 |
|
| Where to Download |
|
|
By thym - Posted on October 2, 2006 - 02:29:36 (#20699)
Current version when comment was posted: 1.4 |
|
|
Hi,but where to download this pretty software for BEOS 5?
The link displayed here maybe a bad one.
|
|
| Fixes my own vlc issues |
|
|
By MYOB - Posted on February 14, 2006 - 19:17:22 (#19065)
Current version when comment was posted: 1.4 |
|
|
Never thought of using this, assumed my Celeron was fine time wise, but installed it on remembering what it did for the "jitter" on Athlons
And it works fine - no more message window full of resampling warnings (except on damaged files). Yay!
|
|
| problem with vlc |
|
|
By whaka - Posted on November 29, 2005 - 20:08:04 (#18477)
Current version when comment was posted: 1.3 |
|
|
|
| Just wait for the new release... |
|
|
By kraton. - Posted on November 29, 2005 - 05:48:18 (#18468)
Current version when comment was posted: 1.3 |
|
|
read more here:
http://beos.mlotz.ch/documents/cpu_fix.html
http://www.iscomputeron.com/
|
|
| problem with vlc |
|
|
By whaka - Posted on November 29, 2005 - 01:35:15 (#18467)
Current version when comment was posted: 1.3 |
|
|
Hi, i use the cpu fix for Exp/Dano it work fine for the media kit, but vlc don't like it the audio always accelerate.
Whaka.
|
|
| Woohoo |
|
|
By tqh - Posted on November 28, 2005 - 11:31:30 (#18459)
Current version when comment was posted: 1.3 |
|
|
|
| Update interesting for SpeedStep users |
|
|
By mmlr - Posted on November 28, 2005 - 09:17:03 (#18458)
Current version when comment was posted: 1.3 |
|
|
I suppose I have some good news for anyone that seeks to fix the timing problems when using the SpeedStep driver (http://www.bebits.com/app/4178). The next version of cpu_fix will bring an additional fix and a new kernel and libroot patch will be released. These components will work together to enable SpeedStep users to switch their CPU frequency and fix the timing issues at runtime. I plan to release the update / patches tonight.
I don't really like to do it by patching system files, but it's the best I can offer.
For the full story, read more at http://beos.mlotz.ch/documents/cpu_fix.html
|
|
| Celeron D 2.4 |
|
|
By fer_all4JC - Posted on June 22, 2005 - 09:44:06 (#16952)
Current version when comment was posted: 1.3 |
|
|
With this fix Zeta Neo SP1 displays correctly my cpu clock, i.e., 2.4 GHZ...
|
|
| great! |
|
|
By nutela - Posted on June 9, 2005 - 09:15:28 (#16847)
Current version when comment was posted: 1.3 |
|
|
A bit late but it works great!
; )
=================\\
Please vote people, I just did (9)
|
|
| file mirror |
|
|
By Kirrilian - Posted on November 19, 2004 - 09:10:19 (#14803)
Current version when comment was posted: 1.3 |
|
|
im going to mirror these files unless someone has a problem with that, its the least i can do to help.
thanks very much for your hard work on this patch, it has made this os usable for me again. :)
|
|
| placement and evidence |
|
|
By s_d - Posted on November 14, 2004 - 07:47:03 (#14737)
Current version when comment was posted: 1.3 |
|
|
Should cpu_fix to be visible in syslog and/or /dev filesystem? I don't see such name there
(Also, about IDE replacement, I noticed it installs in */dev only in case of "user-install". In case of "system-install" it goes usual way with */bin and */dev/symlink)
|
|
| It has a sense |
|
|
By mmlr - Posted on November 14, 2004 - 07:29:08 (#14736)
Current version when comment was posted: 1.3 |
|
|
Try moving it to bin and add a symlink and you will see that it isn't loaded as early as it should (sysinit 2). The same goes for the ide replacement driver, that's the reason why it is installed directly to dev too.
|
|
| Re: disk trick: dev vs bin |
|
|
By mmu_man - Posted on November 14, 2004 - 05:35:44 (#14731)
Current version when comment was posted: 1.3 |
|
|
no reason for that.
it should be in bin/ with a symlink from dev/disk/
|
|
| disk trick: dev vs. bin |
|
|
By s_d - Posted on November 13, 2004 - 15:06:45 (#14726)
Current version when comment was posted: 1.3 |
|
|
Has it any special sense that binary is placed in drivers/dev folder instead usual drivers/bin with symlink to dev?
|
|
| P4 2800Mc |
|
|
By rudolfc - Posted on November 7, 2004 - 04:26:44 (#14666)
Current version when comment was posted: 1.3 |
|
|
Hi,
Thanks a lot for this fix. I just finally installed it over here (a few days ago) on my ASUS MB 533FSB, P4 2800Mhz running Dano, and it's working fully as it should AFAICT. Media seems fine and the speed is correctly Identified.
Best regards,
Rudolf.
|
|
| Re: s_d |
|
|
By BiPolar - Posted on November 6, 2004 - 12:56:55 (#14663)
Current version when comment was posted: 1.3 |
|
|
I just build it from sources, and moved cpu_fix to [...]/drivers/dev/disk/ . I didn't moved the cpu_fix_settings file, so both the cv_factor and the get_cpu_info() patches are applied.
I'm using PE R5.0.4 (5.0.3 + media_kit_update) if that makes any difference.
|
|
| 2 BiPolar |
|
|
By s_d - Posted on November 6, 2004 - 04:48:06 (#14660)
Current version when comment was posted: 1.3 |
|
|
Did you apply both parts of patches (media fix and CPU clock indicator fix)?
|
|
| Holy Cow! this patch even makes good coffee!!! |
|
|
By BiPolar - Posted on November 6, 2004 - 00:52:04 (#14659)
Current version when comment was posted: 1.3 |
|
|
Holy Cow! this patch even makes good coffee!!! :-P
After reading tb100's post, I've tried this patch on my K7@900.
Guess what... it fixed an issue I was having with the SiS7018 sound card driver made by Siarzhuk Zharski. The problem was that you got nasty sound-glitches every XX seconds (24-40 usually).
Now they seems to be gone! Thanks a lot Michael!
Details:
KERN 'BOOT': IBM PC Compatible platform
KERN 'BOOT': CPU speed is 898MHZ
[...]
KERN 'sysinit2'[3]: cpu_fix: fixing cv_factor...
KERN 'sysinit2'[3]: cpu_fix: CPU at 896.684 MHz
|
|
| 2^ vs 10^ |
|
|
By mmu_man - Posted on November 5, 2004 - 17:32:18 (#14655)
Current version when comment was posted: 1.3 |
|
|
Hertz were never a binary unit, thus never meant to be used with powers of 2, there is no reason ever why a KHz would be 1024 Hz.
The only units that use that are bits and bytes.
Some mindless chaps even proposed to use new prefixes instead (use MB for 10^6 bytes and MiB for 2^10 bytes).
|
|
| power of 2:) |
|
|
By s_d - Posted on November 5, 2004 - 12:42:17 (#14649)
Current version when comment was posted: 1.3 |
|
|
Heh.
I'm staring suspect that CPU producers and software writers now are in same marketing/development dilemma, how to measure something - by power of 10 or power of 2.
so 1 KHz isn't anymore 1000 Hz, but 1024!
|
|
| Some syslog details |
|
|
By tb100 - Posted on November 5, 2004 - 09:49:11 (#14645)
Current version when comment was posted: 1.3 |
|
|
Here's some possibly interesting bits of my syslog - there doesn't seem to be an int32 overflow problem, but Be's calculation still seems a bit wrong:
KERN 'BOOT': IBM PC Compatible platform
KERN 'BOOT': CPU speed is 1993MHZ
KERN 'BOOT': PCI: initiating new PCI scan
[...]
KERN 'sysinit2'[3]: cpu_fix: patching get_cpu_info...
KERN 'sysinit2'[3]: cpu_fix: fixing cv_factor...
KERN 'sysinit2'[3]: cpu_fix: CPU at 2.002 GHz
I wonder if all CPUs are reportedly slightly inaccurately by Be's function?
|
|
| Awesome! |
|
|
By tb100 - Posted on November 5, 2004 - 09:25:16 (#14642)
Current version when comment was posted: 1.3 |
|
|
My CPU is only a Athlon XP 2400+ (runs at 2GHz) - so it is reported right in pulse and the clock seems reasonably accurate.
However, a lot of media players seemed to struggle with sync - Be's MediaPlayer was terrible (noticably out after about 20 secs), vlc did the whole regular pitch-shifting thing, mplayer worked but isn't nice to use and ate up way too much CPU.
With this driver, magically vlc can now play in perfect sync, without any pitch shifting. Think I've borked my MediaPlayer install after playing with some Haiku media kit stuff, so haven't been able to test that.
Also, my feeling that the Haiku emuxki driver sounded a bit off has completely disappeared and music is as crisp as it should be.
In short -woohoo!!! Fantastic work, I recommend every one installs this driver, even if you thought your CPU was too slow to be affected!
/Me wonders if it's an AMD-related problem?
Very happy :D
Simon
|
|
| good job |
|
|
By rayman - Posted on November 4, 2004 - 14:28:36 (#14636)
Current version when comment was posted: 1.3 |
|
|
It work on my Dan0 now. ( P4 2,8Ghz HT )
thanks.
Just frequency is correct in the "about panel", but not the processor type. I know that it's not the principal fonction of this patch.
Another thanks.
|
|
| thanks |
|
|
By thaflo1 - Posted on November 4, 2004 - 13:38:47 (#14633)
Current version when comment was posted: 1.3 |
|
|
it works fine an my system too.
2,66 Pentium (Laptop)
shows correct MH and video is sync. correctly.
Zeta neo.
Thanks,
Florian
|
|
| h4ck3rz 0wNz0R |
|
|
By mmu_man - Posted on November 3, 2004 - 17:36:54 (#14612)
Current version when comment was posted: 1.3 |
|
|
|
| 1.3 runs fine under NEO |
|
|
By ljr - Posted on November 3, 2004 - 13:42:19 (#14606)
Current version when comment was posted: 1.3 |
|
|
|
| Found the problem on Dano |
|
|
By mmlr - Posted on November 3, 2004 - 13:01:34 (#14605)
Current version when comment was posted: 1.3 |
|
|
Ok, I found it.
As I took another close look I checked the libroot and found something interesting: The Dano libroot does not have the cv_factor directly, but rather a cv_factor_ptr. With that information I could check for the pointed value and found out that it was the wrong (unpatched) cv_factor. With a closer look at the Dano kernel I found out that userspace variables as system time stuff is exported in a special struct and accessed from the libroot. So the app_server (and all the other servers most applications too for that matter) always used the original cv_factor and was terribly out of sync with the kernel on all timings.
I've now built a special Dano-version of the cpu_fix, that "reexports" the cv_factor to the kernel_export_data struct after fixing it.
Works fine for me, writing this in FireFox on Zeta RC3, with mouse working, videos playing normal and listening to music that stops at the real end of the file.
Upload will be ready within a few minutes. Hope it works for everyone now!
Michael Lotz
|
|
| pulse... |
|
|
By mmu_man - Posted on November 3, 2004 - 10:50:57 (#14603)
Current version when comment was posted: 1.3 |
|
|
it's registrar that generates PULSE messages...
|
|
| Not for Dano |
|
|
By mmlr - Posted on November 3, 2004 - 10:27:52 (#14602)
Current version when comment was posted: 1.2 |
|
|
I said it may work on Zeta, but it seems not to.
The mouse problem is a problem with the app_server of Dano. As Zeta and probably PhOS do use the Dano app_server, the problem will happen there too.
Normally this is caused by patching to late, i.e. when patching after the app_server is started, because then the pulse system and the system_time get out of sync. You can see that the clock is ok, but it sometimes skips a digit. This is because the redraw of the clock uses a different time source.
The mouse is actually working (try clicking around a bit and you will see that the pointer is where it should be) it's just that the cursor is not redrawn as often as it should, so you don't see the movement.
I don't know why this happens, it might be that the app_server has it's own means of determining a cv_factor or it may use another kernel variable / function to do its timings.
I am investigating that problem, but it probably won't be solved that soon.
An updated package with the interrupt patch (thanks Nathan for pointing this out) will be uploaded shortly.
|
|
| works with NathanW's changes |
|
|
By Sir Mik - Posted on November 3, 2004 - 07:45:58 (#14601)
Current version when comment was posted: 1.2 |
|
|
installed a version with the changes suggested by NathanW. Seems to work ok now! great work!
regards,
Tim
|
|
| Modified |
|
|
By ljr - Posted on November 3, 2004 - 03:38:55 (#14600)
Current version when comment was posted: 1.2 |
|
|
Got same problem as NathanW with modified version (mouse blocked every seconds) but i could boot with all options activated. Otherwise standard version makes my system go to kdl if cv_factor is enabled.
It's a pity this mouse problem :-(
|
|
| Re: interrupts |
|
|
By shatty - Posted on November 2, 2004 - 14:31:48 (#14594)
Current version when comment was posted: 1.2 |
|
|
Using a version of cpu_fix patched as Nathan recommended,
my machine works much better now. No freeze at boot and
the clock sync is much improved.
|
|
| Interrupts |
|
|
By NathanW - Posted on October 29, 2004 - 12:34:51 (#14567)
Current version when comment was posted: 1.2 |
|
|
To function properly, interrupts need to be disabled in the calculate_cv_factor() routine. Adding a disable_interrupts() call at the top, and a restore_interrupts() call at the bottom made the cpu speed function properly. As a side note, I'm still having some trouble with my mouse -- it begins to move jerkily after an hour or so.
|
|
| syslog parts |
|
|
By Sir Mik - Posted on October 29, 2004 - 04:12:34 (#14564)
Current version when comment was posted: 1.2 |
|
|
Hi,
I looked at my syslog and here are some parts of it:
KERN 'BOOT':
Be Intel Pentium Kernel version of May 11 2001 time 15:40:59
KERN 'BOOT': (C) Copyright 1991-97 Be, Inc. All Rights Reserved.
KERN 'BOOT': Total Memory 512 MB
KERN 'BOOT': CPU features: 240
KERN 'BOOT': IBM PC Compatible platform
KERN 'BOOT': CPU speed is 2418MHZ <- this seems allright!
and the most interesting part:
KERN 'sysinit2'[3]: cpu_fix: patching get_cpu_info...
KERN 'sysinit2'[3]: cpu_fix: fixing cv_factor...
KERN 'sysinit2'[3]: cpu_fix: CPU at 9.525 MHz <- this doesn't :S
KERN 'sysinit2'[3]: can't add boot driver cpu_fix to /dev
regards,
Tim
|
|
| Problems with cv_factor |
|
|
By Sir Mik - Posted on October 29, 2004 - 03:41:17 (#14563)
Current version when comment was posted: 1.2 |
|
|
Hi,
also tried the 1.2 version of this fix. BeOS boots but again ends in KDL, complaining about some problem in the OHCI driver? I then tried to disable cv_factor like suggested: this boots with no problems and the cpu speed is correctly outputted by Pulse. If I disable _get_cpu_info and leave the cv_factor enabled the system again ends up in KDL, shortly after the desktop has been displayed.
Here are my system specs in case you need them(the things I think are important):
Pentium 4 2.4ghz
Asus P4B533 motherboard (Intel 845E)
512 mb ram
Ati Radeon 9700pro
regards,
Tim
|
|
| ati driver problem |
|
|
By mmlr - Posted on October 26, 2004 - 18:25:09 (#14541)
Current version when comment was posted: 1.2 |
|
|
I can't imagine that this driver should have an impact on the ati graphics driver. I am running an ATI Radeon 9700 with the driver over here and it works fine.
If you have installed version 1.0 please completely remove it by deleting the cpu_fix file from "/boot/home/config/add-ons/kernel/drivers/bin" and it's link in "../dev/misc".
Also edit the system bootscript at "/boot/beos/system/boot/Bootscript" and remove the lines:
#
# patch kernel for timing issues
#
if [ "$SAFEMODE" != "yes" ]
then
ls /dev/misc
fi
Try installing version 1.2 afterwards. It won't change the bootscript anymore and there should not be any reason why it should give problems with the radeon driver anymore.
Please tell me where it did hang. If it hang after the last boot icon, the app_server isn't started and the problem was probably the bootscript-change. If it hanges after the screen is cleared it may just be servers failing to load because of too small sleep times in the original bootscript.
|
|
| patch breaks ati driver |
|
|
By ljr - Posted on October 26, 2004 - 17:52:02 (#14538)
Current version when comment was posted: 1.2 |
|
|
As i reinstalled an ATI card on my machine in place of an nv43, i got Be hanging at boot time. After half an hour to find the problem, i disable this driver and i could have my *fast* desktop again ... Seems not all is fine ?!
|
|
| cv_factor |
|
|
By mmlr - Posted on October 26, 2004 - 16:31:42 (#14537)
Current version when comment was posted: 1.0 |
|
|
Please note that the cv_factor is the one that corrects timing issues. So if you have clocks/videos/audio running too fast but do not use the cv_factor fix, it won't make a big difference.
Of course if you only want the CPU reporting corrected you can disable cv_factor fixing. I'll add a configuration to the installer that will set the correct settings.
|
|
| ok with disable_cv_factor |
|
|
By ljr - Posted on October 26, 2004 - 14:25:59 (#14534)
Current version when comment was posted: 1.0 |
|
|
ReTest with disable_cv_factor in seetings file and i can boot know with an AMD3500+ showed at 2175mhz (should be 2200 but it might be a clock set by asus board...)
|
|
| Great work! |
|
|
By shintaro - Posted on October 26, 2004 - 11:39:16 (#14530)
Current version when comment was posted: 1.0 |
|
|
Initially, I had issues with my mouse when installing this app.
After uncommenting "disable_cd_factor" mouse now works fine, and CPU speed is correctly shown. (P4 3GHz)
All I need now is for someone do get Intels Hyperthreading to work under BeOS/Zeta :)
|
|
|
|
|
|
Comment Pages:
<< prev | 1 | 2 | next >> |
|
|
