Phenom 9500 w/ MSI K9A2 Platinum

[aGeoM]

Hi KTE

The reason of high voltages, have to do with the reboot problems reported, this board boots fine with less voltages but if after program install it ask for reboot, the board refuse to boot and after a reset shows overclocking failure.
After playing around with all settings the ones I used was the best ones for that problem, anyway that was the last run on that board, I send it back to shop bcz I had order the Wifi/Ap version and they had send the normal Deluxe, I had to wait till now bcz shop closed for holidays, got my cash back and I will order another board, ATM and after the ASUS experience (OC/BIOS/LAYOUT) don't know witch one to order, but must be with eeprom removable since I like to mess with bios, and with the MSI I can't (I have an Serial cable 2 SPI chip but no program I know recognize it). That was the reason I bough ASUS, to be able to recover from a bad flash/bios.

About ASUS layout you can't use 4 dual slot graphic cards, like the 2 I got (with MSI it's possible) SATA ports are well placed and horizontal connection is cool, if MSI had those it will be "almost" perfect.

On the MSI side I can't use the Power/Reset buttons bcz 2nd card are on top.

ATM I'm using the MSI still have 1.31 bios, just plugged CPU and all the staff, 2800Mhz without enter in CMOS setup, need to check the settings I left couple weeks ago

BTW you deserve that free parts you are receiving, you are doing an real good job, and I'm sure like me others are buying RD790 systems relying on your expertise, including tictac, polygon and others.

I'm grateful to all of you guys, keep it up

[jonspd]

KTE you had stated that the random freezes where NB related? I think it's more cpu related is the reason I ask.

[justapost]

Why though?

Did not work. Blanker lottery style said.

http://www.xtremesystems.org/forums/...73#post2767073
http://www.xtremesystems.org/forums/...54#post2766954

[Oldguy932]

Yeah, it is nice that I finally got closer to my air clocks, but I just can't figure out what's going on now to keep me from getting any higher stable. I have a feeling it might be something around the cpu that doesn't like the cold rather than the cpu, I felt to the side of the mount towards the ram and the board was cold, but I dunno if that means anything. My last tricks to try are going to be lowering the nb and ht-link, but I don't think it'll work, although increasing the ht voltage did help stability a slight bit at 259, got me to the 3rd iteration of 8k instead of the second. I am going to put the 9500 under before I switch back to air, its just that I use the XP-90c with the vantec tornado, and 50 dba is not pleasant sounding, which is why I bought the vapo in the first place. Cold temps, future proof, and can be made relatively quiet. I wish I had more voltage readings, would make everything a heck of alot easier tweaking-wise.

[KTE]

I will tell you what the error I'm experiencing is, if anyone can help then please do so because I've tried near enough everything and I can't fix it. I have 6 drives here and all are full of OSes with none I can delete so I'm stuck. Phenom system is still no boot (OS).

I can use the system though as it's not in error apart from the drive, I'm running a LiveCD of Ubuntu with the Phneom system right now and was running UBDC4Win with it earlier.

It's 0x7b 0xC000034 STOP error at bootup exactly when the Windows XP loading screen should appear.

The error decodes to "Inaccessible Boot Drive" i.e.

The Stop 0x7B message indicates that Windows XP has lost access to the system partition or boot volume during the startup process. Installing incorrect device drivers when installing or upgrading storage adapter hardware typically causes stop 0x7B errors. Stop 0x7B errors could also indicate possible virus infection.

The second parameter is very important because it can indicate whether the 0x7B Stop message was caused by file system issues or problems with storage hardware and drivers. Values of 0xC000034 or 0xC000000E typically indicate:

• Disks or storage controllers that are failing, defective, or improperly configured.

• Storage-related

Cause: I did nothing but a reboot after using AOD and since then it is giving me this error. It is something drive related. I installed no software/drivers anew since over 2 tweaks apart from Catalyst 8.32 2 days ago so it can't be driver related either.

Steps taken to resolve:
-It is not a registry error since I have replaced the registry with a known working one and a new one and I still get the same error far before the registry is called.
-I have removed all extra drivers and it still will not boot.
-It's not the SATA port nor the PSU connector.
-The Boot.ini is perfect as before.
-Pagefile is good as before.
-Drive has zero fragmentation or checkdisk errors and works with other systems fully (SMART status is very healthy).
-It may be the MBR, but that should signal a different error. I'm reluctant to try this as I may get a no boot situation totally.
-I have tried a XP Repair, no luck.

Hi KTE

The reason of high voltages, have to do with the reboot problems reported, this board boots fine with less voltages but if after program install it ask for reboot, the board refuse to boot and after a reset shows overclocking failure.

Hmm I can boot and run some benches at 2832MHz and have only given it upto 1.445V yet and it was unstable. Maybe I should increase the volts too?

About ASUS layout you can't use 4 dual slot graphic cards, like the 2 I got (with MSI it's possible) SATA ports are well placed and horizontal connection is cool, if MSI had those it will be "almost" perfect.

Yeah for Gfx MSI board is better but the SATA ports layout, I have a different taste due to my setup.

On the MSI side I can't use the Power/Reset buttons bcz 2nd card are on top.

Which cards?

The 2nd card need to be a single slot less than 23cm to be able to use both buttons. Anything bigger than say the X1900XTX in CF will make the two buttons out of reach.

BTW you deserve that free parts you are receiving, you are doing an real good job, and I'm sure like me others are buying RD790 systems relying on your expertise, including tictac, polygon and others.

I wish...
http://www.xtremesystems.org/forums/...23#post2772823

Thank you for the words though.

KTE you had stated that the random freezes where NB related? I think it's more cpu related is the reason I ask.

Not necessarily. Only in specific cases where your NB has low volts for the speeds and you had given enough CPU voltage.

Did not work. Blanker lottery style said.

http://www.xtremesystems.org/forums/...73#post2767073
http://www.xtremesystems.org/forums/...54#post2766954

Oh, I see, thanks. Pretty bad that.

Yeah, it is nice that I finally got closer to my air clocks, but I just can't figure out what's going on now to keep me from getting any higher stable. I have a feeling it might be something around the cpu that doesn't like the cold rather than the cpu, I felt to the side of the mount towards the ram and the board was cold, but I dunno if that means anything. My last tricks to try are going to be lowering the nb and ht-link, but I don't think it'll work, although increasing the ht voltage did help stability a slight bit at 259, got me to the 3rd iteration of 8k instead of the second. I am going to put the 9500 under before I switch back to air, its just that I use the XP-90c with the vantec tornado, and 50 dba is not pleasant sounding, which is why I bought the vapo in the first place. Cold temps, future proof, and can be made relatively quiet. I wish I had more voltage readings, would make everything a heck of alot easier tweaking-wise.

Use a DMM and measure the voltages around your board, load/idle. The ASUS I tried had a large vdroop and vdrop, the MSI doesn't though. Feed it some PLL voltage too.

[KTE]

I will tell you what the error I'm experiencing is, if anyone can help then please do so because I've tried near enough everything and I can't fix it. I have 6 drives here and all are full of OSes with none I can delete so I'm stuck. Phenom system is still no boot (OS).

Fixed!

[Oldguy932]

Great job on getting it fixed KTE. And sadly, the only pll that gets a voltage is the nb pcie pll, nothing else.

[justapost]

Fixed!

Congrats, How?

[KTE]

Great job on getting it fixed KTE. And sadly, the only pll that gets a voltage is the nb pcie pll, nothing else.

Yep, that's the one I'm talking about. Because the SB is where the PLL lies and that's why usually increasing PCIe a little bit (108MHz) tends to help high end stability and oc. If you try the Autoclock feature or Novice AOD Performance Mode, you'll see it tries to clock the PCIe higher up.

Congrats, How?

Oh man, don't ask! What a pain.

So many XP repairs, software reinstalls/installs (manual), bootups, swapping processors and other hardware, hours in recovery console, system restores, manual registry editing (around 20 hours stuck in there), copying/pasting/deleting over, checking, testing, manual driver deleting, driver/hard disk MBR verifying, installing/reinstalling SP2/SP3 manually (registry editing, copying files, making batch and automated files), everything done and then, I find out it's nothing that can be resolved. No Safe Mode was allowed, remember.
So the only thing I had forgotten was testing the first bootup registry and driver loading. Turned out my user profile had become corrupt by a random and rare Windows bug there since XP was first released which then disables the user Admin access to load the registry hives and boot-up, specifically the user profile. It creates random named empty profiles and starts using them in place of the original (original still intact although renamed to random chars). So I had no choice but to repair XP and then mod everything (registry/drivers/files/file paths - must have been past 1200 files alone easily) from another 2 systems to cross-check and restore my documents, software and user profiles back. Had hordes of uncountable problems, can still not install Catalyst 8.2 but have deleted everything excessive now and it's stripped down to the bare minimum goods which are fully working and easier to debug. It is now fully working even better than before with newly retrofitted registries and C:\ directory files.

What a nightmare. That's the worst issue I've ever faced on XP, toughest one ever to debug and fix. Very relieved, still have all software/docs there working. It really did catch me at the worst time, I was about to delete it all in a day.

[justapost]

Yep, that's the one I'm talking about. Because the SB is where the PLL lies and that's why usually increasing PCIe a little bit (108MHz) tends to help high end stability and oc. If you try the Autoclock feature or Novice AOD Performance Mode, you'll see it tries to clock the PCIe higher up.

Oh man, don't ask! What a pain.

So many XP repairs, software reinstalls/installs (manual), bootups, swapping processors and other hardware, hours in recovery console, system restores, manual registry editing (around 20 hours stuck in there), copying/pasting/deleting over, checking, testing, manual driver deleting, driver/hard disk MBR verifying, installing/reinstalling SP2/SP3 manually (registry editing, copying files, making batch and automated files), everything done and then, I find out it's nothing that can be resolved. No Safe Mode was allowed, remember.
So the only thing I had forgotten was testing the first bootup registry and driver loading. Turned out my user profile had become corrupt by a random and rare Windows bug there since XP was first released which then disables the user Admin access to load the registry hives and boot-up, specifically the user profile. It creates random named empty profiles and starts using them in place of the original (original still intact although renamed to random chars). So I had no choice but to repair XP and then mod everything (registry/drivers/files/file paths - must have been past 1200 files alone easily) from another 2 systems to cross-check and restore my documents, software and user profiles back. Had hordes of uncountable problems, can still not install Catalyst 8.2 but have deleted everything excessive now and it's stripped down to the bare minimum goods which are fully working and easier to debug. It is now fully working even better than before with newly retrofitted registries and C:\ directory files.

What a nightmare. That's the worst issue I've ever faced on XP, toughest one ever to debug and fix. Very relieved, still have all software/docs there working. It really did catch me at the worst time, I was about to delete it all in a day.

Hmm have an HP laptop here whom does the same as your pc did afer reboots during automated software installation. No safemode was possible here also.
I can fix this issue here simply by choosing use last known working profile in the boot menu.
You must have had serious causes to recover this setup. Do you have a link or error nuber at hand to that xp error from the ms support site?

[KTE]

Achim has kindly updated his 1st post to now show both the method which does work on all boards and it's derivative batch file method (easier+quicker): http://www.xtremesystems.org/forums/...d.php?t=171105

This works for all AM2+ systems that I've seen so far.

MSI MB/BIOS: Most of you know near enough all of the latest BIOSes have the patch either enabled without option to disable or you're provided an option to choose what you would like yourself. However, as we have noticed the patch is still enabled with either choices. Using the following setup (BIOS P0F):



When you choose TLB Fix = Disabled in the BIOS and you boot-up setting AOD button = Red, you still get low performance, as below:



Why you get that is because the patch is disabled only on the first core. Hence, to completely disable it, you follow the above procedure outlined in the link. MSR registers needed to edit for the MSI RD790 are these two:

For each core:

Read MSR ->
Write MSR values ->

-

And now you get->



Near +170% increase. The above routine works perfectly to disable the patch.

BTW, that's AMD Power Monitor there, the latest version I linked earlier. Works much better.

Hmm have an HP laptop here whom does the same as your pc did afer reboots during automated software installation. No safemode was possible here also.
I can fix this issue here simply by choosing use last known working profile in the boot menu.
You must have had serious causes to recover this setup. Do you have a link or error nuber at hand to that xp error from the ms support site?

Yup, first thing I tried was LKGC but it failed as well as everything else.
MS Support site doesn't have any information on this. I used to be a software dev and MS beta tester a few years ago so I have relations there and talking to a few friends who are MVP's they told me something they didn't know had happened and so only way is to reinstall because it would be too much hassle. But I digress.

Reasons were, I had everything I needed migrated and installed on it temporarily and much work that I have to migrate over to another drive was installed on here for my continuous editing (PhD/science work). Don't have the CD's for those software either so I had to rescue it all. Took me 3 days non-stop.

[aGeoM]

Hmm I can boot and run some benches at 2832MHz and have only given it upto 1.445V yet and it was unstable. Maybe I should increase the volts too?

It helps in my case, and also that random freezes, but with MSI is a different story, vCPU=1.440; vNB=1.200; vHTT=1250.

Yeah for Gfx MSI board is better but the SATA ports layout, I have a different taste due to my setup.
Which cards?

Sapphire HD3870.

I wish...
http://www.xtremesystems.org/forums/...23#post2772823

Thank you for the words though.

Lucky man, I wish to have an uncle like yours , soon some MFG will realize that giving away some parts to the right guys will be more profitable than expending thousands $ in doubtful marketing campaigns.

[Oldguy932]

My Phenom 9500, sitting on my laptop covered in stickers.



In the socket before anything is put on, the rods are for keeping the clamshell in place.



The insulation for the top of the board is on and the cpu is all lathered up.



The bottom of the clamshell, it doesn't look smooth, but it is, trust me.



Everything put together, that orange fan is for keeping my chipset on the cool side.



Sadly, this one is cold bugged. I forgot to turn on the power supply, so I turned on the vapo and it got down to -58C evap, and it wouldn't start. Thankfully it does at -21C cause thats where the startup temp happens to be. Looks like I'm going to be using the heater this time around.

[Manicdan]

KTE i know im late, but i had the same problem, it wouldnt install windows right nomater what i did, after a few full reformats i finally have XP pro working, but im scared to work on getting vista back, ill probably by a new harddrive just to get that working cause i lost everything thanks to this motherboard, and i know its the mobo cause we had the same exact problem and thats the only thing we have in common.

[jonspd]

Can you disable a single core with a pin mod or any other method that would make the system only work off 3 cores even at start up?

So disabling the tbl fix would work on any bios that has the fix built into it?

[Oldguy932]

MSI should be ble to write a bios to down core, my asus board has that option. Oh and I think my cb is in the -30C to -40C range, I'll have to check on it later.

Edit: I'm priming the 9500 at 254x11. Do you guys want an 8 hours screen or wait till I get higher?

[KTE]

aGeoM: I'll try higher volts. AOD is working quite flawlessly with P0F and P0H, so far.

Sapphire HD3870, yeah two in CF would block access to the buttons unfortunately.

Oldguy932: Thanks for the pics, pity about the CB. Don't you think that base needs some lapping and cleaning?
Keep going as high as you can on that stability testing.

Manicdan: Fingers crossed, I don't want it to happen now, again. About to reboot in a few...

jonspd: Doubt it yet, although they can add this BIOS option. Only way I've tried to disable a core is by using the Boot.ini (easily can disable any amounts of core before bootup).
Yep, disabling fix will work on all the MSI BIOSes, because the fix is the same one.

I discovered a bug earlier, it's an application bug affecting all monitoring applications but AOD when AOD is used to oc - it's temporary for CPU-Z, fixes itself after a while but most of the time it won't load the driver hence it reads wrong values. Check this out and the values it gives:




Bug remains in all other software but in CPU-Z/AOD after 15 minutes:




I believe this happens due the reason I've mentioned earlier-> opening more than on monitoring tool at the same time. I knew that was still 2550 because I had ran wPrime 32M and compared with 2530 and 2486 pics.

[jonspd]

msconfig/boot.ini/advanced options numproc?

I wish we could get a bios moder around here I think we ran of tic tac toe

[Oldguy932]

Its giving me a steady -7C load from small fft, evap is at -39C average. I think I'm going to bump it up to 256 now. vCore is at 1.3, vCore-NB is 1.3, vNB is 1.4, ram at 2.1, and everything else is auto. If you really think lapping the head of it would do me any good I could do it, I never have bothered before though.

[jonspd]

I lap anything copper that touches my cores

Where you have vCore-NB is 1.3, vNB is 1.4 i don't understand which is which vcore-nb = nb and vcore? vnb = nb vid?

[Oldguy932]

vCore-NB is CPU-NB voltage, and vNB is just northbridge voltage. Oh and it looks like my 9500 maxed out at 2.8ghz on 1.3v. I wonder how much better it can do with more voltage.

[KTE]

msconfig/boot.ini/advanced options numproc?

Yep or I just edit the file myself.

I wish we could get a bios moder around here I think we ran of tic tac toe

They're here but whether they can do something for this board is a different case.

Oldguy932: you don't need vNB, it has nothing to do with clocking the CPU/IMC/HT/RAM. You'll be burning it up for no reason, stock will do. RD790 isn't like MCH with Intel CPU's to need any voltage.

In my case, I've never needed vHT above +0.05V either, all the way to 272HT and 2596MHz.

[jonspd]

well can you quote code me a copy of what your file would look like to disable core #3 on that line or the whole thing either way.

1
2
3-that one
4


Yea well just maybe if the b3's do something we may get a few better bios options floating around

[KTE]

well can you quote code me a copy of what your file would look like to disable core #3 on that line or the whole thing either way.

1
2
3-that one
4

Can't do that I'm afraid, can only specify core count and the OS decides which core to disable all by it's own (sequentially) as far as I'm aware. Thus 3 cores enabled would be 4th one disabled.

MSI may add this to the BIOS yet though. The thing that'll reserve and delay it all is the rev. 2.0 of the board, once it's out, they'll focus on to that and usually quit producing much effort for the latter iteration. I hope that board gets delayed longer yet or they give us some free RMA's for it.

[Daveburt714]

aGeoM: Are you sure you need that high a VCore? Trust me those temps on air will be very high. Also, why is it you have the vNB and vHTT that high? Try +0.05V on HT and stock on NB, trust me, you will most likely still have it stable.

KTE: I hate to disagree with you Bro, but on my chip there have been a number of times where upping the vHTT to 1.20 (+0.05) wasn't enough, but bringing it up to 1.25v (+0.10) smoothed everything out...

253x10 @ HTT of 1.20v = Unstable
253x10 @ HTT of 1.25v = Stable

1.3v didn't seem to offer anything extra, but I'm thinking that is getting close to the limit of my cores (2.53 Ghz)...