Commentary section update in sha512-x86_64.pl.

#

#

# sha256/512_block procedure for x86_64.

# sha256/512_block procedure for x86_64.

#

#

# 40% improvement over compiler-generated code on Opteron. No magical

# 40% improvement over compiler-generated code on Opteron. On EM64T

# sha256 was observed to run >80% faster and sha512 - >40%. No magical

# tricks, just straight implementation... I really wonder why gcc

# tricks, just straight implementation... I really wonder why gcc

# [being armed with inline assembler] fails to generate as fast code.

# [being armed with inline assembler] fails to generate as fast code.

# The only thing which is cool about this module is that it's very

# The only thing which is cool about this module is that it's very

#

#

# Special note on Intel EM64T. While Opteron CPU exhibits perfect

# Special note on Intel EM64T. While Opteron CPU exhibits perfect

# perfromance ratio of 1.5 between 64- and 32-bit flavors [see above],

# perfromance ratio of 1.5 between 64- and 32-bit flavors [see above],

# [currently available] EM64T CPUs apparently are far from it. 64-bit

# [currently available] EM64T CPUs apparently are far from it. On the

# version, sha512_block, is hardly faster than 32-bit one. This is

# contrary, 64-bit version, sha512_block, is ~30% *slower* than 32-bit

# presumably because 64-bit shifts/rotates apparently are not atomic

# sha256_block:-( This is presumably because 64-bit shifts/rotates

# instructions, but implemented in microcode.

# apparently are not atomic instructions, but implemented in microcode.

$output=shift;

$output=shift;

open STDOUT,"| $^X ../perlasm/x86_64-xlate.pl $output";

open STDOUT,"| $^X ../perlasm/x86_64-xlate.pl $output";