On top of the prior SLS mitigation options for LLVM (the initial hardening pass - AArch
LLVM Adds Additional Protections For Arm's SLS Speculation Vulnerability Mitigation
On top of the prior SLS mitigation options for LLVM (the initial hardening pass - AArch
POCL 1.6 Released For Portable OpenCL Atop CPUs, Other Accelerators
POCL 1.6 is out as the latest feature release and continues providing OpenCL 1.2 support and a subset of OpenCL 2.0 functionality. POCL is most well known for OpenCL on CPUs but thanks to LLVM also allows targeting NVIDIA GPUs with CUDA, AMD GPUs with HSA, and other possible accelerator targets. POCL makes use
OpenBLAS 0.3.13 Released With A RISC-V Port, POWER10 Optimizations
With OpenBLAS 0.3.13 some of the release highlights include:
- Detection for the Fujitsu Fortran compiler.
- A RISC-V port for the C910V has been added. This initial OpenBLAS RISC-V target is catering to the RISC-V Vector Extension 0.7.1. This is the first RISC-V port by OpenBLAS.
AMD AOMP 11.12 Released For OpenMP Offloading To Radeon GPUs
AOMP continues maturing as the company's downstream of LLVM Clang that allows OpenMP offloading to Radeon hardware. AMD has been working to upstream their Radeon OMP patches into LLVM, but at least until that's all perfect, AOMP
Classic OSMesa Retires In Mesa 21.0 As The Worst Of The Software Rendering Paths
Those wanting Mesa software rendering in 2020 and beyond should really be using LLVMpipe or otherwise Softpipe should LLVM not be available for your software/hardware platform. LLVMpipe offers much better performance not to mention OpenGL 4.6 and is actually maintained. With classic OSMesa code just rotting and being of minimal use these days for off-screen rendering, the classic code
Intel AMX Programming Model Lands In LLVM Compiler
AMX is Intel's new programming paradigm with a focus on better AI performance both for training and inference. AMX is built around the concept of "tiles" as a set of two-
Intel Opens Up "IMF LA" As A GPU Compute Speed Boost To Better Compete With Windows
Released today was IGC 1.0.5761 . This routine update to the Intel Graphics Compiler has a number of low-level compiler additions and other changes as usual. All quite low level but then
POCL 1.6-RC1 Released With Better CUDA Performance
On Wednesday marked the release of POCL 1.6-RC1 as the test release for the next update to the Portable Computing Language.
POCL continues leveraging LLVM/Clang for doing much of the heavy lifting and supporting a diverse variety of back-ends for CPU execution
Clang LTO Support For The Linux Kernel Spun Up A Seventh Time
Google's team has done a good job not only working on the mainline Clang support for the Linux kernel across the likes of AArch64 and x86_64, but also with other related features of interest to them like the Clang LTO abilities to which internally they already leverage extensively. This upstreaming work has been ongoing
(PR) LLNL's New 'Ruby' Supercomputer Taps Intel for COVID-19 Research
Intel today announced that Lawrence Livermore National Laboratory (LLNL) will leverage Intel Xeon Scalable processors in "Ruby," its latest high performance computing cluster. The Ruby system will be used for unclassified programmatic work in support of the National Nuclear Security Administration's (NNSA) stockpile stewardship mission, for researching therapeutic drugs and designer antibodies against SARS-CoV-2, the virus that causes COVID-19, and for other open science work at LLNL.
Ruby was built in collaboration with Intel, LLNL, Supermicro and
