From: Travis Boraten <travis.boraten@amd.com>
Date: Thu, 18 Apr 2019 22:03:14 +0000 (-0400)
Subject: arch-gcn3: Fix VOP3 V_LDEXP_F64
X-Git-Tag: v20.1.0.0~428
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=e1d10c3894978015fb20e7247f67b1e5d23d2811;p=gem5.git

arch-gcn3: Fix VOP3 V_LDEXP_F64

Replaced !std::isnormal with std::fpclassify because std::isnormal
is not specific enough. !std::isnormal was incorrectly catching
NaN, Inf, 0.0, and subnormals (aka denormals), where as it was only
suppose to catch subnormals.

The return value and error handling spec of std::ldexp listed on
cppreference.com appears to match up in nearly all cases after
making these changes. If std::ldexp handled subnormals as described
in the GCN3 2016 guide, we could have used vdst[lane] = std::ldexp
and not need to check for any corner cases.

Change-Id: I4c77af77c3b7798f86d40442610cef1296a28441
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/29966
Maintainer: Anthony Gutierrez <anthony.gutierrez@amd.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
---

diff --git a/src/arch/gcn3/insts/instructions.cc b/src/arch/gcn3/insts/instructions.cc
index 2b992b165..302dad4eb 100644
--- a/src/arch/gcn3/insts/instructions.cc
+++ b/src/arch/gcn3/insts/instructions.cc
@@ -30282,10 +30282,11 @@ namespace Gcn3ISA
 
         for (int lane = 0; lane < NumVecElemPerVecReg; ++lane) {
             if (wf->execMask(lane)) {
-                if (std::isnan(src1[lane]) || std::isinf(src1[lane])) {
+                if (std::isnan(src0[lane]) || std::isinf(src0[lane])) {
                     vdst[lane] = src0[lane];
-                } else if (!std::isnormal(src1[lane])) {
-                    if (std::signbit(src1[lane])) {
+                } else if (std::fpclassify(src0[lane]) == FP_SUBNORMAL
+                           || std::fpclassify(src0[lane]) == FP_ZERO) {
+                    if (std::signbit(src0[lane])) {
                         vdst[lane] = -0.0;
                     } else {
                         vdst[lane] = +0.0;