[Mesa-dev] [PATCH 01/17] i965/vec4/fs: Count loops in shader debug

[Matt Turner]

On Mon, Aug 18, 2014 at 5:17 AM, Abdiel Janulgue
<abdiel.janulgue at linux.intel.com> wrote:
> Signed-off-by: Abdiel Janulgue <abdiel.janulgue at linux.intel.com>
> ---
>  src/mesa/drivers/dri/i965/brw_fs_generator.cpp   | 6 ++++--
>  src/mesa/drivers/dri/i965/brw_vec4.h             | 1 +
>  src/mesa/drivers/dri/i965/brw_vec4_generator.cpp | 6 ++++--
>  3 files changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
> index a243003..c4e6c6d 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
> @@ -1483,6 +1483,7 @@ void
>  fs_generator::generate_code(exec_list *instructions)
>  {
>     int start_offset = p->next_insn_offset;
> +   int loop_count = 0;
>
>     struct annotation_info annotation;
>     memset(&annotation, 0, sizeof(annotation));
> @@ -1743,6 +1744,7 @@ fs_generator::generate_code(exec_list *instructions)
>
>        case BRW_OPCODE_WHILE:
>          brw_WHILE(p);
> +         loop_count++;
>          break;
>
>        case SHADER_OPCODE_RCP:
> @@ -1970,9 +1972,9 @@ fs_generator::generate_code(exec_list *instructions)
>           fprintf(stderr, "Native code for blorp program (SIMD%d dispatch):\n",
>                   dispatch_width);
>        }
> -      fprintf(stderr, "SIMD%d shader: %d instructions. Compacted %d to %d"
> +      fprintf(stderr, "SIMD%d shader: %d instructions. %d loops. Compacted %d to %d"
>                        " bytes (%.0f%%)\n",
> -              dispatch_width, before_size / 16, before_size, after_size,
> +              dispatch_width, before_size / 16, loop_count, before_size, after_size,
>                100.0f * (before_size - after_size) / before_size);
>
>        const struct gl_program *prog = fp ? &fp->Base : NULL;
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4.h b/src/mesa/drivers/dri/i965/brw_vec4.h
> index c59d24f..c333baa 100644
> --- a/src/mesa/drivers/dri/i965/brw_vec4.h
> +++ b/src/mesa/drivers/dri/i965/brw_vec4.h
> @@ -699,6 +699,7 @@ private:
>
>     void *mem_ctx;
>     const bool debug_flag;
> +   int loop_count;

Instead of adding a loop_count variable to the class, could you inline
the body of generate_vec4_instruction() into generate_code(), like we
do in the fs backend (in a separate patch before this)?

>  };
>
>  } /* namespace brw */
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> index 1b1e647..b8948c3 100644
> --- a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> @@ -1188,6 +1188,7 @@ vec4_generator::generate_vec4_instruction(vec4_instruction *instruction,
>
>     case BRW_OPCODE_WHILE:
>        brw_WHILE(p);
> +      loop_count++;
>        break;
>
>     case SHADER_OPCODE_RCP:
> @@ -1318,6 +1319,7 @@ vec4_generator::generate_code(exec_list *instructions)
>  {
>     struct annotation_info annotation;
>     memset(&annotation, 0, sizeof(annotation));
> +   loop_count = 0;
>
>     cfg_t *cfg = NULL;
>     if (unlikely(debug_flag))
> @@ -1372,9 +1374,9 @@ vec4_generator::generate_code(exec_list *instructions)
>        } else {
>           fprintf(stderr, "Native code for vertex program %d:\n", prog->Id);
>        }
> -      fprintf(stderr, "vec4 shader: %d instructions. Compacted %d to %d"
> +      fprintf(stderr, "vec4 shader: %d instructions. %d loops. Compacted %d to %d"
>                        " bytes (%.0f%%)\n",
> -              before_size / 16, before_size, after_size,
> +              before_size / 16, loop_count, before_size, after_size,
>                100.0f * (before_size - after_size) / before_size);
>
>        dump_assembly(p->store, annotation.ann_count, annotation.ann, brw, prog);
> --
> 1.9.1

Otherwise looks good.