3 ms·

Could programmable AI chips compete with Graphics Cards and TPUs or is it futile to try?

• ninjaoxygen 6y ago
  In the stock trading world, HFT on FPGA is closer to the edge than GPU / TPU solutions and they are using ML models, if you count that as AI. When the logic and the NIC are on the same hardware, it's really fast. An ASIC would be even faster, but you can't really iterate on that.
• heliophobicdude 6y ago
  Perhaps they would not make good competition. FPGAs have been known to be slower than ASICs. But then again, perhaps some other company will find a good use for rapidly changing IC design.
• galangalalgol 6y ago
  Their new versal line puts a tpu on the die with the fpga. Great for inference, especially if you want to use the fpga to quickly extract features in the fpga and infer from feature space.