https://bugs.kde.org/show_bug.cgi?id=438640

--- Comment #14 from Mark Wielaard <[email protected]> ---
(In reply to mcermak from comment #11)
> Created attachment 193045 [details]
> proposed patch

About the addition tests in  none/tests/x86/sse4-x86.c with results in
none/tests/x86/sse4-x86.stdout.exp
The results (sse4-x86.stdout.exp) when generated by running the program
natively
none/tests/x86/sse4-x86 > none/tests/x86/sse4-x86.stdout.exp
differ from the results when ran natively
./vg-in-place -q none/tests/x86/sse4-x86 > none/tests/x86/sse4-x86.stdout.exp
Which means the crc32 implementation in VEX is wrong (but I haven't looked yet
where the bug is).

Secondly I don't understand this comment:
+      // OF, SF, ZF, AF, CF, PF are all cleared.
+      // ZF is set if SRC = 0, otherwise ZF is cleared
I don't believe that is true. And the tests seem to show the the flags
(block[3]) are always zero?

Finally, amd64 does have crc32 testcases, but they aren't in 
none/tests/amd64/sse4-64 but in
./none/tests/amd64/crc32.vgtest. Could we have a ./none/tests/x86/crc32
variant. With maybe
the new tests from sse4-x86 included?

The none/tests/x86/sse4-x86 is getting really big, lets only add more tests to
it if they are also
already in none/tests/amd64/sse4-64.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to