NVIDIA open-sources CUDA Tile IR

NVIDIA has released the source code for CUDA Tile IR, making it available to the open-source community. This intermediate representation tool for CUDA is now freely accessible.

NVIDIA announced the open-sourcing of CUDA Tile IR, an intermediate representation used in CUDA development. The move allows developers and researchers to access and contribute to the tool previously proprietary.

Phoronix reported on this development, highlighting its significance for Linux hardware and open-source graphics communities. CUDA Tile IR supports advanced GPU programming tasks, and its open-source status could foster broader innovation in parallel computing.

This release aligns with ongoing efforts to enhance accessibility in GPU technologies. Developers can now explore and modify the codebase, potentially leading to improvements in performance and compatibility across platforms.

Liittyvät artikkelit

Researchers at NVIDIA have developed a system where teams of AI coding agents autonomously train robots to perform complex tasks such as inserting graphics cards and cutting zip ties.

Raportoinut AI

Red Hat and NVIDIA have achieved industry-leading results in the latest MLPerf Inference v6.0 benchmarks for vision, speech, and reasoning models. The companies optimized layers from the RHEL kernel to the vLLM engine. This work aims to help enterprises reduce costs per token on H200 and B200 GPUs.

Tämä verkkosivusto käyttää evästeitä

Käytämme evästeitä analyysiä varten parantaaksemme sivustoamme. Lue tietosuojakäytäntömme tietosuojakäytäntö lisätietoja varten.
Hylkää