On 2023-09-28, Richard Henderson <richard.hender...@linaro.org> wrote: > On 9/24/23 01:03, Nick Bowler wrote: >> All of the VIS subtraction instructions are documented to subtract the >> second input operand from the first. This is also consistent with how >> the instructions actually work on a real UltraSparc II. >> >> But the emulator is implementing the subtraction in the wrong order, >> subtracting the first input from the second, so the results are wrong >> in all nontrivial cases. >> >> Signed-off-by: Nick Bowler <nbow...@draconx.ca> >> --- >> target/sparc/vis_helper.c | 18 +++++++++--------- >> 1 file changed, 9 insertions(+), 9 deletions(-) > > While this patch works, better to use > > void tcg_gen_vec_add16_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b); > void tcg_gen_vec_add16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b); > void tcg_gen_vec_add32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b); > > void tcg_gen_vec_sub16_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b); > void tcg_gen_vec_sub16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b); > void tcg_gen_vec_sub32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b); > > from "tcg/tcg-op-gvec.h" and remove the sparc helpers.
OK, I will try to respin this one using these. Thanks, Nick
