diff options
author | Roland Scheidegger <[email protected]> | 2016-02-02 03:36:19 +0100 |
---|---|---|
committer | Roland Scheidegger <[email protected]> | 2016-02-02 05:58:19 +0100 |
commit | dc16086e3b616f767ab31c86505502c4df7739ac (patch) | |
tree | f3f30e58c69beeea8a20a49df12b8267b0c0b299 /src/gallium/auxiliary/tgsi | |
parent | 99bd96abbb62d2c7da60c6102661b590e05bf143 (diff) |
tgsi: add PK2H/UP2H support
The util functions handle the half-float conversion.
Note that piglit won't like it much due to:
a) The util functions use magic float mul conversion but when run inside
softpipe/llvmpipe, denorms are flushed to zero, therefore when the conversion
is from/to f16 denorm the result will be zero. This is a bug which should be
fixed in these functions (should not rely on denorms being available), but
will happen elsewhere just the same (e.g. conversion to f16 render targets).
b) The util functions use trunc round mode rather than round-to-nearest. This
is NOT a bug (as it is a d3d10 requirement). This will result of rounding not
representable finite values to MAX_F16 rather than INFINITY. My belief is the
piglit tests are wrong here but it's difficult to tell (generally glsl
rounding mode is undefined, however I'm not sure if rounding mode might need
to be consistent for different operations). Nevertheless, for gl it would be
better to use round-to-nearest, but using different rounding for GL and d3d10
is an unsolved problem (as it affects things like conversion to f16 render
targets, clear colors, this shader opcode).
Hence for now don't enable the cap bit (so the code is unused).
(Code is from imirkin, comment from sroland)
Signed-off-by: Ilia Mirkin <[email protected]>
Reviewed-by: Roland Scheidegger <[email protected]>
Reviewed-by: Brian Paul <[email protected]>
Diffstat (limited to 'src/gallium/auxiliary/tgsi')
-rw-r--r-- | src/gallium/auxiliary/tgsi/tgsi_exec.c | 44 |
1 files changed, 42 insertions, 2 deletions
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c index f67c16200a9..d898fd66f48 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c @@ -58,6 +58,7 @@ #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_util.h" #include "tgsi_exec.h" +#include "util/u_half.h" #include "util/u_memory.h" #include "util/u_math.h" @@ -3058,6 +3059,45 @@ exec_dp2(struct tgsi_exec_machine *mach, } static void +exec_pk2h(struct tgsi_exec_machine *mach, + const struct tgsi_full_instruction *inst) +{ + unsigned chan; + union tgsi_exec_channel arg[2], dst; + + fetch_source(mach, &arg[0], &inst->Src[0], TGSI_CHAN_X, TGSI_EXEC_DATA_FLOAT); + fetch_source(mach, &arg[1], &inst->Src[0], TGSI_CHAN_Y, TGSI_EXEC_DATA_FLOAT); + for (chan = 0; chan < TGSI_QUAD_SIZE; chan++) { + dst.u[chan] = util_float_to_half(arg[0].f[chan]) | + (util_float_to_half(arg[1].f[chan]) << 16); + } + for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + store_dest(mach, &dst, &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_UINT); + } + } +} + +static void +exec_up2h(struct tgsi_exec_machine *mach, + const struct tgsi_full_instruction *inst) +{ + unsigned chan; + union tgsi_exec_channel arg, dst[2]; + + fetch_source(mach, &arg, &inst->Src[0], TGSI_CHAN_X, TGSI_EXEC_DATA_UINT); + for (chan = 0; chan < TGSI_QUAD_SIZE; chan++) { + dst[0].f[chan] = util_half_to_float(arg.u[chan] & 0xffff); + dst[1].f[chan] = util_half_to_float(arg.u[chan] >> 16); + } + for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) { + if (inst->Dst[0].Register.WriteMask & (1 << chan)) { + store_dest(mach, &dst[chan & 1], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT); + } + } +} + +static void exec_scs(struct tgsi_exec_machine *mach, const struct tgsi_full_instruction *inst) { @@ -4339,7 +4379,7 @@ exec_instruction( break; case TGSI_OPCODE_PK2H: - assert (0); + exec_pk2h(mach, inst); break; case TGSI_OPCODE_PK2US: @@ -4425,7 +4465,7 @@ exec_instruction( break; case TGSI_OPCODE_UP2H: - assert (0); + exec_up2h(mach, inst); break; case TGSI_OPCODE_UP2US: |