std.math: Implement more optimal floorImpl for ieeeSingle#5820
std.math: Implement more optimal floorImpl for ieeeSingle#5820dlang-bot merged 2 commits intodlang:masterfrom
Conversation
|
Thanks for your pull request, @ibuclaw! Bugzilla referencesYour PR doesn't reference any Bugzilla issue. If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog. Testing this PR locallyIf you don't have a local development environment setup, you can use Digger to test this PR: dub fetch digger
dub run digger -- build "master + phobos#5820" |
ibuclaw
force-pushed
the
floatfloor
branch
2 times, most recently
from
c048355 to
5f81d28
Compare
|
Ping? |
|
Pinging people who are qualified to review this from git blame @yebblies @wilzbach @don-clugston-sociomantic @klickverbot |
|
Or you know, you could just request a review. :) |
don-clugston-sociomantic
left a comment
don-clugston-sociomantic
left a comment
There was a problem hiding this comment.
I think the code itself is probably OK but it's so cryptic that it's hard to review. I think this low-level stuff really needs copious comments.
| floatBits y = void; | ||
| y.rv = x; | ||
|
|
||
| // Find the exponent (power of 2) |
There was a problem hiding this comment.
This deserves a much better comment since this code is actually quite tricky.
// Find the exponent (power of 2)
// Do this by shifting the raw value so that the exponent lies in the low bits.
// Then mask out the sign bit, and subtract the bias
std/math.d
Outdated
| exp -= 16; | ||
| if (exp < (T.mant_dig - 1)) | ||
| { | ||
| const uint i = F.MANTISSAMASK_INT >> exp; |
There was a problem hiding this comment.
This variable deserves a better name. I think this is actually fraction_mask, the bits of the mantissa which form the fraction.
| if ((y.vi & i) != 0) | ||
| { | ||
| if (y.vi < 0) | ||
| y.vi += 0x00800000 >> exp; |
There was a problem hiding this comment.
This REALLY needs a comment. My guess is that this is the float value 0.5, added so that it rounds up towards zero? Or something like that.
There was a problem hiding this comment.
This is subtracting one, adding comment.
|
@redstar - And this you might be interested in also. |
5 participants
Less branching, and doesn't care about endian-ness. Tested against PPC64/LE.
For comparison between old (source view 1) and new (source view 2) https://explore.dgnu.org/g/fTvttz
Probably better to review with https://github.com/dlang/phobos/pull/5820/files?w=1