See my above comment - the choice of metrics in table 3.2 doesn't really make any sense. Also, they revived some ancient 1998 IBM chip, and report on that, for no clear reason. They call it a 2004 benchmark, but what…
It continues to boggle my mind that this guy chose to get fired over this paper, which is the dumbest paper I have read in a long time. As an example, his paper reports *median* congestion! I literally laughed out loud…
Some tweets by someone who replicated their results: https://twitter.com/sguada/status/1521680121412808704 https://twitter.com/sguada/status/1521587406385807361 Tweets by other people who know the researchers:…
Seems pretty clear to me, tbh: 1) Google terminated this guy "with cause" and then said so publicly. Google definitely does not do that lightly. 2) The TPU team used this method to help make TPUs. This is a flagship…
See my above comment - the choice of metrics in table 3.2 doesn't really make any sense. Also, they revived some ancient 1998 IBM chip, and report on that, for no clear reason. They call it a 2004 benchmark, but what…
It continues to boggle my mind that this guy chose to get fired over this paper, which is the dumbest paper I have read in a long time. As an example, his paper reports *median* congestion! I literally laughed out loud…
Some tweets by someone who replicated their results: https://twitter.com/sguada/status/1521680121412808704 https://twitter.com/sguada/status/1521587406385807361 Tweets by other people who know the researchers:…
Seems pretty clear to me, tbh: 1) Google terminated this guy "with cause" and then said so publicly. Google definitely does not do that lightly. 2) The TPU team used this method to help make TPUs. This is a flagship…