Meta has done something that will get Nvidia and AMD very, very worried — it gave up on GPU and CPU to take a RISC-y route for AI training and inference acceleration

Key Takeaways:

– Meta is planning to begin mass development of hardware and products with RISC-V technology at its core.
– RISC-V is an alternative Instruction Set Architecture (ISA) to x86 and is dominated by Arm.
– RISC-V is an open-source alternative that lowers the barrier of entry and eliminates the need for licensing fees.
– Companies like Alibaba and Qualcomm have announced plans to migrate to the RISC-V ecosystem.
– Meta is migrating to RISC-based components for their power efficiency, performance, lower latency, and flexibility.
– Meta has already built video transcoding hardware on RISC-V, replacing 85% of CPUs.
– Meta is also considering using RISC-V for developing AI training and inference chips instead of Nvidia GPUs.
– Meta’s first in-house RISC-V silicon for AI acceleration operates at 800MHz, has 128MB on-chip memory, and supports up to 128GB LPDDR5 RAM.

TechRadar:

The RISC-V movement has been given a massive boost after Meta revealed it is primed to begin mass development of hardware and products with the technology at its core.

RISC, an alternative Instruction Set Architecture (ISA) to the x86 architecture used by the likes of Intel and AMD in their best CPUs, is dominated by Arm. This ISA is largely able to replicate many of the functionalities of the best GPUs and processors out there. 

Source link

AI Eclipse TLDR:

Meta, a company formerly known as Facebook, has announced plans to begin mass development of hardware and products based on the RISC-V technology. RISC-V is an alternative Instruction Set Architecture (ISA) to the x86 architecture used by Intel and AMD. Unlike x86, RISC-V is an open-source alternative that does not require organizations to pay licensing fees to use CPUs. This has led to increasing interest in RISC-V, with companies like Alibaba and Qualcomm announcing plans to migrate to the technology. Meta is migrating away from CPUs to RISC-based components due to the power efficiency, performance, lower latency, and flexibility of the RISC-V ISA. The company has already built video transcoding hardware, called the Meta Scalable Video Processor, using RISC-V, replacing 85% of the CPUs. Meta is also considering using RISC-V for developing AI processors, aiming to skip the need for Nvidia GPUs. Meta’s first in-house RISC-V silicon for AI acceleration is a 7nm component with 128MB on-chip memory and support for up to 128GB LPDDR5 RAM. It is currently dedicated to accelerating models in both training and inference.