Meta Launches AI Software Tools Offering Flexibility Between NVIDIA & AMD Chips

  • Facebook parent Meta Platforms Inc META unveiled a new set of free software tools for artificial intelligence applications helping developers smoothly switch between different underlying chips.
  • Meta's software would help to quickly swap between chips without being locked in, Reuters reports.
  • Meta's new open-source AI platform is based on an open-source machine learning framework called PyTorch
  • It can help code run up to 12 times faster on NVIDIA Corp's NVDA flagship A100 chip or up to four times faster on Advanced Micro Devices Inc's AMD MI250 GPU systems.
  • Also Read: US Offers Relaxations To Nvidia For AI Chip Development In China
  • But just as important as the speed boost is the flexibility the software can provide, the report added.
  • The software has become a key battleground for chipmakers seeking to build up an ecosystem of developers to use their chips. 
  • CUDA, NVIDIA's Compute Unified Device Architecture, has been the most popular so far for artificial intelligence work.
  • The unified GPU back-end support gives deep learning developers more hardware vendor choices with minimal migration costs, Meta said.
  • In March, Qualcomm Inc QCOM courted Meta to make Facebook's owner the flagship customer for Qualcomm's first AI data center chip, the AI 100.
  • Meta questioned whether the software that accompanied the chip was mature enough to wring the best performance from the chip on future tasks. 
  • Meta decided to stick with its existing chips as it evaluated its options.
  • Price Action: META shares closed higher by 2.16% at $138.61 on Monday.
Posted In: NewsTechMediaBriefs