news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

According to Fast Technology on July 18, British startup Spectral Compute recently launched a GPGPU programming toolkit called "SCALE", which successfully achieved seamless operation of NVIDIA's CUDA software on AMD GPUs, and is expected to break NVIDIA's monopoly in the field of GPU computing.

CUDA is a parallel computing platform and programming model launched by NVIDIA in 2007. It is widely used in fields such as high-performance computing and deep learning.

Due to its deep binding with NVIDIA GPU hardware, the richness of the CUDA ecosystem makes it difficult for other manufacturers to compete.

Spectral Compute's SCALE toolkit enables developers to run CUDA programs natively on AMD GPUs through a CUDA-compatible toolchain without relying on NVIDIA's assembly.

Spectral Compute CEO said that GPUs should have an open source environment, similar to modern CPUs, and there should be interconnectivity between various platforms.

The SCALE toolkit, which took seven years to develop, aims to remove exclusivity constraints in the market and promote interoperability between hardware platforms.

SCALE has been tested in multiple applications, including Blender, Llama-cpp, XGboost, FAISS, GOMC, STDGPU, Hashcat, and Nvidia Thrust, and can be used with AMD's RDNA 3 and RDNA 2 architectures.

In addition, major companies such as Qualcomm, Google and Intel also plan to work together to create a new AI software platform to provide an alternative to CUDA, further challenging Nvidia's market position.