[Libclc-dev] [PATCH 1/1] Provide vstore_half helper to workaround clc restrictions

[Jan Vesely via Libclc-dev]

clang won't accept half precision loads and stores without cl_khr_fp16 since r281904
Tested on clover carrizo and iceland

Signed-off-by: Jan Vesely <jan.vesely at rutgers.edu>
---
 generic/lib/SOURCES                       |  1 +
 generic/lib/shared/vstore.cl              | 53 +++++++++++++++++++------------
 generic/lib/shared/vstore_half.inc        | 12 +++----
 generic/lib/shared/vstore_half_helpers.ll | 35 ++++++++++++++++++++
 4 files changed, 75 insertions(+), 26 deletions(-)
 create mode 100644 generic/lib/shared/vstore_half_helpers.ll

diff --git a/generic/lib/SOURCES b/generic/lib/SOURCES
index 423a50b..ecd2e73 100644
--- a/generic/lib/SOURCES
+++ b/generic/lib/SOURCES
@@ -142,6 +142,7 @@ shared/max.cl
 shared/min.cl
 shared/vload.cl
 shared/vstore.cl
+shared/vstore_half_helpers.ll
 workitem/get_global_id.cl
 workitem/get_global_size.cl
 image/get_image_dim.cl
diff --git a/generic/lib/shared/vstore.cl b/generic/lib/shared/vstore.cl
index ebc9446..2838384 100644
--- a/generic/lib/shared/vstore.cl
+++ b/generic/lib/shared/vstore.cl
@@ -52,32 +52,45 @@ VSTORE_TYPES()
 #endif
 
 /* vstore_half are legal even without cl_khr_fp16 */
+#define DECLARE_HELPER(STYPE, AS) void __clc_vstore_half_##STYPE##_helper##AS(STYPE, AS half *);
 
-#define VEC_STORE1(val) mem[offset++] = val;
-#define VEC_STORE2(val) \
-	VEC_STORE1(val.lo) \
-	VEC_STORE1(val.hi)
-#define VEC_STORE3(val) \
-	VEC_STORE1(val.s0) \
-	VEC_STORE1(val.s1) \
-	VEC_STORE1(val.s2)
-#define VEC_STORE4(val) \
-	VEC_STORE2(val.lo) \
-	VEC_STORE2(val.hi)
-#define VEC_STORE8(val) \
-	VEC_STORE4(val.lo) \
-	VEC_STORE4(val.hi)
-#define VEC_STORE16(val) \
-	VEC_STORE8(val.lo) \
-	VEC_STORE8(val.hi)
+DECLARE_HELPER(float, __private);
+DECLARE_HELPER(float, __global);
+DECLARE_HELPER(float, __local);
 
-#define __FUNC(SUFFIX, VEC_SIZE, TYPE, AS) \
+#ifdef cl_khr_fp64
+#pragma OPENCL EXTENSION cl_khr_fp64 : enable
+DECLARE_HELPER(double, __private);
+DECLARE_HELPER(double, __global);
+DECLARE_HELPER(double, __local);
+#endif
+
+
+#define VEC_STORE1(STYPE, AS, val) __clc_vstore_half_##STYPE##_helper##AS (val, &mem[offset++]);
+#define VEC_STORE2(STYPE, AS, val) \
+	VEC_STORE1(STYPE, AS, val.lo) \
+	VEC_STORE1(STYPE, AS, val.hi)
+#define VEC_STORE3(STYPE, AS, val) \
+	VEC_STORE1(STYPE, AS, val.s0) \
+	VEC_STORE1(STYPE, AS, val.s1) \
+	VEC_STORE1(STYPE, AS, val.s2)
+#define VEC_STORE4(STYPE, AS, val) \
+	VEC_STORE2(STYPE, AS, val.lo) \
+	VEC_STORE2(STYPE, AS, val.hi)
+#define VEC_STORE8(STYPE, AS, val) \
+	VEC_STORE4(STYPE, AS, val.lo) \
+	VEC_STORE4(STYPE, AS, val.hi)
+#define VEC_STORE16(STYPE, AS, val) \
+	VEC_STORE8(STYPE, AS, val.lo) \
+	VEC_STORE8(STYPE, AS, val.hi)
+
+#define __FUNC(SUFFIX, VEC_SIZE, TYPE, STYPE, AS) \
   _CLC_OVERLOAD _CLC_DEF void vstore_half##SUFFIX(TYPE vec, size_t offset, AS half *mem) { \
     offset *= VEC_SIZE; \
-    VEC_STORE##VEC_SIZE(vec) \
+    VEC_STORE##VEC_SIZE(STYPE, AS, vec) \
   }
 
-#define FUNC(SUFFIX, VEC_SIZE, TYPE, AS) __FUNC(SUFFIX, VEC_SIZE, TYPE, AS)
+#define FUNC(SUFFIX, VEC_SIZE, TYPE, STYPE, AS) __FUNC(SUFFIX, VEC_SIZE, TYPE, STYPE, AS)
 
 #define __CLC_BODY "vstore_half.inc"
 #include <clc/math/gentype.inc>
diff --git a/generic/lib/shared/vstore_half.inc b/generic/lib/shared/vstore_half.inc
index 8ed03a0..fee52bc 100644
--- a/generic/lib/shared/vstore_half.inc
+++ b/generic/lib/shared/vstore_half.inc
@@ -1,10 +1,10 @@
 
 #ifdef __CLC_VECSIZE
-  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __private);
-  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __local);
-  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __global);
+  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __private);
+  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __local);
+  FUNC(__CLC_VECSIZE, __CLC_VECSIZE, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __global);
 #else
-  FUNC(, 1, __CLC_GENTYPE, __private);
-  FUNC(, 1, __CLC_GENTYPE, __local);
-  FUNC(, 1, __CLC_GENTYPE, __global);
+  FUNC(, 1, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __private);
+  FUNC(, 1, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __local);
+  FUNC(, 1, __CLC_GENTYPE, __CLC_SCALAR_GENTYPE, __global);
 #endif
diff --git a/generic/lib/shared/vstore_half_helpers.ll b/generic/lib/shared/vstore_half_helpers.ll
new file mode 100644
index 0000000..e958664
--- /dev/null
+++ b/generic/lib/shared/vstore_half_helpers.ll
@@ -0,0 +1,35 @@
+define void @__clc_vstore_half_float_helper__private(float %data, half addrspace(0)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc float %data to half
+  store half %res, half addrspace(0)* %ptr
+  ret void
+}
+
+define void @__clc_vstore_half_float_helper__global(float %data, half addrspace(1)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc float %data to half
+  store half %res, half addrspace(1)* %ptr
+  ret void
+}
+
+define void @__clc_vstore_half_float_helper__local(float %data, half addrspace(3)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc float %data to half
+  store half %res, half addrspace(3)* %ptr
+  ret void
+}
+
+define void @__clc_vstore_half_double_helper__private(double %data, half addrspace(0)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc double %data to half
+  store half %res, half addrspace(0)* %ptr
+  ret void
+}
+
+define void @__clc_vstore_half_double_helper__global(double %data, half addrspace(1)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc double %data to half
+  store half %res, half addrspace(1)* %ptr
+  ret void
+}
+
+define void @__clc_vstore_half_double_helper__local(double %data, half addrspace(3)* nocapture %ptr) nounwind alwaysinline {
+  %res = fptrunc double %data to half
+  store half %res, half addrspace(3)* %ptr
+  ret void
+}
-- 
2.7.4