Increasingly greater chances of the RISC-V GPU. The CUDA library has a port

You’ve probably heard about RISC-V, an ISA instruction set architecture that anyone can access for free. This is why so much has been mentioned about it in recent times and why even Apple is interested in it. If you’ve been following this topic, you surely know how important the step that has just been taken is, which could one day bring the world a widely available RISC-V GPU.

At the moment, RISC-V is mainly used for the production of central processing units (CPUs), but once a general-purpose graphics processor was created based on this architecture. He is the protagonist here, because the engineers behind the RISC-V GPGPU project called Vortex extended its capabilities beyond the OpenCL 1.2 API support, using some CUDA calculations from NVIDIA.

Also read: The premiere of Turing Pi V2. It is a minicomputer compatible with NVIDIA Jetson

NVIDIA’s CUDA is a unique computing and API platform. Programming the application for its support means that each time the system detects a graphics processor based on CUDA, it gains a huge increase in performance in applications optimized for this purpose.

Also read: NVIDIA CMP 170HX tested. This crypto card kicks like crazy

[…] in this project, we propose and build a pipeline to support comprehensive CUDA migration: the pipeline accepts CUDA source codes as input and executes them on the extended RISC-V GPU architecture. consists of several steps: translate CUDA source code to NVVM IR, convert NVVM IR to SPIR-V IR, pass SPIR-V IR to POCL to get RISC-V binary, and finally execute binary on RISC extended GPU- V.

– the engineers explain in their extensive publication about the fruit of their work.

HEADING