Daily Archives: 2020-04-24

Article note: That's pretty heartening. I kind of expect the system to crush people who admit the rush to publish has caused a problem. Then again, this is in psychology where 97% of original studies and 36% of replications find a significant result, so they're already in up to their eyeballs.

Article note: Oooh, that's exciting. Shitty support for specified precision is a _major_ problem in most languages and toolchains for everyone looking to do anything about power or performance right now. I couldn't use LLVM for my MS work in part because at he time it scrubbed almost all the precision information in the IR.