Matt Arsenault
5bbfce06e0
ValueTracking: Special case fmul by llvm.amdgcn.trig.preop ( #183373 )
...
This is another instance of the logic from #183159 . If we know
one source is not-infinity, and the other source is less than or
equal to 1, this cannot overflow. Special case llvm.amdgcn.trig.preop,
as a substitute for proper range tracking. This almost enables pruning
edge case handling in trig function implementations, if not for the
recursion depth limit (but that's a problem for another day).
2026-02-26 20:42:26 +01:00
..
2026-02-26 20:42:26 +01:00
2025-11-04 12:26:17 +08:00
2026-01-23 10:34:00 +00:00
2026-02-05 13:31:26 +00:00
2025-04-08 22:15:50 +07:00
2025-11-04 12:26:17 +08:00
2025-04-07 23:59:53 +07:00
2025-04-07 23:59:53 +07:00
2026-02-05 13:31:26 +00:00
2025-12-02 08:45:33 +00:00
2025-08-08 11:09:34 +02:00
2025-08-08 11:09:34 +02:00
2025-04-24 21:41:55 +02:00
2025-08-08 11:09:34 +02:00
2025-07-23 15:05:43 +02:00
2025-08-08 11:09:34 +02:00
2025-09-17 11:27:06 +02:00
2025-04-17 12:44:57 +02:00
2026-02-05 13:31:26 +00:00
2025-12-29 14:24:24 +00:00
2026-02-05 13:31:26 +00:00
2026-01-30 11:48:36 +01:00
2026-02-26 17:49:51 +01:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-01-13 20:36:00 +01:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2026-02-21 11:16:17 +01:00
2026-01-13 20:36:00 +01:00
2026-02-05 13:31:26 +00:00
2026-02-05 13:31:26 +00:00
2025-11-04 12:00:44 +00:00
2025-12-02 08:45:33 +00:00
2025-11-04 12:00:44 +00:00
2025-04-07 23:59:53 +07:00
2025-08-08 11:09:34 +02:00
2025-10-25 10:36:31 -04:00
2025-07-05 16:47:03 +02:00
2025-10-20 08:50:09 +00:00
2025-04-07 23:59:53 +07:00
2025-07-01 17:46:19 -04:00
2025-12-02 08:45:33 +00:00
2025-07-05 16:47:03 +02:00
2025-12-02 08:45:33 +00:00