From 156f565f9eb36dad3cd959952724bc54f9ff21ea Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Fri, 21 Nov 2014 10:47:37 -0800 Subject: [PATCH] i965/vec4: Extract depctrl hazards Move this to a separate function so that we can begin to add other little caveats without making too big a mess. NOTE: There is some desire to improve this function eventually, but we need to fix a bug first. v2: Use const for the inst for the hazard check (Matt) Invert safe logic to get rid of the double negative (Matt) Add PRM reference for predicates (Matt) Add note about empirical evidence for math (Matt) Signed-off-by: Ben Widawsky Reviewed-by: Matt Turner --- src/mesa/drivers/dri/i965/brw_vec4.cpp | 49 +++++++++++++++++++--------------- 1 file changed, 27 insertions(+), 22 deletions(-) diff --git a/src/mesa/drivers/dri/i965/brw_vec4.cpp b/src/mesa/drivers/dri/i965/brw_vec4.cpp index df589b87c79..655299472ff 100644 --- a/src/mesa/drivers/dri/i965/brw_vec4.cpp +++ b/src/mesa/drivers/dri/i965/brw_vec4.cpp @@ -840,6 +840,32 @@ vec4_visitor::move_push_constants_to_pull_constants() pack_uniform_registers(); } +/* Conditions for which we want to avoid setting the dependency control bits */ +static bool +is_dep_ctrl_unsafe(const vec4_instruction *inst) +{ + /* + * mlen: + * In the presence of send messages, totally interrupt dependency + * control. They're long enough that the chance of dependency + * control around them just doesn't matter. + * + * predicate: + * From the Ivy Bridge PRM, volume 4 part 3.7, page 80: + * When a sequence of NoDDChk and NoDDClr are used, the last instruction that + * completes the scoreboard clear must have a non-zero execution mask. This + * means, if any kind of predication can change the execution mask or channel + * enable of the last instruction, the optimization must be avoided. This is + * to avoid instructions being shot down the pipeline when no writes are + * required. + * + * math: + * Dependency control does not work well over math instructions. + * NB: Discovered empirically + */ + return (inst->mlen || inst->predicate || inst->is_math()); +} + /** * Sets the dependency control fields on instructions after register * allocation and before the generator is run. @@ -885,28 +911,7 @@ vec4_visitor::opt_set_dependency_control() assert(inst->src[i].file != MRF); } - /* In the presence of send messages, totally interrupt dependency - * control. They're long enough that the chance of dependency - * control around them just doesn't matter. - */ - if (inst->mlen) { - memset(last_grf_write, 0, sizeof(last_grf_write)); - memset(last_mrf_write, 0, sizeof(last_mrf_write)); - continue; - } - - /* It looks like setting dependency control on a predicated - * instruction hangs the GPU. - */ - if (inst->predicate) { - memset(last_grf_write, 0, sizeof(last_grf_write)); - memset(last_mrf_write, 0, sizeof(last_mrf_write)); - continue; - } - - /* Dependency control does not work well over math instructions. - */ - if (inst->is_math()) { + if (is_dep_ctrl_unsafe(inst)) { memset(last_grf_write, 0, sizeof(last_grf_write)); memset(last_mrf_write, 0, sizeof(last_mrf_write)); continue; -- 2.11.0