i965/fs: Lower 32x32 bit multiplication on BXT.

[mesa.git] / src / mesa / drivers / dri / i965 / brw_vec4_cse.cpp

diff --git a/src/mesa/drivers/dri/i965/brw_vec4_cse.cpp b/src/mesa/drivers/dri/i965/brw_vec4_cse.cpp
index 489248ee351cc493b8e3cc71335416a2232c3622..5a277f74c4401a9c7e54994aac643a121cdd3f02 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_vec4_cse.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_cse.cpp
@@ -62,6 +62,7 @@ is_expression(const vec4_instruction *const inst)
    case BRW_OPCODE_CMPN:
    case BRW_OPCODE_ADD:
    case BRW_OPCODE_MUL:
+   case SHADER_OPCODE_MULH:
    case BRW_OPCODE_FRC:
    case BRW_OPCODE_RNDU:
    case BRW_OPCODE_RNDD:
@@ -72,6 +73,8 @@ is_expression(const vec4_instruction *const inst)
    case BRW_OPCODE_MAD:
    case BRW_OPCODE_LRP:
    case VEC4_OPCODE_UNPACK_UNIFORM:
+   case SHADER_OPCODE_FIND_LIVE_CHANNEL:
+   case SHADER_OPCODE_BROADCAST:
       return true;
    case SHADER_OPCODE_RCP:
    case SHADER_OPCODE_RSQ:
@@ -112,8 +115,16 @@ instructions_match(vec4_instruction *a, vec4_instruction *b)
 {
    return a->opcode == b->opcode &&
           a->saturate == b->saturate &&
+          a->predicate == b->predicate &&
+          a->predicate_inverse == b->predicate_inverse &&
           a->conditional_mod == b->conditional_mod &&
+          a->flag_subreg == b->flag_subreg &&
           a->dst.type == b->dst.type &&
+          a->offset == b->offset &&
+          a->mlen == b->mlen &&
+          a->base_mrf == b->base_mrf &&
+          a->header_size == b->header_size &&
+          a->shadow_compare == b->shadow_compare &&
           a->dst.writemask == b->dst.writemask &&
           a->force_writemask_all == b->force_writemask_all &&
           a->regs_written == b->regs_written &&
@@ -232,13 +243,7 @@ vec4_visitor::opt_cse_local(bblock_t *block)
              * more -- a sure sign they'll fail operands_match().
              */
             if (src->file == GRF) {
-               assert((unsigned)(src->reg * 4 + 3) < (alloc.count * 4));
-
-               int last_reg_use = MAX2(MAX2(virtual_grf_end[src->reg * 4 + 0],
-                                            virtual_grf_end[src->reg * 4 + 1]),
-                                       MAX2(virtual_grf_end[src->reg * 4 + 2],
-                                            virtual_grf_end[src->reg * 4 + 3]));
-               if (last_reg_use < ip) {
+               if (var_range_end(var_from_reg(alloc, *src), 4) < ip) {
                   entry->remove();
                   ralloc_free(entry);
                   break;