diff options
author | Dave Airlie <[email protected]> | 2011-04-24 08:33:19 +1000 |
---|---|---|
committer | Dave Airlie <[email protected]> | 2011-04-24 08:33:19 +1000 |
commit | 77dc4c154cc6a8c3bc2369112c46f0be6ac444ae (patch) | |
tree | ec01a0f1988f7c722d395854ffa35a58a20c22e4 /src | |
parent | 22d5a71e4dedfc047e1a14082b7d0d287cfe910e (diff) |
r600g: fix bank swizzle calcs for scalar only operations.
In the initial code if we had nothing in the vector slots r would
never get reset to 0, so we'd fail to compile shaders, after the previous
commit this would happen for the LIT tests. When I fixed that we did a lot
of unnecessary loops through all the vector states when we had no vector
slots filled. So this patch optimises thing for the scalar only state.
This fixes the 3 LIT piglit tests on r600g.
Signed-off-by: Dave Airlie <[email protected]>
Diffstat (limited to 'src')
-rw-r--r-- | src/gallium/drivers/r600/r600_asm.c | 40 |
1 files changed, 25 insertions, 15 deletions
diff --git a/src/gallium/drivers/r600/r600_asm.c b/src/gallium/drivers/r600/r600_asm.c index 4cdd1ed698f..efb22fdb8ee 100644 --- a/src/gallium/drivers/r600/r600_asm.c +++ b/src/gallium/drivers/r600/r600_asm.c @@ -725,13 +725,15 @@ static int check_and_set_bank_swizzle(struct r600_bc *bc, struct alu_bank_swizzle bs; int bank_swizzle[5]; int i, r = 0, forced = 0; - - for (i = 0; i < 5; i++) + boolean scalar_only = true; + for (i = 0; i < 5; i++) { if (slots[i] && slots[i]->bank_swizzle_force) { slots[i]->bank_swizzle = slots[i]->bank_swizzle_force; forced = 1; } - + if (i < 4 && slots[i]) + scalar_only = false; + } if (forced) return 0; @@ -742,13 +744,17 @@ static int check_and_set_bank_swizzle(struct r600_bc *bc, bank_swizzle[4] = SQ_ALU_SCL_210; while(bank_swizzle[4] <= SQ_ALU_SCL_221) { init_bank_swizzle(&bs); - for (i = 0; i < 4; i++) { - if (slots[i]) { - r = check_vector(bc, slots[i], &bs, bank_swizzle[i]); - if (r) - break; + if (scalar_only == false) { + for (i = 0; i < 4; i++) { + if (slots[i]) { + r = check_vector(bc, slots[i], &bs, bank_swizzle[i]); + if (r) + break; + } } - } + } else + r = 0; + if (!r && slots[4]) { r = check_scalar(bc, slots[4], &bs, bank_swizzle[4]); } @@ -760,12 +766,16 @@ static int check_and_set_bank_swizzle(struct r600_bc *bc, return 0; } - for (i = 0; i < 5; i++) { - bank_swizzle[i]++; - if (bank_swizzle[i] <= SQ_ALU_VEC_210) - break; - else - bank_swizzle[i] = SQ_ALU_VEC_012; + if (scalar_only) { + bank_swizzle[4]++; + } else { + for (i = 0; i < 5; i++) { + bank_swizzle[i]++; + if (bank_swizzle[i] <= SQ_ALU_VEC_210) + break; + else + bank_swizzle[i] = SQ_ALU_VEC_012; + } } } |