| View previous topic :: View next topic |
| Author |
Message |
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
 Posted: Thu May 05, 2005 12:02 am Post subject: |
 |
|
Little bit out of topic but I just run onto this by luck today :
cat /usr/src/linux/arch/i386/Makefile
...
CFLAGS += -pipe -msoft-float
...
Why -msoft-float is there? According to the gcc manpage :
| Code: |
-msoft-float
Generate output containing library calls for floating
point. Warning: the requisite libraries are not part
of GCC. Normally the facilities of the machine's
usual C compiler are used, but this can't be done
directly in cross-compilation. You must make your own
arrangements to provide suitable library functions for
cross-compilation.
On machines where a function returns floating point
results in the 80387 register stack, some floating
point opcodes may be emitted even if -msoft-float is
used.
|
But from i486 and then, *every* x86 processor has an FPU. Why gcc is forced to use library functions to perform math functions and not use native cpu instructions?
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
 |
rmh3093
Advocate
Joined: 06 Aug 2003
Posts: 2138
Location: Albany, NY
|
| Posted: Thu May 05, 2005 1:17 am Post subject: |
|
|
have you tried removing it and seeing what happens?
_________________
Do not meddle in the affairs of wizards, for they are subtle and quick to anger. |
|
| Back to top |
|
|
HecHacker1
Apprentice
Joined: 26 Jun 2003
Posts: 213
Location: UCSD
|
| Posted: Thu May 05, 2005 7:58 am Post subject: |
|
|
i have this problem when trying to compile:
CC arch/i386/kernel/time.o
CC arch/i386/kernel/ioport.o
CC arch/i386/kernel/ldt.o
CC arch/i386/kernel/setup.o
arch/i386/kernel/setup.c: In function `setup_bootmem_allocator':
arch/i386/kernel/setup.c:1150: error: `__PHYSICAL_START' undeclared (first use in this function)
arch/i386/kernel/setup.c:1150: error: (Each undeclared identifier is reported only once
arch/i386/kernel/setup.c:1150: error: for each function it appears in.)
make[1]: *** [arch/i386/kernel/setup.o] Error 1
make: *** [arch/i386/kernel] Error 2 |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Thu May 05, 2005 10:27 am Post subject: |
|
|
No problem. Everything compiles and runs perfectly. Maybe a placebo performance increase, I don't know. But since its working...
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
bollucks
l33t
Joined: 27 Oct 2004
Posts: 606
|
| Posted: Thu May 05, 2005 11:41 am Post subject: |
|
|
| The kernel is integer only code so the fpu is never used. If it is used in the kernel it's a disaster because there is no way of locking the data and you can get races. |
|
| Back to top |
|
|
rmh3093
Advocate
Joined: 06 Aug 2003
Posts: 2138
Location: Albany, NY
|
| Posted: Thu May 05, 2005 1:37 pm Post subject: |
|
|
-skunk4 will prob be out this evening with, -mm3, nicksched (maybe plugsched), and the latest Suspend2
_________________
Do not meddle in the affairs of wizards, for they are subtle and quick to anger. |
|
| Back to top |
|
|
HecHacker1
Apprentice
Joined: 26 Jun 2003
Posts: 213
Location: UCSD
|
| Posted: Thu May 05, 2005 6:22 pm Post subject: |
|
|
| HecHacker1 wrote: | i have this problem when trying to compile:
CC arch/i386/kernel/time.o
CC arch/i386/kernel/ioport.o
CC arch/i386/kernel/ldt.o
CC arch/i386/kernel/setup.o
arch/i386/kernel/setup.c: In function `setup_bootmem_allocator':
arch/i386/kernel/setup.c:1150: error: `__PHYSICAL_START' undeclared (first use in this function)
arch/i386/kernel/setup.c:1150: error: (Each undeclared identifier is reported only once
arch/i386/kernel/setup.c:1150: error: for each function it appears in.)
make[1]: *** [arch/i386/kernel/setup.o] Error 1
make: *** [arch/i386/kernel] Error 2 |
I disabled 1 GIG low mem support and I got past this error, but now I won't be able to use all my ram. Please fix in your next release. Thanks |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Thu May 05, 2005 8:07 pm Post subject: |
|
|
Try using HIGHMEM. Plus you get vmware working 
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Thu May 05, 2005 9:40 pm Post subject: |
|
|
| bollucks wrote: | | The kernel is integer only code so the fpu is never used. If it is used in the kernel it's a disaster because there is no way of locking the data and you can get races. |
That has nothing to do with the fact that it is compiled using software floating point instructions. Plus only process syncronization code (spinlocks, mutexes, semaphores etc.) is (and has to be) integer only, not audio drivers and such that do use floating point arithmetic. So the question is why is software floating point arithmetic used when native hardware instructions exist.
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
HecHacker1
Apprentice
Joined: 26 Jun 2003
Posts: 213
Location: UCSD
|
| Posted: Thu May 05, 2005 9:40 pm Post subject: |
|
|
i don't care to use VmWare in linux because I have it on my XP RAID partition that has much faster disk access.
And I know HighMem works, but there is overhead associated with it just to get the extra 200MB or so that the 1G patch allows. |
|
| Back to top |
|
|
rmh3093
Advocate
Joined: 06 Aug 2003
Posts: 2138
Location: Albany, NY
|
| Posted: Fri May 06, 2005 12:39 am Post subject: |
|
|
I started -skunk4 but i need more beer so im going to take a break and go to the store, it will be out tonight, definately with nicksched, times running out for last minute requests
_________________
Do not meddle in the affairs of wizards, for they are subtle and quick to anger. |
|
| Back to top |
|
|
bollucks
l33t
Joined: 27 Oct 2004
Posts: 606
|
| Posted: Fri May 06, 2005 5:55 am Post subject: |
|
|
| priestjim wrote: | | bollucks wrote: | | The kernel is integer only code so the fpu is never used. If it is used in the kernel it's a disaster because there is no way of locking the data and you can get races. |
That has nothing to do with the fact that it is compiled using software floating point instructions. Plus only process syncronization code (spinlocks, mutexes, semaphores etc.) is (and has to be) integer only, not audio drivers and such that do use floating point arithmetic. So the question is why is software floating point arithmetic used when native hardware instructions exist. |
Find some floating point code in the drivers for me. |
|
| Back to top |
|
|
HecHacker1
Apprentice
Joined: 26 Jun 2003
Posts: 213
Location: UCSD
|
| Posted: Fri May 06, 2005 7:08 am Post subject: |
|
|
mmmmmm, beer. That reminds me, I need to pick some up too. It's almost friday! (and I'm almost done with midterms) |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Fri May 06, 2005 9:04 am Post subject: |
|
|
| bollucks wrote: | | priestjim wrote: | | bollucks wrote: | | The kernel is integer only code so the fpu is never used. If it is used in the kernel it's a disaster because there is no way of locking the data and you can get races. |
That has nothing to do with the fact that it is compiled using software floating point instructions. Plus only process syncronization code (spinlocks, mutexes, semaphores etc.) is (and has to be) integer only, not audio drivers and such that do use floating point arithmetic. So the question is why is software floating point arithmetic used when native hardware instructions exist. |
Find some floating point code in the drivers for me. |
cd /usr/src/linux
grep -R "float " *
grep -R "double " *
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
playfool
l33t
Joined: 01 Jun 2004
Posts: 688
Location: Ã
rhus, Denmark
|
| Posted: Fri May 06, 2005 10:20 am Post subject: |
|
|
| priestjim wrote: | | swsusp2? Lately noone seems to care about that and why is beyond me... |
You haven't read the code I see |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Fri May 06, 2005 2:24 pm Post subject: |
|
|
Hey no!  Well you're right, I haven't read the code but "care" wasn't going to the developers but to the patchset makers. CKO and love haven't had swsusp2 in their latest releases.
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
feld
Guru
Joined: 29 Aug 2004
Posts: 593
Location: WI, USA
|
| Posted: Fri May 06, 2005 5:07 pm Post subject: |
|
|
looks like i'll be trying out your kernel patchset
For fun i decided to try gentoo-sources once again. Let me just say that I almost threw my keyboard against the wall. The interactivity sucks. I cant compile anything and do any other cpu intensive tasks no matter how NICED portage is without it going laggy.
In my experience with my past few kernels I have found that nothing compares to Genetic AS and Genetic Nicksched. I can play Americas Army in another X session (takes up about 93-97% of cpu), and in my other X i can run KDE, konsole with irssi in it, internet radio via beep media player, start emerging something, and burn a cd/dvd with my system still being 100% responsive.
I see you have the genetic AS and the rest of your kernel looks tight so we'll see how it goes 
Here's to hopin you can fill this gap for me *raises glass of beer*
-Feld
_________________
< bmg505> I think the first line in reiserfsck is
if (random(65535)< 65500) { hose(partition); for (i=0;i<100000000;i++) print_crap(); } |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Sat May 07, 2005 1:51 pm Post subject: |
|
|
Would it be rude to ask where is skunk4? Been waiting for 2 days 
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
rmh3093
Advocate
Joined: 06 Aug 2003
Posts: 2138
Location: Albany, NY
|
| Posted: Sat May 07, 2005 3:23 pm Post subject: |
|
|
No sorry, here's where Im at right now.... Im sorta sick of Staircase, and Nicksched just sits idle there is no development of it. I sorta agree with Love that since Ingosched is in constant development, its probably the best choice. Plus since I try to use -mm things should be faily quick if the kernel actually boots for you. I was considereing try out the Realtime patch since that gets updated nearly every day. I till be out tonight definately!! This is my last week of class before exams come, so the next 2 weeks will be a little slow for me on here but once thats out of the way
_________________
Do not meddle in the affairs of wizards, for they are subtle and quick to anger. |
|
| Back to top |
|
|
rmh3093
Advocate
Joined: 06 Aug 2003
Posts: 2138
Location: Albany, NY
|
| Posted: Sat May 07, 2005 3:28 pm Post subject: |
|
|
and now -rc4 is out so thats cool
_________________
Do not meddle in the affairs of wizards, for they are subtle and quick to anger. |
|
| Back to top |
|
|
priestjim
Tux's lil' helper
Joined: 09 Jan 2005
Posts: 100
Location: Never never land
|
| Posted: Sun May 08, 2005 9:22 am Post subject: |
|
|
What's wrong with Staircase? And because NickSched is not being actively developed it means that it's not good? Both Staircase and Nicksched are very good schedulers so users should have a choice (hint: plugsched)
_________________
My soul is my antenna, I am the instrument + the guitar is my amplifier.
-- Ottmar Liebert |
|
| Back to top |
|
|
|
Unsupported Software
