-
-
Replying to @cmuratori
@cmuratori It's the calculation of U and V - the PixelPy calcs and anything that just depends on them and constants.1 reply 0 retweets 0 likes -
Replying to @rygorous
@cmuratori You actually moved them out of the loop in last night's stream, then moved them back in "because it didn't help".1 reply 0 retweets 0 likes -
Replying to @rygorous
@cmuratori The reason it didn't help was because the compiler was already doing it for you anyway! :)1 reply 0 retweets 0 likes -
Replying to @cmuratori
@rygorous I feel like it is mostly just multiple issue, like I was saying, right - the adds and muls, for example can overlap.1 reply 0 retweets 0 likes -
Replying to @cmuratori
@cmuratori The problem is that your throughput calc assumes that some types of superscalar issue always happen and some never happen.1 reply 0 retweets 0 likes -
Replying to @rygorous
@cmuratori Instruction tables say ANDPS has tpt of 1/3 cycle. Correct - if you issue nothing but ANDPS, machine will run 3 per cycle.1 reply 0 retweets 0 likes -
Replying to @rygorous
@cmuratori (provided they're independent) because there's 3 units that can handle them.1 reply 0 retweets 0 likes -
Replying to @rygorous
@cmuratori The tables also say ADDPS is 1 cycle tpt. Again correct, if you have nothing but ADDPS, because there's only one FP adder unit.2 replies 0 retweets 0 likes
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.