fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-01-08 17:10:11 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

i965/fs: Move the old gen4 bspec-based scheduling info to a helper func.

Browse source 
For gen7 everything changes, and we have actual information on latency.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
This commit is contained in:
Eric Anholt 2012-12-05 15:24:07 -08:00
parent 461a29783a
commit 6255fc7426
1 changed files with 41 additions and 33 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

74
src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
View file

@ -67,41 +67,11 @@ public:
this->parent_count = 0;
this->unblocked_time = 0;
int chans = 8;
int math_latency = 22;
switch (inst->opcode) {
case SHADER_OPCODE_RCP:
this->latency = 1 * chans * math_latency;
break;
case SHADER_OPCODE_RSQ:
this->latency = 2 * chans * math_latency;
break;
case SHADER_OPCODE_INT_QUOTIENT:
case SHADER_OPCODE_SQRT:
case SHADER_OPCODE_LOG2:
/* full precision log. partial is 2. */
this->latency = 3 * chans * math_latency;
break;
case SHADER_OPCODE_INT_REMAINDER:
case SHADER_OPCODE_EXP2:
/* full precision. partial is 3, same throughput. */
this->latency = 4 * chans * math_latency;
break;
case SHADER_OPCODE_POW:
this->latency = 8 * chans * math_latency;
break;
case SHADER_OPCODE_SIN:
case SHADER_OPCODE_COS:
/* minimum latency, max is 12 rounds. */
this->latency = 5 * chans * math_latency;
break;
default:
this->latency = 2;
break;
}
set_latency_gen4();
}
void set_latency_gen4();
fs_inst *inst;
schedule_node **children;
int *child_latency;
@ -112,6 +82,44 @@ public:
int latency;
};
void
schedule_node::set_latency_gen4()
{
int chans = 8;
int math_latency = 22;
switch (inst->opcode) {
case SHADER_OPCODE_RCP:
this->latency = 1 * chans * math_latency;
break;
case SHADER_OPCODE_RSQ:
this->latency = 2 * chans * math_latency;
break;
case SHADER_OPCODE_INT_QUOTIENT:
case SHADER_OPCODE_SQRT:
case SHADER_OPCODE_LOG2:
/* full precision log. partial is 2. */
this->latency = 3 * chans * math_latency;
break;
case SHADER_OPCODE_INT_REMAINDER:
case SHADER_OPCODE_EXP2:
/* full precision. partial is 3, same throughput. */
this->latency = 4 * chans * math_latency;
break;
case SHADER_OPCODE_POW:
this->latency = 8 * chans * math_latency;
break;
case SHADER_OPCODE_SIN:
case SHADER_OPCODE_COS:
/* minimum latency, max is 12 rounds. */
this->latency = 5 * chans * math_latency;
break;
default:
this->latency = 2;
break;
}
}
class instruction_scheduler {
public:
instruction_scheduler(fs_visitor *v, void *mem_ctx, int grf_count,