[Git][ghc/ghc][wip/supersven/riscv-vectors] simd013 test for RISC-V

[Sven Tennie (@supersven)]

Sven Tennie pushed to branch wip/supersven/riscv-vectors at Glasgow Haskell Compiler / GHC


Commits:
eb03f43d by Sven Tennie at 2025-03-09T18:10:23+01:00
simd013 test for RISC-V

- - - - -


2 changed files:

- testsuite/tests/simd/should_run/all.T
- testsuite/tests/simd/should_run/simd013C.c


Changes:

=====================================
testsuite/tests/simd/should_run/all.T
=====================================
@@ -31,7 +31,8 @@ def riscvVlen():
 # produce ABI-incompatible code, e.g. when compiling usage of YMM registers
 # with or without -mavx2.
 setTestOpts([
-    when(arch('riscv64'), extra_hc_opts('-mriscv-vlen' + str(riscvVlen()) + " -opta=-march=rv64gv"))
+    # TODO: -optc and -opta should not be required, but provided by the GHC pipeline
+    when(arch('riscv64'), extra_hc_opts('-mriscv-vlen' + str(riscvVlen()) + " -optc=-march=rv64gv -opta=-march=rv64gv"))
   ])
 
 test('simd_insert_baseline', [], compile_and_run, [''])
@@ -81,7 +82,9 @@ test('simd011', [ req_fma_cpu
 test('simd012', [], compile_and_run, [''])
 test('simd013',
      [ req_c
-     , unless((arch('x86_64') or arch('riscv64')), skip) # because the C file uses architecture-dependent intrinsics
+     , unless((arch('x86_64') or arch('riscv64')), skip)
+      # ^ Because the C file uses architecture-dependent intrinsics.
+      # Requires GCC >= 14 on RISC-V.
      ],
      compile_and_run, ['simd013C.c'])
 test('simd014',


=====================================
testsuite/tests/simd/should_run/simd013C.c
=====================================
@@ -1,4 +1,4 @@
-
+#if defined(__x86_64__) || defined(_M_X64)
 #include <immintrin.h>
 
 __m128d sub(__m128d x, __m128d y)
@@ -10,3 +10,22 @@ __m128d add7(__m128d x1, __m128d x2, __m128d x3, __m128d x4, __m128d x5, __m128d
 {
   return _mm_add_pd(x1,_mm_add_pd(x2,_mm_add_pd(x3,_mm_add_pd(x4,_mm_add_pd(x5,_mm_add_pd(x6, x7))))));
 }
+#elif defined(__riscv_v)
+#if __riscv_v_intrinsic >= 12000
+#include <riscv_vector.h>
+
+size_t vl = 2;
+
+vfloat64m1_t sub(vfloat64m1_t x, vfloat64m1_t y)
+{
+  return __riscv_vfsub(x,y,vl);
+}
+
+vfloat64m1_t add7(vfloat64m1_t x1, vfloat64m1_t x2, vfloat64m1_t x3, vfloat64m1_t x4, vfloat64m1_t x5, vfloat64m1_t x6, vfloat64m1_t x7)
+{
+  return __riscv_vfadd(x1,__riscv_vfadd(x2,__riscv_vfadd(x3,__riscv_vfadd(x4,__riscv_vfadd(x5,__riscv_vfadd(x6, x7, vl),vl),vl),vl),vl),vl);
+}
+#else
+#error "RISC-V vector target, but current intrinsics not supported."
+#endif
+#endif



View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/eb03f43de920c454c4bbb639f1bffc7aa4f7ff47

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/eb03f43de920c454c4bbb639f1bffc7aa4f7ff47
You're receiving this email because of your account on gitlab.haskell.org.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20250309/f7f48544/attachment-0001.html>