Subzero: Add support for SSE4.1 instructions.

crosstest/runtests.sh

Index: crosstest/runtests.sh
diff --git a/crosstest/runtests.sh b/crosstest/runtests.sh
index 7b81df12b5d2cb455ed86771105e34e01e48ae7f..0b79f48aa4b706a8fd078c2f71c9b3d2ed3fe487 100755
--- a/crosstest/runtests.sh
+++ b/crosstest/runtests.sh
@@ -6,116 +6,144 @@
 set -eux
 
 OPTLEVELS="m1 2"
+ATTRIBUTES="sse2 sse4.1"
 OUTDIR=Output
 # Clean the output directory to avoid reusing stale results.
 rm -rf "${OUTDIR}"
 mkdir -p "${OUTDIR}"
 
 for optlevel in ${OPTLEVELS} ; do
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=simple_loop.c \
-        --driver=simple_loop_main.c \
-        --output=simple_loop_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=mem_intrin.cpp \
-        --driver=mem_intrin_main.cpp \
-        --output=mem_intrin_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_arith.cpp \
-        --test=test_arith_frem.ll \
-        --test=test_arith_sqrt.ll \
-        --driver=test_arith_main.cpp \
-        --output=test_arith_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_bitmanip.cpp --test=test_bitmanip_intrin.ll \
-        --driver=test_bitmanip_main.cpp \
-        --output=test_bitmanip_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_cast.cpp --test=test_cast_to_u1.ll \
-        --driver=test_cast_main.cpp \
-        --output=test_cast_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_fcmp.pnacl.ll \
-        --driver=test_fcmp_main.cpp \
-        --output=test_fcmp_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_global.cpp \
-        --driver=test_global_main.cpp \
-        --output=test_global_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_icmp.cpp --test=test_icmp_i1vec.ll \
-        --driver=test_icmp_main.cpp \
-        --output=test_icmp_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_select.ll \
-        --driver=test_select_main.cpp \
-        --output=test_select_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_stacksave.c \
-        --driver=test_stacksave_main.c \
-        --output=test_stacksave_O${optlevel}
-
-    # Compile the non-subzero object files straight from source
-    # since the native LLVM backend does not understand how to
-    # lower NaCl-specific intrinsics.
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-       --dir="${OUTDIR}" \
-       --llvm-bin-path="${LLVM_BIN_PATH}" \
-       --test=test_sync_atomic.cpp \
-       --crosstest-bitcode=0 \
-       --driver=test_sync_atomic_main.cpp \
-       --output=test_sync_atomic_O${optlevel}
-
-    ./crosstest.py -O${optlevel} --prefix=Subzero_ --target=x8632 \
-        --dir="${OUTDIR}" \
-        --llvm-bin-path="${LLVM_BIN_PATH}" \
-        --test=test_vector_ops.ll \
-        --driver=test_vector_ops_main.cpp \
-        --output=test_vector_ops_O${optlevel}
-
+    for attribute in ${ATTRIBUTES} ; do

[jvoung (off chromium), 2014/07/30 04:30:11]

We might want to start parallelizing some of these compiles, once we add more
attributes =)

Maybe I'll look into that, if you guys haven't already started looking into it.

[Jim Stichnoth, 2014/07/30 04:45:47]

Right, I forgot to mention the same in my review. :)

By parallelizing, you mean adding them to the lit framework, right?  That's on
"the list", but I didn't have any immediate plans to do it yet.

[jvoung (off chromium), 2014/07/31 23:38:17]

Ah sorry for the delay in the reply and I didn't get it done before vacation,
though I volunteered...

Yes, we could try to fit it into the lit framework. Some things to watch out
for, but hadn't considered fully:

(*) How to be able to loop through attributes, arches, and optlevels in the lit
framework, without have to unroll the loops?  Currently, our .ll files have
multiple run lines, and we end up with some duplication. That's okay for the
.ll, since we often have different CHECK lines and check-prefixes. These cross
tests just check the exit status though, so it's more amenable to just looping.

Perhaps this can be handled, if each test case has a "RUN" line that invokes a
script, and the script can then loop through. Then at least there is some
parallelism between the test cases, but not the loops within the test cases.

(*) Naming the test cases / setting up "config.suffixes": I think the test cases
would have to use a suffix that is not ".cpp" or ".ll", since one test case
consists of multiple .cpp/.ll files. That's a bit different from the usual use
of lit, where each .ll and .cpp are individual test cases. Maybe it doesn't
matter and we just make the individual tests have .py suffixes.

Otherwise, I had considered a quick and dirty use of bash "&" plus "wait"...

+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=simple_loop.c \
+            --driver=simple_loop_main.c \
+            --output=simple_loop_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=mem_intrin.cpp \
+            --driver=mem_intrin_main.cpp \
+            --output=mem_intrin_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_arith.cpp \
+            --test=test_arith_frem.ll \
+            --test=test_arith_sqrt.ll \
+            --driver=test_arith_main.cpp \
+            --output=test_arith_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_bitmanip.cpp --test=test_bitmanip_intrin.ll \
+            --driver=test_bitmanip_main.cpp \
+            --output=test_bitmanip_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_cast.cpp --test=test_cast_to_u1.ll \
+            --driver=test_cast_main.cpp \
+            --output=test_cast_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_fcmp.pnacl.ll \
+            --driver=test_fcmp_main.cpp \
+            --output=test_fcmp_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_global.cpp \
+            --driver=test_global_main.cpp \
+            --output=test_global_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_icmp.cpp --test=test_icmp_i1vec.ll \
+            --driver=test_icmp_main.cpp \
+            --output=test_icmp_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_select.ll \
+            --driver=test_select_main.cpp \
+            --output=test_select_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_stacksave.c \
+            --driver=test_stacksave_main.c \
+            --output=test_stacksave_O${optlevel}_${attribute}
+
+        # Compile the non-subzero object files straight from source
+        # since the native LLVM backend does not understand how to
+        # lower NaCl-specific intrinsics.
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ \
+            --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_sync_atomic.cpp \
+            --crosstest-bitcode=0 \
+            --driver=test_sync_atomic_main.cpp \
+            --output=test_sync_atomic_O${optlevel}_${attribute}
+
+        ./crosstest.py -O${optlevel} --mattr ${attribute} \
+            --prefix=Subzero_ --target=x8632 \
+            --dir="${OUTDIR}" \
+            --llvm-bin-path="${LLVM_BIN_PATH}" \
+            --test=test_vector_ops.ll \
+            --driver=test_vector_ops_main.cpp \
+            --output=test_vector_ops_O${optlevel}_${attribute}
+
+    done
 done
 
 for optlevel in ${OPTLEVELS} ; do
-    "${OUTDIR}"/simple_loop_O${optlevel}
-    "${OUTDIR}"/mem_intrin_O${optlevel}
-    "${OUTDIR}"/test_arith_O${optlevel}
-    "${OUTDIR}"/test_bitmanip_O${optlevel}
-    "${OUTDIR}"/test_cast_O${optlevel}
-    "${OUTDIR}"/test_fcmp_O${optlevel}
-    "${OUTDIR}"/test_global_O${optlevel}
-    "${OUTDIR}"/test_icmp_O${optlevel}
-    "${OUTDIR}"/test_select_O${optlevel}
-    "${OUTDIR}"/test_stacksave_O${optlevel}
-    "${OUTDIR}"/test_sync_atomic_O${optlevel}
-    "${OUTDIR}"/test_vector_ops_O${optlevel}
+    for attribute in ${ATTRIBUTES}; do
+        "${OUTDIR}"/simple_loop_O${optlevel}_${attribute}
+        "${OUTDIR}"/mem_intrin_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_arith_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_bitmanip_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_cast_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_fcmp_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_global_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_icmp_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_select_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_stacksave_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_sync_atomic_O${optlevel}_${attribute}
+        "${OUTDIR}"/test_vector_ops_O${optlevel}_${attribute}
+    done
 done