Phenom 9500 w/ MSI K9A2 Platinum

[Rammsteiner]

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.

Hmmm, lets hope it's a BIOS issue then and they fix it soon.

But thus far I really dont like it lol. But we'll see, first got to see if this is finally stable heh.

[KTE]

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

Vcore or VID? I'm talking about VID

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

What did you change it to and from? What values?

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...

Like I said a long time back, Prime is only FPU/INT and load stability, whereas with complex architectures, no more complex than Phenom, you have many different components to test for stability. Whatever it is, this is a first for me and its been across all 6x Phenoms I've tried. I am perplexed as to exactly what it is though...

BUT - it doesn't mean no setting above stock is unstable, infact, I'm running 2700 now and its perfectly stable. What does that mean?
Load-bench may be fine up to 3155MHz in my case but real stability is only fine <2800 so far.

I can't see how its BIOS or board, tried many boards and BIOSes and RAM, and CPUs - its just Phenoms, that's the bare fact.

[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.

Yea I said it a long while back - I don't count it unless you can run it error free for a few days uptime and messing - which includes regular tasks without a hiccup.

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

Hey Chris
Only BIOS that we kow can change it is 1.13 and it does work fine for B3s although officially not supported - its a BETA

~November 16th release IIRC.

[ChrisTucker2004]

Hey Chris
Only BIOS that we kow can change it is 1.13 and it does work fine for B3s although officially not supported - its a BETA

~November 16th release IIRC.

Thx KTE

Then i flash now 1,13

EDIT: Bios 1,13B is named 1,1B3 from 11/16/2007, is it correct?

[jonspd]

I thought vid is vcore on the cpu part?

21 to 19 I believe which changed from 1.288 to 1.30 nb vid.


did divide by?
vid voltage?
fid muilti?

[KTE]

Thx KTE

Then i flash now 1,13

EDIT: Bios 1,13B is named 1,1B3 from 11/16/2007, is it correct?

Does it work for you, as in bootup and run ok?

You're very welcome.

I thought vid is vcore on the cpu part?

Forgot it all in 3 weeks



They're different, Vcore you can change but VID never changed for me.
VID is seen in AMD Power Monitor, the offset boundary to supply the voltage [higher or lower].
Vcore is what it says, can be seen in AOD as VDDC.
The rest is as you mentioned.

21 to 19 I believe which changed from 1.288 to 1.30 nb vid.

Hmm.. yeah, stock is 20 = 1.300VID so that is about right. Have you changed it higher than 1.3VID, i.e. 18?

Check this link out, another AMD controversy: Motherboard ills leave AMD users with Phenomenal headache

Various news sources have been pouncing on AMD's acknowledgment of potential problems when a high-end Phenom X4 processor is paired with certain 780G motherboards, but some of these articles have been focusing on the wrong part of the equation. The reported overheating and stability issues aren't being caused by Phenom, but by motherboard manufacturers' failure to design/certify their solutions for a Phenom processor with a 125W TDP.

This type of issue actually isn't all that rare; a $49 motherboard is $49 for a reason. Even if the motherboard's chipset is technically capable of running a high-end processor, the motherboard may not be designed to handle the CPU's power draw and thermal output. That's essentially what has happened here, but a combination of other factors has made the situation more important than it might otherwise be.

And BTW, AMD's recommended tested boards can be found here: AMD Recommended Motherboards for K10 CPU's

[Suosaaski]

Quick and simple testing on my 9650 at 11,5*218=2.5GHz. 1,3 volts, nb and ht is at 1960MHz, memory 872MHz. This is the first OC I tried with this chip (not the highest that proved stable). So don't make any conclusions about the overclockability of this chip because I only have it at 2,5GHz.

Uptime is only 1 hour or so (first tests), so stability is not guaranteed, but I did some youtubing while copying a 2.5GB file from disk to disk. Also tried playing a game for a while (5-10 mins). I was gonna test a lot more today but was in a bit of hurry, so I think I'll go to bed now and leave this idling over night to see the status in the morning.

If it proves to be stable tomorrow too, maybe I will try to enable cool n quiet and see how that goes. Or maybe I'll try and get further. Remains to be seen.

Other components:

MSI K9A2 Platinum (version 1)
4 * 1GB OCZ Platinum DDR2-800
MSI 3870 Overclocked Edition
Antec NeoHE 550W

Running Windows Vista (32-bit).

Screenshot: http://www.aijaa.com/img/b/00609/2020458.jpg

Edit: I just found out that the stock voltage for this chip is 1,15 volts. Need to do some testing to see how much is actually even needed for 2,5GHz.

[jonspd]

Yea I read about the 780g problem kinda sucks...


http://www.theinquirer.net/gb/inquir...t-comptibility




you lost me on the vcore vid talk

yea the voltage options have never changed for me also except for ram. but the p-states all work.

[Suosaaski]

Update, now that I figured out mine is an 1,15V chip, I decided to do some more testing. Same volts, 226*11,5 = 2,6GHz, NB and HT at 2034MHz, mem at 904MHz. Remains to be seen how it goes.

[KTE]

For 9650, 2600 is good - good start, let us know how it goes on

Ok Jon

Check some news out.......

AMD gains slightly in processor market in 4Q07, says iSuppli

Although AMD's struggles have dominated recent headlines, the company actually managed to turn around its fortunes slightly in the fourth quarter of 2007, gaining some share in the market for all types of processors, according to market research firm iSuppli.

http://www.digitimes.com/news/a20080423PR201.html
http://www.digitimes.com/mobos/a20080430PR201.html



What did I say despite the hay-o glooms of lopsided geeks?
What do you think will happen with K10s only really shipping since Q2?
Like I and pretty much everyone in the industry said - you have something out there, it will get bought and you make revenue however little. If its nearby in costs and cheap, it will win many - they don't overclock like us.

AMD: Next CPU architecture will be completely different: http://www.custompc.co.uk/news/60251...ifferent.html#

AMD [45nm] desktop CPU lineup revealed: http://www.digitimes.com/NewsShow/Ma...ges=PD&seq=219

AMD is planning to launch two 45nm Phenom X4 CPUs (Deneb) in 2008 with core frequencies of between 2.5-2.8GHz and 2.4-2.7GHz, with both adopting 6MB L3 Cache and having a TDP rating of 95W, according to sources at motherboard makers.

AMD will announce the final order date for its 125W Phenom X4 9750 by the end of the second quarter and the CPU will be replaced by a 95W version – which is available this quarter – while the 95W Phenom X4 9850 will appear before January of 2009. The company is also planning to launch an updated version of its upcoming Phenom X4 9950 (schedule to launch in June) at the end of the fourth quarter.

In addition to the current triple-core CPUs, AMD will launch the Phenom X3 8550 with a core frequency of 2.2GHz in the third quarter (production beginning in the second quarter) and the company will launch more triple-core CPUs in the fourth quarter.

The company will begin production and take orders for the low-power Phenom X4 9350e (Agena) in the third quarter, while the low-power 8450e (Tollman) will see production begin in the second quarter, though order availability will not begin until the third quarter. The Phenom X4 9150e, which was originally planned to be launched in the second quarter, will not be available for orders until the third quarter, along with the 9350e. In the fourth quarter, AMD will launch another low-power CPU, while in the first half of 2009, the 45nm low-power quad-core Phenom CPUs (Propus) will appear in the market.

The company will stop taking orders for the Athlon 64 FX-74 in the beginning of the fourth quarter.

AMD's 45nm CPUs are already in EVT testing and motherboard makers should receive their samples by August or September this year, if the process goes smoothly, 45nm Phenom X4 CPUs should appear in the market by the end of November, added the sources.

AMD indicated that it declines to comment on unannounced products.

BTW->

[Suosaaski]

For 9650, 2600 is good - good start, let us know how it goes on

6 hours later (I left it idling) everything is still stable. Now I'll start doing some stability tests under normal usage and load. I MIGHT try 2,7GHz at the current 1,3 voltage, but I have decided I am not going to raise the volts any more.

Edit: Decided to lower the volts from 1,315 in bios to 1,281, I'll see how it goes.

running prime failed (system froze) after 10 mins, I'll put the voltages back up and try again...

Edit: Even when I lovered the ht link to 8x multi and upped the voltage to 1,349 the system kept freezing (and rebooted once). I think I'll go back to 2,5GHz and try running Prime at that first...

[KTE]

Hmm.. the freezing eh.
Did you actually use it for a day or so at those settings doing usual browsing and work?
That's what'll show the "idling instability" problem quicker.

Right now for me... C&P from DFI thread:-
I've just ran the Ubuntu 7.10 32b LiveCD, installed it, then updated it, modified it, which took about two hours, and then decided to upgrade to 8.04 LTS - took a few hours on all this. All at stock volts (1.288v) 2.700GHz and left it for 9 hours now.

Just installed some functions and had to restart - uptime is about 2 hours now but many ways to get or change the CPU frequency are not working. Here's with 25% load on it:



Using

cat /proc/cpuinfo | grep -i Mhz

You can access the current speed.

Also have compiz and my GPU drivers working perfectly first time

Tried enabling various frequencies by writing a shell script which works if supported on the processor: did not work
Tried Powernowd to do it: did not work.
Tried to install PTS v0.5.0 which has just been released but it failed download.

[jonspd]

great news to here about the new stuff there gonna be trying and hopefully doing, glad to see they may be starting to gain instead of the same trickle lose effect...

also did you see this

http://www.fudzilla.com/index.php?op...=7033&Itemid=1

BTW: I like the new chip it seems pretty nice but not gonna say anything else untill further testing

[ChrisTucker2004]

Does it work for you, as in bootup and run ok?

You're very welcome.

[/B][/URL]

Yeah the Bios is running, but i Cant test it with my Phenom 9650. I sent my CPU for speed tests to a friend in another City, my stable speed was 2600 mhz, my friend runs 2800mhz on an Asus M3N HT 780a.

[KTE]

Saw more than that Jon, many pics of the board and its chips
I never check FUD/Inq/DT, last resort if I know the info to be true, or never.

The fact that they have a new one coming rebadging its launch means the current boards have a hardware problem, pretty simple.

Chris, are you sure?
Ask him to post settings or screenshots of benchmarks and the voltages, etc please.

[ChrisTucker2004]

KTE:

Here is a Screenshot, my CPU runs maybe @ 3000mhz, but 1,55v is to high

http://www.abload.de/image.php?img=phenom29ur.jpg

An HT @ 280 is stable with a Multi of 5,75. I think that board is buggy but good

[Rammsteiner]

Saw more than that Jon, many pics of the board and its chips
I never check FUD/Inq/DT, last resort if I know the info to be true, or never.

The fact that they have a new one coming rebadging its launch means the current boards have a hardware problem, pretty simple.

Chris, are you sure?
Ask him to post settings or screenshots of benchmarks and the voltages, etc please.

Oh lol, I linked in another thread the discussion about this new board to you lol.

But you already saw it.

Anyway, I wonder if the 'issues to run 3Ghz' are like, 3Ghz unstable, not able to get upto 3Ghz at all (like me) etc. I mean, Im still having problems to get 2.7Ghz without freezing... So I wonder if the board would fix 1) the freezing and 2) eventual the scaling?

How many Volts I throw at this chip, it just doesnt want to do anything above 2.8Ghz. 2.9Ghz I think I got it to boot but almost insta freeze, I dont call that 'it wanted to do something'. And 2.8Ghz is like very fast freezing. And tbh, my I never saw Windows @ 3Ghz.

But the fact Voltage doesnt matter shows for me something else is wrong here. But as I said, whether the new chip would actually get me all the way up there... I would be SO happy, but Im afraid I still have a crap batch and I might instead get maybe 2.8Ghz stable and there it would stop...

Are there any dates known when we might expect this SB? I dont hope it would take another 3 months or so, a month or eventual 2 months ok... But just fast

[Suosaaski]

Hmm.. the freezing eh.
Did you actually use it for a day or so at those settings doing usual browsing and work?

Well, if it freezes running Prime, I don't consider it stable, so I didn't find it necessary to try browsing etc. I left it idle for a few hours while I was asleep (can't really do any browsing or work while sleeping), but did not try to find out the instability during idle for more than 30 mins at first.

It might be that I'll have to revert to an older BIOS in order to change the NB multi to something lower, just to find out if it was NB causing the problems or what. But I don't like messing around with P-States, I wish there was just a simple selection for the multi.

[KTE]

Oh man, XGL and Compiz on Hardy give a pain!
Doesn't seem to like being uninstalled. Looks fine ATM, I have all the Compiz desktop effects working.

KTE:
Here is a Screenshot, my CPU runs maybe @ 3000mhz, but 1,55v is to high
http://www.abload.de/image.php?img=phenom29ur.jpg

An HT @ 280 is stable with a Multi of 5,75. I think that board is buggy but good

Very good for one of those - stable?
280HT is max (if not, what is, try it)?

Can your friend with the 780a tell us if it has CPU-NB Multi and CPU-NB volt options in BIOS?

Oh lol, I linked in another thread the discussion about this new board to you lol.
But you already saw it.

AFAIK late May to early June is the "planned" date for that and 790GX but that's not hard info.

Anyway, I wonder if the 'issues to run 3Ghz' are like, 3Ghz unstable, not able to get upto 3Ghz at all (like me) etc. I mean, Im still having problems to get 2.7Ghz without freezing... So I wonder if the board would fix 1) the freezing and 2) eventual the scaling?

Kinda doubt that. Because there's a point where the CPU just will not go, and mine is at around 317xMHz. It'll fly all the way up to 3GHz on 1.3v but then you have to add 1.38v for 3100MHz and suddenly, things start to crash and the limit is very easily seen - it just does not boot fully.

If it's a CPU limit, then it just won't go there.
If it's possible something other, it will go there fine, bench it, load test fine - but not stable for daily use. It might even have runs where it doesn't crash for 2 days, mine did it for 5 days, a record but it was still unstable. A simple boot after leaving off for 7 days and it froze within 25 minutes.

I wish macci could help with this, its issues like these we all want help for, being lost as to what it could be and they do have more Phenoms and motherboards than us to be able to test better, it being their daily job.

How many Volts I throw at this chip, it just doesnt want to do anything above 2.8Ghz. 2.9Ghz I think I got it to boot but almost insta freeze, I dont call that 'it wanted to do something'. And 2.8Ghz is like very fast freezing. And tbh, my I never saw Windows @ 3Ghz.

The not-possible MHz are different to the idling unstable.
Test whatever MHz you can benchmark in everything, boot repeatedly, install Windows/Linux with and then try and find something lower than that for stability.
If you have the freezing issues, they will appear with a very frustrating pattern similar to this although different MHz/Volts:

3200MHz 1.30v load/bench stable many hours [say 20hrs]
3200MHz 1.30v idling unstable
3200MHz 1.35v idling unstable
3200MHz 1.40v idling unstable
3200MHz 1.45v idling unstable
3200MHz 1.50v idling unstable
3200MHz 1.55v idling unstable
3200MHz 1.60v idling unstable
3100MHz 1.55v idling unstable
3000MHz 1.55v idling unstable
2900MHz 1.55v idling unstable
2800MHz 1.55v idling unstable
2800MHz 1.50v idling unstable
2800MHz 1.45v idling unstable
2800MHz 1.40v idling unstable
2800MHz 1.35v idling unstable
2800MHz 1.30v idling unstable
2800MHz 1.25v idling unstable
2700MHz 1.30v idling stable



You see how weird that is compared to all other CPUs we know about before?
IMO this is process variation more than anything - one tiny property becomes very enlarged as you increase MHz/Volts and if its slightly different to elsewhere in the chip, it can cause erratic instability, as we're seeing. I'm not sure what the SB600 clockgen has to do with this, it could be to some degree but not massively.

Your "batch" is not bad BTW, it's just your chip is most likely not as good as some others. Have you thrown up voltages to about 1.55v and benched Cinebench 10 multiCPU at say, 3000MHz or 3100MHz?
I have a similar batch click-> <-me

Suosaask you can just do it with BAR_EDIT, no need to swap BIOSes.
If it was the NB, then you would have the problem at stock too though.

OC - if its stable, use it daily without worry. Then you'll see whether it really is stable

[Andi64]

I'm having problems with idle stability too.

It was working at 2.9Ghz 1.3V for 10hs at full load with Prime95, when I closed it and opened MSN the system freezes

Stable so far @ 2.8Ghz 1.3V (1.28 load).

I don't get it


System:
K9A2 Platinum BIOS 1.4
Phenom 9850BE 0810BPMW

[Rammsteiner]

I didnt try all the way up at 1.55V though, because that simply felt 'wrong' lol.

But for now Phenom is indeed strange. Something is simply not right, clear as that. I dont care to play so much around, it's fun. But the fun degrades quickly if you notice there's a general problem which prevents you from getting good results.

Everytime you go into BIOS and try a new setting you go like 'THIS MUST BE IT!', and you startup a game and it freezes. No matter what Voltage or what speeds (above a certain area). Also as you noticed, being 24/7 stable for a few days, every reboot might change things rapidly.

Since we dont know when it's REALLY stable it's kind of a lost case at the moment. You can try and try, but then this and then that messes things up. I mean, 5 days of stability, a reboot and a crash withing 25 minutes... What does this tell us? Nothing. Where it might have crashed before the 5 days in 10 minutes, you find a setting which delays the crash and then at once you get it to 5 days and think it's stable. Next start up it's after 25 minutes a crash so all the work, notes and settings from before the 5 day stability were useless and worthless.

And that part just, yeah, pisses me off to name it that way. That makes one wonder how long he should test for and then still you get suprise buttsex at next startup.

I really hope SB750 throws some more fun in. I really dont care if my chip wont clock as good as Tony's. Of course I'd like to get 3Ghz or so, but 2.8~2.9, where Ive seen Windows, Im fine. But just in general, not for my poor results thus far, I think this issue has to be solved very quick. Unstable is unstable really, but as I said before, this makes no sense.

[KTE]

Yeah, that same problem
I can't help because I don't know what it is.

Although I do know I have 2700MHz 1.288v fully stable

http://picsorban.com/upload/screenshot-1.png

Now benching Phoronix.. including Super Pi in Linux.

[jonspd]

The issue is not known other then it is with the Phenom chip itself some way shape or form. I wish for it to be fixed also but as much testing that has been done there is nothing that points to what it could be doing to cause the system to freeze/

[Suosaaski]

Suosaask you can just do it with BAR_EDIT, no need to swap BIOSes.
If it was the NB, then you would have the problem at stock too though.

OC - if its stable, use it daily without worry. Then you'll see whether it really is stable

I would like my OC to be enabled all the time, not by some program.

What I meant by NB, is that it ran at 2034MHz where as stock is 1800MHz, so I don't quite understand why I would have the problem also at stock...

Well, I already figured out my OC wasn't stable at 2,6GHz, but at least one person at a Finnish forum also had problem with NB climbing too high with his 9750, he has lowered the multi so that NB and HT results at about 1750MHz and as far as I understand, 3GHz seems stable for him so far. His motherboard is MSI K9A2 Platinum, and he also reverted back to BIOS 1.1 to get P-States and NB multi working again.

By the way, does anyone happen to know the default volts for NB and HT? Would be nice to know.

PS. Of course, I could try to raise the volts for my memory too. Had it at 2.0V.

[KTE]

I would like my OC to be enabled all the time, not by some program.

That isn't what I answered to. You said:
"It might be that I'll have to revert to an older BIOS in order to change the NB multi to something lower, just to find out if it was NB causing the problems or what."

And to that I said use Bar_Edit and you can find out very easily by dropping the multi.
Takes less than 10 seconds to change it and then to reboot, the reboot time and you'll have a lower multi to see if that hinders you or otherwise.

By the way, does anyone happen to know the default volts for NB and HT? Would be nice to know.

When you say NB, what do you mean, CPU-NB or MB-NB?

MB NB is 1.10v
HT 1.20v

[Suosaaski]

That isn't what I answered to. You said:
"It might be that I'll have to revert to an older BIOS in order to change the NB multi to something lower, just to find out if it was NB causing the problems or what."

And to that I said use Bar_Edit and you can find out very easily by dropping the multi.
Takes less than 10 seconds to change it and then to reboot, the reboot time and you'll have a lower multi to see if that hinders you or otherwise.
When you say NB, what do you mean, CPU-NB or MB-NB?

MB NB is 1.10v
HT 1.20v

The one that needs to be adjusted via P-states. so that would be CPU NB

And when I sead "OC to be enabled all the time" I meant via BIOS settings, not by program like bar_edit, but you are correct, bar edit would be an easy way to check the nb multi.