5 ms·
0.007109375 minute per Forth-gen TPU (1.82 / 256) VS 0.000095214 minutes per third-gen TPU (0.39 / 4096) Is it only me or Third-gen seem better lol
by skyde 5y ago
0.007109375 minute per Forth-gen TPU (1.82 / 256)
VS
0.000095214 minutes per third-gen TPU (0.39 / 4096)
Is it only me or Third-gen seem better lol
- codefined 5y agoYour maths is incorrect, dividing by the number of devices doesn't make sense in this case, you should be multiplying them. E.g. - 466 minutes of fourth-generation TPU time. - 1,600 minutes of third-generation TPU time. Using this logic, fourth generation TPUs are 3.4x better. But, comparing different cluster sizes is pointless. These things don't scale linearly.
- skyde 5y agoyou are most likely correct! I read ´Training took 1.82 minutes with 256 fourth-gen TPUs’ to mean Using all 256 TPU it took 1.82 wall clock time to do the training not 466 minutes. This mean each TPU contributed to (1/256) assuming it scale linearly. Even if it doesn’t the 256 TPU ran in parallel not sequentially right ?
- skyde 5y agoits only slightly (74 times) slower :)
- ml_hardware 5y agoI think your math is backwards. The training workload W is the same, and the time to complete it is: W / (256 * speed_v4) = 1.82 W / (4096 * speed_v3) = 0.39 speed_v4 / speed_v3 = (4096 * 0.39) / (256 * 1.82) = 3.43 Note that this assumes that training speed is perfectly linear with the # of accelerators which is not true as you get to very large #s (like 4096!). So the true number should be smaller than the 3.43x above, and the reported 2.7x makes sense.