[Mesa-dev] [PATCH 3/3] mesa: don't check mapped buffers in every draw call if drivers allow it

[Thomas Helland]

Would this be a win for a game limited by the winsys/buffer handling?
I played some Hitman the other day, and noticed my gpu-utilization
was hovering at about 80%. Some profiling later and I concluded
that one thread was pegged at 100% doing buffer handling work.
amdgpu_cs_add_buffer, amdgpu_lookup_buffer, the cso_hash,
and a couple other functions related to the winsys where the hottest.
Should I expect that to be mitigated by your threaded_gallium work,
or by this patch series? If so, I will not delve into that much further =)

Oh, and excuse me if I'm totally of track here. Just starting to make
sense of the gallium stuff, as I've just recently acquired a couple
radeon powered PC's =)

2017-05-15 16:41 GMT+02:00 Marek Olšák <maraeo at gmail.com>:
> From: Marek Olšák <marek.olsak at amd.com>
>
> Before: DrawElements (16 VBOs) w/ no state change: 4.34 million/s
> After:  DrawElements (16 VBOs) w/ no state change: 8.80 million/s
>
> This inefficiency was uncovered by Timothy Arceri's no_error work.
> ---
>  src/mesa/main/api_validate.c | 15 ++++++++++++++-
>  1 file changed, 14 insertions(+), 1 deletion(-)
>
> diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c
> index cbb2361..b804e05 100644
> --- a/src/mesa/main/api_validate.c
> +++ b/src/mesa/main/api_validate.c
> @@ -236,21 +236,34 @@ _mesa_valid_to_render(struct gl_context *ctx, const char *where)
>  /**
>   * Check if OK to draw arrays/elements.
>   */
>  static bool
>  check_valid_to_render(struct gl_context *ctx, const char *function)
>  {
>     if (!_mesa_valid_to_render(ctx, function)) {
>        return false;
>     }
>
> -   if (!_mesa_all_buffers_are_unmapped(ctx->Array.VAO)) {
> +   /* Section 6.3.2 from the GL 4.5:
> +    * "Any GL command which attempts to read from, write to, or change
> +    *  the state of a buffer object may generate an INVALID_OPERATION error if
> +    *  all or part of the buffer object is mapped ... However, only commands
> +    *  which explicitly describe this error are required to do so. If an error
> +    *  is not generated, such commands will have undefined results and may
> +    *  result in GL interruption or termination."
> +    *
> +    * Only some buffer API functions require INVALID_OPERATION with mapped
> +    * buffers. No other functions list such an error, thus it's not required
> +    * to report INVALID_OPERATION for draw calls with mapped buffers.
> +    */
> +   if (!ctx->Const.AllowMappedBuffersDuringExecution &&
> +       !_mesa_all_buffers_are_unmapped(ctx->Array.VAO)) {
>        _mesa_error(ctx, GL_INVALID_OPERATION,
>                    "%s(vertex buffers are mapped)", function);
>        return false;
>     }
>
>     /* Section 11.2 (Tessellation) of the ES 3.2 spec says:
>      *
>      * "An INVALID_OPERATION error is generated by any command that
>      *  transfers vertices to the GL if the current program state has
>      *  one but not both of a tessellation control shader and tessellation
> --
> 2.7.4
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev