Board index Linux

On a 386? Re: Linux-1.0-inline-asm uploaded

On a 386? Re: Linux-1.0-inline-asm uploaded

Post by Jeff Epl » Thu, 24 Mar 1994 02:46:51




> 1) speed up some code sequences which are inefficient on the i486
>    and pentium compared to the i386. In particular, the string move
>    instructions (movsb,lodsb,stosb) are slower on the new processors.
>    The gain in speed is about 20-30% on the single functions; on the
>    whole kernel it is much difficult to measure, but certainly smaller
>    and not normally noticeable.

I still have a 386 computer.  Are these optimizations going to affect
speed for better or for worse on my system?

Inquiring minds want to know.

Jeff

--

____ "Nuke the unborn gay whales" -- Never seen on a protest sign
\bi/ |umop apisdn| First year comp sci major.  CRPG *.
 \/  1.5<kinsey<2.5 IRC Synger  Running Linux 0.99.15d. DOS is evil

 
 
 
Top

On a 386? Re: Linux-1.0-inline-asm uploaded

Post by Byron Thomas Fab » Thu, 24 Mar 1994 23:34:09



>I still have a 386 computer.  Are these optimizations going to affect
>speed for better or for worse on my system?
>Inquiring minds want to know.
>Jeff
>--

>____ "Nuke the unborn gay whales" -- Never seen on a protest sign
>\bi/ |umop apisdn| First year comp sci major.  CRPG *.
> \/  1.5<kinsey<2.5 IRC Synger  Running Linux 0.99.15d. DOS is evil

                                         ^^^^^^^^^^^^^^

Umm..it'll probably cause you problems with that.  ;)

Byron
--
PGP 2.3 key available (in plan file) at:        Support public code:



 
 
 
Top

1. Linux-1.0-inline-asm uploaded

Hi Linuxers.
We have uploaded lx10inline.tgz on sunsite. It is a set of patches for 1.0
kernel (and maybe 0.99.15) for speeding up and optimizing some code sequen-
ces used by the kernel, notably those in /linux/include/linux/string.h.

The purpose of these patches is:

 1) speed up some code sequences which are inefficient on the i486
    and pentium compared to the i386. In particular, the string move
    instructions (movsb,lodsb,stosb) are slower on the new processors.
    The gain in speed is about 20-30% on the single functions; on the
    whole kernel it is much difficult to measure, but certainly smaller
    and not normally noticeable.
 2) help gcc performing a better register allocation and produce smaller
    code, as a consequence of avoiding as much as possible any
    instruction which force register use (lods,stos,movs,loop); this
    allows replacing the "c","S","D"... asm constraints with the more
    general forms "q" and "r", letting gcc choose the best register
    placement strategy.
 3) fix the behaviour of the functions in string.h, which are now
    ANSI-compliant.

Here is the lsm entry:

Begin
Title        = Linux-1.0-inline.asm.diff
Version      = 1.0  Mar 18,1994
Desc1        = A set of patches for the Linux 1.0 kernel to speed up
Desc2        = string operations and allow better register allocation by
Desc3        = gcc.
Author       = Alberto Vignani, Davide Parodi

Site1        = sunsite.unc.edu
Path1        = kernel/misc.patches
File1        = lx10inline.tgz
FileSize1    = 6656
Required1    = Linux 1.0 kernel sources
CopyPolicy1  = same as kernel sources
Keywords     = kernel, gcc, assembler, inline, pentium
Entered      = 18 Mar 1994
EnteredBy    = Alberto Vignani
End

Many thanks to Linus Torvalds for his useful suggestions.

-----------------------------------------------------------------------------



[ If mail bounces back please retry ]

2. Vmware Express on FBSD

3. SCSI Problem: DC2964F (AM53C974)

4. inline asm errors when compiling linux-1.2.13 with gcc-2.7.1

5. I can't get hooked to my ISP

6. compiling error: filds 386 asm instruction

7. send(3n) vs send(3xn)

8. 2.5.49-ac1 : include/asm-386/io_apic.h

9. other->386 asm compiler ?

10. 386 disassembler nearly ready for upload

11. inline asm

12. gcc inline asm - short question


All times are UTC
Board index