View previous topic :: View next topic |
Author |
Message |
fizz Guru
Joined: 31 Aug 2003 Posts: 309 Location: Florida
|
Posted: Mon Dec 29, 2003 7:07 pm Post subject: |
|
|
so to take full advantage of the nvidia agp patch, would i NOT include support for agpgart? i havent compiled that in the kernel for a long time anywayz. _________________ Athlon 64 3200, MSI NEO NForce 3, 1Gig PC3700, EVGA Geforce 6800 GT |
|
Back to top |
|
|
Wedge_ Advocate
Joined: 08 Aug 2002 Posts: 3614 Location: Scotland
|
Posted: Mon Dec 29, 2003 7:33 pm Post subject: |
|
|
fizz wrote: | so to take full advantage of the nvidia agp patch, would i NOT include support for agpgart? |
Nope. Unless I'm thinking of the wrong nvidia-agp patch all that it does is add a function to the existing code in the kernel which seems to break AGP on Epox 8RDA boards if it isn't there. I have the same board as you, and it doesn't seem to have any effect either way on the A7N8X. Set the AGP options in the kernel the same way you normally would. _________________ Per Ardua Ad Astra
The Earth is the cradle of the mind, but we cannot live forever in a cradle - Konstantin E. Tsiolkovsky
Gentoo Radeon FAQ |
|
Back to top |
|
|
relyt Apprentice
Joined: 29 Aug 2002 Posts: 238 Location: Massachusetts
|
Posted: Mon Dec 29, 2003 10:34 pm Post subject: Re: 2.6.0-love1 (and 2.6.0-love2) |
|
|
steel300 wrote: | Yet To Do:
Runtime Selectable Block Scheduler |
I have fixed these patches for -love2 (and -mm2) and combined them into one patch. I can send this to people, but maybe there's someone out there with web space who will post it? |
|
Back to top |
|
|
neenee Veteran
Joined: 20 Jul 2003 Posts: 1786
|
Posted: Mon Dec 29, 2003 11:18 pm Post subject: |
|
|
i could host it - send it to neeneelaatmaar@gmx.net;
i'll post a link once it's up. |
|
Back to top |
|
|
fca Guru
Joined: 22 Feb 2003 Posts: 346 Location: Netherlands
|
Posted: Mon Dec 29, 2003 11:19 pm Post subject: Re: 2.6.0-love1 (and 2.6.0-love2) |
|
|
relyt wrote: | steel300 wrote: | Yet To Do:
Runtime Selectable Block Scheduler |
I have fixed these patches for -love2 (and -mm2) and combined them into one patch. I can send this to people, but maybe there's someone out there with web space who will post it? |
I could host them. I already have a variety of ebuilds and such on my server. But also send it to steel300, he can incorporate it into the love patchset, which I cannot. |
|
Back to top |
|
|
neenee Veteran
Joined: 20 Jul 2003 Posts: 1786
|
|
Back to top |
|
|
dedeaux Apprentice
Joined: 19 Jun 2002 Posts: 183 Location: Nairobi, Kenya
|
Posted: Tue Dec 30, 2003 1:04 am Post subject: performance |
|
|
hey folks!
thanks for the hard work on this patchset. I am currently running the love2 set.
I was just wanting to inquire as to how to see the performance boost on the desktop.
I use grub as my boot loader and have this line:
Code: |
kernel (hd0,1)/boot/bzImage root=/dev/hda4 elevator=cfq
|
in my grub.conf
Is this correct? I have used both gentoo 2.6 and love2 patched 2.6 and see no improvement.
I am glad to see it working as earlier test kernels wouldn't work.
I still get random lockups on boot or kernel panics on boot, but overall the system is stable.
Thanks for the help. |
|
Back to top |
|
|
dabooty Guru
Joined: 15 May 2003 Posts: 482 Location: Belgium
|
Posted: Tue Dec 30, 2003 1:23 am Post subject: |
|
|
since you're all very much concerned with kernel performance, you might wanna have a look at this thread i just started _________________ registered user #284425
get yourself counted
http://counter.li.org
------
#emerge -pv solves a lot of questions beforehand |
|
Back to top |
|
|
dedeaux Apprentice
Joined: 19 Jun 2002 Posts: 183 Location: Nairobi, Kenya
|
Posted: Tue Dec 30, 2003 1:36 am Post subject: |
|
|
thanks dabooty, I read your thread and will reply.... but I was asking about the kernel flag 'elevator=cfq' as it has been said to offer the best performance. I know this is all relative to our systems and setup -- I just wanted confirmation that it was correct as I see no difference. |
|
Back to top |
|
|
neenee Veteran
Joined: 20 Jul 2003 Posts: 1786
|
Posted: Tue Dec 30, 2003 11:38 am Post subject: |
|
|
yes. elevator=cfq in the kernel line is correct
like that. that scheduler is not really for per-
formance, but for good responsiviness under
high load i believe; watching a movie while
emerging world and such.
(please correct me if i am wrong) |
|
Back to top |
|
|
asimon l33t
Joined: 27 Jun 2002 Posts: 979 Location: Germany, Old Europe
|
Posted: Tue Dec 30, 2003 4:44 pm Post subject: |
|
|
I just asked myself if anyone does actually review all those patches? Just wondering how easy it would be to get some evil(tm) code into the love-sources... |
|
Back to top |
|
|
relyt Apprentice
Joined: 29 Aug 2002 Posts: 238 Location: Massachusetts
|
Posted: Tue Dec 30, 2003 6:24 pm Post subject: |
|
|
asimon wrote: | I just asked myself if anyone does actually review all those patches? Just wondering how easy it would be to get some evil(tm) code into the love-sources... |
You make a good point. The answer is that most of the 'review' done at this level comes from being discriminating about where you accept patches from.
To make people feel better about my patch (in particular), here it is, presented as a patch to the elv-select.patch that Nick has posted:
Code: | --- elv-select.patch 2003-12-30 12:54:02.000000000 -0500
+++ elv-select.patch 2003-12-30 13:08:12.000000000 -0500
@@ -77,7 +77,7 @@
return;
- elevator_exit(q);
-+ if (e->elevator)
++ if (q->elevator)
+ elevator_exit(q);
del_timer_sync(&q->unplug_timer);
@@ -740,10 +740,8 @@
ad->current_batch_expires = jiffies + ad->batch_expire[REQ_SYNC];
ad->write_batch_count = ad->batch_expire[REQ_ASYNC] / 10;
-@@ -1823,8 +1786,58 @@ static int as_init(request_queue_t *q, e
+@@ -1823,6 +1786,56 @@ static int as_init(request_queue_t *q, e
ad->write_batch_count = 2;
-
- ad->new_success = 512;
+}
- return 0; |
Apply this patch to Nick Piggin's patch, then apply it the your tree, then apply Nick's elv-init-cleanup patch and you'll have what I made. You can see that all I did was fix the e/q mistake, and remove two lines that are no longer in one of the the source files. |
|
Back to top |
|
|
PrakashP Veteran
Joined: 27 Oct 2003 Posts: 1249 Location: C.C.A.A., Germania
|
Posted: Wed Dec 31, 2003 11:17 am Post subject: |
|
|
2.6.1-rc1-mm1 is out... |
|
Back to top |
|
|
lems n00b
Joined: 07 Nov 2003 Posts: 25
|
|
Back to top |
|
|
asimon l33t
Joined: 27 Jun 2002 Posts: 979 Location: Germany, Old Europe
|
Posted: Wed Dec 31, 2003 1:23 pm Post subject: |
|
|
Yes, apart from 5 small rejects it applies cleanly to mm2. The rejected stuff is already in mm2. |
|
Back to top |
|
|
didl Retired Dev
Joined: 09 Sep 2003 Posts: 1106 Location: Pittsburgh, PA
|
Posted: Wed Dec 31, 2003 2:29 pm Post subject: |
|
|
2.6.1-rc1-mm1 is out |
|
Back to top |
|
|
asimon l33t
Joined: 27 Jun 2002 Posts: 979 Location: Germany, Old Europe
|
Posted: Wed Dec 31, 2003 2:42 pm Post subject: |
|
|
I just booted into 2.6.1-rc-1-mm1 with alsa-bk-2003-12-30 and Con's 2.6.0-am-5. |
|
Back to top |
|
|
didl Retired Dev
Joined: 09 Sep 2003 Posts: 1106 Location: Pittsburgh, PA
|
Posted: Wed Dec 31, 2003 3:00 pm Post subject: |
|
|
asimon wrote: |
I just booted into 2.6.1-rc-1-mm1 with alsa-bk-2003-12-30 and Con's 2.6.0-am-5. |
Alsa doesn't apply very nicely to 2.6.1-rc-1-mm1 for me! A few hunks fail.
Did you do anything to fix this?? |
|
Back to top |
|
|
asimon l33t
Joined: 27 Jun 2002 Posts: 979 Location: Germany, Old Europe
|
Posted: Wed Dec 31, 2003 3:42 pm Post subject: |
|
|
didl wrote: |
Alsa doesn't apply very nicely to 2.6.1-rc-1-mm1 for me! A few hunks fail. Did you do anything to fix this?? |
No, I did nothing to "fix" it. It's the same as above with 2.6.0-mm2. The few hunks which fail are already part of 2.6.0-mm2 and 2.6.1-rc-1-mm1 (looks like these are mostly spinlock fixes):
./sound/i2c/i2c.c.rej
./sound/isa/gus/gus_pcm.c.rej
./sound/core/sound.c.rej
./sound/sound_core.c.rej
./include/sound/i2c.h.rej
Sound is working fine here. |
|
Back to top |
|
|
Roguelazer Veteran
Joined: 10 Feb 2003 Posts: 1233 Location: San Francisco, CA
|
Posted: Wed Dec 31, 2003 8:59 pm Post subject: |
|
|
So... anyone found 2.6.1-rc1-love1 yet? _________________ Registered Linux User #263260 |
|
Back to top |
|
|
Dinini Apprentice
Joined: 11 Jun 2003 Posts: 236
|
Posted: Thu Jan 01, 2004 12:45 pm Post subject: |
|
|
bootsplash-3.1.3 applies to 2.6.1-rc1-mm1 with one small change.
(The "struct device *dev; /* pointer to the device for this fb */" line is new and is what causes the sole reject.)
Code: | --- linux/include/linux/fb.h.orig 2004-01-01 05:49:13.422361696 -0600
+++ linux/include/linux/fb.h 2004-01-01 05:54:38.732906976 -0600
@@ -414,6 +414,14 @@
int currcon; /* Current VC. */
void *pseudo_palette; /* Fake palette of 16 colors */
struct device *dev; /* pointer to the device for this fb */
+#ifdef CONFIG_BOOTSPLASH
+ struct splash_data *splash_data;
+ unsigned char *splash_pic;
+ int splash_pic_size;
+ int splash_bytes;
+ char *silent_screen_base; /* real screen base */
+ char fb_cursordata[64];
+#endif
/* From here on everything is device dependent */
void *par;
}; |
|
|
Back to top |
|
|
nepenthe Guru
Joined: 14 Jul 2002 Posts: 317 Location: Sanford, ME
|
Posted: Fri Jan 02, 2004 5:02 pm Post subject: |
|
|
miss you love-sources... |
|
Back to top |
|
|
|