public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug target/105246] New: [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode
@ 2022-04-12 17:29 burnus at gcc dot gnu.org
2022-04-13 6:35 ` [Bug target/105246] " rguenth at gcc dot gnu.org
2022-04-13 8:29 ` ams at gcc dot gnu.org
0 siblings, 2 replies; 3+ messages in thread
From: burnus at gcc dot gnu.org @ 2022-04-12 17:29 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105246
Bug ID: 105246
Summary: [amdgcn] Use library call for SQRT with -ffast-math +
provide additional option to use single-precsion
opcode
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Keywords: documentation, wrong-code
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: burnus at gcc dot gnu.org
CC: ams at gcc dot gnu.org
Target Milestone: ---
Target: amdgcn-amdhsa
AMD GCN hardware has opcodes which operate on double-precision variables as
input/output but internally only do single-precision operation.
This affects (currently) only "sqrt" which for -funsafe-math-optimizations
(implied by -Ofast / -ffast-math) uses AMDGCN's "v_sqrt".
Namely gcc/config/gcn/gcn-valu.md has:
(define_insn "sqrt<mode>2<exec>"
...
"flag_unsafe_math_optimizations"
"v_sqrt%i0\t%0, %1"
Thus: while "v_sqrt" works on double-precision variables, it only calculates
with 23bits (as with float32) instead of 52bits (as float64 provides) for the
fractional part of the floating-point number.
PROBLEM: In many cases, this loss of precision by an order of 100,000,000 (10⁸
/ 2²⁹) is
very unexpected and too much for code which requires double precision. An ULP
of 4 is
expected not an ULP of 10⁸!
In particular: In order to permit several optimizations, -Ofast or --fast-math
is
commonly recommended and the precision loss is unexpected.
In terms of testsuites, OvO's sqrt examples are effected, requiring a way
higher
OVO_TOL_ULP to pass (→ https://github.com/TApplencourt/OvO )
But the issue really came up when discussion with HPC code users.
EXPECTED:
- By default, with -ffast-math, do the double-precision operation by a library
call
- Provide some (GCN-specific) -m... flag to do those calculations in single
precision.
For instance something like:
-mpermit-reduced-precision
Use hardware intrinsics instead of library even if they provide a much
reduced precision. Example: use v_sqrt with double-precision variables
even though the hardware only provides single-precision results for the
fractional part of the floating-point variable. (Default: disabled)
^ permalink raw reply [flat|nested] 3+ messages in thread
* [Bug target/105246] [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode
2022-04-12 17:29 [Bug target/105246] New: [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode burnus at gcc dot gnu.org
@ 2022-04-13 6:35 ` rguenth at gcc dot gnu.org
2022-04-13 8:29 ` ams at gcc dot gnu.org
1 sibling, 0 replies; 3+ messages in thread
From: rguenth at gcc dot gnu.org @ 2022-04-13 6:35 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105246
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
can one improve the precision by employing some additional NR step or so, using
v_sqrt only for the initial guess? I agree that just v_sqrt isn't good for
-funsafe-math-optimizations.
^ permalink raw reply [flat|nested] 3+ messages in thread
* [Bug target/105246] [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode
2022-04-12 17:29 [Bug target/105246] New: [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode burnus at gcc dot gnu.org
2022-04-13 6:35 ` [Bug target/105246] " rguenth at gcc dot gnu.org
@ 2022-04-13 8:29 ` ams at gcc dot gnu.org
1 sibling, 0 replies; 3+ messages in thread
From: ams at gcc dot gnu.org @ 2022-04-13 8:29 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105246
--- Comment #2 from Andrew Stubbs <ams at gcc dot gnu.org> ---
When we first coded this we only had the GCN3 ISA manual, which says nothing
about the accuracy.
Now I look in the Vega manual (GCN5) I see:
Square root with perhaps not the accuracy you were hoping for --
(2**29)ULP accuracy. On the upside, denormals are supported.
The most recent CDNA2 manual is a bit less verbose:
Square root. Precision is (2**29) ULP, and supports denormals.
The compiler already emits Newton Raphson iterations for division with
-ffast-math, so I'm sure it can be done, but I'm not too clear on the
mathematics myself.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-04-13 8:29 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-12 17:29 [Bug target/105246] New: [amdgcn] Use library call for SQRT with -ffast-math + provide additional option to use single-precsion opcode burnus at gcc dot gnu.org
2022-04-13 6:35 ` [Bug target/105246] " rguenth at gcc dot gnu.org
2022-04-13 8:29 ` ams at gcc dot gnu.org
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).