @@ +++ title = "CUDA.jl 5.1: Unified memory and cooperative groups" author = "Tim Besard" -external = true abstract = """ CUDA.jl 5.1 greatly improves the support of two important parts of the CUDA toolkit: unified memory, for accessing GPU memory on the CPU and vice-versa, and cooperative diff --git a/post/2024-04-26-cuda_5.2_5.3.md b/post/2024-04-26-cuda_5.2_5.3.md index bf020be..9bf1fee 100644 --- a/post/2024-04-26-cuda_5.2_5.3.md +++ b/post/2024-04-26-cuda_5.2_5.3.md @@ -1,7 +1,6 @@ +++ title = "CUDA.jl 5.2 and 5.3: Maintenance releases" author = "Tim Besard" -external = true abstract = """ CUDA.jl 5.2 and 5.3 are two minor release of CUDA.jl that mostly focus on bug fixes and minor improvements, but also come with a number of interesting new diff --git a/post/2024-05-24-oneapi_1.5.md b/post/2024-05-24-oneapi_1.5.md index ce58502..f293a27 100644 --- a/post/2024-05-24-oneapi_1.5.md +++ b/post/2024-05-24-oneapi_1.5.md @@ -1,7 +1,6 @@ +++ title = "oneAPI.jl 1.5: Ponte Vecchio support and oneMKL improvements" author = "Tim Besard" -external = true abstract = """ oneAPI.jl v1.5 is a significant release that brings many new features, from extended hardware support to greatly improved wrappers of the oneMLK math library.""" diff --git a/post/2024-05-28-cuda_5.4.md b/post/2024-05-28-cuda_5.4.md index c79f868..b13a6b4 100644 --- a/post/2024-05-28-cuda_5.4.md +++ b/post/2024-05-28-cuda_5.4.md @@ -1,7 +1,6 @@ +++ title = "CUDA.jl 5.4: Memory management mayhem" author = "Tim Besard" -external = true abstract = """ CUDA.jl 5.4 comes with many memory-management related changes that should improve performance of memory-heavy applications, and make it easier to work with heterogeneous From 09ab63fbdf075864b98a617674818cc6c8fab60b Mon Sep 17 00:00:00 2001 From: Tim Besard Date: Wed, 18 Sep 2024 16:20:58 +0200 Subject: [PATCH 2/4] Add release notes for CUDA 5.5. --- post/2024-09-18-cuda_5.5.md | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) create mode 100644 post/2024-09-18-cuda_5.5.md diff --git a/post/2024-09-18-cuda_5.5.md b/post/2024-09-18-cuda_5.5.md new file mode 100644 index 0000000..70450ad --- /dev/null +++ b/post/2024-09-18-cuda_5.5.md @@ -0,0 +1,33 @@ ++++ +title = "CUDA.jl 5.5: Maintenance release" +author = "Tim Besard" +abstract = """ + CUDA.jl 5.5 is a minor release that comes with a couple of small improvements + and new features.""" ++++ + +{{abstract}} + +The only important change is that the minimal required Julia version has been +bumped to 1.10, in anticipation of it becoming the next LTS release. + + +## New features + +- Support for the upcoming Julia 1.11 release has been added, as well as for + [CUDA 12.6 (Update 1)](https://github.com/JuliaGPU/CUDA.jl/pull/2461). +- Launch overhead has been reduced by [avoiding double argument + conversions](https://github.com/JuliaGPU/CUDA.jl/pull/2472). Note that this + does not apply to kernels that are obtained using `@cuda launch=false`. +- CUSOLVER's dense wrappers have been improved by [Ben + Arthur](https://github.com/bjarthur), [now caching workspace + buffers](https://github.com/JuliaGPU/CUDA.jl/pull/2465). This should greatly + reduce the number of allocations needed for repeated calls. +- [Alexis Montoison](https://github.com/amontoison) has improved the CUSPARSE + wrappers, adding [conversions between sparse vectors and sparse + matrices](https://github.com/JuliaGPU/CUDA.jl/pull/2489) that enable [a + version of `gemv`](https://github.com/JuliaGPU/CUDA.jl/pull/2488) which + preserves sparsity of the inputs. +- CUDA.jl's CUFFT wrappers [now support + `Float16`](https://github.com/JuliaGPU/CUDA.jl/pull/2430), thanks to [Erik + Schnetter](https://github.com/eschnett). From 7c3094189162f332b418fc54e1b882731a217c68 Mon Sep 17 00:00:00 2001 From: Tim Besard Date: Wed, 18 Sep 2024 16:22:28 +0200 Subject: [PATCH 3/4] Update note on OpenCL.jl. --- other.md | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/other.md b/other.md index 7dde54a..f12b208 100644 --- a/other.md +++ b/other.md @@ -9,14 +9,23 @@ support as the NVIDIA and AMD back-ends. ## OpenCL -Programming OpenCL GPUs in Julia is much more limited than other supported platforms. On recent versions of Julia, only [OpenCL.jl](https://github.com/JuliaGPU/OpenCL.jl) is available. This package can be used to compile and execute GPU kernels written in OpenCL C. - +Support for OpenCL is available through the +[OpenCL.jl](https://github.com/JuliaGPU/OpenCL.jl) package, which is currently +undergoing a rewrite to enable native support for Julia. ## ArrayFire ArrayFire is a general-purpose software library that targets CPUs, GPUs, and other -accelerator hardware. The [ArrayFire.jl](https://github.com/JuliaGPU/ArrayFire.jl) package provides a Julia interface to this library, and makes it possible to program accelerators using an array abstraction built on the ArrayFire library. +accelerator hardware. The +[ArrayFire.jl](https://github.com/JuliaGPU/ArrayFire.jl) package provides a +Julia interface to this library, and makes it possible to program accelerators +using an array abstraction built on the ArrayFire library. ## SX-Aurora -The [NEC SX-Aurora Tsubasa](https://www.nec.com/en/global/solutions/hpc/sx/index.html) is a PCIe card which works as a Vector Computer. It can be programmed from Julia using the [VectorEngine.jl](https://github.com/sx-aurora-dev/VectorEngine.jl) package, which at the moment requires a custom Julia build using a LLVM fork. Support is expected to improve due NECs involement. +The [NEC SX-Aurora +Tsubasa](https://www.nec.com/en/global/solutions/hpc/sx/index.html) is a PCIe +card which works as a Vector Computer. It can be programmed from Julia using the +[VectorEngine.jl](https://github.com/sx-aurora-dev/VectorEngine.jl) package, +which at the moment requires a custom Julia build using a LLVM fork. Support is +expected to improve due NECs involement. From eb9ff54e7238e47d4b9d7c345a4b5be99357c95e Mon Sep 17 00:00:00 2001 From: Tim Besard Date: Wed, 18 Sep 2024 16:25:41 +0200 Subject: [PATCH 4/4] Mention the office hours. --- index.md | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) diff --git a/index.md b/index.md index c7f6cc9..af21349 100644 --- a/index.md +++ b/index.md @@ -59,8 +59,14 @@ activities, we would be grateful if you could cite our work: ## Community -If you need help, or have questions about GPU programming in Julia, you can find members of -the community at: - -- Julia Discourse, with a dedicated [GPU section](https://discourse.julialang.org/c/domain/gpu/11) -- Julia Slack ([register here](https://slackinvite.julialang.org/)), on the [#gpu channel](https://julialang.slack.com/messages/C689Y34LE/) \ No newline at end of file +If you need help, or have questions about GPU programming in Julia, you can find +members of the community at: + +- Julia Discourse, with a dedicated [GPU + section](https://discourse.julialang.org/c/domain/gpu/11) +- Julia Slack ([register here](https://slackinvite.julialang.org/)), on the + [#gpu channel](https://julialang.slack.com/messages/C689Y34LE/) +- [JuliaGPU office + hours](https://mit.zoom.us/j/92410737378?pwd=MWkrSkIzMEtRWkZHdm1XQmpNbm90UT09), + every other week at 2PM CET (check the [Julia community + calendar](https://julialang.org/community/#events) for more details).