@FioraAeterna why don't GPUs use fixed-point?
-
-
Replying to @whitequark
@whitequark@FioraAeterna because fixed point is an absolute clusterfuck that no one wants to use for actual programs.4 replies 0 retweets 2 likes -
Replying to @stephentyrone
@stephentyrone@whitequark@FioraAeterna Fixed point makes sense when your quantities are all coordinates in a uniformly-discrete space.1 reply 0 retweets 1 like -
Replying to @RichFelker
@stephentyrone@whitequark@FioraAeterna But fixed point is utterly awful when you also have to represent anything beyond that.1 reply 0 retweets 0 likes -
Replying to @RichFelker
@RichFelker@whitequark@FioraAeterna and woe unto you when you need to make the coordinate space a little bit bigger.1 reply 0 retweets 0 likes -
Replying to @stephentyrone
@stephentyrone@whitequark@FioraAeterna Well the same problem can happen with float but it's more subtle (and thus more ugly/dangerous).1 reply 0 retweets 0 likes -
Replying to @RichFelker
@stephentyrone@whitequark@FioraAeterna Increasing the size of your float coordinate space gives "more discrete than others" regions.1 reply 0 retweets 0 likes
@stephentyrone @whitequark @FioraAeterna For [in-bound-]translation-invariance, float spaces should be centered around 0x1.8pN for N big.
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.