Open Source LLM

High Performance Mold Linker Begins Adding macOS, CMake Build Support

Last year marked the release of Mold 1.0 as a modern, high-speed linker alternative to the likes of GNU Gold and LLVM LLD. Mold was started by Rui Ueyama who previously worked on the LLVM linker. Mold has continued seeing new features added and out today is version 1.4.1 where "alpha" support for macOS is added and also beginning to work on CMake build system integration.

While just a point release, Mold 1.4.1 is notable in that it has alpha-

Open Source LLM

LZ4 v1.9.4 Achieves 20~70% Speedups For Some CPUs & Configurations

LZ4 v1.9.4 is out today as the first point release in nearly two years for this BSD-licensed, speedy, lossless compression algorithm.

Yann Collet announced the release today of LZ v1.9.4 as while it's a "maintenance" release, there are around 350 patches and a number of improvements -- including performance optimizations. LZ4 v1.9.4 brings a ~20% speedup for data decompression on ARM64 platforms. The

Open Source LLM

MGLRU v14 Released For Improving Linux Low-Memory Performance

While Linux 6.0 will bring a lot of shiny new features , Multi-Gen LRU (MGLRU) is one of the anticipated changes that isn't going to land now until Linux 6.1. But in the interim, MGLRU v14 was posted today that re-bases the code against 6.0-rc1 to help facilitate more testing of this kernel change that primarily helps Linux systems under memory pressure.

MGLRU has been performing very well in my own testing and the performance benchmarks are looking

Open Source LLM

Classroom activities to discuss machine learning accuracy and ethics | Hello World #18

In Hello World issue 18, available as a free PDF download, teacher Michael Jones shares how to use Teachable Machine with learners aged 13–14 in your classroom to investigate issues of accuracy and ethics in machine learning models. Machine learning: Accuracy and ethics The landscape for working with machine learning/AI/deep learning has grown considerably over…

The post Classroom activities to discuss machine learning accuracy and ethics | Hello World #18 appeared first on Raspberry Pi .


Open Source LLM|Intel

OpenBLAS 0.3.21 Brings Support For Newer Arm CPUs, More Optimizations

OpenBLAS as the high performance, open-source BLAS / LAPACK implementation debuted a new version on Sunday with more CPU optimizations and expanded processor coverage.

OpenBLAS 0.3.21 features support for more CPUs, particularly on the Arm side. There is also support for more compilers, build system enhancements, and more. The OpenBLAS 0.3.21 release highlights include:

- Support for building OpenBLAS with the Intel IFX, Fujitsu FCC, and Cray C/Fortran compilers are now supported.

- Initial support
Continue reading...

Open Source LLM

MGLRU & Maple Tree Miss Out On Linux 6.0 But Will Aim For Linux 6.1

A lot of great features are landing for Linux 6.0 but two more are now confirmed to partake in this next major kernel version: Multi-Gen LRU (MGLRU) and the Maple Leaf data structure.

We have been looking quite forward to MGLRU being mainlined since it's already widely-used by Google with Chrome OS and Android with showing very promising performance results. MGLRU overhauls the handling for Linux's page reclamation code and makes it much faster. The benchmark results are looking great and there was hope

Open Source LLM

OpenMandriva Lx ROME Technical Preview Released For This Rolling-Release Linux Distro

The OpenMandriva crew behind this long ago Mandriva/Mandrake-derived distribution announced their Lx ROME technical preview release as their rolling release version of the operating system.

The OpenMandriva Lx ROME technical preview was announced today for those wanting to test this rolling release version of OpenMandriva. The rolling release flow of new packages though has been temporarily stalled while OpenMandriva developers made more invasive toolchain and system packages to OpenMandriva.

At the moment the OpenMandriva Lx ROME state consists of a Clang-built Linux 5.18 kernel, Python 3.

Open Source LLM

Oracle's GraalVM 22.2 Brings More Performance Optimizations

Oracle has issued its newest quarterly release of GraalVM as its high performance Java JVM/JDK that also supports additional execution models and programming languages. GraalVM 22.2 is this newest version and contains a number of different optimizations for its growing number of components.

GraalVM 22.2 is another big release both for the open-source "Community Edition" and Oracle's paid enterprise alternative. With GraalVM 22.2 some of the big changes include:

- GraalVM itself is much smaller for its base binary
Continue reading...

Open Source LLM|Intel

AOCL 3.2 Released As AMD Optimizing CPU Libraries Now With LibMEM & Crypto

Last week AMD quietly released AOCL 3.2 as the newest version of their optimized CPU software libraries for use across Ryzen, Ryzen Threadripper, and EPYC platforms.

The AMD Optimizing CPU Libraries (AOCL) started off as core math libraries optimized for AMD Zen-based processors and has grown to encompass more libraries with time. AOCL includes implementations for libM, FFTW, BLAS, BLIS, libFLAME, ScaLAPACK, and more.

Now with the AOCL 3.2 release AOCL-Cryptography has been added for AES encryption/decryption
Continue reading...

Open Source LLM|Intel

AOCL 3.2 Released As AMD Optimizing CPU Libraries Now With LibMEM & Crypto

Last week AMD quietly released AOCL 3.2 as the newest version of their optimized CPU software libraries for use across Ryzen, Ryzen Threadripper, and EPYC platforms.

The AMD Optimizing CPU Libraries (AOCL) started off as core math libraries optimized for AMD Zen-based processors and has grown to encompass more libraries with time. AOCL includes implementations for libM, FFTW, BLAS, BLIS, libFLAME, ScaLAPACK, and more.

Now with the AOCL 3.2 release AOCL-Cryptography has been added for AES encryption/decryption
Continue reading...