| Author | Post |
|---|
billycar
Member
|
Posted: Fri Nov 18th, 2005 05:28 am |
|
I'm trying to install MACH Speed Control on my PowerPC 9500, under XPostFacto 4 / OS X 10.3.9. I'm hoping to insure the 512KB onboard L2 cache is disabled.
It keeps crashing at startup. If I remove the installed kext and startup files, it doesn't ...
I removed all CPU Director 2.1 files, and tried restarting with option-apple-x-r hoping to disable the onboard cache (in my 6 slot machine), so the Sonnet G3/500 I have installed will run with its 1MB L2 cache. I repaired permissions, re-installed the MACH Speed Control files, to no avail.
In addition to the G3/500 cpu upgrade, I have a Sonnet FW/USB, ixmicro ultimateRez video, and M-Audio 2496 Sound PCI cards installed, and 832MB RAM.
Any suggestions?
The 9500 setup appeared to be working fine with CPU Director 2.1, but I'm not sure the 512kb motherboard cache had been disabled, giving free reign to the upgrade card's L2 cache.
(Additionally, it seems to be working in my PowerPC 8500 with a Sonnet G4/450, except the System Profiler no longer reports the L2 cache, and neither does XBench ... although MACH Speed Control does ... ???)
|
billycar
Member
|
Posted: Sat Nov 19th, 2005 03:37 pm |
|
Do I need to correct Ownership and/or Permission characteristics?
My installation of Panther on my PowerMac 9500 went ok, except after the Disk 1 was finished, the Reboot didn't go (probably because of the ixmicro ultimateRez video card). I rebooted through XPF from the OS 9.1 partition, and the installation resumed, and continued with Disk 2.
Then, because the OS X Installer didn't like the video card, it wouldn't allow the Setup Assistant to continue, and said I could set my Preferences using the Menu items.
So ... now I have an Administrator with no password, and I set up an individual account through System Preferences.
Are my Permission and Ownerships confusing to my Account? I've tried to Install MACH Speed Control and it crashes at Startup ... I think it might be my Ownership/Permissions characteristics. I've tried Repairing Permissions, and re-installing to no avail.
Do I need to use Get Info some way, to clear up Ownership? and what do I need to do?
_________________
MUSIC PPC9500 10.3.9
SonnetG3/500
FW400/USB1.1
M-AUDIOAudiophile2496
IntSeagateLVD50GB & ExtSegateBarracuda18GB SCSIHDs
ixmicro ultimateRez
ART PPC 8500 10.3.9
SonnetG4/450
FW400/USB2/ATA133-2 SgateATA100s
ATI Rad7KMacEd
Astra2400S
|
billycar
Member
|
Posted: Sun Nov 20th, 2005 05:45 pm |
|
I reinstalled Panther, up to 10.3.9, by borrowing my ATI Radeon 7000 Mac Edition video card for the installation process.
Reinstalled Mach Speed Control after Repairing Permissions and still cannot get past the startup. Don't get to the startup window ... just a few flashes of monitor adjustment activity, a blue screen, then the window-shade with multilingual interpretations of "You need to restart your computer".
What gives Daystar? I know its the weekend before Thanksgiving and the Christmas bells are already ringing ...
Is ther something wrong with the Ownership/Permission of your files? My accounts use of them needs to be Read and Write? not Read only?Last edited on Mon Nov 21st, 2005 12:28 am by billycar
|
Gary
Administrator
|
Posted: Mon Nov 21st, 2005 05:15 pm |
|
billycar wrote:
I'm trying to install MACH Speed Control on my PowerPC 9500, under XPostFacto 4 / OS X 10.3.9. I'm hoping to insure the 512KB onboard L2 cache is disabled.
It keeps crashing at startup. If I remove the installed kext and startup files, it doesn't ...
(Additionally, it seems to be working in my PowerPC 8500 with a Sonnet G4/450, except the System Profiler no longer reports the L2 cache, and neither does XBench ... although MACH Speed Control does ... ???)
Okay, it sounds like there are a couple things going on here. One is that you may not be running the latest MAChSpeed Control.. I will send a link for 3.4.2.
Also the M-Audio is problematic. Try removing their kext and card just to ensure that it is not fighting for control
Then, on boot, the command-option-x-r keys do not work in OSX. The reset procedure is to boot in safe boot (holding the shift key down), then open the MAChSpeed Control App. This will bring up a reset dialog. Then reset and reboot.
Alternatively, you can trash the XLR8 Preferences in Library Preferences, and reboot, but then you will need to re-authenticate with the serial number.
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
Gary
Administrator
|
Posted: Mon Nov 21st, 2005 05:18 pm |
|
billycar wrote:
I reinstalled Panther, up to 10.3.9, by borrowing my ATI Radeon 7000 Mac Edition video card for the installation process.
Reinstalled Mach Speed Control after Repairing Permissions and still cannot get past the startup. Don't get to the startup window ... just a few flashes of monitor adjustment activity, a blue screen, then the window-shade with multilingual interpretations of "You need to restart your computer".
?
The "restart you computer" message is a GUI for a "kernal panic". You can open console and review the logs for possible culprits, but most often, the logs are not helpful.
This is not a permissions issue, so repairing premissions doesn't help.
You can check to see if some piece of hardware is interferring. You can also trash the XLR8 Preferences and reboot (in the event that they are corrupt).
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
billycar
Member
|
Posted: Mon Nov 21st, 2005 08:00 pm |
|
I have stripped down my 9500 to nothing more than the ixmicro ultimateRez video card in slot 4 (D2). CUDA button pressed ... and I'm using 3.4.2.
That means I removed the M-Audio 2496 (F2) and Sonnet FW/USB (A1) cards I had installed. I also Un-Installed the M-Audio driver (which by the way had to be v 1.2.1, as 2.0.3 or 2.0.4 don't work ... and then, I had to use the Uninstaller from 2.0.3 ...). I also Uninstalled the Logitec Control Center that might be competing as well ... it ran my wireless KB and Mouse.
I still get the KernalPanic at startup. I've Repaired Permissions, installed the MSC driver and friends, to no avail. I''ve tried Reset after Safe Boot, and trashing the Prefs from OS 9.1.
Suggestions?Last edited on Mon Nov 21st, 2005 08:01 pm by billycar
|
billycar
Member
|
Posted: Mon Nov 21st, 2005 08:37 pm |
|
I received and downloaded the combo, and have installed again the MCS onto OS X 10.3.9, but get the same kernal panic. (probably the same 3.4.2 I tried before).
I have it running under OS 9 now (v 2.8.2) ... but it is interesting to note, it will not allow me to disable the on-board cache. It says it is there, but the Disable (Default) button keeps jumping back to the Enabled.
Is this the bug? CPU Director 2.1 seems to handle things along with XPF ... although it doesnt mention the onboard cache ... my Sonnet G3/500/1MB seems to run ok ... software removal of onboard L2 cache is the main reason I am investigating your software.
I have to sign off for awhile (dinner and a music lesson). Will check back later this evening.
Thanks for your reponses.Last edited on Mon Nov 21st, 2005 08:47 pm by billycar
|
billycar
Member
|
Posted: Tue Nov 22nd, 2005 03:09 am |
|
| I''ll hold off replacing my M-Audio 2496, Sonnet FW/USB and Logitech hard/softwares until I hear from you ...
|
Gary
Administrator
|
Posted: Tue Nov 22nd, 2005 07:29 pm |
|
billycar wrote:
I have it running under OS 9 now (v 2.8.2) ... but it is interesting to note, it will not allow me to disable the on-board cache. It says it is there, but the Disable (Default) button keeps jumping back to the Enabled.
Well, then I guess it is all moot then. There is a problem. MAChSpeed control is unique from all of the other cache controllers in that it actually checks to see if the cache is actually disabled in realtime. If enabled, then it resets the button. That is what you are seeing. For some reason, the software is not able to disable the cache the on your motherboard. My guess, (since we all use the same Apple code) is that no one can actually disable it on your system via software.
If you really want it disabled, and know someone with good solder ability, I can send you the info for removing the controlling resistor. Just send email direct to gary@daystar-tech.com.
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
billycar
Member
|
Posted: Wed Nov 23rd, 2005 07:03 pm |
|
The reason I invested in MACH Speed Control was because it claimed to be able to Disable onboard cache.
I have since physically removed R31, and MSC now will boot up on my PowerPC 9500 under XPF 4 and OS 10.3.9.
The System Profiler and XBench neither report L2 cache of my Sonnet G3/500/1MB on my 9500, and same for my G4/450/1MB on my 8500.
Since MSC reports these, can I assume they are being used by my OS 10.3.9 and apps?
|
billycar
Member
|
Posted: Thu Nov 24th, 2005 09:15 am |
|
Additional comments:
I finally removed the R31 resistor, turning off access to the onboard L2 cache.
Mach Speed Control 3.4.2 now runs on the 9500, and seems to do a better job of "cache tuning" over CPU Director 2.1, if you can trust XBench results.
Even though my 9500 ran with the cache onboard, memory usage appears to be faster when it is removed, and the L2 cache of the upgrade card is utilized. This was illustrated by CPU Director 2.1 / XBench 1.2 results comparisons before and after R31 was removed, and then comparing MSC 3.4.2 / XBench 1.2 to CPUD 2.1 / XBench 1.2 after R31 was removed.
Still not thrilled with System Profiler of 10.3.9 not reporting the L2 cache of my Sonnet upgrade card when using MSC. It did so when using CPUD.
If anyone's interested in the test results comparison data, let me know. I have a pdf file I can send.
|
Gary
Administrator
|
Posted: Thu Nov 24th, 2005 02:13 pm |
|
Thanks for he post. You should see slightly faster results with MSC installed on Xbench. MSC has slightly faster memory access settings.
We can't really do much about correct reporting on the Sonnet card. ASP actually gets its info from the IOReg. We patch it for our cards. You can always verify wth MSC which always shows the truth.
Be sure not to run any benchmarks with the MSC Utility open. The animation is actually a representation of the active CPU/RAM/Cache test being performed. it does tax the CPU when open.
I should note that your Apps/System do not care what ASP shows. if MSC shows that you have a backside cache, then everything on your system actively working will use the backside cache.
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
Gary
Administrator
|
Posted: Thu Nov 24th, 2005 02:13 pm |
|
Thanks for he post. You should see slightly faster results with MSC installed on Xbench. MSC has slightly faster memory access settings.
We can't really do much about correct reporting on the Sonnet card. ASP actually gets its info from the IOReg. We patch it for our cards. You can always verify wth MSC which always shows the truth.
Be sure not to run any benchmarks with the MSC Utility open. The animation is actually a representation of the active CPU/RAM/Cache test being performed. it does tax the CPU when open.
I should note that your Apps/System do not care what ASP shows. if MSC shows that you have a backside cache, then everything on your system actively working will use the backside cache.
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
Gary
Administrator
|
Posted: Thu Nov 24th, 2005 02:17 pm |
|
Thanks for the post. You should see slightly faster results with MSC installed on Xbench. MSC has slightly faster memory access settings. Remember that Xbench (like any other benchmark utility) is accurate only in a relative mode, comparing it on the same system with controlled variables.
We can't really do much about correct reporting on the Sonnet card. ASP actually gets its info from a "static" database which is set up very early. Aple does this so that users on PowerBooks don't notice when their systems are running on half speed during boot and "low performance mode".
We patch ASP for our XLR8 cards. You can always verify wth MSC which always shows the truth.
Be sure not to run any benchmarks with the MSC Utility open. The animation is actually a representation of the active CPU/RAM/Cache test being performed. It does tax the CPU when open.
I should note that your Apps/System do not care what ASP shows. if MSC shows that you have a backside cache, then everything on your system actively working will use the backside cache.
____________________
Gary Dailey
Daystar Technology
http://Daystar-Tech.com, http://Daystar-Store.com
email: gary@daystartechnology.com, chat: garydailey@mac.com
|
billycar
Member
|
Posted: Thu Nov 24th, 2005 10:26 pm |
|
I see where you allow pdf files to be posted.
Here's the results of my tests. I believe all variables are constant, except I switch between having R31 and not having it, showing CPU Director 2.1 cpu controlled results. Then I compare between CPU Director 2.1 and MACH Speed Control 3.4.2, having no R31 connection to the onboard cache.
I try to have nothing I control running in the background, including no mouse activity. (Since the machine would not startup with MSC having R31 and onboard cache, I of course report no results)
Just FYI.Attachment: R31Comparison.pdf (Downloaded 61 times)
|
Current time is 04:05 pm | |
|
