On 15/9/26 01:08, Richard Henderson wrote:
Use movcond instead of a pair of branches to perform the computation, packed in a new gen_cps_i32.Signed-off-by: Richard Henderson <[email protected]> --- target/mips/tcg/mxu_translate.c | 42 +++++++++------------------------ 1 file changed, 11 insertions(+), 31 deletions(-)
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
