b-sumner

Results 144 comments of b-sumner

Relax the required accuracy of tan(half)?

Are any opposed to making this change?

Relax the required accuracy of tan(half)?

I added the agenda label hoping to get final approval of this change.

[Issue]: how to close or bypass L1 cache

@JNYXR your situation is not clear to me. But if you are writing HIP source code, then I would recommend using `__builtin_nontemporal_{load, store}`. This will work on all targets.

"Cannot register Static Global Var" Error when loading HIP code

Is your shared library linked to the hip runtime, i.e. are all of its dependencies explicitly linked in?

"Cannot register Static Global Var" Error when loading HIP code

But is it NEEDED? "readelf -d your_shared.so | grep NEEDED"

"Cannot register Static Global Var" Error when loading HIP code

Thanks! I guess we need to dig deeper. Can you provide a minimal reproducer?

"Cannot register Static Global Var" Error when loading HIP code

Thank you! We'll take a look.

Herbie does not improve `sin(pi x)`

I wasn't sure if I should start a new issue or add to this one. It would be nice if herbie would accept sinpi(x) and cospi(x) in expressions since these...

Herbie does not improve `sin(pi x)`

Nice!

very slow atomic operations over unified memory

@zjin-lcf that code has every single thread hammering on a small set of locations in very few cache lines. Code doing that should be expected to be "slow". However, we...