From: Junyan He <junyan...@linux.intel.com>
Because of the HW limitation, vertical stride is at
least aligned to 2. For 1D array image, the data has interval.
The image size is just twice as big as the buffer size we think.
Use clEnqueueWriteImage is safe and fix this bug.
Signed-off-by: Junyan He <junyan...@linux.intel.com>
---
utests/compiler_fill_image_1d_array.cpp | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/utests/compiler_fill_image_1d_array.cpp
b/utests/compiler_fill_image_1d_array.cpp
index cc7cf0a..67f9643 100644
--- a/utests/compiler_fill_image_1d_array.cpp
+++ b/utests/compiler_fill_image_1d_array.cpp
@@ -10,6 +10,11 @@ static void compiler_fill_image_1d_array(void)
size_t origin[3] = { };
size_t region[3];
uint32_t* dst;
+ uint32_t* src;
+
+ region[0] = w;
+ region[1] = array;
+ region[2] = 1;
memset(&desc, 0x0, sizeof(cl_image_desc));
memset(&format, 0x0, sizeof(cl_image_format));
@@ -26,9 +31,9 @@ static void compiler_fill_image_1d_array(void)
OCL_CREATE_IMAGE(buf[0], 0, &format, &desc, NULL);
- OCL_MAP_BUFFER_GTT(0);
- memset(buf_data[0], 0, sizeof(uint32_t) * w * array);
- OCL_UNMAP_BUFFER_GTT(0);
+ src = (uint32_t*)malloc(w*array*sizeof(uint32_t));
+ memset(src, 0, sizeof(uint32_t) * w * array);
+ OCL_WRITE_IMAGE(buf[0], origin, region, src);
// Run the kernel
OCL_SET_ARG(0, sizeof(cl_mem), &buf[0]);
@@ -39,9 +44,6 @@ static void compiler_fill_image_1d_array(void)
OCL_NDRANGE(2);
// Check result
- region[0] = w;
- region[1] = array;
- region[2] = 1;
dst = (uint32_t*)malloc(w*array*sizeof(uint32_t));
OCL_READ_IMAGE(buf[0], origin, region, dst);
--
1.9.1
_______________________________________________
Beignet mailing list
Beignet@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/beignet
