AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Support wave32/wave64 scratch allocations on gfx10

Browse Source 
- Use new buffer resource descriptor layout
- Handle wave32 scratch allocation error from CP
- Make wavefront size a property of scratch allocation requests
- Repurpose wave64-specific amd_queue_t.scratch_workitem_byte_size field
- Clear index_stride field in V# on gfx10, calculated per-dispatch by CP

Change-Id: If2acdf6430772abd4d6a8c792fc8c11260764dda
This commit is contained in:
Jay Cornwall
2018-09-18 15:09:08 -05:00
committed by Chris Freehill
parent 08841faf4c
commit f8d0ccd159
5 changed files with 103 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files
runtime/hsa-runtime/core/inc/amd_gpu_agent.h
+1
View File
@@ -67,6 +67,7 @@ struct ScratchInfo {
void* queue_base;
size_t size;
size_t size_per_thread;
uint32_t lanes_per_wave;
ptrdiff_t queue_process_offset;
bool large;
bool retry;
runtime/hsa-runtime/core/inc/registers.h
+38
View File
@@ -85,6 +85,10 @@ BUF_NUM_FORMAT_RESERVED_6__VI = 0x00000006,
BUF_NUM_FORMAT_FLOAT = 0x00000007,
} BUF_NUM_FORMAT;
typedef enum BUF_FORMAT {
BUF_FORMAT_32_UINT = 0x00000014,
} BUF_FORMAT;
typedef enum SQ_SEL_XYZW01 {
SQ_SEL_0 = 0x00000000,
SQ_SEL_1 = 0x00000001,
@@ -201,4 +205,38 @@ SQ_SEL_W = 0x00000007,
float f32All;
};
union SQ_BUF_RSRC_WORD3_GFX10 {
struct {
#if defined(LITTLEENDIAN_CPU)
unsigned int DST_SEL_X : 3;
unsigned int DST_SEL_Y : 3;
unsigned int DST_SEL_Z : 3;
unsigned int DST_SEL_W : 3;
unsigned int FORMAT : 7;
unsigned int RESERVED1 : 2;
unsigned int INDEX_STRIDE : 2;
unsigned int ADD_TID_ENABLE : 1;
unsigned int RESOURCE_LEVEL : 1;
unsigned int RESERVED2 : 3;
unsigned int OOB_SELECT : 2;
unsigned int TYPE : 2;
#elif defined(BIGENDIAN_CPU)
unsigned int TYPE : 2;
unsigned int OOB_SELECT : 2;
unsigned int RESERVED2 : 3;
unsigned int RESOURCE_LEVEL : 1;
unsigned int ADD_TID_ENABLE : 1;
unsigned int INDEX_STRIDE : 2;
unsigned int RESERVED1 : 2;
unsigned int FORMAT : 7;
unsigned int DST_SEL_W : 3;
unsigned int DST_SEL_Z : 3;
unsigned int DST_SEL_Y : 3;
unsigned int DST_SEL_X : 3;
#endif
} bitfields, bits;
unsigned int u32All;
signed int i32All;
float f32All;
};
#endif // header guard