Phenom stepping/batch overclocking database

[SocketMan]

I am wondering how AMD could "cook up" so many 9850's
in just 1 week,the 0810 (The tenth week of 2008)


Also the last 4 letters on the Littleboy's Phenom are APMW(0810APMW), which we don't have in the table.

[KTE]

Also the last 4 letters on the Littleboy's Phenom are APMW(0810APMW), which we don't have in the table.

Mine is

[duploxxx]

got a new stepping

hd985zxaj4bg
jaafb aa 0811bpjw

will try on asus 780 H-hdmi and 790 msi k9a plat board and put it against 5000be

[JumpingJack]

You probably should not enter this into the database just yet, not my top clock.... also, very minor test for stability -- not prime95 tested.

CPU: 9850
Batch: JAAFB AA 0810 GPBW
Clock: 3.0 GHz
Voltage: 1.30 (stock VID -- KTE likes to report with droop -- 1.296 )
http://valid.x86-secret.com/show_oc.php?id=357500

EDIT: Stability update... no problems, several runs of CB10, UT3, prime95 (about 10-15 minutes) ... just no issues. Runs a bit warm, but expected at this clock.

[KTE]

^ Your CPU 1.296v stock idle is 1.296v under load too Jack?

1.296v is stock Vdd supply for 9850 BE, vdroop is different on boards, some overshoot, some under, some you can disable [not recommended] -- TDP is decided by vdroop Vdd supply, as will stability at a fixed current.

Also, please test with some uptime - preferably 10-15 hours minimum with something light running, such as "stability tester" which runs Pi looped on any number of cores you decided. I'd like to see if you get any idle freezing at high MHz because many of us are benching 3G 1.296-1.350v but cannot get this daily stable at all.

[JumpingJack]

^ Your CPU 1.296v stock idle is 1.296v under load too Jack?

1.296v is stock Vdd supply for 9850 BE, vdroop is different on boards, some overshoot, some under, some you can disable [not recommended] -- TDP is decided by vdroop Vdd supply, as will stability at a fixed current.

Also, please test with some uptime - preferably 10-15 hours minimum with something light running, such as "stability tester" which runs Pi looped on any number of cores you decided. I'd like to see if you get any idle freezing at high MHz because many of us are benching 3G 1.296-1.350v but cannot get this daily stable at all.

I just threw it into the socket a couple of hours ago.... just a quick run up with multi to see if 3 was doable... to my delight .. 3 GHz was simple, stock V -- I haven't put it through the full ringers yet.

I will set up the test tomorrow (idle freeze) ... right now I am measuring power at stock, VID is 1.3 V, CPUID reports 1.296 idle, and I am now loading with OCCT and it is still 1.296... weird.

It really, really sucks the power though ...

9600BE (stock) full load drew 289 W (using a 8800GTX -- power sucker -- also this is from memory, need to find the graph), the 9850BE (stock) full load is drawing 333 W, this is using OCCT to stress. (EDIT: 289 W is correct).

This is same everything -- just pop the 9600 out and pop the 9850 in....

EDIT: Added the power curve -- loaded with OCCT 2.0 for 20 minutes, the app idles first minute, and last 4 minutes. Both processors on same platform (asus M3A32-MVP deluxe, 8800GTX, 2 HDs, one optical, Zalman 9700 fan).

[QuietIce]

M3A32-MVP
According to CPU-Z & AOD I have the same idle and load vCore too (1.344), which is different than the BIOS setting (1.3625).


Finally running at 200x15 two solid days now (24/7) under SETI load. No bad work units as yet - all that have been validated (about half) as good. HT Bus was turned down to 1.4 GHz (NB still at 2.0 GHz), increased CPU-NB to 1.3375v, and increased Mem vDDQ from 1.8v to 1.93v. Stock vDIMM for this RAM is 2.10v and I guess vDDQ needed to be a little higher because of that ...

[KTE]

Thanks for the data

Yup, very similar to mine. Difference between 9500, 9600, 9600BE and 9850BE for me in the same system was (roughly from the top of my head, can't access excel here);

Model - Measured Wall [CPU Max set in register]

9500 - 165W AC [89W max TDP]
9600 - 180W AC [93W max TDP]
9600 - 185W AC [95W max TDP]
9600BE - 175W AC [91W max TDP]
9600BE - 185W AC [95W max TDP]
9850BE - 215W AC [121W max TDP]

Tried to find max load giving app after 8 hours, used S&M finally. Current supply is also different between them, hence the TDP changes.

MSI 790FX 1.296v idle = 1.288/280v load
GBT 780G 1.296v idle = 1.316v load
GBT 740G 1.296v idle = 1.320v load
Abit 770 1.296v idle = 1.296v load

Haven't tried B3 in other boards that worked.

[JumpingJack]

^^^ Haven't tried S&M yet (same utility that Xbitlabs uses as I recall), I posted another profile a few days ago in another thread looking at a few apps. OCCT pushing to within 2-3 Watts of Prime95, but from all that I have tested (except S&M), Prime95 reigns supreme on this chip.

[KTE]

I'm trying the same tests. Let me know what you find

[Particle]

Update needed: My OC was tested for 34 hours--not 2.

[Nedjo]

LIKMARK could you add column for OS. I think it's important knowing is OC done under 32bit or 64bit Win (Vista).

[QuietIce]

Water cooling to come nexy weekend (I hope) so this may get better!

Model: 9850BE
Batch: JAAFB AA 0810GPBW
Speed: 3000 MHz (nominal)
Vcore: 1.344 (by CPU-Z; 1.3625v BIOS)
Motherboard/chipset: M3A32-MVP
Cpuid validation url: http://valid.x86-secret.com/show_oc.php?id=359072
Cooling: Air

Over 100 CPU hours of SETI work units validated by other machines in the past 5 days.
No off switch!->

Mem vDDQ = 1.93; DDR-800 (4-4-4-12) vDIMM = 2.1v stock voltage.
Wouldn't stay stable above ~2.9 GHz with a 200 MHz clock until I turned up the Mem vDDQ.


Suicide run for grins

[LIKMARK]

LIKMARK could you add column for OS. I think it's important knowing is OC done under 32bit or 64bit Win (Vista).

Os can be included. Update coming soon, I'm a bit busy atm, got some area outside my house that's supposed to become a garden this summer sorry

[Aussie FX]

Jack those temperatures look high for the 9850. Are they at load?

I have my 9850 @ stock now (2.5) and it's showing 43°c at load.

[JumpingJack]

Jack those temperatures look high for the 9850. Are they at load?

I have my 9850 @ stock now (2.5) and it's showing 43°c at load.

Yeah, they are at load... also I would not trust the temps.... I have no idea how OCCT is calibrating their numbers. EDIT: also, when I look at the actual numbers it is around 48-49 deg C, and 5 degrees could easily be explained by HSF used or ambient (especially ambient), I had 7 rigs fired up (5 quads, 2 singles) all at the same time -- the room was toasty

Also, this was taken within literally 4-5 hours of installation and the TIM likely has not had a chance to settle.

The power is higher but not unmanageable, and it is inflated by an nVidia power sucking GPU. The proc itself runs like a charm, and I can get 3.0 GHz easy peasy at stock.

Jack

[Oldguy932]

Come on guys, we need to try and keep this thread going.



There's my overclock on my 9850, stepping is in my sig. Just won 250 at afterprom yesterday so I might head over to microcenter later and pick another one up...

[QuietIce]

Nice! I may have to give that setting a try ...!

[Oldguy932]

I did get another 9850, does 2.9ghz cpu, 2.32ghz nb stable with 1.3v on each, haven't found anything past that yet. 3ghz boots with 1.3v, but isn't stable. Not sure what's causing the problem yet, setting the ht link down to 7x doesn't do anything really. And its a 811 CPFW, but I'm not even going to take a screenie till I beat my other 9850.

Edit: Nevermind, 3ghz cpu, 2.4ghz nb works at 1.325v on each. Looking to be a much better clocker than my other one, just hope asus can release an even better bios than the one we have right now.

[Particle]

I've been running at 2.8GHz using 1.34V for the past 190 hours without turning off my machine. It does everything without any stability problems, including games, programming, video encoding, etc.

[JumpingJack]

M3A32-MVP
According to CPU-Z & AOD I have the same idle and load vCore too (1.344), which is different than the BIOS setting (1.3625).

This is normal -- there is always a voltage drop when the supply is applied to a load. Some boards will 'correct' for this to maintain a constant Vcore, but that takes additional circuit/monitoring and engineering.

[QuietIce]

This is normal -- there is always a voltage drop when the supply is applied to a load. Some boards will 'correct' for this to maintain a constant Vcore, but that takes additional circuit/monitoring and engineering.

I expected the slight vDrop between BIOS and CPU-Z. What I didn't expect (and have seldom seen) is a 0.00v droop when all four cores are loaded. Even the A8N32's (also 8-phase power) have ~0.013v droop per loaded core ...

[Oldguy932]

My M3A32 board drops down one setting from the bios to windows, but only has vdroop on certain settings that I've noticed. Most settings there is no drop however.

[The Weedman]

Unfortunately this cpu has 2 pretty much bum cores for overclocking but i was able to bench 2 cores at 3.2ghz with the other 2 at 2.7ghz. Heres the 3dm06 with a 3870x2 http://service.futuremark.com/compare?3dm06=6754455

Right now i am running it stock at 1.2v while i wait for another 9850, which should come tomorrow

Model: 9850BE
Batch: JAAFB AA 0811BPKW
Speed: 2.5ghz
Vcore: 1.2v
Motherboard: M3A32-MVP
Cpuid validation url: http://valid.x86-secret.com/show_oc.php?id=364406
Cooling: W/C





Edit: Looks like i got a 0812GPKW so this weekend i will see how it clocks. Hopefully the newer ones perform better.

[svende]

Phenom 9600 Black Edition
HD960ZWCJ4BGD
CAAWB AA 0748MPMW
1939104L70407

DFI LanParty DK 790FX-M2RS
Air cooling: Pentagram Karakorum

max screen: 3264,6MHz 1,536V



max stable: 2940MHz 1,424V 4h primestable