Building LLVM 23.1.1 from llvm.org on Discoverer CPU cluster

Building LLVM 23.1.1 from llvm.org on Discoverer CPU cluster

Comments 0 Comment

Overview This document describes the build procedure for the LLVM 23.1.1 source distribution on an AMD EPYC 7H12 (Zen 2, x86_64) based CPU-only compute cluster (Discoverer CPU cluster). The build is performed from source using a shell-based recipe that loads Environment Modules, configures LLVM with CMake and Ninja, and installs the resulting tool chain into a versioned prefix. The recipe and build log are published at: https://gitlab.discoverer.bg/vkolev/recipes/-/blob/main/llvm/23/llvmorg-23.1.1.recipe https://gitlab.discoverer.bg/vkolev/recipes/-/blob/main/llvm/23/llvmorg-23.1.1.log Build environment and approach The build runs on AMD EPYC 7H12 compute…

Read More Read More

Global Arrays 5.9.2: analysis of three test suite failures on Discoverer CPU cluster

Global Arrays 5.9.2: analysis of three test suite failures on Discoverer CPU cluster

Comments 1 Comment

Global Arrays underpins a number of computational chemistry codes, NWChem among them, which makes the correctness of its distributed operations a practical concern rather than an academic one. We recently built GA 5.9.2 on an AMD EPYC compute node with GCC, Open MPI 5.0.10 and OpenBLAS, added the test suite to our build recipe for the first time, and watched three of the seventy-nine tests fail. This post walks through how we established that the failures were upstream defects rather…

Read More Read More

Running llama.cpp across multiple CPU nodes on Discoverer: possibilities and expectations

Running llama.cpp across multiple CPU nodes on Discoverer: possibilities and expectations

Comments 0 Comment

Most people running llama.cpp are familiar with its single-node CPU mode, where inference is spread across cores using multithreading. What is less commonly known is that llama.cpp can also be distributed across multiple machines — but understanding what that actually means in practice is essential before building a cluster setup around it. The built-in RPC backend llama.cpp includes an RPC feature that connects multiple nodes over TCP. A master node holds the model file and coordinates inference, while worker nodes…

Read More Read More

In most cases, avoid def main() in Python — It’s not Java or C++

In most cases, avoid def main() in Python — It’s not Java or C++

Comments 0 Comment

A pattern that keeps appearing in Python scripts, tutorials, and AI-generated code: This is cargo-culted from C and Java, where a main() function is a syntactic requirement. Python has no such rule. The above is functionally identical to: When def main() actually makes sense There is one legitimate use case: when you need a named entry point for Python packaging. If your pyproject.toml defines a console script, it must reference a callable: Here, def main() earns its place. But this is a specific packaging concern — not a general…

Read More Read More

Enhanced SSH Security: HSM and PIV Device Authentication for Discoverer and Discoverer+ Login Nodes

Enhanced SSH Security: HSM and PIV Device Authentication for Discoverer and Discoverer+ Login Nodes

Comments 0 Comment

We’re excited to announce comprehensive documentation for enabling HSM and PIV device SSH authentication for Discoverer and Discoverer+ login nodes. Our new guide enables users to leverage their existing state and EU-issued HSM and PIV devices (smart cards) for SSH authentication on Discoverer login nodes. Instead of generating new SSH keys, users can now use the cryptographic tokens they already possess from state and EU-recognized PKI certificate authorities for secure access to our computing infrastructure. Moreover, this way, the SSH…

Read More Read More

Presenting Discoverer abilities to forecast weather during the COP29 conference

Presenting Discoverer abilities to forecast weather during the COP29 conference

Comments 0 Comment

The abilities of the Discoverer Petascale Supercomputer to predict the climate changes through a high-resolution weather forecasts, were demonstrated at the annual UN Climate Change Conference in Baku (November 24, 2024). It is our commitment to accelerate the green transition. Here’s a video of the presentation:

Python 3.13.0 built with LLVM 19 and full LTO

Python 3.13.0 built with LLVM 19 and full LTO

Comments 0 Comment

Finally! We managed to compile the code of Python 3.13.0 using LLVM 19, with full link-time optimization (LTO): The installation passes all 44 standard optimization tests: Using random seed: 4347316840:00:00 load avg: 1.72 Run 44 tests sequentially in a single process0:00:00 load avg: 1.72 [ 1/44] test_array0:00:00 load avg: 1.72 [ 2/44] test_base640:00:00 load avg: 1.72 [ 3/44] test_binascii0:00:00 load avg: 1.72 [ 4/44] test_binop0:00:00 load avg: 1.72 [ 5/44] test_bisect0:00:00 load avg: 1.72 [ 6/44] test_bytes0:00:03 load avg: 1.67…

Read More Read More

WordPress Appliance - Powered by TurnKey Linux