Phenom 9500 w/ MSI K9A2 Platinum

[justapost]

Sandra reads the BIOS and CPU register temps and volts perfectly Achim
Only problem is, they are slightly under load temps.

Are those screenies both idle (desktop vs CnQ)? If so desktop temp seems abit high.
What do you get under load?

[SocketMan]

If this is true i am surprised we all missed it and I tested 1.4 and 1.41.

is anyone here on 1.4 or 1.41 and able to test this with a 9850?

the HT rate/multi setting gave the clue, since NB must be >= then HT.
I figured adjusting the HT ratio/multy will make the board adjust the NB as well or
it will never boot to windows, or even post..
Check the picks

Oh, also I think (this one I am not 100%) the ganged/unganged setting is messed up:there are 2 options there but both give me unganged
I am using this 2x2048 Balistix kit, that I just picked up..
CPUz,Everest and AOD all show NB @ 2200 (multi of 11 in bios for HT)

The HT stays @ 1800 on the boot up, but is changeable with AOD

[KTE]

I was curious about onboard graphic results so I ran 3DMark06 on four different Systems also added our previous results:

Whoa man thanks for the testing!

My scores the lowest, yep
I can't open GPU-Z and ATiTool etc, my GFX crashes
It should not be 57MHz though, I'm pretty sure that's 2D mode clocks.

Note: M2A-VM shows 57MHz for the GPU also, the Turion Notebook with the same gfx chipset shows a correct frequency. I used 8.4 CAT on both 1250 systems.

Hmm.. I tracked a program running at startup which set "ATi mode changes" - possibly something to do with that. Your scores are still very close by though for both runs.

So that means 1250 chip is equal to 740G I guess.

Are those screenies both idle (desktop vs CnQ)? If so desktop temp seems abit high.
What do you get under load?

Yep, Sandra has to put load on or it can't fetch temps, thus what you see is slightly loaded temps [+2-3C from idle] - ambient was high.

I've been able to troubleshoot and found EVEREST also picks up BIOS temps, CPU and GFX-NB temps, as does EasyTune, it just labels them wrong way round. Here's real idle temps in both states->

"Motherboard" in EVE is CPU in BIOS as you can see through EasyTune and "Aux" in EVE is GFX-NB labeled as System in BIOS/ET.
+2.5v rail is actually the DRAM voltage, real is slightly lower [-0.016v]

-
Stock max fan - CnQ max fan [ambient is 18C]


CnQ + fanspeed = below 100RPM

BTW, two of my old gen PATA Maxtor drives, still living, one of them beats all of my WD SATA II-> [all settings are same]

- -
4x WD SATA II [all same] - Old Maxtor PATA 1 - Old Maxtor PATA 2

the HT rate/multi setting gave the clue, since NB must be >= then HT.
I figured adjusting the HT ratio/multy will make the board adjust the NB as well or
it will never boot to windows, or even post..

Thanks for the tests mate
Problem is they are independent settings, like CPU and NB speeds, even though they have a relationship

It shouldn't be changing, if MSI found out, they'll take BIOS down and replace it with one that doesn't touch NB. I tried to check it out yesterday but the MSI didn't want to boot 9850 so I left it.

Oh, also I think (this one I am not 100%) the ganged/unganged setting is messed up:there are 2 options there but both give me unganged

You disabled Unganged mode?
Its worked in all BIOSes since the first but I've not tried the latest.

The HT stays @ 1800 on the boot up, but is changeable with AOD

Then MSI messed up; they labeled it HT link multi but the register its changing is NB multi not the HT multi

Any more confirmations that BIOS 1.4 has NB multi?

[KTE]

Ah ok, this is the difference between them, no wonder

Review 780G vs 690G vs 7050

AMD Chipsets: http://en.wikipedia.org/wiki/Comparison_of_AMD_chipsets

RX740 - 740
RS740 - 740G aka 690G+SB700 aka X700 IGP aka RV410/X1250 @ 400MHz
RX780 - 780
RS780 - 780G
RS780M - Griffin [Puma - probably best IGP of the lot]
RD780 - 780X

Best IGP is 790GX and best Southbridge is SB750, both highest end IGP

So X1250 should be same as 740G in perf. 740G is just slightly better.




's' stands for server.

With M3A+HD2400pro+4850e and The GBT Odin 550 i measured ~45W DC (idle) and 90W DC (load) at stock voltage.

Did power measurements, many repeats until I found consistency.

CPU
3.58W DC CnQ idle from CPU 12V rail
9.56W DC stock idle from CPU 12V rail
39.435W DC full load from CPU 12V rail

System
25W DC idle with CnQ+4 USB connections+4 HDDs+1 HSF+1 PATA ODD
36W DC idle stock with 4 USB connections+4 HDDs+1 HSF+1 PATA ODD
76W DC full load with 4 USB connections+4 HDDs+1 HSF+1 PATA ODD

I've got Abit 770 MB here too, I'll check it out later.

BTW: How did you get the HD running again?

IDK TBH I just sent it to a business contact of my uncles employer 35 minutes away and he said he'll sort it out
Fixed both of them inside 3 hours, so it'll do me

[justapost]

Whoa man thanks for the testing!
My scores the lowest, yep

I did not want to prove that. I'm curious what intel will bring to the table, currently their onborad gfx chipset is more or less on par with 1250/HD2100 beside intel also can run the SM3 stuff.

Hmm.. I tracked a program running at startup which set "ATi mode changes" - possibly something to do with that. Your scores are still very close by though for both runs.

I run an about a year old winxp installation on the M2A-VM so a software problem caused by old drivers or registry settings are quite possible here.

Yep, Sandra has to put load on or it can't fetch temps, thus what you see is slightly loaded temps [+2-3C from idle] - ambient was high.

Ah ok.

I've been able to troubleshoot and found EVEREST also picks up BIOS temps, CPU and GFX-NB temps, as does EasyTune, it just labels them wrong way round. Here's real idle temps in both states->

"Motherboard" in EVE is CPU in BIOS as you can see through EasyTune and "Aux" in EVE is GFX-NB labeled as System in BIOS/ET.
+2.5v rail is actually the DRAM voltage, real is slightly lower [-0.016v]

Great, thank you it's similar to what i get on the Sapphire with Sapphire bios.

BTW, two of my old gen PATA Maxtor drives, still living, one of them beats all of my WD SATA II-> [all settings are same]

No big difference between SB600 and SB700 with a normal disc it seems. That PATA drive was faster on the MSI also if i remember correct.
Thank you for the power measurement. I'll connect a few discs and the GBT PSU and post results here for comparison.
The 4850e run's at stock frequency (2.5GHz) with 1.175V have you tested the 4450e in that area?

[KTE]

I'll try undervolting now, just managed some free time.

[KTE]

1.280v idle does 2.507GHz very easily without much idling power increase [27W].
Also onboard GFX clocks are picked up wrongly, basically, the Default clock is the real-time clock. Check CCC for real clocks i.e. (running 450/400 here)



I'm running that for about 28 hours now without problems, even gaming, 3 hour multimedia decoding and benchmarks.

Will edit to add 2k6 score...

EDIT:

Slight improvement.

CBID seems to show current wattage very accurately....

[jonspd]

9850 is in the house just got back from frys with a week 0811 BPRW.

Probably want be able to test till later this week.


KTE is there a bios I should think about using other then the 1.13 seeing as thats whats already installed

[justapost]

1.280v idle does 2.507GHz very easily without much idling power increase [27W].
Also onboard GFX clocks are picked up wrongly, basically, the Default clock is the real-time clock. Check CCC for real clocks i.e. (running 450/400 here)



I'm running that for about 28 hours now without problems, even gaming, 3 hour multimedia decoding and benchmarks.

Congrats, your onpar with the Mobility Radeon 9600 now:

Will edit to add 2k6 score...

EDIT:

Slight improvement.

CBID seems to show current wattage very accurately....

I wonder what register the software is reading, skimming over the BKDG atm and looking for current amps. I saw that the itc sensor chip on the M3A has an amp sensor pin beside ones for cpu, nb voltage etc, so it might be possible to read the amps via i2c/smbus.

EDIT: This just reflects the limit for the current p-state and not the current used value.

[KTE]

9850 is in the house just got back from frys with a week 0811 BPRW.

KTE is there a bios I should think about using other then the 1.13 seeing as thats whats already installed

Great! Try 1.13 and 1.4

Have you booted with 113 yet? I'd be wary since its a November BIOS and it may not support them properly - not officially anyway.

Congrats, your onpar with the Mobility Radeon 9600 now:

Yeah....
Don't worry, most likely replacing it with ASRock 780G with DisplayPort, can get it just as cheap

I wonder what register the software is reading, skimming over the BKDG atm and looking for current amps. I saw that the itc sensor chip on the M3A has an amp sensor pin beside ones for cpu, nb voltage etc, so it might be possible to read the amps via i2c/smbus.

EDIT: This just reflects http://www.abload.de/thumb/iddl69.jpg the limit for the current p-state and not the current used value.

Hmm.. for K10, the current maximum limit is set using the above registers. Does that apply to K8 as well?

nForce 750a runs very hot at idle->
http://my.ocworkbench.com/bbs/showth...454#post430454

Edit: the ASRock supports 9850 BTW, 5-phase

[KTE]

Anyone willing to try new BETA BIOS 151?

Here it is: http://www.yourfilehost.com/media.ph...376AMS.151.zip

BTW, GBT 780G MB is more power hungry than the ASRock 780G
ASRock board also has many better features.

I found out there are B3 step Phenoms which are classed as 9600/8600 with the xx00 suffix, but they are only rated for the Business Class marked with a 'B'.
I also came across some new CPUs, namely Intel quad cores, 3GHz and 3.33GHz that are coming out soon - judging by Intels 45nm record, don't take my word for it though.

Anyway, 780G vs 8200 vs G35 preview: http://www.anandtech.com/cpuchipsets...spx?i=3288&p=1

[jonspd]

Gonna try it with 1.13 for starters. Are you saying that 1.4 is the best official bios?

Sry haven't been keeping up with everything to much lately.

[KTE]

OK, keep us up to date though

BTW in case people are confused (I know websites are) ->

780V = Radeon 3100 @ 350MHz core
780G = Radeon 3200 @ 500MHz core



And Achim, 1xHDD power idling at 2507MHz stock volts with CnQ is 44W AC. But my board has started acting erratic now so can't test more - looks like CnQ driver.

[Suosaaski]

Should start testing with a Phenom 9650 tomorrow hopefully. More likely on Friday. Too bad I am a lowsy overclocker, but I am still hoping to get past 2,7GHz :|

[KTE]

Mine is back up and running perfect with 113 .. seems its Crucial and Geil 1066 that it hates causing C1 at boot
Using Corsair 800 RAM now and its doing perfect.

So far today in 1hr;

Lowest MHz = 100MHz
Lowest multi = 0.5x
Highest MHz = 3100MHz
Highest multi = 15.5x

System idle stock = 83W DC
System idle CnQ = 65W DC

[dr_drache]

Anyone willing to try new BETA BIOS 151?

Here it is: http://www.yourfilehost.com/media.ph...376AMS.151.zip

I have been running 1.5b1 sence the last time you posted it... it has nothing special about it that i can tell, call it a renamed 1.4.

no speed differnces in WinRAR/HD speeds.
does have TLB in bios, no p-states, nothing other than the version string can i tell a differnce.

[KTE]

1.13 BETA works stable with B3

Notes:-

*FID/DID values are exactly the same as with B2 step and B3 step CPUs, they work.
*However VID values do not work for 9850 BE, they will not change from stock, be careful
*OC is better;
I could not get 2800MHz CPU and 2300MHz NB even 1.38v stable with P0J/P0E/P0H before but... with 113 now, it is perfectly stable at stock 1.288v for 2800MHz CPU and 2400MHz NB -- I have been benching it too and it passed all my WISE tests so far.
*Perf is also better on this BIOS.

Resolved and Findings:-

*Finally after so much screwing around and PAIN, I've worked out what causes the dreaded C1 and no boot - AKA "Oh no I have a dead Phenom!"
It is 1066 RAM which causes it as soon as you clear CMOS or switch the power off! Crucial and Geil 1066 kits I have, both showed this behavior on a few MBs. Tried them = fail. Tried Corsair 800 MEM = perfect boot first time. Failed oc, cleared CMOS and tried again = perfect boot again

Issues:-

*2900MHz passed LD testing 1.440v ID / 1.403v LD but failed WISE during 6 youtube video playbacks - 35 minutes.
*3000MHz also passed load with the same voltages but failed.
*The perfectly load stable but WISE unstable error still exists.

The royalty screenshots of my new boot settings.. stock volts.

Stable before was ->



Stable now ->







Idle is 84-85W DC and 105-107W AC system power draw at those settings
Full load is 185W DC and 196W AC system power draw at those settings

This is same settings but 100MHz difference; 9850 BE 2500MHz vs 9600 BE 2600MHz in X264 benchmark:



You can see 9600 BE still wins on MHz.

[KTE]

I have been running 1.5b1 sence the last time you posted it... it has nothing special about it that i can tell, call it a renamed 1.4.

no speed differnces in WinRAR/HD speeds.
does have TLB in bios, no p-states, nothing other than the version string can i tell a differnce.

Thanks for testing it

So.. no apparent change eh


Stability Update:

Above setting failed WISE testing after ~14 hours uptime.
So now started testing more, 1.3v failed and so did 1.35v.

Right now at 2800MHz 1.368v ID / 1.344v LD for 4-5 hours, going strong.

EDIT1:
2800 1.368v ID / 1.344v LD failed WISE after another hour [5.5hrs].
Testing 1.392v ID / 1.368v LD now...

EDIT2:
Wow, 1.392v ID / 1.368v LD failed WISE stability
1.408v ID / 1.388v LD also failed

Back down to 2700MHz testing now...

[jonspd]

What is WISE? sry to ask but I went back 20 pages and couldn't figure out what the wise test is

Also can you not change vid on the cpu or not on the cpu and NB with b3 installed?

[KTE]

What is WISE? sry to ask but I went back 20 pages and couldn't figure out what the wise test is

http://www.xtremesystems.org/forums/...8&#post2897618

Easy start, open 10 youtube videos to play at once switching between them and reloading them whilst doing some cross-drive file transfer.

Also can you not change vid on the cpu or not on the cpu and NB with b3 installed?

Both

BTW, tested a SATA II in 740G SB700 [4450e] vs 790FX SB600 [9850BE].
SB700 is slightly better (blue) ->



IDE drives work fine with the system.. 7200RPM IDE (blue) perform better overall than 7200RPM SATA II, although IDE add ~20-30s to boot

[jonspd]

strange I had no problem changing my vcore from 1.42 which I was running on the 9600be to 1.35 on the 9850 seeing as I forgot to set that back to stock before chip swap

also the nb vid works as I check after I realized that the cpu vid was working.


also It may take me a few days to get back with results. gonna shot for 12x250...


2 for burn-in and 2 for finding the sweet spot

[Rammsteiner]

I really find that WISE stability thing annoying. It looks like it's never going to be stable either. Once you think it's going good and you think you got the right setting it freezes...

Im wondering if this is really our fault as OC instability or maybe some sort of issue with Phenom/BIOS/whatever? How could a CPU be 14+ hours stable at Prime and then freeze while gaming

[EDIT]Also that makes me wonder if those good OC's achieved by for example Tony are REALLY stable. I mean, what are we doing wrong then.

[ChrisTucker2004]

Hello from Germany,

Yesterday i wrote an email, Since Bios 1.2 I cant Change the Northbridge Multi, thats the answer.



Dear Christian,

As adjusting NB multiplier will cause not booting, sorry we can not supply bios to adjust it according to MSI BIOS spec. Please follow our suggestion not to overclock too much for sytem stability.

Best regards,
MSI Technical Support Team




Could anybody say me, how i can change my NB Multi, Bios 1.2 isnt for my Phenom B3 9650. I cant oc in high regions when the NB Frequency is about 2100

best regards

Chris

[jonspd]

ramm

I think it has been talked about in this thread before. It may be a driver or bios issue but untill the fix is release we want know. However it is instability, I was upset also when I could run my 9600 at 2.7 full load anything but 2.56+ idle was a no go.



chris


There is bios with the NB adjustment out there I'm currently using 1.13b with my 9850 however do so at your own risk


MSI has started to suck on support atleast for this board as we have all found out the hard way.

[ChrisTucker2004]

hey jonspd

also Bios 1,13B have the NB Multi? Is it a mod or an official MSI Bios for the B2s?

Could you say me the Date of the release of Bios 1,13B? I thougt that the Phenom B3 was since Bios 1.4 in the support List???

Wonderful ^^