Redfire75369’s avatarRedfire75369’s Twitter Archive—№ 1,234

    1. Intel seems to have tricked me slightly in this slide. The numbers they give are per Xe-core, I thought it was per Vector/Matrix Engine. Operations/Clock per Engine: Vector: 32 FP32/FP64, 64 FP16 Matrix: 256 TF32, 512 FP16/BF16, 1024 INT8
      oh my god twitter doesn’t include alt text from images in their API
  1. …in reply to @Redfire75369
    I only noticed because PVC would have to be running at 172 MHz for 45 TFLOPs FP32/FP64... Calculating with Xe-cores instead of Vector/Matrix makes it 1.373 GHz, which is far more reasonable.