I would like to thank all my Twitter followers for ensuring in this thread that, indeed, I *did* in fact die on this hill, repeatedly and pointlessly.https://twitter.com/chandlerc1024/status/1261006073789480960 …
-
-
Replying to @chandlerc1024
Mental note made to use that word next chance I get to hang our with ya
1 reply 0 retweets 1 like -
Replying to @jonmasters @chandlerc1024
Uh oh... Looks like I used this word in an upcoming presentation that I already recorded
pic.twitter.com/j09BQxPwBi
1 reply 0 retweets 1 like -
Replying to @epakskape @jonmasters
I will find some way to reap retribution on you both.... Ok, now time to design a compiler that introduces *more* speculative execution vulnerabilities into code, and somehow hoodwink MS into using it.....
1 reply 0 retweets 3 likes -
I'm sure you can come up with a compiler pass that makes indirect branches more performant by adding a branch and direct call for the most likely callee. (And sneak in some new spectre-v1 type confusions that way.)
1 reply 0 retweets 4 likes -
Replying to @tehjh @chandlerc1024 and
if you place the branch the right way around, you can probably even make the type confusion happen on BTB miss, so no messy training will be necessary
1 reply 0 retweets 3 likes
Measure performance in a microbenchmark carefully crafted such that the branch predictor state is gone on every call and the function pointer is evicted from at least L1D.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.