radeonsi/gfx9: reduce max threads per block to 1024 on gfx9+

The number of supported waves per thread group has been reduced to 16
with gfx9. Trying to use 32 waves causes hangs, and barriers might
not work correctly with > 16 waves.

Cc: [email protected]
Reviewed-by: Marek Olšák <[email protected]>
(cherry picked from commit a0e6b9a2db5aa5f06a4f60d270aca8344e7d8b3f)
[Emil Velikov: add a HAVE_LLVM check, as applicable in branch]
Signed-off-by: Emil Velikov <[email protected]>

Conflicts:
	src/gallium/drivers/radeon/r600_pipe_common.c

1 files changed, 27 insertions, 15 deletions

diff --git a/src/gallium/drivers/radeon/r600_pipe_common.c b/src/gallium/drivers/radeon/r600_pipe_common.c
index 3b490404d39..2019ecdd577 100644
--- a/src/gallium/drivers/radeon/r600_pipe_common.c
+++ b/src/gallium/drivers/radeon/r600_pipe_common.c
@@ -1014,6 +1014,28 @@ const char *r600_get_llvm_processor_name(enum radeon_family family)
 	}
 }
 
+static unsigned get_max_threads_per_block(struct r600_common_screen *screen,
+					  enum pipe_shader_ir ir_type)
+{
+	if (ir_type != PIPE_SHADER_IR_TGSI)
+		return 256;
+
+	if (HAVE_LLVM < 0x309)
+		return 256;
+
+	/* Only 16 waves per thread-group on gfx9. */
+	if (screen->chip_class >= GFX9)
+		return 1024;
+
+	/* Up to 40 waves per thread-group on GCN < gfx9. Expose a nice
+	 * round number.
+	 */
+	if (screen->chip_class >= SI)
+		return 2048;
+
+	return 256;
+}
+
 static int r600_get_compute_param(struct pipe_screen *screen,
         enum pipe_shader_ir ir_type,
         enum pipe_compute_cap param,
@@ -1068,27 +1090,17 @@ static int r600_get_compute_param(struct pipe_screen *screen,
 	case PIPE_COMPUTE_CAP_MAX_BLOCK_SIZE:
 		if (ret) {
 			uint64_t *block_size = ret;
-			if (rscreen->chip_class >= SI && HAVE_LLVM >= 0x309 &&
-			    ir_type == PIPE_SHADER_IR_TGSI) {
-				block_size[0] = 2048;
-				block_size[1] = 2048;
-				block_size[2] = 2048;
-			} else {
-				block_size[0] = 256;
-				block_size[1] = 256;
-				block_size[2] = 256;
-			}
+			unsigned threads_per_block = get_max_threads_per_block(rscreen, ir_type);
+			block_size[0] = threads_per_block;
+			block_size[1] = threads_per_block;
+			block_size[2] = threads_per_block;
 		}
 		return 3 * sizeof(uint64_t);
 
 	case PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK:
 		if (ret) {
 			uint64_t *max_threads_per_block = ret;
-			if (rscreen->chip_class >= SI && HAVE_LLVM >= 0x309 &&
-			    ir_type == PIPE_SHADER_IR_TGSI)
-				*max_threads_per_block = 2048;
-			else
-				*max_threads_per_block = 256;
+			*max_threads_per_block = get_max_threads_per_block(rscreen, ir_type);
 		}
 		return sizeof(uint64_t);
 	case PIPE_COMPUTE_CAP_ADDRESS_BITS: