Phenom stepping/batch overclocking database

[aGeoM]

For comparison

9950BE at 3000 (15x200) 1.3625V

Same temps running everest stress test.


Higher temps with prime95 and coredamage(60°C)

EDIT Prime95



Coredamage 45°C

[LIKMARK]

Am I missing something from my post to not be added?.. I it was updated since I posted...

No you are not, it's just me postphoning updating it. Sorry

[charged3800z24]

No you are not, it's just me postphoning updating it. Sorry

OK.. I just wanted to make sure.. So I could add it if needed.

[justapost]

EDIT Prime95
Coredamage 45°C

Thanks for the comparison aGeom, I rearranged the D-Tec and TIM a dozend times today and settled at ~55°C Coredamage temp (worst 72°C). My water temps are ~30°C so i think there is still room for improvement.
1.4825V seems to be a chip or board limit here. I can run 3.2GHz fine at 1.48135 but the system pwoers off the screen after few minutes coredamage at 1.5V.
Can be the temps are the limit with bad adjusted D-tec the system reached 70° at 1.4V and powered off at 1.425V.

[Psychlone]

Model: Phenom 9850BE
Batch: JAAFB AA 0316GPMW
Frequency: 3.15GHz
Vcore: 1.32V
Motherboard/chipset: M3A32-MVP Deluxe Wifi/AP / 790FX
Cpuid validation url: http://valid.x86-secret.com/show_oc.php?id=389547
Cooling: ThermalRight-120 Extreme w/ 2 Scythe S-Flex 1600RPM fans






This stepping has proven stable with a 12 hour run of 2 instances of Orthos StressPrime at priority10 at 3.15GHz, 1.32V - with full-load temps of 50*C, which is a bit higher than the other 3 9850BE's I've tested, but I'd bet that after the AS5 cures a couple more days, that it'll drop down to a reasonable level.

Psychlone

**EDIT: 24 Hour run of dual instances of Orthos StressPrime completed - Full load is now 47*C

[cdawall]

phenom 9500 CAAZB AA 0745GPIW
@2807 (255x11) 1.37v
Biostar TA780G M2+

http://valid.x86-secret.com/show_oc.php?id=389701

only done one mod to the mobo so far



noww the HD3200 is running @750mhz

[daltona]

Model: Phenom 9850BE
Batch: JAAFB AA 0810 BPMW
Frequency: 3.214GHz
Vcore: 1.37V
Motherboard/chipset:Asus CrosHair II/nVidia 780A
Cpuid validation url:
Cooling: Water

[daltona]

Model: Phenom 9850BE
Batch: JAAFB AA 0810 BPMW
Frequency: 3.315GHz
Vcore: 1.40V
Motherboard/chipset:Asus CrosHair II/nVidia 780A
Cpuid validation url:
Cooling: Water

[Particle]

We need an update. It's been nearly a month and a half.

[daltona]

Model: Phenom 9850BE
Batch: JAAFB AA 0810 BPMW
Frequency: 3.415GHz
Vcore: 1.41V
Motherboard/chipset:Asus CrosHair II/nVidia 780A
Cpuid validation url:
Cooling: Water

[leoy]

@ Daltona
Up to what frequency is this stable?
Is this the first phenom trying at this mobo?

I have the crosshair and am very anxious about buying a phenom...
I am afraid it won't do this good.

[daltona]

@ leoy
For now its stable at 3400mhz. have been running seti@home for 20 hours now, and it's still going.

I have tried a 9600 but it did not clock very well.

No, there is no garanty, but then again maybe you get a better one.

[Final8ty]

@Daltona that is one of the best chips so fare..N1

[QuietIce]

For now its stable at 3400mhz. have been running seti@home for 20 hours now, and it's still going.

I have tried a 9600 but it did not clock very well.

No, there is no garanty, but then again maybe you get a better one.

Excellent OC!!!

Your 24 hour output should be in the lower 5k's at that speed ...

[Psychlone]

Model: Phenom 9850BE
Batch: JAAFB AA 0316GPMW
Frequency: 3.485GHz
Vcore: 1.375V
Motherboard/chipset: M3A32-MVP Deluxe WiFi/AP / 790FX
Cpuid validation url: (see below)
Cooling: TRUE-120 w/ 2 Scythe S-Flex 1600RPM fans

Optional:
3.485GHz Orthos StressPrime Priority10 stable for 10 hours


Here's what's possible with a good LOW-Voltage CPU...I'm starting to think that some Phenoms are much like the dual-core Opterons in that a few of them don't require much voltage to go much further (just like Opterons!)
At the stock 2.5GHz speed, I can pass 10 hours of Orthos StressPrime Priority10 at 1.14V (this is it's lowest stable voltage at stock speed)

http://valid.x86-secret.com/show_oc.php?id=391337


http://valid.x86-secret.com/show_oc.php?id=391343


http://valid.x86-secret.com/show_oc.php?id=391344


Now, 3.15 is only drawing 1.32V, but to get to 3.485GHz, it took 1.375, and to get to 3.51GHz, it was 1.45V - so you can begin to see the limitations of the motherboard and CPU combo, along with the point of diminishing returns happening right between 3.4 and 3.5GHz. Even though it's (roughly) 96% stable at 3.51GHz, it did BSOD just after 8 hours of Orthos StressPrime at priority10, but that's not all - pumping 1.45V though the CPU was idling at ~55*C and full-load was 62*C, well out of my comfort zone.

So, I've settled at the 3.485GHz at 1.375V for now...idle at 46*C and full-load at 55*C - well within range of my optimum thermal threshold, and passes 10 hours of Orthos StressPrime priority10.

I'm doing a bunch more tests to see what different variables do to the bandwidth and throughput, but for now, it spits out SuperPi 1M in 22.039 seconds, and has MB/s in Everest Ultimate Cache and Memory Benchmark of 10311R, 9457C, 11739W (which is just *screaming* fast - fastest I've tested yet!)

Also, when I can get ahold of the new 780G board, I'll be redoing the entire thing again...

Psychlone

[Phenomenalspeed]

Psychlone your batch number is inconsistent will all other Phenom 9850 numbers.
All the ones I have seen have 08 year followed by the week number.

[Psychlone]

Yep, sorry. Appears that I've been confused more than once...between all these different Phenoms, apparently I got mixed up a bit:
It's actually a JAAHB AA 0816 GPMW



And, here's one with 3 of them together while the 4th (which is actually the 3rd one) was still in my computer:




Sorry for the obvious mistake, I need to go around the net and change that info and my sig in several places. Thank you for pointing that out!

And, so far, my 3.485GHz has now passed 15 hours of Orthos Stressprime priority10 - but for some reason, it's load temp is now at 60*C...not sure if it's my house, as it is the hottest part of the day now, or if the AS5 has been pushed too far...

Psychlone

[Phenomenalspeed]

No problem. It was just weird that you seemed to have one that was made in 2003!

Week 16? Unusual.
You do have an excellent CPU there.

If I'm not very much mistaken I believe the only ones we have seen so far have been week 9, 10 11 or 12. All of a sudden an (apparrent) 4 week gap. Interesting.
Maybe they were on holiday.
On a more serious note perhaps this is not a coincidence that your CPU overclocks so well. Maybe something significant changed in the manuafacturing process. Perhaps CPUs from week 16 onwards are different/better builds. This is all just speculation and cojecture at this point. We need others with Week 16+ plus 9850 Black editions to test.

Has anyone seen or does anyone have a 9850 Black edition with a batch of weeks 13-15 of 2008??

Just out of interest Psychlone how well/poorly do your other two Phenom 9850 Black editions overclock with the same setup?

[hstuehmeyer2000]

It's actually a JAAHB AA 0816 GPMW

well after i saw your first post i was like, wow I wish i could get a phenom like that and i was thinking about buying one anyways so i went to micro center and bought one and this is what i got,


JAAHB AA 0816GPMW

now i just need to find the right mobo for it

[QuietIce]

Model: Phenom 9850BE
Batch: JAAFB AA 0316GPMW
Frequency: 3.485GHz
Vcore: 1.375V
Motherboard/chipset: M3A32-MVP Deluxe WiFi/AP / 790FX
Cpuid validation url: (see below)
Cooling: TRUE-120 w/ 2 Scythe S-Flex 1600RPM fans

Psychlone

Nice OC!
Not that I doubt you at all but with all the other CPU-Z validations listed previously didn't you forget the important one for your stable and running OC?


Hey, I think I figured out why yours is such a golden chip - it's a custom built Phenom from '03 that slipped out ...!!!

[Psychlone]

No problem. It was just weird that you seemed to have one that was made in 2003!

Week 16? Unusual.
You do have an excellent CPU there.

If I'm not very much mistaken I believe the only ones we have seen so far have been week 9, 10 11 or 12. All of a sudden an (apparrent) 4 week gap. Interesting.
Maybe they were on holiday.
On a more serious note perhaps this is not a coincidence that your CPU overclocks so well. Maybe something significant changed in the manuafacturing process. Perhaps CPUs from week 16 onwards are different/better builds. This is all just speculation and cojecture at this point. We need others with Week 16+ plus 9850 Black editions to test.

Has anyone seen or does anyone have a 9850 Black edition with a batch of weeks 13-15 of 2008??

I would have to put in the old addage of the 'Wednesday Car'...the analogy goes like this:
Monday - everyone is still thinking about that party they went to on Sat or Sun and is probably still nursing a hangover - quality down the drain
Tuesday - everyone is starting to get back into the swing, but dreading work - quality is coming up, but not the best
Wednesday - everyone is at the top of their game - quality at max
Thursday - everyone is beginning to think about the weekend ahead - quality lowering
Friday - everyone is ready to get the Hell out of work to party - quality at it's lowest
(...repeat)

I would suspect, like you, that the missing weeks were time-off...perhaps there was a problem at the plant, maybe they all earned vacation, could be that they were tweaking something to make it better...

In any case, in one of the databases that I've seen, there are no steppings from weeks 14-16 (there was a single JAAHB 0813 APMW 9550 that I found while skimming real quick, but nothing at all for 14-16), then picks back up again at week 17 and goes to week 21 (for an X2 4600+) - may be that they weren't focusing on the Phenom plates at that timeframe and were pumping out some other or older cores just to get rid of the dies...I really don't know, and this is speculation.

Just out of interest Psychlone how well/poorly do your other two Phenom 9850 Black editions overclock with the same setup?

Well, let's break them down by stepping:

JAAFB AA 0809APMW (2.75GHz stable at 1.28V but wasn't stable at higher freq without over 1.45V)
JAAFB AA 0810BPAW (2.92GHz stable at 1.25V but wasn't stable at higher freq without over 1.45V)
JAAFB AA 0812EPMW (2.88GHz stable at 1.24V but wouldn't clock higher no matter how much voltage I threw at it)
And then the *winner*:
JAAHB AA 0816GPMW (3.49GHz stable at 1.375V and will go higher, but I don't dare give it much more voltage - this one I'm keepin'! )

Psychlone

[Oliverda]

Psychlone

Great result! Which OS do you use?

[nuts24]

OK, i know my system still needs a few upgrades, but heres my overclock at 72hrs and still going.

Model: 9850 BE (B3)
Batch: JAAFB AA 0812 BPCW
Frequency: 3029.82
Vcore: 1.275
Motherboard/chipset: Asus M2N32-SLI Delux/NVIDIA nForce 590 SLI
Cpuid validation url: http://valid.x86-secret.com/show_oc.php?id=390862
Cooling: Air (Scythe Mugen)

[gOtVoltage]

Model :9950 BE (B3)

Batch : JAAWB AA 0820 BPMW

Freqency: 3450mhzCPU/2600+mhzNB


Vcore: Testing -Pushing the envolope High 1.56v+ (some one needs too)
Note: 3.3ghz @1.31volts default!!!!!!!


Mobo: TA770/RX780/SB600 with Mod Bios 612.XXX) No TLBfix (Disabled) and All options for Optimal speed Enabled.

Cooling: D-Tek 2xBIX3's with 655(non chilled) Idle 26c after 6hours of Hell







I BELIEVE THIS IS GOOD.. i will test high FSB and low Multis soon enuff. AMD is on the right track.

[nuts24]

Hi,

I have a M2N32-SLI Delux motherboard with 2x 2GB 800 MHz ddr2 ram and an AMD 9850 CPU fitted.

I have hit 3.0 GHz with just playing with the CPU multiplier and the system was stable, but when I move the multiplier up to the next level I will hit 3.1 GHz, Windows Vista will work for about 5 mins then I get blue screen of death.

Can any one give me any tips on what else I can set and change & try to see if I can get my system stable at 3.1 GHz or to get a higher clock?

At this time I can not buy a new motherboard, so I am stuck with the Asus M2n32-SLI Delux.

Many thanks for any help on this