Phenom 9500 w/ MSI K9A2 Platinum

[Hosser]

I recommend disable HPET if you're not using Windows Vista.

[jonspd]

I left it as it was defaulted to. Maybe KTE can explain more or why or why not I have a pci-e 2.0 card so I do what it to run the correct bandwidth.

[madfaze]

HPET has nothing to do with performance, it's just new high resolution hardware interrupt timer. Can only be used by Windows Vista and some modern Linux kernels. Windows XP does not make use of it.

Nice cooling solution, though I'd imagine it's quite loud? How has it affected temperatures/overclockability?

thanks!

noise is tolerable when you get to use of it... (8k rpm)

for temps im not quite sure about it however can you monitor this?...

[Hosser]

thanks!

noise is tolerable when you get to use of it... (8k rpm)

for temps im not quite sure about it however can you monitor this?...

I doubt there are any software monitoring methods available; could use a finger maybe? Check to see if they still get hot?

I left it as it was defaulted to. Maybe KTE can explain more or why or why not I have a pci-e 2.0 card so I do what it to run the correct bandwidth.

HPET is seperate from PCI-e 2.0 and not required for its operation.

[madfaze]

I doubt there are any software monitoring methods available; could use a finger maybe? Check to see if they still get hot?

nah its not that warm anymore...

[KTE]

tictac: what was that about left shift+tab F1 right shift+tab F2 or something you were saying... ?

anyone has an idea on what size are those circupipes?
http://www.sunonusa.com/pdf/MMF-AD-PR.pdf
you think this would works?

Hehe, nice find. But they give no specs we're after there.

You only need one slow big fan really, low noise and moves decent air.

His username is: d412k5t412

Should be approved very soon.

KTE do all 4 pci-e gens settings need to be enabled? or just the hpet setting?

Only the PCIe you use (1 usually) needs to be put to Auto and HPET timer is best disabled.

noise is tolerable when you get to use of it... (8k rpm)

I have a 60mm 8k RPM fan, I know damn well how much screaming noise they make!

Why not mount a 1k RPM 120mm fan near the back of the case instead, low noise, good cooling? You guys over in Asia do have far higher temps than we do here though, so you need it much more.

for temps im not quite sure about it however can you monitor this?...

Stick a thermocouple inside the slots in the MOSFET heatsink until you feel it hits a MOSFET. Be running something like P95 at that time, the temps will fly high, usually plus 65C (which is still OK).

BTW that above setting I had to switch off (game) at 20hrs P95 stable ambient varied from 30-20C:

[KTE]

Have been trying to finally find high end stability at 22C ambient...

2.8G (200x14) boots easily at 1.35V but it's unstable.
1.40V = unstable
1.450V = unstable
1.50V = unstable
1.56V = unstable

Never tried more, this is air, and for air even that was too high. MHz limit @ cooling (stock cooler).

As a test I tried 1.47V at 2.3G (idles 35C, max fanspeed) and it was also unstable. So we see, Phenom hates high voltage on air at these ambients, at least mine does. 1.456V was max which would run without giving errors.

Tried (207x13.5) 2.794G next;
1.40V = unstable
1.450V = unstable
1.488V = unstable

Tried (206x13.5) 2.78G next;
1.40V = unstable
1.450V = unstable
1.488V = unstable

Stopped right there. Previously I've had upto 2.756G fully stable. So now I'm trying the same MHz again to see if it is still stable.
Right now testing 1.39V idle/1.368V load and it's 2.756G stable.

I'll drop volts next to find lowest it's stable at... but you can already see that at 1.27V easily stable for 2.63G, for just 135MHz more (+5.13%) you require more than +0.2V (1.488V -- +17.17%) and you still cannot get perfect stability (fails after ~2-3 hours P95 small FFT). That's what you call a processor limit.

On H2O, I do think you can get 2.8G stable with lower volts with this chip and possibly bench +2.9G.

EDIT!:
-2.756G passed 1.392V 5hr and then, whilst gaming, it froze the system = unstable
-2.7G used to be fully stable for anything around 1.320V around 10 days back for over a week that I tested: Original Stability Report
Now, unfortunately it failed P95 at 2hr20min and repeatedly failed playing a simple Youtube video at 1.360V.

This may however have something to do with the temps. Phenom may not like these temps much, before I was using the same cooler but with a good mount and solid paste. Now I was idling over 10C higher since the paste was 1/3 of original left and reused. Will throw my Zalman onto it to check how it does... 50-54C are not temps which will make Phenom core error, so if it is finding instability at >50C it'll be because the core temp probe is reading minimum -10-15C from the real value. This'll have to be tested...

Coming up next...

At what <monitoring tool> temperatures does a stable Phenom at stock start to error?

[jonspd]

Coming up next...

At what <monitoring tool> temperatures does a stable Phenom at stock start to error?

What are you gonna do add heat


Great testing you are still doing to this day. Glad that someone is still trying to figure out the ropes of this quad...

[Manicdan]

or just turn down the fan speed

[KTE]

Or just... have no fan at all.

Heatink with some passive air keeps it relatively "up there" IMO.

[tictac]

or hair dryer on top of cpu fan.. hehe..

[KTE]

Well well.. we're making good progress.

See the above 2.6G 20hr stable?
I run that setting daily... replicated it just now at 30C, 35C, 40C, 45C, 50C and 55C BIOS (70% load)/EVEREST -> results:

30C idle/load -> stable
35C idle/load -> stable
40C idle/load -> stable
45C idle -> shaky
45C load -> stable
50C idle > no boot!
50C load -> shaky
52C idle -> no boot!
52C load -> mostly errors
55C idle -> no boot!
55C load -> full of errors!

Shaky = Errorful on and off but altogether unstable (freezes/hard locks)

We have established a temperature bottleneck with Phenom guys. Achim (justapost) experienced something very similar and I've never had high temp readings to be able to tell but today my temps were simulated as high as his were before and what he found is exactly what I'm finding...

Freezing/hard locks are known as a very possible temp issue on any setup.

I could not measure the IHS temp nor did I have probes with me but I used the normal method made for us to use: software monitoring tools. The temperatures reported were the above with software/BIOS.

However, I do not think that was the real temperature. To me it looks highly likely MSI board/Phenom is reading 10-15C below the actual core temps as we know and there is no CPU I've seen which fails POST 20C below max danger temp limit. In Phenoms case, that's 70C, anything below that (real) should run fine. Those temps of 50C must have been 60-65C real IMHO.

Jack (JumpingJack) briefly mentioned relatively high stock temps with his Phenom sample too and that I couldn't corroborate with Gigabyte/MSI/DFI+Phenom testings but I kept checked around and found two other ASUS+Phenom pairings and they both idled/load pretty high up according to software at same MHz/Volts as us, like him. Thus, to me it seems highly likely that the ASUS boards have the correct temp. calibration for Phenoms and that the MSI boards show 10-15C below actual. Otherwise, 50C will never fail POST and freeze whilst in the BIOS.

One other thing I tested but could not measure for lack of equipment with me was MOS/inductor cooling: yes you do need it or you'll fail stability testing at stable MHz easy. Especially at load, typically lower ambient/PWM/CPU temp will get good stability it seems. Guess I better get my skating shoes on.

Gack I was certainly surprised.

[Manicdan]

how do u get it to be at 30C load? if its not having problems reading the temps, then your doing a great job cooling it, or is that some kind of weird chiiller setup you used to do that

[KTE]

-2C ambients 3x Delta 220CFM 1x Delta 102CFM 1x 20inch fan do the trick very well.

I'm thinking of testing it under more cold now. Depends on how the Wolfdale pan out. If the new ones they send us are also bugged with temps then will have to stick with this yet because my boss won't accept it.

My last 9600BE and 9500 didn't have this temp problem. I'd been upto 60C software reading with those stable.

I just booted the 9600 BE while ambient was 27C, very interesting that it booted up faster than usual straight into the BIOS inside I would say 3 seconds and it read 19C (and moving up quick) in HWM within the BIOS. This is very likely 29C actual, thus a -10C deficit since processor started off at 27C (ambient) and can't be less. I modified my 9500 to verify its temps but I don't want to do the same with this.

[d412k5t412]

Wow, Thanks KTE for helping me get privledges!!! I've been reading this forum for a while now and I've found it very helpful!

But I just have a broken 9600BE or something because I can't overclock worth anything. The most I've booted was 12x on the multiplier. Thats it. I dont know what I'm doing wrong. I'm back to the 1.1 bios. Didn't like the beta ones. Maybe you guys can help out!

Heres my rig:
MSI K9A2 Platinum 1.1 bios
Phenom 9600BE 2.3ghz (stock)
eVGA Geforce 8800gt SC
2x2gb G.skill DDR2-1000
Seagate Barracuda 7200.10 250gb
Antec Earthwatts 500W PSU

i think thats it...

[Daveburt714]

D4!!

Pretty sure everyone would agree the 1.13 Betas are by far the best Bios, although I haven't used the 1.1's, I did try 1.2 and went right back to 1.13..

What the... Tornado Sirens are going off here, better check the weather channel...

[KTE]

Wow, Thanks KTE for helping me get privledges!!! I've been reading this forum for a while now and I've found it very helpful!

But I just have a broken 9600BE or something because I can't overclock worth anything. The most I've booted was 12x on the multiplier. Thats it. I dont know what I'm doing wrong. I'm back to the 1.1 bios. Didn't like the beta ones. Maybe you guys can help out!

Heres my rig:
MSI K9A2 Platinum 1.1 bios
Phenom 9600BE 2.3ghz (stock)
eVGA Geforce 8800gt SC
2x2gb G.skill DDR2-1000
Seagate Barracuda 7200.10 250gb
Antec Earthwatts 500W PSU

i think thats it...

Welcome.

I would move over to the 1.13 BETA as Dave suggested, it is much better for oc and will give good performance compared to the others.

Have you modified VID/Volts with your oc's yet?
What about trying the HT ref.?
What CPU cooling are you using and what are the case ambient temps like?
What is your stock VID/Voltage?
Does AOD work for you?

When you oc, leave the RAM in Ganged DCT mode for starters.

Dave, if a tornado strikes be sure to pack off your Phenom to me before that!

[madfaze]

does pcie speed lock @ 100mhz on this board?cant see any option at the bios...

[Daveburt714]

Dave, if a tornado strikes be sure to pack off your Phenom to me before that!

That was wierd!! I'm no pup, and I've never seen weather like that in February before, it actually got pretty scary for a few minutes..

D4, the thing that helped me most was lowering CPU VID to 24 (1.25v). Stock on my machine was 28 (1.20v). Every Phenom seems to have it's own personality, but If I had a BE I would leave CPU VDDC at Auto and just lower CPU VID when needed, although I'm not sure anything under 18 (1.325v) would offer much extra...

KTE posted a table of VIDS/FIDS a few pages back (Very Helpful!)

I actually typed a more detailed message but my machine locked up...
I've been running these settings for a few days and I really thought it was stable...

[KTE]

does pcie speed lock @ 100mhz on this board?cant see any option at the bios...

At boot-up, yes.

[madfaze]

At boot-up, yes.

how about when you oveclock?!

is it still locked @ 100mhz?

[KTE]

Yes, unless you change it manually.

[madfaze]

Yes, unless you change it manually.

are there any choices @ the bios? i cant seem to find one...

[Daveburt714]

I was gonna start a new thread, but I'm buzzed again, so I thought I would run it by you folks in here first... Your all cool people!!

There are 3 things that the Phenom Platform needs:

1. A better power system for the Mobo's, voltages fluctuate alot, and Phenom seems to be real sensitive to that.. I think the K9A2 keeps the MOSFETS cool enough, but those Inductors get way too hot!! (I'd consider an aftermarket option...)

2. A little more work on the BIOS, especially the option to disable the TLB Fix... And I mean totally disable it!! TLB will go down in history as the most overblown eratta in history... I can see where it would affect Servers, but on the desktop? I can't say for sure, but I don't see myself ever running Virtual Machines...

3. We need a piece of monitoring software that shows ALL the options, NB/CPU VIDS/FIDS, NB/CPU/HTT Multis, the VDDC Tweaks applied, HTT Ref./Volts, temps from all the sensors and what they apply too... I bet the guy who did CrystalCPU could whip this out in a few days..

There are so many options on Phenom, #3 may very well be the most useful, at least you could send one ss that would show folks all your settings..

@ MADFAZE, I really don't think the PCI-e clock is gonna make much difference anyway, I know it didn't do anything accept cause problems on the K8, so I haven't even messed with it on K10 yet, but If you want to play with it, I'd be interested in the results...

[KTE]

No PCIe choices in the BIOS's so far madfaze.

PCIe speed will improve perf. in GPU benches. Up-to 140MHz I know will show gain but use an IDE drive if you move it past 120MHz and not a SATA.

Dave, I reckon this board needs some revamping. The PWM area is weak for plus 2.7G oc with Phenom IMO and then starts to need much extra cooling.

The options you're wanting to see can all except NB VID/Multi be seen in AOD you know. Even CPU VID/Vcore is shown there. But the best half of what you mention is the multi's and VIDs selectable within the tool and shown in digital form. That would be best.

My Phenom reports wrong temps now, this is a new thing since the last couple of days. Idling at 2kRPM fanspeed 1.272V 2600/2400/533 (MHz), it is showing 24C idle across all cores whilst the room ambient is 25C and the heatsink ambient is 28C. I would call that a minimum real temp. of 35C of all cores since a) core IHS is always hotter than heatsink ambient b) at over 70W TDP in even idle c) core internal is always minimum 10C even hotter than the Tcase d) most parameters including voltage is increased much so imagine what the stock temperature readout would be -> sub-20C for sure. The image speaking most words =>




THUS from now on, I'm only going to report all temps +10C offset from all s/w readouts.