Phenom 9500 w/ MSI K9A2 Platinum

You know something, my BSoD's relating to the GPU I only figured out after lengthy troubleshooting, the stacktrace and analysis logs didn't show anything but the CPU or RAM at error. Typically I had to uninstall and reinstall GPU driver many times to figure this out.
I was stable 1.28V 2.7GHz at the beginning as I've shown, which was then moved up to 1.344V for 2.754GHz. But now that is impossible without those dreaded freezes. The max I've had without freezes was 2613MHz. It's now at 1.28V idle at 2613MHz stable for many hours even benchmarking, priming and gaming without the random freezes. Still far off from 1.232V 2.6GHz stable as beginning.

I don't want to spread unsubstantiated FUD but I will tell you the honest facts to keep in mind. The TLB errata doesn't have to crash your system alone.. it can make you lose hard drive data, cause drive fragmentations, loss of files and corrupt files especially those held in system memory i.e. VGA SYS and DLL files included. :yepp:

But you really shouldn't rely on that PSU. It's not even a PSU, honestly advising you, it's a total rip off for what it's doing to your components.

Whats the stepping on that 60Ghz Phenom KTE!! :rofl:

That is wierd man! You think it's a software problem? Seems like if it was an actual problem with the clocks it would have crashed long before that....

It's the CPU MSR's and/or PLL feedbacks and any software which depends on them. Actual clocks do not change with these bugs, core limitations are well known by now. :D

Once I opened up Crystal CPUID to check my MSR values and a lot of them were totally blank (0's) where they shouldn't be at all. :confused:
It's the same with BarEdit/WPCREDIT access of NB registers, they totally whack out many times. The last time I checked that also was reading 0b for everything in the NB FID D4 register, whereas it should not be and was still running 9x multi as it showed. At those instances L3 performance and access becomes very bad and slower than stock.

If I could capture it, I'd show you some much higher MHz to show you the scale of the problem. Dilemma is, when the [fake] high MHz show up in all software the system freezes for 2-3 seconds at a time and so the PrntScrn is never saved until after it restores from freezing. The way to test the bug is to test the system on multiple boards really.

BTW what config are you now running on the GBT? Any more oc on HT, NB or CPU?

Quote:

---

Originally Posted by KTE

BTW what config are you now running on the GBT? Any more oc on HT, NB or CPU?

---

The best I've gotten so far with any stability (not really stressed) was 2.76Ghz (240x11.5x8x8) using AOD, those setting would not boot windows when set in Bios though... :( Looks like 2.7 may wind up being doable, while my best stable on the MSI board was 2.5...

So far it looks much better than K9A2, I can boot at 13x multi now.
To be fair to the MSI board, I never really got a chance to play with it much using XP, most of my testing was done using Vista 64....

A short story of how I wound up with the DS5:
I have a 9600std and a 9600be, when my K9A2 died I thought I'd be shady :shakes: and go buy another one switch them out, return it, then I'd have 2 boards right?
Well, the board I bought actually WAS bad!! When I went back to the store they were out of K9A2's and all they had were DS5/DQ6 and the Asus boards, so it wound up costing me an extra $45!!

Poetic Justice I guess.. :yepp:
I'm just not good at being bad..

Your chip looked 2.7G doable pretty easily to me as I stated once before. You looked limited by Vista on the MSI. ;)

The DS5 doesn't clock better than the MSI IME, it doesn't have the BIOS options to, not if you use all we have made clear regarding VID/FID/DID or the 1.13 BETA. The DQ6 will. You can't oc the NB/RAM with the DS5 much. Try plus 600 on RAM and plus 2.4G on NB for instance.

So you're telling me that you tried to do some fraud? :p:

Quote:

---

Originally Posted by Daveburt714

I have a 9600std and a 9600be, when my K9A2 died I thought I'd be shady :shakes: and go buy another one switch them out, return it, then I'd have 2 boards right?
Well, the board I bought actually WAS bad!! When I went back to the store they were out of K9A2's and all they had were DS5/DQ6 and the Asus boards, so it wound up costing me an extra $45!!

---

What goes around comes around :yepp:

I got my new ASUS board and couldnt get video so I installed a 5000 and she fired right up so my Phenom is dead as well RMA-ing it tommorow :(

Quote:

---

Originally Posted by KTE

So you're telling me that you tried to do some fraud? :p:

---

Fraud is a pretty harsh word, I NEVER do that kind of stuff. But I had bought both chips from there plus the board and a cheapo CoolerMaster PSU which I thought had caused the problem (apparently not). Plus the fact that the board was only a couple weeks past thier normal return policy and I bought another board, a decent PSU and another set of RAM...
Oh well, just trying to justify it in my own mind, guess I got what I deserved. ;)

As far as the DS5 goes, I think the BIOS may have gotten alot better since you tested it. I'm running the F4f version and they seem very solid. Here are the pros & Cons so far:

PROS: TLB disable works across all 4 cores (fully disables)
Temps report much more accurately than the MSI
HTT/HTR voltage settings (although it doesn't seem to offer much)
SB Voltage control
Voltages seem alot more stable (Less fluctuation)

CONS: No voltage option for the IMC (NB Vid)
Looks like max HTT Ref is around 235 (250 was stable on the MSI)
Voltage options are a little more limited (idiot proof), which may be a good thing in my case... :ROTF:
If you want to see some BIOS pics let me know.

Quote:

---

Originally Posted by jesteronyer6

I got my new ASUS board and couldnt get video so I installed a 5000 and she fired right up so my Phenom is dead as well RMA-ing it tommorow :(

---

Sorry to hear about your Phenom Jest... My CPU is fine, guess it was just some weird coincidence...

oh yeah, my phenom's batch number is:
CAAWB AA 0747MPMW

Decided to reseat my HSF since when I had originally installed it, the thumb push down latch was right up against the sercupipe on my board, so I switched the mounting latch around to the opposite direction.

Quote:

---

Originally Posted by KTE

Guys, the BIOS file P0H aGeoM attached is fine, checksum is also perfect. I've flashed it without even setting defaults although I did follow the safe known procedure and did use the flash parameters.

---

what parameters did you used during update of bios?what does this do?:confused:

Quote:

---

Originally Posted by Daveburt714

Fraud is a pretty harsh word, I NEVER do that kind of stuff. But I had bought both chips from there plus the board and a cheapo CoolerMaster PSU which I thought had caused the problem (apparently not). Plus the fact that the board was only a couple weeks past thier normal return policy and I bought another board, a decent PSU and another set of RAM...
Oh well, just trying to justify it in my own mind, guess I got what I deserved. ;)

---

:stick: :fact: :D

Quote:

---

As far as the DS5 goes, I think the BIOS may have gotten alot better since you tested it. I'm running the F4f version and they seem very solid. Here are the pros & Cons so far:

PROS: TLB disable works across all 4 cores (fully disables)
Temps report much more accurately than the MSI
HTT/HTR voltage settings (although it doesn't seem to offer much)
SB Voltage control
Voltages seem alot more stable (Less fluctuation)

---

The first thing is awesome! Nope that wasn't in the earlier BIOS I tried. The rest of the volts aren't important as we know, only NB VID/Volts is the main importance after CPU. What are your idle/load temps now, any test ss?

Quote:

---

CONS: No voltage option for the IMC (NB Vid)
Voltage options are a little more limited (idiot proof), which may be a good thing in my case... :ROTF:

---

This is what made me stay away from it frankly. NB is central to control for a good oc here.

Quote:

---

If you want to see some BIOS pics let me know.

---

Yes please (just don't make them huge) :D

Quote:

---

Originally Posted by madfaze

what parameters did you used during update of bios?what does this do?:confused:

---

Example for Flash Utility 418 with P0H:

Code:

---

afud418.exe a7376ams.p0h /p /b /n /c

---

Some type /pbnc instead, it's best to clear everything properly when flashing.

basically what it does with those parameters?

i mean even without those parameters can you still update the bios normally like i do...

Parameters flash over the boot-strap of your BIOS.

As I've said before, I would never flash without those using a BETA, you are going to end up with random problems many times and even unbootable systems.

geez... now i have second thoughts on flashing BETA bios ... dont wanna end up a brick mobo...

but when the bios is official release is it still good to run it without those parameters?right?

It's always best to do that, always... always. :yepp:

Well that is what BETA is about, you know. :p:

how about the command "/reboot" will it work too?

KTE: Here some Bios/Temp shots from the DS5.

http://i30.photobucket.com/albums/c3...Temps_0228.jpg

As you can see those temps look much more realistic than what I was getting from the K9A2. Alot of times the MSI board would show temps below ambient, and never went over 32c. One thing I never noticed on the MSI board was the somewhat broad Voltage swing when you put the CPU under load. CPU V's are set at 1.3v's but when I load the processor it jumps to 1.36 as shown in the ss, as soon as I turn off Prime4 it drops back down to 1.296v. Seems a little odd but it does seem to help stability... :)
(BTW, Ambient is around 71f right now)

http://i30.photobucket.com/albums/c3...h_DSCN0194.jpg

We don't need no stinking TLB Patch! Sweet... :D

http://i30.photobucket.com/albums/c3...h_DSCN0197.jpg

Haven't had a chance to play with the AutoExpress thing yet, but it looks interesting. I'm just looking for stability right now...

http://i30.photobucket.com/albums/c3...h_DSCN0199.jpg

The only thing I really miss in this Bios is the ability to adjust NB v's. I wonder if they just didn't feel the need to include it, or if they cheaped out on the power section of the board (I doubt that)...

KTE, If you and Achim can figure out a way to adjust that setting through registers let me know. If it's not too scary I might try it... ;)

Quote:

---

Originally Posted by madfaze

how about the command "/reboot" will it work too?

---

I don't recall what that's for or if it works. Load the flash utility and type /? or /help and you should get a list of parameters which do work and what they do.

Quote:

---

Originally Posted by Daveburt714

KTE: Here some Bios/Temp shots from the DS5.

---

Thank you! :)

Quote:

---

As you can see those temps look much more realistic than what I was getting from the K9A2. Alot of times the MSI board would show temps below ambient, and never went over 32c.

---

I'm working through this but the temperatures reported by all software including the BIOS are TCONTROL, and TCONTROL is not a temperature value but an arbitrary value in decimals to allow better control of thermal management. It is designed to read sub-ambient values and even negative values whilst ambient is plus 10C as it does not literally show temperature values in degrees but shows a TCONTROL value that is set to always stay below TCONTROL max; which is calibrated to be consistent with the actual processors TCASE max. ;)

The TCONTROL correlating to the TCASE can be found in the thermal management guide.

My 9500 and last 9600BE didn't read this value as this but once or twice. They were reading the same temp as the TCASE with software and a thermocouple. My current one and 9600s reacted different though and often read sub-ambient. This is by design. ;)

The values for temps you have above are also not real temp values (they are very low for real values, lower than my 9600BE is at stock).

Here's the pertinent explanation of AMD K10h temperature workings:

http://img132.imageshack.us/img132/8700/amdtempev2.png

Quote:

---

One thing I never noticed on the MSI board was the somewhat broad Voltage swing when you put the CPU under load. CPU V's are set at 1.3v's but when I load the processor it jumps to 1.36 as shown in the ss, as soon as I turn off Prime4 it drops back down to 1.296v. Seems a little odd but it does seem to help stability... :)
(BTW, Ambient is around 71f right now)

---

That's good and bad actually.

Good, as in no Vdroop as on MSI which causes mass instability many times, especially if more than 0.02V.
Bad as in you'll have higher load power/temps and current going to your chip, which is not healthy for it. The Vdroop is actually by design a feature for the CPU safety. :yepp:

Very good options in BIOS though, just a trimmed version of the DQ6 one.

Quote:

---

The only thing I really miss in this Bios is the ability to adjust NB v's. I wonder if they just didn't feel the need to include it, or if they cheaped out on the power section of the board (I doubt that)...

KTE, If you and Achim can figure out a way to adjust that setting through registers let me know. If it's not too scary I might try it... ;)

---

You can do it on the fly in Windows, yup. But it will revert after restart.

Check mine and Achim's posts out towards the end of the DFI RD790 thread, they'll help. The register that controls the current processor state is this following and you need to change and write the following value to whatever you want using Crystal CPUID:

Attachment 73166

The 48h control NB VID.
The 340h controls CPU VID.

48h = 1.100V
340h = 1.225V

i.e.

Attachment 73167

I had to put load on it to show you the specs because I'm running CnQ. Check what we're idling with CnQ in the DFI thread. ;)

Just posting some K10h related roadmaps I was about to a day ago but was delayed. Don't mind me... :D

BTW, Phenom FX maybe mentioned there you know. :p:

http://img352.imageshack.us/img352/5728/17888812pg4.png

http://img84.imageshack.us/img84/937...4026kb3.th.png http://img352.imageshack.us/img352/6...0689bi3.th.jpg http://img213.imageshack.us/img213/4...0875cj0.th.png http://img206.imageshack.us/img206/3715/25fp9.th.png

http://img207.imageshack.us/img207/5616/53549237wh7.png

Hmm....2 version of 9750 with diff TBD?

I wonder if the availability of the FX82 will lead to a FX70/72/74 style dual CPU socked motherboard. If there is a chance AMD will get a dual socket motherboard for the Phenom FX series before September I'll put off my computer build.

Quote:

---

Originally Posted by ragnarok95

Hmm....2 version of 9750 with diff TBD?

---

Probably one with L3 cahce and one with out the L3 cache.

Quote:

---

Originally Posted by ragnarok95

Hmm....2 version of 9750 with diff TBD?

---

Different TDP's (2.4GHz).

Quote:

---

Originally Posted by spicypixel

I wonder if the availability of the FX82 will lead to a FX70/72/74 style dual CPU socked motherboard.

---

Don't count on it, believe me, there won't be. Not until after Deneb 2009 anyway. 1P you will get though. :yepp:

http://www.lavalys.com:8081/beta/eve...l9xdsv2crt.zip

Just gets better and better. ;)

Any of you miss Barcelona benchmarked with Xeon (Core 2 and Penryn)?

Independent world famous professional research found a while ago, that due to FBDIMMs Intel uses, Barcelona servers beat similarly configured Xeons based on Core 2 and Penryn in power usage and performance efficiency: http://www.worlds-fastest.com/wfz986.html

Test were conducted far better than SPEC_CPU 2006 or TPC can, these saturate each RAM module, network, disk and core far more. These are also not configured by separate entities, flags or components, but are purely apple to apple comparisons.

Full details of the testing:
Power Usage at various transaction levels: http://www.worlds-fastest.com/wf3400.html
Throughput: http://www.worlds-fastest.com/wf2400.html

That's basically why Penryn was rushed and released as Intel knew, and as any market analyst would tell you. In the major server market with strong loads, Woodcrest would get trumped for power/cost/efficiency/performance by Barcelona. Let alone that, even 2.33GHz Harpertown vs. 2.0GHz Barcelona get beat overall when you weight the overall cost:performance down. That's also why someone would take Barcelona over them for supercomputers.

Never would have thought that was the case, Barcelona on par with Penryn. :cool:

I´d like to say that I have seen that before somewhere, but I´m not sure where.... maybe in the news section..