> I had this problem when I wanted to use SIMD instructions
> and registers for the PIII, the kernel was not adapted !
> So, I don't think using P4 will enhance anything more than PIII
> or Athlon at same speed .
I've been using the SIMD/mmx registers and instructions without a problem
on both the Athlon, the Intel-Pro-III. It may be kind of nice that the
kernel and gcc doesn't know about the registers and instructions in
the P4 yet.
That means neither the kernel nor gcc is using them and therefore those
register do not need to be saved during a task switch unless you are using
them in multiple tasks. There are so many new FP registers ( I think 80 )
that different tasks could simply have their own subset and avoid
save/restore during task switch.
Unfortunately the basic mmx and basic floating point
registers share the same space in the Athlon and the P3. I don't think
that is the case with the extended P4 fp registers. The new FP
registers implement SIMD (Single Instruction Multiple Data) for floating
point operations.
FYI, using the SIMD/mmx stuff and being paying attention to how data are
cached I have been able to increase the speed of certain functions by as
much as 500%, or 5 times faster. I've been using the nasm assembler to
code in assembly and link with gcc C programs.
-Wayne
--<<< We are committed to exceeding requirements >>>----->-+
C.W. Wright,[EMAIL PROTECTED], Ph:757-824-1698
http://lidar.wff.nasa.gov Fax:603-925-6886
NASA, Goddard Space Flight Center
Building N-159 Room E117, Code 972,Wallops Flight Facility
Wallops Island, Va. 23337 Telecommute phone: 410-742-7333
-------+-<-------------------------------------------------
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
--
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
