From 4e1d51786e0657c7430d731ac464f2a73e32eddf Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Fri, 23 Oct 2009 13:49:04 +0100 Subject: gallium: remove noise opcodes Provide a dummy implementation in the GL state tracker (move 0.5 to the destination regs). At some point, a motivated person could add a better implementation of noise. Currently not even the nvidia binary drivers do anything more than this. In any case, the place to do this is in the GL state tracker, not the poor driver. --- src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h | 4 ---- 1 file changed, 4 deletions(-) (limited to 'src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h') diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h index e7bcf4bf75..d321f013d0 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h +++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h @@ -139,10 +139,6 @@ OP00_LBL(BGNLOOP) OP00(BGNSUB) OP00_LBL(ENDLOOP) OP00(ENDSUB) -OP11(NOISE1) -OP11(NOISE2) -OP11(NOISE3) -OP11(NOISE4) OP00(NOP) OP11(NRM4) OP01(CALLNZ) -- cgit v1.2.3 From b9cb74c7f826dfd320f5e5b54aa933898f7ddd3d Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Fri, 23 Oct 2009 14:31:24 +0100 Subject: gallium: remove the swizzling parts of ExtSwizzle These haven't been used by the mesa state tracker since the conversion to tgsi_ureg, and it seems that none of the other state trackers are using it either. This helps simplify one of the biggest suprises when starting off with TGSI shaders. --- src/gallium/auxiliary/draw/draw_vs_aos.c | 30 ++--------- src/gallium/auxiliary/tgsi/tgsi_build.c | 27 +--------- src/gallium/auxiliary/tgsi/tgsi_build.h | 4 -- src/gallium/auxiliary/tgsi/tgsi_dump.c | 27 ---------- src/gallium/auxiliary/tgsi/tgsi_dump_c.c | 26 --------- src/gallium/auxiliary/tgsi/tgsi_exec.c | 34 ++++-------- src/gallium/auxiliary/tgsi/tgsi_info.c | 2 +- src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h | 1 - src/gallium/auxiliary/tgsi/tgsi_ppc.c | 26 +++------ src/gallium/auxiliary/tgsi/tgsi_scan.c | 5 -- src/gallium/auxiliary/tgsi/tgsi_sse2.c | 34 +++--------- src/gallium/auxiliary/tgsi/tgsi_text.c | 28 ++-------- src/gallium/auxiliary/tgsi/tgsi_util.c | 74 ++------------------------ src/gallium/auxiliary/tgsi/tgsi_util.h | 12 +---- src/gallium/drivers/cell/ppu/cell_gen_fp.c | 18 ++----- src/gallium/drivers/cell/spu/spu_exec.c | 27 +++------- src/gallium/drivers/cell/spu/spu_util.c | 48 +---------------- src/gallium/drivers/i915/i915_fpc_translate.c | 20 ++----- src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c | 26 +++------ src/gallium/drivers/nv30/nv30_fragprog.c | 24 ++------- src/gallium/drivers/nv40/nv40_fragprog.c | 18 ++----- src/gallium/drivers/nv40/nv40_vertprog.c | 18 ++----- src/gallium/drivers/nv50/nv50_program.c | 28 +++------- src/gallium/drivers/r300/r300_tgsi_to_rc.c | 9 ++-- src/gallium/include/pipe/p_shader_tokens.h | 17 +----- src/mesa/state_tracker/st_mesa_to_tgsi.c | 2 - 26 files changed, 96 insertions(+), 489 deletions(-) (limited to 'src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h') diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c index 645d7cccba..88bc790b62 100644 --- a/src/gallium/auxiliary/draw/draw_vs_aos.c +++ b/src/gallium/auxiliary/draw/draw_vs_aos.c @@ -537,19 +537,10 @@ static struct x86_reg fetch_src( struct aos_compilation *cp, unsigned abs = 0; for (i = 0; i < 4; i++) { - unsigned swizzle = tgsi_util_get_full_src_register_extswizzle( src, i ); + unsigned swizzle = tgsi_util_get_full_src_register_swizzle( src, i ); unsigned neg = tgsi_util_get_full_src_register_sign_mode( src, i ); - switch (swizzle) { - case TGSI_EXTSWIZZLE_ZERO: - case TGSI_EXTSWIZZLE_ONE: - AOS_ERROR(cp, "not supporting full swizzles yet in tgsi_aos_sse2"); - break; - - default: - swz |= (swizzle & 0x3) << (i * 2); - break; - } + swz |= (swizzle & 0x3) << (i * 2); switch (neg) { case TGSI_UTIL_SIGN_TOGGLE: @@ -632,23 +623,10 @@ static void x87_fld_src( struct aos_compilation *cp, src->SrcRegister.File, src->SrcRegister.Index); - unsigned swizzle = tgsi_util_get_full_src_register_extswizzle( src, channel ); + unsigned swizzle = tgsi_util_get_full_src_register_swizzle( src, channel ); unsigned neg = tgsi_util_get_full_src_register_sign_mode( src, channel ); - switch (swizzle) { - case TGSI_EXTSWIZZLE_ZERO: - x87_fldz( cp->func ); - break; - - case TGSI_EXTSWIZZLE_ONE: - x87_fld1( cp->func ); - break; - - default: - x87_fld( cp->func, x86_make_disp(arg0, (swizzle & 3) * sizeof(float)) ); - break; - } - + x87_fld( cp->func, x86_make_disp(arg0, (swizzle & 3) * sizeof(float)) ); switch (neg) { case TGSI_UTIL_SIGN_TOGGLE: diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c index e0cfc54420..98d36f43e4 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_build.c +++ b/src/gallium/auxiliary/tgsi/tgsi_build.c @@ -692,12 +692,8 @@ tgsi_build_full_instruction( tgsi_default_src_register_ext_swz() ) ) { struct tgsi_src_register_ext_swz *src_register_ext_swz; - /* Use of the extended swizzle requires the simple swizzle to be identity. + /* Use of the extended negate requires the simple negate to be identity. */ - assert( reg->SrcRegister.SwizzleX == TGSI_SWIZZLE_X ); - assert( reg->SrcRegister.SwizzleY == TGSI_SWIZZLE_Y ); - assert( reg->SrcRegister.SwizzleZ == TGSI_SWIZZLE_Z ); - assert( reg->SrcRegister.SwizzleW == TGSI_SWIZZLE_W ); assert( reg->SrcRegister.Negate == FALSE ); if( maxsize <= size ) @@ -707,10 +703,6 @@ tgsi_build_full_instruction( size++; *src_register_ext_swz = tgsi_build_src_register_ext_swz( - reg->SrcRegisterExtSwz.ExtSwizzleX, - reg->SrcRegisterExtSwz.ExtSwizzleY, - reg->SrcRegisterExtSwz.ExtSwizzleZ, - reg->SrcRegisterExtSwz.ExtSwizzleW, reg->SrcRegisterExtSwz.NegateX, reg->SrcRegisterExtSwz.NegateY, reg->SrcRegisterExtSwz.NegateZ, @@ -1048,10 +1040,7 @@ tgsi_default_src_register_ext_swz( void ) struct tgsi_src_register_ext_swz src_register_ext_swz; src_register_ext_swz.Type = TGSI_SRC_REGISTER_EXT_TYPE_SWZ; - src_register_ext_swz.ExtSwizzleX = TGSI_EXTSWIZZLE_X; - src_register_ext_swz.ExtSwizzleY = TGSI_EXTSWIZZLE_Y; - src_register_ext_swz.ExtSwizzleZ = TGSI_EXTSWIZZLE_Z; - src_register_ext_swz.ExtSwizzleW = TGSI_EXTSWIZZLE_W; + src_register_ext_swz.Padding0 = 0; src_register_ext_swz.NegateX = 0; src_register_ext_swz.NegateY = 0; src_register_ext_swz.NegateZ = 0; @@ -1074,10 +1063,6 @@ tgsi_compare_src_register_ext_swz( struct tgsi_src_register_ext_swz tgsi_build_src_register_ext_swz( - unsigned ext_swizzle_x, - unsigned ext_swizzle_y, - unsigned ext_swizzle_z, - unsigned ext_swizzle_w, unsigned negate_x, unsigned negate_y, unsigned negate_z, @@ -1088,20 +1073,12 @@ tgsi_build_src_register_ext_swz( { struct tgsi_src_register_ext_swz src_register_ext_swz; - assert( ext_swizzle_x <= TGSI_EXTSWIZZLE_ONE ); - assert( ext_swizzle_y <= TGSI_EXTSWIZZLE_ONE ); - assert( ext_swizzle_z <= TGSI_EXTSWIZZLE_ONE ); - assert( ext_swizzle_w <= TGSI_EXTSWIZZLE_ONE ); assert( negate_x <= 1 ); assert( negate_y <= 1 ); assert( negate_z <= 1 ); assert( negate_w <= 1 ); src_register_ext_swz = tgsi_default_src_register_ext_swz(); - src_register_ext_swz.ExtSwizzleX = ext_swizzle_x; - src_register_ext_swz.ExtSwizzleY = ext_swizzle_y; - src_register_ext_swz.ExtSwizzleZ = ext_swizzle_z; - src_register_ext_swz.ExtSwizzleW = ext_swizzle_w; src_register_ext_swz.NegateX = negate_x; src_register_ext_swz.NegateY = negate_y; src_register_ext_swz.NegateZ = negate_z; diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.h b/src/gallium/auxiliary/tgsi/tgsi_build.h index 17d977b059..5e0062a96f 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_build.h +++ b/src/gallium/auxiliary/tgsi/tgsi_build.h @@ -241,10 +241,6 @@ tgsi_compare_src_register_ext_swz( struct tgsi_src_register_ext_swz tgsi_build_src_register_ext_swz( - unsigned ext_swizzle_x, - unsigned ext_swizzle_y, - unsigned ext_swizzle_z, - unsigned ext_swizzle_w, unsigned negate_x, unsigned negate_y, unsigned negate_z, diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c index 111d95b666..3a584a10a1 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_dump.c +++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c @@ -148,15 +148,6 @@ static const char *texture_names[] = "SHADOWRECT" }; -static const char *extswizzle_names[] = -{ - "x", - "y", - "z", - "w", - "0", - "1" -}; static const char *modulate_names[TGSI_MODULATE_COUNT] = { @@ -446,24 +437,6 @@ iter_instruction( ENM( src->SrcRegister.SwizzleZ, swizzle_names ); ENM( src->SrcRegister.SwizzleW, swizzle_names ); } - if (src->SrcRegisterExtSwz.ExtSwizzleX != TGSI_EXTSWIZZLE_X || - src->SrcRegisterExtSwz.ExtSwizzleY != TGSI_EXTSWIZZLE_Y || - src->SrcRegisterExtSwz.ExtSwizzleZ != TGSI_EXTSWIZZLE_Z || - src->SrcRegisterExtSwz.ExtSwizzleW != TGSI_EXTSWIZZLE_W) { - CHR( '.' ); - if (src->SrcRegisterExtSwz.NegateX) - TXT("-"); - ENM( src->SrcRegisterExtSwz.ExtSwizzleX, extswizzle_names ); - if (src->SrcRegisterExtSwz.NegateY) - TXT("-"); - ENM( src->SrcRegisterExtSwz.ExtSwizzleY, extswizzle_names ); - if (src->SrcRegisterExtSwz.NegateZ) - TXT("-"); - ENM( src->SrcRegisterExtSwz.ExtSwizzleZ, extswizzle_names ); - if (src->SrcRegisterExtSwz.NegateW) - TXT("-"); - ENM( src->SrcRegisterExtSwz.ExtSwizzleW, extswizzle_names ); - } if (src->SrcRegisterExtMod.Complement) CHR( ')' ); diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c index 4a9c02b141..4f59ed22b5 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c +++ b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c @@ -167,16 +167,6 @@ static const char *TGSI_SRC_REGISTER_EXTS[] = "SRC_REGISTER_EXT_TYPE_MOD" }; -static const char *TGSI_EXTSWIZZLES[] = -{ - "EXTSWIZZLE_X", - "EXTSWIZZLE_Y", - "EXTSWIZZLE_Z", - "EXTSWIZZLE_W", - "EXTSWIZZLE_ZERO", - "EXTSWIZZLE_ONE" -}; - static const char *TGSI_WRITEMASKS[] = { "0", @@ -560,22 +550,6 @@ dump_instruction_verbose( EOL(); TXT( "\nType : " ); ENM( src->SrcRegisterExtSwz.Type, TGSI_SRC_REGISTER_EXTS ); - if( deflt || fs->SrcRegisterExtSwz.ExtSwizzleX != src->SrcRegisterExtSwz.ExtSwizzleX ) { - TXT( "\nExtSwizzleX: " ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleX, TGSI_EXTSWIZZLES ); - } - if( deflt || fs->SrcRegisterExtSwz.ExtSwizzleY != src->SrcRegisterExtSwz.ExtSwizzleY ) { - TXT( "\nExtSwizzleY: " ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleY, TGSI_EXTSWIZZLES ); - } - if( deflt || fs->SrcRegisterExtSwz.ExtSwizzleZ != src->SrcRegisterExtSwz.ExtSwizzleZ ) { - TXT( "\nExtSwizzleZ: " ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleZ, TGSI_EXTSWIZZLES ); - } - if( deflt || fs->SrcRegisterExtSwz.ExtSwizzleW != src->SrcRegisterExtSwz.ExtSwizzleW ) { - TXT( "\nExtSwizzleW: " ); - ENM( src->SrcRegisterExtSwz.ExtSwizzleW, TGSI_EXTSWIZZLES ); - } if( deflt || fs->SrcRegisterExtSwz.NegateX != src->SrcRegisterExtSwz.NegateX ) { TXT( "\nNegateX : " ); UID( src->SrcRegisterExtSwz.NegateX ); diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c index d9661c75a0..14f0fc4e38 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c @@ -210,9 +210,8 @@ tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst) uint channelsWritten = 0x0; FOR_EACH_ENABLED_CHANNEL(*inst, chan) { /* check if we're reading a channel that's been written */ - uint swizzle = tgsi_util_get_full_src_register_extswizzle(&inst->FullSrcRegisters[i], chan); - if (swizzle <= TGSI_SWIZZLE_W && - (channelsWritten & (1 << swizzle))) { + uint swizzle = tgsi_util_get_full_src_register_swizzle(&inst->FullSrcRegisters[i], chan); + if (channelsWritten & (1 << swizzle)) { return TRUE; } @@ -338,7 +337,7 @@ tgsi_exec_machine_bind_shader( /* XXX we only handle SOA dependencies properly for MOV/SWZ * at this time! */ - if (opcode != TGSI_OPCODE_MOV && opcode != TGSI_OPCODE_SWZ) { + if (opcode != TGSI_OPCODE_MOV) { debug_printf("Warning: SOA dependency in instruction" " is not handled:\n"); tgsi_dump_instruction(&parse.FullToken.FullInstruction, @@ -1130,10 +1129,10 @@ fetch_src_file_channel( union tgsi_exec_channel *chan ) { switch( swizzle ) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch( file ) { case TGSI_FILE_CONSTANT: assert(mach->Consts); @@ -1201,14 +1200,6 @@ fetch_src_file_channel( } break; - case TGSI_EXTSWIZZLE_ZERO: - *chan = mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]; - break; - - case TGSI_EXTSWIZZLE_ONE: - *chan = mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]; - break; - default: assert( 0 ); } @@ -1367,7 +1358,7 @@ fetch_source( */ } - swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); + swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); fetch_src_file_channel( mach, reg->SrcRegister.File, @@ -1689,10 +1680,8 @@ exec_kil(struct tgsi_exec_machine *mach, uint kilmask = 0; /* bit 0 = pixel 0, bit 1 = pixel 1, etc */ union tgsi_exec_channel r[1]; - /* This mask stores component bits that were already tested. Note that - * we test if the value is less than zero, so 1.0 and 0.0 need not to be - * tested. */ - uniquemask = (1 << TGSI_EXTSWIZZLE_ZERO) | (1 << TGSI_EXTSWIZZLE_ONE); + /* This mask stores component bits that were already tested. */ + uniquemask = 0; for (chan_index = 0; chan_index < 4; chan_index++) { @@ -1700,7 +1689,7 @@ exec_kil(struct tgsi_exec_machine *mach, uint i; /* unswizzle channel */ - swizzle = tgsi_util_get_full_src_register_extswizzle ( + swizzle = tgsi_util_get_full_src_register_swizzle ( &inst->FullSrcRegisters[0], chan_index); @@ -2031,7 +2020,6 @@ exec_instruction( break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: if (inst->Flags & SOA_DEPENDENCY_FLAG) { /* Do all fetches into temp regs, then do all stores to avoid * intermediate/accidental clobbering. This could be done all the diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c index fe8b0bdce3..be375cabb8 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_info.c +++ b/src/gallium/auxiliary/tgsi/tgsi_info.c @@ -149,7 +149,7 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] = { 0, 1, 0, 0, 0, 0, "BREAKC", TGSI_OPCODE_BREAKC }, { 0, 1, 0, 0, 0, 0, "KIL", TGSI_OPCODE_KIL }, { 0, 0, 0, 0, 0, 0, "END", TGSI_OPCODE_END }, - { 1, 1, 0, 0, 0, 0, "SWZ", TGSI_OPCODE_SWZ } + { 0, 0, 0, 0, 0, 0, "", 118 } /* removed */ }; const struct tgsi_opcode_info * diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h index d321f013d0..b34263da48 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h +++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h @@ -146,7 +146,6 @@ OP01(IFC) OP01(BREAKC) OP01(KIL) OP00(END) -OP11(SWZ) #undef OP00 diff --git a/src/gallium/auxiliary/tgsi/tgsi_ppc.c b/src/gallium/auxiliary/tgsi/tgsi_ppc.c index 4b1c7d4e01..617fd7f6be 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_ppc.c +++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.c @@ -283,14 +283,14 @@ emit_fetch(struct gen_context *gen, const struct tgsi_full_src_register *reg, const unsigned chan_index) { - uint swizzle = tgsi_util_get_full_src_register_extswizzle(reg, chan_index); + uint swizzle = tgsi_util_get_full_src_register_swizzle(reg, chan_index); int dst_vec = -1; switch (swizzle) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch (reg->SrcRegister.File) { case TGSI_FILE_INPUT: { @@ -349,16 +349,6 @@ emit_fetch(struct gen_context *gen, assert( 0 ); } break; - case TGSI_EXTSWIZZLE_ZERO: - ppc_vzero(gen->f, dst_vec); - break; - case TGSI_EXTSWIZZLE_ONE: - { - int one_vec = gen_one_vec(gen); - dst_vec = ppc_allocate_vec_register(gen->f); - ppc_vmove(gen->f, dst_vec, one_vec); - } - break; default: assert( 0 ); } @@ -418,8 +408,8 @@ equal_src_locs(const struct tgsi_full_src_register *a, uint chan_a, return FALSE; if (a->SrcRegister.Index != b->SrcRegister.Index) return FALSE; - swz_a = tgsi_util_get_full_src_register_extswizzle(a, chan_a); - swz_b = tgsi_util_get_full_src_register_extswizzle(b, chan_b); + swz_a = tgsi_util_get_full_src_register_swizzle(a, chan_a); + swz_b = tgsi_util_get_full_src_register_swizzle(b, chan_b); if (swz_a != swz_b) return FALSE; sign_a = tgsi_util_get_full_src_register_sign_mode(a, chan_a); @@ -635,7 +625,6 @@ emit_unaryop(struct gen_context *gen, struct tgsi_full_instruction *inst) ppc_vlogefp(gen->f, v1, v0); /* v1 = log2(v0) */ break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: if (v0 != v1) ppc_vmove(gen->f, v1, v0); break; @@ -1119,7 +1108,6 @@ emit_instruction(struct gen_context *gen, switch (inst->Instruction.Opcode) { case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: case TGSI_OPCODE_ABS: case TGSI_OPCODE_FLR: case TGSI_OPCODE_FRC: diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c b/src/gallium/auxiliary/tgsi/tgsi_scan.c index 0db4481a3d..f9c16f1b6c 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_scan.c +++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c @@ -228,11 +228,6 @@ tgsi_is_passthrough_shader(const struct tgsi_token *tokens) src->SrcRegister.SwizzleZ != TGSI_SWIZZLE_Z || src->SrcRegister.SwizzleW != TGSI_SWIZZLE_W || - src->SrcRegisterExtSwz.ExtSwizzleX != TGSI_EXTSWIZZLE_X || - src->SrcRegisterExtSwz.ExtSwizzleY != TGSI_EXTSWIZZLE_Y || - src->SrcRegisterExtSwz.ExtSwizzleZ != TGSI_EXTSWIZZLE_Z || - src->SrcRegisterExtSwz.ExtSwizzleW != TGSI_EXTSWIZZLE_W || - dst->DstRegister.WriteMask != TGSI_WRITEMASK_XYZW) { tgsi_parse_free(&parse); diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c index 5f6b83b236..a96fc94c7a 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c +++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c @@ -1260,13 +1260,13 @@ emit_fetch( const struct tgsi_full_src_register *reg, const unsigned chan_index ) { - unsigned swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); + unsigned swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); switch (swizzle) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch (reg->SrcRegister.File) { case TGSI_FILE_CONSTANT: emit_const( @@ -1308,22 +1308,6 @@ emit_fetch( } break; - case TGSI_EXTSWIZZLE_ZERO: - emit_tempf( - func, - xmm, - TGSI_EXEC_TEMP_00000000_I, - TGSI_EXEC_TEMP_00000000_C ); - break; - - case TGSI_EXTSWIZZLE_ONE: - emit_tempf( - func, - xmm, - TEMP_ONE_I, - TEMP_ONE_C ); - break; - default: assert( 0 ); } @@ -1582,13 +1566,13 @@ emit_kil( /* This mask stores component bits that were already tested. Note that * we test if the value is less than zero, so 1.0 and 0.0 need not to be * tested. */ - uniquemask = (1 << TGSI_EXTSWIZZLE_ZERO) | (1 << TGSI_EXTSWIZZLE_ONE); + uniquemask = 0; FOR_EACH_CHANNEL( chan_index ) { unsigned swizzle; /* unswizzle channel */ - swizzle = tgsi_util_get_full_src_register_extswizzle( + swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); @@ -1772,7 +1756,6 @@ emit_instruction( break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 4 + chan_index, 0, chan_index ); } @@ -2938,8 +2921,7 @@ tgsi_emit_sse2( * the result in the cases where the code is too opaque to * fix. */ - if (opcode != TGSI_OPCODE_MOV && - opcode != TGSI_OPCODE_SWZ) { + if (opcode != TGSI_OPCODE_MOV) { debug_printf("Warning: src/dst aliasing in instruction" " is not handled:\n"); tgsi_dump_instruction(&parse.FullToken.FullInstruction, 1); diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c b/src/gallium/auxiliary/tgsi/tgsi_text.c index d438450b1e..87d9cd7b3f 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_text.c +++ b/src/gallium/auxiliary/tgsi/tgsi_text.c @@ -538,13 +538,11 @@ static boolean parse_optional_swizzle( struct translate_ctx *ctx, uint swizzle[4], - boolean *parsed_swizzle, - boolean *parsed_extswizzle ) + boolean *parsed_swizzle ) { const char *cur = ctx->cur; *parsed_swizzle = FALSE; - *parsed_extswizzle = FALSE; eat_opt_white( &cur ); if (*cur == '.') { @@ -562,15 +560,8 @@ parse_optional_swizzle( else if (uprcase( *cur ) == 'W') swizzle[i] = TGSI_SWIZZLE_W; else { - if (*cur == '0') - swizzle[i] = TGSI_EXTSWIZZLE_ZERO; - else if (*cur == '1') - swizzle[i] = TGSI_EXTSWIZZLE_ONE; - else { - report_error( ctx, "Expected register swizzle component `x', `y', `z', `w', `0' or `1'" ); - return FALSE; - } - *parsed_extswizzle = TRUE; + report_error( ctx, "Expected register swizzle component `x', `y', `z', `w', `0' or `1'" ); + return FALSE; } cur++; } @@ -595,7 +586,6 @@ parse_src_operand( uint swizzle[4]; boolean parsed_ext_negate_paren = FALSE; boolean parsed_swizzle; - boolean parsed_extswizzle; if (*ctx->cur == '-') { cur = ctx->cur; @@ -690,16 +680,8 @@ parse_src_operand( /* Parse optional swizzle. */ - if (parse_optional_swizzle( ctx, swizzle, &parsed_swizzle, &parsed_extswizzle )) { - if (parsed_extswizzle) { - assert( parsed_swizzle ); - - src->SrcRegisterExtSwz.ExtSwizzleX = swizzle[0]; - src->SrcRegisterExtSwz.ExtSwizzleY = swizzle[1]; - src->SrcRegisterExtSwz.ExtSwizzleZ = swizzle[2]; - src->SrcRegisterExtSwz.ExtSwizzleW = swizzle[3]; - } - else if (parsed_swizzle) { + if (parse_optional_swizzle( ctx, swizzle, &parsed_swizzle )) { + if (parsed_swizzle) { src->SrcRegister.SwizzleX = swizzle[0]; src->SrcRegister.SwizzleY = swizzle[1]; src->SrcRegister.SwizzleZ = swizzle[2]; diff --git a/src/gallium/auxiliary/tgsi/tgsi_util.c b/src/gallium/auxiliary/tgsi/tgsi_util.c index 71f8a6ca40..6120a2da94 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_util.c +++ b/src/gallium/auxiliary/tgsi/tgsi_util.c @@ -69,59 +69,15 @@ tgsi_util_get_src_register_swizzle( return 0; } -unsigned -tgsi_util_get_src_register_extswizzle( - const struct tgsi_src_register_ext_swz *reg, - unsigned component ) -{ - switch( component ) { - case 0: - return reg->ExtSwizzleX; - case 1: - return reg->ExtSwizzleY; - case 2: - return reg->ExtSwizzleZ; - case 3: - return reg->ExtSwizzleW; - default: - assert( 0 ); - } - return 0; -} unsigned -tgsi_util_get_full_src_register_extswizzle( +tgsi_util_get_full_src_register_swizzle( const struct tgsi_full_src_register *reg, unsigned component ) { - unsigned swizzle; - - /* - * First, calculate the extended swizzle for a given channel. This will give - * us either a channel index into the simple swizzle or a constant 1 or 0. - */ - swizzle = tgsi_util_get_src_register_extswizzle( - ®->SrcRegisterExtSwz, + return tgsi_util_get_src_register_swizzle( + ®->SrcRegister, component ); - - assert (TGSI_SWIZZLE_X == TGSI_EXTSWIZZLE_X); - assert (TGSI_SWIZZLE_Y == TGSI_EXTSWIZZLE_Y); - assert (TGSI_SWIZZLE_Z == TGSI_EXTSWIZZLE_Z); - assert (TGSI_SWIZZLE_W == TGSI_EXTSWIZZLE_W); - assert (TGSI_EXTSWIZZLE_ZERO > TGSI_SWIZZLE_W); - assert (TGSI_EXTSWIZZLE_ONE > TGSI_SWIZZLE_W); - - /* - * Second, calculate the simple swizzle for the unswizzled channel index. - * Leave the constants intact, they are not affected by the simple swizzle. - */ - if( swizzle <= TGSI_SWIZZLE_W ) { - swizzle = tgsi_util_get_src_register_swizzle( - ®->SrcRegister, - swizzle ); - } - - return swizzle; } void @@ -148,30 +104,6 @@ tgsi_util_set_src_register_swizzle( } } -void -tgsi_util_set_src_register_extswizzle( - struct tgsi_src_register_ext_swz *reg, - unsigned swizzle, - unsigned component ) -{ - switch( component ) { - case 0: - reg->ExtSwizzleX = swizzle; - break; - case 1: - reg->ExtSwizzleY = swizzle; - break; - case 2: - reg->ExtSwizzleZ = swizzle; - break; - case 3: - reg->ExtSwizzleW = swizzle; - break; - default: - assert( 0 ); - } -} - unsigned tgsi_util_get_src_register_extnegate( const struct tgsi_src_register_ext_swz *reg, diff --git a/src/gallium/auxiliary/tgsi/tgsi_util.h b/src/gallium/auxiliary/tgsi/tgsi_util.h index 21eb656327..bf3f20ca6c 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_util.h +++ b/src/gallium/auxiliary/tgsi/tgsi_util.h @@ -45,13 +45,9 @@ tgsi_util_get_src_register_swizzle( const struct tgsi_src_register *reg, unsigned component ); -unsigned -tgsi_util_get_src_register_extswizzle( - const struct tgsi_src_register_ext_swz *reg, - unsigned component); unsigned -tgsi_util_get_full_src_register_extswizzle( +tgsi_util_get_full_src_register_swizzle( const struct tgsi_full_src_register *reg, unsigned component ); @@ -61,12 +57,6 @@ tgsi_util_set_src_register_swizzle( unsigned swizzle, unsigned component ); -void -tgsi_util_set_src_register_extswizzle( - struct tgsi_src_register_ext_swz *reg, - unsigned swizzle, - unsigned component ); - unsigned tgsi_util_get_src_register_extnegate( const struct tgsi_src_register_ext_swz *reg, diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c index b6b2f885af..19e3ab0844 100644 --- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c +++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c @@ -231,7 +231,7 @@ static boolean is_register_src(struct codegen *gen, int channel, const struct tgsi_full_src_register *src) { - int swizzle = tgsi_util_get_full_src_register_extswizzle(src, channel); + int swizzle = tgsi_util_get_full_src_register_swizzle(src, channel); int sign_op = tgsi_util_get_full_src_register_sign_mode(src, channel); if (swizzle > TGSI_SWIZZLE_W || sign_op != TGSI_UTIL_SIGN_KEEP) { @@ -271,23 +271,14 @@ get_src_reg(struct codegen *gen, const struct tgsi_full_src_register *src) { int reg = -1; - int swizzle = tgsi_util_get_full_src_register_extswizzle(src, channel); + int swizzle = tgsi_util_get_full_src_register_swizzle(src, channel); boolean reg_is_itemp = FALSE; uint sign_op; assert(swizzle >= TGSI_SWIZZLE_X); - assert(swizzle <= TGSI_EXTSWIZZLE_ONE); + assert(swizzle <= TGSI_SWIZZLE_W); - if (swizzle == TGSI_EXTSWIZZLE_ONE) { - /* Load const one float and early out */ - reg = get_const_one_reg(gen); - } - else if (swizzle == TGSI_EXTSWIZZLE_ZERO) { - /* Load const zero float and early out */ - reg = get_itemp(gen); - spe_xor(gen->f, reg, reg, reg); - } - else { + { int index = src->SrcRegister.Index; assert(swizzle < 4); @@ -1758,7 +1749,6 @@ emit_instruction(struct codegen *gen, case TGSI_OPCODE_ARL: return emit_ARL(gen, inst); case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: return emit_MOV(gen, inst); case TGSI_OPCODE_ADD: case TGSI_OPCODE_SUB: diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c index 725a72b326..4c32b2d06d 100644 --- a/src/gallium/drivers/cell/spu/spu_exec.c +++ b/src/gallium/drivers/cell/spu/spu_exec.c @@ -346,10 +346,10 @@ fetch_src_file_channel( union spu_exec_channel *chan ) { switch( swizzle ) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch( file ) { case TGSI_FILE_CONSTANT: { unsigned i; @@ -413,14 +413,6 @@ fetch_src_file_channel( } break; - case TGSI_EXTSWIZZLE_ZERO: - *chan = mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]; - break; - - case TGSI_EXTSWIZZLE_ONE: - *chan = mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]; - break; - default: ASSERT( 0 ); } @@ -500,7 +492,7 @@ fetch_source( } } - swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); + swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); fetch_src_file_channel( mach, reg->SrcRegister.File, @@ -610,10 +602,8 @@ exec_kil(struct spu_exec_machine *mach, uint kilmask = 0; /* bit 0 = pixel 0, bit 1 = pixel 1, etc */ union spu_exec_channel r[1]; - /* This mask stores component bits that were already tested. Note that - * we test if the value is less than zero, so 1.0 and 0.0 need not to be - * tested. */ - uniquemask = (1 << TGSI_EXTSWIZZLE_ZERO) | (1 << TGSI_EXTSWIZZLE_ONE); + /* This mask stores component bits that were already tested. */ + uniquemask = 0; for (chan_index = 0; chan_index < 4; chan_index++) { @@ -621,7 +611,7 @@ exec_kil(struct spu_exec_machine *mach, uint i; /* unswizzle channel */ - swizzle = tgsi_util_get_full_src_register_extswizzle ( + swizzle = tgsi_util_get_full_src_register_swizzle ( &inst->FullSrcRegisters[0], chan_index); @@ -909,7 +899,6 @@ exec_instruction( break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); STORE( &r[0], 0, chan_index ); diff --git a/src/gallium/drivers/cell/spu/spu_util.c b/src/gallium/drivers/cell/spu/spu_util.c index af25dd3718..25a7a71133 100644 --- a/src/gallium/drivers/cell/spu/spu_util.c +++ b/src/gallium/drivers/cell/spu/spu_util.c @@ -26,59 +26,15 @@ tgsi_util_get_src_register_swizzle( return 0; } -unsigned -tgsi_util_get_src_register_extswizzle( - const struct tgsi_src_register_ext_swz *reg, - unsigned component ) -{ - switch( component ) { - case 0: - return reg->ExtSwizzleX; - case 1: - return reg->ExtSwizzleY; - case 2: - return reg->ExtSwizzleZ; - case 3: - return reg->ExtSwizzleW; - default: - ASSERT( 0 ); - } - return 0; -} unsigned tgsi_util_get_full_src_register_extswizzle( const struct tgsi_full_src_register *reg, unsigned component ) { - unsigned swizzle; - - /* - * First, calculate the extended swizzle for a given channel. This will give - * us either a channel index into the simple swizzle or a constant 1 or 0. - */ - swizzle = tgsi_util_get_src_register_extswizzle( - ®->SrcRegisterExtSwz, + return tgsi_util_get_src_register_swizzle( + reg->SrcRegister, component ); - - ASSERT (TGSI_SWIZZLE_X == TGSI_EXTSWIZZLE_X); - ASSERT (TGSI_SWIZZLE_Y == TGSI_EXTSWIZZLE_Y); - ASSERT (TGSI_SWIZZLE_Z == TGSI_EXTSWIZZLE_Z); - ASSERT (TGSI_SWIZZLE_W == TGSI_EXTSWIZZLE_W); - ASSERT (TGSI_EXTSWIZZLE_ZERO > TGSI_SWIZZLE_W); - ASSERT (TGSI_EXTSWIZZLE_ONE > TGSI_SWIZZLE_W); - - /* - * Second, calculate the simple swizzle for the unswizzled channel index. - * Leave the constants intact, they are not affected by the simple swizzle. - */ - if( swizzle <= TGSI_SWIZZLE_W ) { - swizzle = tgsi_util_get_src_register_swizzle( - ®->SrcRegister, - component ); - } - - return swizzle; } unsigned diff --git a/src/gallium/drivers/i915/i915_fpc_translate.c b/src/gallium/drivers/i915/i915_fpc_translate.c index 1fe5cda956..3074044441 100644 --- a/src/gallium/drivers/i915/i915_fpc_translate.c +++ b/src/gallium/drivers/i915/i915_fpc_translate.c @@ -214,20 +214,11 @@ src_vector(struct i915_fp_compile *p, return 0; } - if (source->SrcRegister.Extended) { - src = swizzle(src, - source->SrcRegisterExtSwz.ExtSwizzleX, - source->SrcRegisterExtSwz.ExtSwizzleY, - source->SrcRegisterExtSwz.ExtSwizzleZ, - source->SrcRegisterExtSwz.ExtSwizzleW); - } - else { - src = swizzle(src, - source->SrcRegister.SwizzleX, - source->SrcRegister.SwizzleY, - source->SrcRegister.SwizzleZ, - source->SrcRegister.SwizzleW); - } + src = swizzle(src, + source->SrcRegister.SwizzleX, + source->SrcRegister.SwizzleY, + source->SrcRegister.SwizzleZ, + source->SrcRegister.SwizzleW); /* There's both negate-all-components and per-component negation. @@ -681,7 +672,6 @@ i915_translate_instruction(struct i915_fp_compile *p, break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: emit_simple_arith(p, inst, A0_MOV, 1); break; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c index 926cc1cc9f..64027de6aa 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c @@ -158,14 +158,14 @@ emit_fetch( const unsigned chan_index ) { const struct tgsi_full_src_register *reg = &inst->FullSrcRegisters[index]; - unsigned swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); + unsigned swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); LLVMValueRef res; switch (swizzle) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch (reg->SrcRegister.File) { case TGSI_FILE_CONSTANT: { @@ -198,14 +198,6 @@ emit_fetch( } break; - case TGSI_EXTSWIZZLE_ZERO: - res = bld->base.zero; - break; - - case TGSI_EXTSWIZZLE_ONE: - res = bld->base.one; - break; - default: assert( 0 ); return bld->base.undef; @@ -394,12 +386,7 @@ emit_kil( unsigned swizzle; /* Unswizzle channel */ - swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); - - /* Note that we test if the value is less than zero, so 1.0 and 0.0 need - * not to be tested. */ - if(swizzle == TGSI_EXTSWIZZLE_ZERO || swizzle == TGSI_EXTSWIZZLE_ONE) - continue; + swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index ); /* Check if the component has not been already tested. */ assert(swizzle < NUM_CHANNELS); @@ -488,7 +475,6 @@ emit_instruction( #endif case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { dst0[chan_index] = emit_fetch( bld, inst, 0, chan_index ); } diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 62c6c76cee..93cf869f58 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -326,21 +326,13 @@ src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc, uint c; for (c = 0; c < 4; c++) { - switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + switch (tgsi_util_get_full_src_register_swizzle(fsrc, c)) { + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: mask |= (1 << c); break; - case TGSI_EXTSWIZZLE_ZERO: - zero_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - case TGSI_EXTSWIZZLE_ONE: - one_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; default: assert(0); } @@ -357,12 +349,6 @@ src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc, if (mask) arith(fpc, 0, MOV, *src, mask, tgsi, none, none); - if (zero_mask) - arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); - - if (one_mask) - arith(fpc, 0, STR, *src, one_mask, *src, none, none); - if (neg_mask) { struct nv30_sreg one = temp(fpc); arith(fpc, 0, STR, one, neg_mask, one, none, none); diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index e3550baa63..4a6b355936 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -329,21 +329,13 @@ src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc, uint c; for (c = 0; c < 4; c++) { - switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + switch (tgsi_util_get_full_src_register_swizzle(fsrc, c)) { + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: mask |= (1 << c); break; - case TGSI_EXTSWIZZLE_ZERO: - zero_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - case TGSI_EXTSWIZZLE_ONE: - one_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; default: assert(0); } diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 0382dbba8f..4898aaa809 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -370,21 +370,13 @@ src_native_swz(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc, uint c; for (c = 0; c < 4; c++) { - switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + switch (tgsi_util_get_full_src_register_swizzle(fsrc, c)) { + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: mask |= tgsi_mask(1 << c); break; - case TGSI_EXTSWIZZLE_ZERO: - zero_mask |= tgsi_mask(1 << c); - tgsi.swz[c] = SWZ_X; - break; - case TGSI_EXTSWIZZLE_ONE: - one_mask |= tgsi_mask(1 << c); - tgsi.swz[c] = SWZ_X; - break; default: assert(0); } diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index c7145bb9be..5c691877e0 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1544,12 +1544,12 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src, sgn = tgsi_util_get_full_src_register_sign_mode(src, chan); - c = tgsi_util_get_full_src_register_extswizzle(src, chan); + c = tgsi_util_get_full_src_register_swizzle(src, chan); switch (c) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: + case TGSI_SWIZZLE_X: + case TGSI_SWIZZLE_Y: + case TGSI_SWIZZLE_Z: + case TGSI_SWIZZLE_W: switch (src->SrcRegister.File) { case TGSI_FILE_INPUT: r = &pc->attr[src->SrcRegister.Index * 4 + c]; @@ -1586,13 +1586,6 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src, break; } break; - case TGSI_EXTSWIZZLE_ZERO: - r = alloc_immd(pc, 0.0); - return r; - case TGSI_EXTSWIZZLE_ONE: - if (sgn == TGSI_UTIL_SIGN_TOGGLE || sgn == TGSI_UTIL_SIGN_SET) - return alloc_immd(pc, -1.0); - return alloc_immd(pc, 1.0); default: assert(0); break; @@ -2005,7 +1998,6 @@ nv50_program_tx_insn(struct nv50_pc *pc, } break; case TGSI_OPCODE_MOV: - case TGSI_OPCODE_SWZ: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; @@ -2189,10 +2181,7 @@ prep_inspect_insn(struct nv50_pc *pc, const struct tgsi_full_instruction *insn) for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - k = tgsi_util_get_full_src_register_extswizzle(src, c); - - if (k > TGSI_EXTSWIZZLE_W) - continue; + k = tgsi_util_get_full_src_register_swizzle(src, c); reg[src->SrcRegister.Index * 4 + k].acc = pc->insn_nr; } @@ -2295,11 +2284,10 @@ nv50_tgsi_scan_swizzle(const struct tgsi_full_instruction *insn, if (!(mask & (1 << chn))) /* src is not read */ continue; - c = tgsi_util_get_full_src_register_extswizzle(fs, chn); + c = tgsi_util_get_full_src_register_swizzle(fs, chn); s = tgsi_util_get_full_src_register_sign_mode(fs, chn); - if (c > TGSI_EXTSWIZZLE_W || - !(fd->DstRegister.WriteMask & (1 << c))) + if (!(fd->DstRegister.WriteMask & (1 << c))) continue; /* no danger if src is copied to TEMP first */ diff --git a/src/gallium/drivers/r300/r300_tgsi_to_rc.c b/src/gallium/drivers/r300/r300_tgsi_to_rc.c index 3d2f6cafee..de599b068f 100644 --- a/src/gallium/drivers/r300/r300_tgsi_to_rc.c +++ b/src/gallium/drivers/r300/r300_tgsi_to_rc.c @@ -142,7 +142,6 @@ static unsigned translate_opcode(unsigned opcode) /* case TGSI_OPCODE_IFC: return RC_OPCODE_IFC; */ /* case TGSI_OPCODE_BREAKC: return RC_OPCODE_BREAKC; */ case TGSI_OPCODE_KIL: return RC_OPCODE_KIL; - case TGSI_OPCODE_SWZ: return RC_OPCODE_SWZ; } fprintf(stderr, "Unknown opcode: %i\n", opcode); @@ -205,10 +204,10 @@ static void transform_srcreg( dst->File = translate_register_file(src->SrcRegister.File); dst->Index = translate_register_index(ttr, src->SrcRegister.File, src->SrcRegister.Index); dst->RelAddr = src->SrcRegister.Indirect; - dst->Swizzle = tgsi_util_get_full_src_register_extswizzle(src, 0); - dst->Swizzle |= tgsi_util_get_full_src_register_extswizzle(src, 1) << 3; - dst->Swizzle |= tgsi_util_get_full_src_register_extswizzle(src, 2) << 6; - dst->Swizzle |= tgsi_util_get_full_src_register_extswizzle(src, 3) << 9; + dst->Swizzle = tgsi_util_get_full_src_register_swizzle(src, 0); + dst->Swizzle |= tgsi_util_get_full_src_register_swizzle(src, 1) << 3; + dst->Swizzle |= tgsi_util_get_full_src_register_swizzle(src, 2) << 6; + dst->Swizzle |= tgsi_util_get_full_src_register_swizzle(src, 3) << 9; dst->Abs = src->SrcRegisterExtMod.Absolute; dst->Negate = src->SrcRegisterExtSwz.NegateX | diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h index 48e6583ada..b01df41b0e 100644 --- a/src/gallium/include/pipe/p_shader_tokens.h +++ b/src/gallium/include/pipe/p_shader_tokens.h @@ -275,7 +275,7 @@ union tgsi_immediate_data #define TGSI_OPCODE_BREAKC 115 #define TGSI_OPCODE_KIL 116 /* conditional kill */ #define TGSI_OPCODE_END 117 /* aka HALT */ -#define TGSI_OPCODE_SWZ 118 + /* gap */ #define TGSI_OPCODE_LAST 119 #define TGSI_SAT_NONE 0 /* do not saturate */ @@ -496,17 +496,7 @@ struct tgsi_src_register_ext * follows. */ -#define TGSI_EXTSWIZZLE_X TGSI_SWIZZLE_X -#define TGSI_EXTSWIZZLE_Y TGSI_SWIZZLE_Y -#define TGSI_EXTSWIZZLE_Z TGSI_SWIZZLE_Z -#define TGSI_EXTSWIZZLE_W TGSI_SWIZZLE_W -#define TGSI_EXTSWIZZLE_ZERO 4 -#define TGSI_EXTSWIZZLE_ONE 5 - /** - * ExtSwizzleX, ExtSwizzleY, ExtSwizzleZ and ExtSwizzleW swizzle the source - * register in an extended manner. - * * NegateX, NegateY, NegateZ and NegateW negate individual components of the * source register. * @@ -518,10 +508,7 @@ struct tgsi_src_register_ext struct tgsi_src_register_ext_swz { unsigned Type : 4; /* TGSI_SRC_REGISTER_EXT_TYPE_SWZ */ - unsigned ExtSwizzleX : 4; /* TGSI_EXTSWIZZLE_ */ - unsigned ExtSwizzleY : 4; /* TGSI_EXTSWIZZLE_ */ - unsigned ExtSwizzleZ : 4; /* TGSI_EXTSWIZZLE_ */ - unsigned ExtSwizzleW : 4; /* TGSI_EXTSWIZZLE_ */ + unsigned Padding0 : 16; /* unused */ unsigned NegateX : 1; /* BOOL */ unsigned NegateY : 1; /* BOOL */ unsigned NegateZ : 1; /* BOOL */ diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c index 1b9d35d353..3d6c215819 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.c +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c @@ -515,8 +515,6 @@ translate_opcode( unsigned op ) return TGSI_OPCODE_SSG; case OPCODE_SUB: return TGSI_OPCODE_SUB; - case OPCODE_SWZ: - return TGSI_OPCODE_SWZ; case OPCODE_TEX: return TGSI_OPCODE_TEX; case OPCODE_TXB: -- cgit v1.2.3 From df0826fba37f38e9ea094180c71f2460e5b50e38 Mon Sep 17 00:00:00 2001 From: Michal Krol Date: Wed, 23 Dec 2009 17:00:15 +0100 Subject: tgsi: Implement new integer opcodes. Update interpreter and ureg. Also, get rid of SHR -- it's actually an alias for ISHR. --- src/gallium/auxiliary/tgsi/tgsi_exec.c | 591 +++++++++++++++++---------- src/gallium/auxiliary/tgsi/tgsi_info.c | 25 +- src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h | 22 +- 3 files changed, 413 insertions(+), 225 deletions(-) (limited to 'src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h') diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c index 517ae67d79..27af0c6057 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c @@ -446,20 +446,6 @@ micro_add( dst->f[3] = src0->f[3] + src1->f[3]; } -#if 0 -static void -micro_iadd( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] + src1->i[0]; - dst->i[1] = src0->i[1] + src1->i[1]; - dst->i[2] = src0->i[2] + src1->i[2]; - dst->i[3] = src0->i[3] + src1->i[3]; -} -#endif - static void micro_and( union tgsi_exec_channel *dst, @@ -536,20 +522,6 @@ micro_div( } } -#if 0 -static void -micro_udiv( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] / src1->u[0]; - dst->u[1] = src0->u[1] / src1->u[1]; - dst->u[2] = src0->u[2] / src1->u[2]; - dst->u[3] = src0->u[3] / src1->u[3]; -} -#endif - static void micro_eq( union tgsi_exec_channel *dst, @@ -564,22 +536,6 @@ micro_eq( dst->f[3] = src0->f[3] == src1->f[3] ? src2->f[3] : src3->f[3]; } -#if 0 -static void -micro_ieq( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1, - const union tgsi_exec_channel *src2, - const union tgsi_exec_channel *src3 ) -{ - dst->i[0] = src0->i[0] == src1->i[0] ? src2->i[0] : src3->i[0]; - dst->i[1] = src0->i[1] == src1->i[1] ? src2->i[1] : src3->i[1]; - dst->i[2] = src0->i[2] == src1->i[2] ? src2->i[2] : src3->i[2]; - dst->i[3] = src0->i[3] == src1->i[3] ? src2->i[3] : src3->i[3]; -} -#endif - static void micro_exp2( union tgsi_exec_channel *dst, @@ -616,19 +572,6 @@ micro_exp2( #endif } -#if 0 -static void -micro_f2ut( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src ) -{ - dst->u[0] = (uint) src->f[0]; - dst->u[1] = (uint) src->f[1]; - dst->u[2] = (uint) src->f[2]; - dst->u[3] = (uint) src->f[3]; -} -#endif - static void micro_float_clamp(union tgsi_exec_channel *dst, const union tgsi_exec_channel *src) @@ -734,38 +677,6 @@ micro_lt( dst->f[3] = src0->f[3] < src1->f[3] ? src2->f[3] : src3->f[3]; } -#if 0 -static void -micro_ilt( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1, - const union tgsi_exec_channel *src2, - const union tgsi_exec_channel *src3 ) -{ - dst->i[0] = src0->i[0] < src1->i[0] ? src2->i[0] : src3->i[0]; - dst->i[1] = src0->i[1] < src1->i[1] ? src2->i[1] : src3->i[1]; - dst->i[2] = src0->i[2] < src1->i[2] ? src2->i[2] : src3->i[2]; - dst->i[3] = src0->i[3] < src1->i[3] ? src2->i[3] : src3->i[3]; -} -#endif - -#if 0 -static void -micro_ult( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1, - const union tgsi_exec_channel *src2, - const union tgsi_exec_channel *src3 ) -{ - dst->u[0] = src0->u[0] < src1->u[0] ? src2->u[0] : src3->u[0]; - dst->u[1] = src0->u[1] < src1->u[1] ? src2->u[1] : src3->u[1]; - dst->u[2] = src0->u[2] < src1->u[2] ? src2->u[2] : src3->u[2]; - dst->u[3] = src0->u[3] < src1->u[3] ? src2->u[3] : src3->u[3]; -} -#endif - static void micro_max( union tgsi_exec_channel *dst, @@ -778,34 +689,6 @@ micro_max( dst->f[3] = src0->f[3] > src1->f[3] ? src0->f[3] : src1->f[3]; } -#if 0 -static void -micro_imax( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] > src1->i[0] ? src0->i[0] : src1->i[0]; - dst->i[1] = src0->i[1] > src1->i[1] ? src0->i[1] : src1->i[1]; - dst->i[2] = src0->i[2] > src1->i[2] ? src0->i[2] : src1->i[2]; - dst->i[3] = src0->i[3] > src1->i[3] ? src0->i[3] : src1->i[3]; -} -#endif - -#if 0 -static void -micro_umax( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] > src1->u[0] ? src0->u[0] : src1->u[0]; - dst->u[1] = src0->u[1] > src1->u[1] ? src0->u[1] : src1->u[1]; - dst->u[2] = src0->u[2] > src1->u[2] ? src0->u[2] : src1->u[2]; - dst->u[3] = src0->u[3] > src1->u[3] ? src0->u[3] : src1->u[3]; -} -#endif - static void micro_min( union tgsi_exec_channel *dst, @@ -818,34 +701,6 @@ micro_min( dst->f[3] = src0->f[3] < src1->f[3] ? src0->f[3] : src1->f[3]; } -#if 0 -static void -micro_imin( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] < src1->i[0] ? src0->i[0] : src1->i[0]; - dst->i[1] = src0->i[1] < src1->i[1] ? src0->i[1] : src1->i[1]; - dst->i[2] = src0->i[2] < src1->i[2] ? src0->i[2] : src1->i[2]; - dst->i[3] = src0->i[3] < src1->i[3] ? src0->i[3] : src1->i[3]; -} -#endif - -#if 0 -static void -micro_umin( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] < src1->u[0] ? src0->u[0] : src1->u[0]; - dst->u[1] = src0->u[1] < src1->u[1] ? src0->u[1] : src1->u[1]; - dst->u[2] = src0->u[2] < src1->u[2] ? src0->u[2] : src1->u[2]; - dst->u[3] = src0->u[3] < src1->u[3] ? src0->u[3] : src1->u[3]; -} -#endif - #if 0 static void micro_umod( @@ -872,20 +727,6 @@ micro_mul( dst->f[3] = src0->f[3] * src1->f[3]; } -#if 0 -static void -micro_imul( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] * src1->i[0]; - dst->i[1] = src0->i[1] * src1->i[1]; - dst->i[2] = src0->i[2] * src1->i[2]; - dst->i[3] = src0->i[3] * src1->i[3]; -} -#endif - #if 0 static void micro_imul64( @@ -951,19 +792,6 @@ micro_neg( dst->f[3] = -src->f[3]; } -#if 0 -static void -micro_ineg( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src ) -{ - dst->i[0] = -src->i[0]; - dst->i[1] = -src->i[1]; - dst->i[2] = -src->i[2]; - dst->i[3] = -src->i[3]; -} -#endif - static void micro_not( union tgsi_exec_channel *dst, @@ -1040,18 +868,6 @@ micro_shl( dst->i[3] = src0->i[3] << src1->i[3]; } -static void -micro_ishr( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] >> src1->i[0]; - dst->i[1] = src0->i[1] >> src1->i[1]; - dst->i[2] = src0->i[2] >> src1->i[2]; - dst->i[3] = src0->i[3] >> src1->i[3]; -} - static void micro_trunc( union tgsi_exec_channel *dst, @@ -1063,20 +879,6 @@ micro_trunc( dst->f[3] = (float) (int) src0->f[3]; } -#if 0 -static void -micro_ushr( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src0, - const union tgsi_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] >> src1->u[0]; - dst->u[1] = src0->u[1] >> src1->u[1]; - dst->u[2] = src0->u[2] >> src1->u[2]; - dst->u[3] = src0->u[3] >> src1->u[3]; -} -#endif - static void micro_sin( union tgsi_exec_channel *dst, @@ -1110,19 +912,6 @@ micro_sub( dst->f[3] = src0->f[3] - src1->f[3]; } -#if 0 -static void -micro_u2f( - union tgsi_exec_channel *dst, - const union tgsi_exec_channel *src ) -{ - dst->f[0] = (float) src->u[0]; - dst->f[1] = (float) src->u[1]; - dst->f[2] = (float) src->u[2]; - dst->f[3] = (float) src->u[3]; -} -#endif - static void micro_xor( union tgsi_exec_channel *dst, @@ -2001,6 +1790,291 @@ exec_declaration(struct tgsi_exec_machine *mach, } } +typedef void (* micro_op)(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src); + +static void +exec_vector_unary(struct tgsi_exec_machine *mach, + const struct tgsi_full_instruction *inst, + micro_op op) +{ + unsigned int chan; + struct tgsi_exec_vector dst; + + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + union tgsi_exec_channel src; + + fetch_source(mach, &src, &inst->Src[0], chan); + op(&dst.xyzw[chan], &src); + } + } + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan); + } + } +} + +static void +exec_vector_binary(struct tgsi_exec_machine *mach, + const struct tgsi_full_instruction *inst, + micro_op op) +{ + unsigned int chan; + struct tgsi_exec_vector dst; + + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + union tgsi_exec_channel src[2]; + + fetch_source(mach, &src[0], &inst->Src[0], chan); + fetch_source(mach, &src[1], &inst->Src[1], chan); + op(&dst.xyzw[chan], src); + } + } + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan); + } + } +} + +static void +exec_vector_trinary(struct tgsi_exec_machine *mach, + const struct tgsi_full_instruction *inst, + micro_op op) +{ + unsigned int chan; + struct tgsi_exec_vector dst; + + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + union tgsi_exec_channel src[3]; + + fetch_source(mach, &src[0], &inst->Src[0], chan); + fetch_source(mach, &src[1], &inst->Src[1], chan); + fetch_source(mach, &src[2], &inst->Src[2], chan); + op(&dst.xyzw[chan], src); + } + } + for (chan = 0; chan < NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan); + } + } +} + +static void +micro_f2i(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = (int)src->f[0]; + dst->i[1] = (int)src->f[1]; + dst->i[2] = (int)src->f[2]; + dst->i[3] = (int)src->f[3]; +} + +static void +micro_idiv(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] / src[1].i[0]; + dst->i[1] = src[0].i[1] / src[1].i[1]; + dst->i[2] = src[0].i[2] / src[1].i[2]; + dst->i[3] = src[0].i[3] / src[1].i[3]; +} + +static void +micro_imax(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] > src[1].i[0] ? src[0].i[0] : src[1].i[0]; + dst->i[1] = src[0].i[1] > src[1].i[1] ? src[0].i[1] : src[1].i[1]; + dst->i[2] = src[0].i[2] > src[1].i[2] ? src[0].i[2] : src[1].i[2]; + dst->i[3] = src[0].i[3] > src[1].i[3] ? src[0].i[3] : src[1].i[3]; +} + +static void +micro_imin(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] < src[1].i[0] ? src[0].i[0] : src[1].i[0]; + dst->i[1] = src[0].i[1] < src[1].i[1] ? src[0].i[1] : src[1].i[1]; + dst->i[2] = src[0].i[2] < src[1].i[2] ? src[0].i[2] : src[1].i[2]; + dst->i[3] = src[0].i[3] < src[1].i[3] ? src[0].i[3] : src[1].i[3]; +} + +static void +micro_ineg(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = -src->i[0]; + dst->i[1] = -src->i[1]; + dst->i[2] = -src->i[2]; + dst->i[3] = -src->i[3]; +} + +static void +micro_isge(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] >= src[1].i[0] ? -1 : 0; + dst->i[1] = src[0].i[1] >= src[1].i[1] ? -1 : 0; + dst->i[2] = src[0].i[2] >= src[1].i[2] ? -1 : 0; + dst->i[3] = src[0].i[3] >= src[1].i[3] ? -1 : 0; +} + +static void +micro_ishr(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] >> src[1].i[0]; + dst->i[1] = src[0].i[1] >> src[1].i[1]; + dst->i[2] = src[0].i[2] >> src[1].i[2]; + dst->i[3] = src[0].i[3] >> src[1].i[3]; +} + +static void +micro_islt(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->i[0] = src[0].i[0] < src[1].i[0] ? -1 : 0; + dst->i[1] = src[0].i[1] < src[1].i[1] ? -1 : 0; + dst->i[2] = src[0].i[2] < src[1].i[2] ? -1 : 0; + dst->i[3] = src[0].i[3] < src[1].i[3] ? -1 : 0; +} + +static void +micro_f2u(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = (uint)src->f[0]; + dst->u[1] = (uint)src->f[1]; + dst->u[2] = (uint)src->f[2]; + dst->u[3] = (uint)src->f[3]; +} + +static void +micro_u2f(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->f[0] = (float)src->u[0]; + dst->f[1] = (float)src->u[1]; + dst->f[2] = (float)src->u[2]; + dst->f[3] = (float)src->u[3]; +} + +static void +micro_uadd(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] + src[1].u[0]; + dst->u[1] = src[0].u[1] + src[1].u[1]; + dst->u[2] = src[0].u[2] + src[1].u[2]; + dst->u[3] = src[0].u[3] + src[1].u[3]; +} + +static void +micro_udiv(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] / src[1].u[0]; + dst->u[1] = src[0].u[1] / src[1].u[1]; + dst->u[2] = src[0].u[2] / src[1].u[2]; + dst->u[3] = src[0].u[3] / src[1].u[3]; +} + +static void +micro_umad(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] * src[1].u[0] + src[2].u[0]; + dst->u[1] = src[0].u[1] * src[1].u[1] + src[2].u[1]; + dst->u[2] = src[0].u[2] * src[1].u[2] + src[2].u[2]; + dst->u[3] = src[0].u[3] * src[1].u[3] + src[2].u[3]; +} + +static void +micro_umax(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] > src[1].u[0] ? src[0].u[0] : src[1].u[0]; + dst->u[1] = src[0].u[1] > src[1].u[1] ? src[0].u[1] : src[1].u[1]; + dst->u[2] = src[0].u[2] > src[1].u[2] ? src[0].u[2] : src[1].u[2]; + dst->u[3] = src[0].u[3] > src[1].u[3] ? src[0].u[3] : src[1].u[3]; +} + +static void +micro_umin(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] < src[1].u[0] ? src[0].u[0] : src[1].u[0]; + dst->u[1] = src[0].u[1] < src[1].u[1] ? src[0].u[1] : src[1].u[1]; + dst->u[2] = src[0].u[2] < src[1].u[2] ? src[0].u[2] : src[1].u[2]; + dst->u[3] = src[0].u[3] < src[1].u[3] ? src[0].u[3] : src[1].u[3]; +} + +static void +micro_umul(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] * src[1].u[0]; + dst->u[1] = src[0].u[1] * src[1].u[1]; + dst->u[2] = src[0].u[2] * src[1].u[2]; + dst->u[3] = src[0].u[3] * src[1].u[3]; +} + +static void +micro_useq(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] == src[1].u[0] ? ~0 : 0; + dst->u[1] = src[0].u[1] == src[1].u[1] ? ~0 : 0; + dst->u[2] = src[0].u[2] == src[1].u[2] ? ~0 : 0; + dst->u[3] = src[0].u[3] == src[1].u[3] ? ~0 : 0; +} + +static void +micro_usge(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] >= src[1].u[0] ? ~0 : 0; + dst->u[1] = src[0].u[1] >= src[1].u[1] ? ~0 : 0; + dst->u[2] = src[0].u[2] >= src[1].u[2] ? ~0 : 0; + dst->u[3] = src[0].u[3] >= src[1].u[3] ? ~0 : 0; +} + +static void +micro_ushr(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] >> src[1].u[0]; + dst->u[1] = src[0].u[1] >> src[1].u[1]; + dst->u[2] = src[0].u[2] >> src[1].u[2]; + dst->u[3] = src[0].u[3] >> src[1].u[3]; +} + +static void +micro_uslt(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] < src[1].u[0] ? ~0 : 0; + dst->u[1] = src[0].u[1] < src[1].u[1] ? ~0 : 0; + dst->u[2] = src[0].u[2] < src[1].u[2] ? ~0 : 0; + dst->u[3] = src[0].u[3] < src[1].u[3] ? ~0 : 0; +} + +static void +micro_usne(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] != src[1].u[0] ? ~0 : 0; + dst->u[1] = src[0].u[1] != src[1].u[1] ? ~0 : 0; + dst->u[2] = src[0].u[2] != src[1].u[2] ? ~0 : 0; + dst->u[3] = src[0].u[3] != src[1].u[3] ? ~0 : 0; +} + static void exec_instruction( struct tgsi_exec_machine *mach, @@ -3073,17 +3147,6 @@ exec_instruction( } break; - case TGSI_OPCODE_SHR: - FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { - FETCH( &r[0], 0, chan_index ); - FETCH( &r[1], 1, chan_index ); - micro_ishr(&d[chan_index], &r[0], &r[1]); - } - FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) { - STORE(&d[chan_index], 0, chan_index); - } - break; - case TGSI_OPCODE_AND: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); @@ -3310,6 +3373,90 @@ exec_instruction( UPDATE_EXEC_MASK(mach); break; + case TGSI_OPCODE_F2I: + exec_vector_unary(mach, inst, micro_f2i); + break; + + case TGSI_OPCODE_IDIV: + exec_vector_binary(mach, inst, micro_idiv); + break; + + case TGSI_OPCODE_IMAX: + exec_vector_binary(mach, inst, micro_imax); + break; + + case TGSI_OPCODE_IMIN: + exec_vector_binary(mach, inst, micro_imin); + break; + + case TGSI_OPCODE_INEG: + exec_vector_unary(mach, inst, micro_ineg); + break; + + case TGSI_OPCODE_ISGE: + exec_vector_binary(mach, inst, micro_isge); + break; + + case TGSI_OPCODE_ISHR: + exec_vector_binary(mach, inst, micro_ishr); + break; + + case TGSI_OPCODE_ISLT: + exec_vector_binary(mach, inst, micro_islt); + break; + + case TGSI_OPCODE_F2U: + exec_vector_unary(mach, inst, micro_f2u); + break; + + case TGSI_OPCODE_U2F: + exec_vector_unary(mach, inst, micro_u2f); + break; + + case TGSI_OPCODE_UADD: + exec_vector_binary(mach, inst, micro_uadd); + break; + + case TGSI_OPCODE_UDIV: + exec_vector_binary(mach, inst, micro_udiv); + break; + + case TGSI_OPCODE_UMAD: + exec_vector_trinary(mach, inst, micro_umad); + break; + + case TGSI_OPCODE_UMAX: + exec_vector_binary(mach, inst, micro_umax); + break; + + case TGSI_OPCODE_UMIN: + exec_vector_binary(mach, inst, micro_umin); + break; + + case TGSI_OPCODE_UMUL: + exec_vector_binary(mach, inst, micro_umul); + break; + + case TGSI_OPCODE_USEQ: + exec_vector_binary(mach, inst, micro_useq); + break; + + case TGSI_OPCODE_USGE: + exec_vector_binary(mach, inst, micro_usge); + break; + + case TGSI_OPCODE_USHR: + exec_vector_binary(mach, inst, micro_ushr); + break; + + case TGSI_OPCODE_USLT: + exec_vector_binary(mach, inst, micro_uslt); + break; + + case TGSI_OPCODE_USNE: + exec_vector_binary(mach, inst, micro_usne); + break; + default: assert( 0 ); } diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c index be375cabb8..c7cdd163ec 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_info.c +++ b/src/gallium/auxiliary/tgsi/tgsi_info.c @@ -119,7 +119,7 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] = { 1, 1, 0, 0, 0, 0, "NOT", TGSI_OPCODE_NOT }, { 1, 1, 0, 0, 0, 0, "TRUNC", TGSI_OPCODE_TRUNC }, { 1, 2, 0, 0, 0, 0, "SHL", TGSI_OPCODE_SHL }, - { 1, 2, 0, 0, 0, 0, "SHR", TGSI_OPCODE_SHR }, + { 0, 0, 0, 0, 0, 0, "", 88 }, /* removed */ { 1, 2, 0, 0, 0, 0, "AND", TGSI_OPCODE_AND }, { 1, 2, 0, 0, 0, 0, "OR", TGSI_OPCODE_OR }, { 1, 2, 0, 0, 0, 0, "MOD", TGSI_OPCODE_MOD }, @@ -149,7 +149,28 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] = { 0, 1, 0, 0, 0, 0, "BREAKC", TGSI_OPCODE_BREAKC }, { 0, 1, 0, 0, 0, 0, "KIL", TGSI_OPCODE_KIL }, { 0, 0, 0, 0, 0, 0, "END", TGSI_OPCODE_END }, - { 0, 0, 0, 0, 0, 0, "", 118 } /* removed */ + { 0, 0, 0, 0, 0, 0, "", 118 }, /* removed */ + { 1, 1, 0, 0, 0, 0, "F2I", TGSI_OPCODE_F2I }, + { 1, 2, 0, 0, 0, 0, "IDIV", TGSI_OPCODE_IDIV }, + { 1, 2, 0, 0, 0, 0, "IMAX", TGSI_OPCODE_IMAX }, + { 1, 2, 0, 0, 0, 0, "IMIN", TGSI_OPCODE_IMIN }, + { 1, 1, 0, 0, 0, 0, "INEG", TGSI_OPCODE_INEG }, + { 1, 2, 0, 0, 0, 0, "ISGE", TGSI_OPCODE_ISGE }, + { 1, 2, 0, 0, 0, 0, "ISHR", TGSI_OPCODE_ISHR }, + { 1, 2, 0, 0, 0, 0, "ISLT", TGSI_OPCODE_ISLT }, + { 1, 1, 0, 0, 0, 0, "F2U", TGSI_OPCODE_F2U }, + { 1, 1, 0, 0, 0, 0, "U2F", TGSI_OPCODE_U2F }, + { 1, 2, 0, 0, 0, 0, "UADD", TGSI_OPCODE_UADD }, + { 1, 2, 0, 0, 0, 0, "UDIV", TGSI_OPCODE_UDIV }, + { 1, 3, 0, 0, 0, 0, "UMAD", TGSI_OPCODE_UMAD }, + { 1, 2, 0, 0, 0, 0, "UMAX", TGSI_OPCODE_UMAX }, + { 1, 2, 0, 0, 0, 0, "UMIN", TGSI_OPCODE_UMIN }, + { 1, 2, 0, 0, 0, 0, "UMUL", TGSI_OPCODE_UMUL }, + { 1, 2, 0, 0, 0, 0, "USEQ", TGSI_OPCODE_USEQ }, + { 1, 2, 0, 0, 0, 0, "USGE", TGSI_OPCODE_USGE }, + { 1, 2, 0, 0, 0, 0, "USHR", TGSI_OPCODE_USHR }, + { 1, 2, 0, 0, 0, 0, "USLT", TGSI_OPCODE_USLT }, + { 1, 2, 0, 0, 0, 0, "USNE", TGSI_OPCODE_USNE } }; const struct tgsi_opcode_info * diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h index b34263da48..a35186c223 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h +++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h @@ -124,7 +124,6 @@ OP11(I2F) OP11(NOT) OP11(TRUNC) OP12(SHL) -OP12(SHR) OP12(AND) OP12(OR) OP12(MOD) @@ -146,6 +145,27 @@ OP01(IFC) OP01(BREAKC) OP01(KIL) OP00(END) +OP11(F2I) +OP12(IDIV) +OP12(IMAX) +OP12(IMIN) +OP11(INEG) +OP12(ISGE) +OP12(ISHR) +OP12(ISLT) +OP11(F2U) +OP11(U2F) +OP12(UADD) +OP12(UDIV) +OP13(UMAD) +OP12(UMAX) +OP12(UMIN) +OP12(UMUL) +OP12(USEQ) +OP12(USGE) +OP12(USHR) +OP12(USLT) +OP12(USNE) #undef OP00 -- cgit v1.2.3 From c34f6faf35c16ff81a1c4420290be4ed7f542121 Mon Sep 17 00:00:00 2001 From: Michal Krol Date: Sat, 2 Jan 2010 11:00:40 +0100 Subject: gallium: Add UMOD TGSI opcode. Either that or have UDIV have two destination operands. --- src/gallium/auxiliary/tgsi/tgsi_exec.c | 14 ++++++++++++++ src/gallium/auxiliary/tgsi/tgsi_info.c | 1 + src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h | 1 + src/gallium/include/pipe/p_shader_tokens.h | 23 ++++++++++++----------- 4 files changed, 28 insertions(+), 11 deletions(-) (limited to 'src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h') diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c index 0a276483c4..4d56cac64d 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c @@ -2089,6 +2089,16 @@ micro_umin(union tgsi_exec_channel *dst, dst->u[3] = src[0].u[3] < src[1].u[3] ? src[0].u[3] : src[1].u[3]; } +static void +micro_umod(union tgsi_exec_channel *dst, + const union tgsi_exec_channel *src) +{ + dst->u[0] = src[0].u[0] % src[1].u[0]; + dst->u[1] = src[0].u[1] % src[1].u[1]; + dst->u[2] = src[0].u[2] % src[1].u[2]; + dst->u[3] = src[0].u[3] % src[1].u[3]; +} + static void micro_umul(union tgsi_exec_channel *dst, const union tgsi_exec_channel *src) @@ -3491,6 +3501,10 @@ exec_instruction( exec_vector_binary(mach, inst, micro_umin); break; + case TGSI_OPCODE_UMOD: + exec_vector_binary(mach, inst, micro_umod); + break; + case TGSI_OPCODE_UMUL: exec_vector_binary(mach, inst, micro_umul); break; diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c index 8e0635677a..de0e09cdba 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_info.c +++ b/src/gallium/auxiliary/tgsi/tgsi_info.c @@ -165,6 +165,7 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] = { 1, 3, 0, 0, 0, 0, "UMAD", TGSI_OPCODE_UMAD }, { 1, 2, 0, 0, 0, 0, "UMAX", TGSI_OPCODE_UMAX }, { 1, 2, 0, 0, 0, 0, "UMIN", TGSI_OPCODE_UMIN }, + { 1, 2, 0, 0, 0, 0, "UMOD", TGSI_OPCODE_UMOD }, { 1, 2, 0, 0, 0, 0, "UMUL", TGSI_OPCODE_UMUL }, { 1, 2, 0, 0, 0, 0, "USEQ", TGSI_OPCODE_USEQ }, { 1, 2, 0, 0, 0, 0, "USGE", TGSI_OPCODE_USGE }, diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h index a35186c223..e4af15c156 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h +++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h @@ -160,6 +160,7 @@ OP12(UDIV) OP13(UMAD) OP12(UMAX) OP12(UMIN) +OP12(UMOD) OP12(UMUL) OP12(USEQ) OP12(USGE) diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h index f8d4a45dfd..550e2abc32 100644 --- a/src/gallium/include/pipe/p_shader_tokens.h +++ b/src/gallium/include/pipe/p_shader_tokens.h @@ -308,17 +308,18 @@ struct tgsi_property_data { #define TGSI_OPCODE_UMAD 131 #define TGSI_OPCODE_UMAX 132 #define TGSI_OPCODE_UMIN 133 -#define TGSI_OPCODE_UMUL 134 -#define TGSI_OPCODE_USEQ 135 -#define TGSI_OPCODE_USGE 136 -#define TGSI_OPCODE_USHR 137 -#define TGSI_OPCODE_USLT 138 -#define TGSI_OPCODE_USNE 139 -#define TGSI_OPCODE_SWITCH 140 -#define TGSI_OPCODE_CASE 141 -#define TGSI_OPCODE_DEFAULT 142 -#define TGSI_OPCODE_ENDSWITCH 143 -#define TGSI_OPCODE_LAST 144 +#define TGSI_OPCODE_UMOD 134 +#define TGSI_OPCODE_UMUL 135 +#define TGSI_OPCODE_USEQ 136 +#define TGSI_OPCODE_USGE 137 +#define TGSI_OPCODE_USHR 138 +#define TGSI_OPCODE_USLT 139 +#define TGSI_OPCODE_USNE 140 +#define TGSI_OPCODE_SWITCH 141 +#define TGSI_OPCODE_CASE 142 +#define TGSI_OPCODE_DEFAULT 143 +#define TGSI_OPCODE_ENDSWITCH 144 +#define TGSI_OPCODE_LAST 145 #define TGSI_SAT_NONE 0 /* do not saturate */ #define TGSI_SAT_ZERO_ONE 1 /* clamp to [0,1] */ -- cgit v1.2.3