Phenom 9850 OC Discussion

hey sup all, I got my 9850 running at 3ghz stable.
I was wondering what others have it at?
I'm running 232 X 13 Multi and 1.32 Vcore.
Temps idle @ 30 & Load @ around 39-42.

I also cant run my old ballistix 800 ram at 1080 speeds like I used to do on my 6000 for some reason.
If anyone else had 9850 clocked post your results.
Sorry if theres a thread already, but I didnt see one.

Hopefully, I'll be posting preliminary results here in the next couple of days. All the parts are here except a good cooling system but I can't wait. I've decided to throw together an OK cooling solution and let 'er rip ...

Wow, nice OC.

I dropped one into a Gigiabyte MA780FX last night. First boot locked up on the DBI success message but the a reboot solved that issue. I ran it for 4 hours doing 35 mins of 2X orthos, then surfed, played a game, worked with a roller coaster design and simulator with no issues all at stock. Just basic testing to check for initial stability before OC. I closed those apps started folding and hit the sack. Checked it this morning expecting to see it happily crunching away but found the log in. Thought OK, stupid windows update must have somehow started but a quick check of the event log showed it rebooted about 2+ hours later with no event recorded. This is usually a RAM type crash and reboot. I don't think the Phenom is at fault but this board has never been right. I'll discuss the rest in the MOBO section.

Stock for this CPU is 2.5G 200x12.5 with RAM underclocked around 357.
Vcore at 1.275.
Idle temps 40C, ambient 70F.
Load temps 52-57C.

Last night, I pulled the HS and put AS5 on with just a little lower results 43C but after a night at idle and some messing with timings this morn it dropped to 38C which I see as close.The small dallop trick with OCZF may not work but I'll try it on another (stable) rig.

I'm still unstable, even under clocked. I dropped it to 2.4G and ran games/graphics all night with no probs them started FAH it rebooted after about 20mins. This is just like it the 9500 and the board I RMAed back in Dec. Intel is starting to interest me again
This morning I found it idle at the Win login. I did a little tweaking and bumped CPU volts 2 notches above norm 1.261 to 1.31 (PC-PNC overvolts a little). I set the ram to SPD spec and DQS on. Reboot failed so I set the RAM back to auto and DQS off. Booted fine. Checked temps and it is now idle at 38C did some email and surfing no isues there, turned it off. Tonight I'll see what it does. I'm wondering if the 9550 and 9650 are better silicon then the 9850BE. I would be shocked to hear that. I went with the 9850 so I would only need a small bump in oc to get near 2.8 or even higher. I may just get another board and send this $250 POS back to Gigabyte.

Must be that gb board.

New word MOBOSUCK! Tried running 200x11 (2.2G) ran just 15 mins more and rebooted. I do not recommend Gigabyte high end boards to anyone for Phenom. To be fair I will say if you are jumping over to sintel then their DS3P and DS4 boards are the best two for a Q6600. I was close to the point of selling all this off and going with a DS4 and sintel chip but I gave AMD another chance and ordered the DFI M2R 790FX. I did not like the pollution of intel posts we had in the AMD threads when we were trying to get the first chips running but it really comes down to the experiences. I dropped a Q6600 into the DSP3 mobo warmed it up at 2.6G for a few days and then hit the gas running it up to 3.3G where it's been for months. I did have a network issue that I thought was hardware but turned out to be a zonealarm issue blocking the port randomly. It now folds 24/7, burns CDs, prints and other production stuff. I really expected to see the Phenoms do this and I expected that with B3 I would be back in business but so far no go.

Update;
Update, Found some RAM timings on the GB790 that needed attention after memtest threw errors. I think these dimms 2x2G need 127ms for the refreash where the board thinks they are 107. Also I found something calls Express Tweaking for AM2+ or something like that. I enabled it and rerunning a full memtest,

Heres what I got so far....

Wicker Man, is that vcore being correctly read by cpu-z? If so, that is looking great man.

Do you think it still need 1.312 or will it run at 1.28?

muddocktor said:

Wicker Man, is that vcore being correctly read by cpu-z? If so, that is looking great man.

Click to expand...

The 9850's are rated at 1.20-1.30v instead of 1.35v like we're all used to for AMD's. Still a VERY respectable showing especially at that vCore! Hope mine works out that well ...

That is correct.

I got it 100.16% I think, I would have to look to make sure, but it the first option off of default....

I took the GA790FX for one last overclocking spin today. 2740 was the highest OC I could get Win boot stable. I did manage to stabilize ram a little getting these Corsairs up to 840 at 5-5-5-18-24 on 1.95V. I've been running at 2600 stable enough to run Prime for an hour but folding brings it down in 5 mins still. I'll just hold of for the DFI now.

My 9850 seems happy at 2.8GHz with 1.34V. It refuses to be stable at 2.9GHz at any vCore up to 1.51V, so that looks like it's going to be the max. At 2.8GHz I've ran Prime95 (64-bit) for over a day now without issue. At 1.30V it would crash eventually.

Because of the odd behavior between 2.8 and 2.9, I plan to bump individual cores to 2.9 and see what I can get away with. Load temps are 61C, which is 37C above the room's ambient and probably half of that above the case temp. A good cook test really lets you know if something is going to be solid.

Particle said:

My 9850 seems happy at 2.8GHz with 1.34V. It refuses to be stable at 2.9GHz at any vCore up to 1.51V, so that looks like it's going to be the max. At 2.8GHz I've ran Prime95 (64-bit) for over a day now without issue. At 1.30V it would crash eventually.

Because of the odd behavior between 2.8 and 2.9, I plan to bump individual cores to 2.9 and see what I can get away with. Load temps are 61C, which is 37C above the room's ambient and probably half of that above the case temp. A good cook test really lets you know if something is going to be solid.

Click to expand...

61C is a little high. Of all the CPUs I have ever owned, none would be stable at that temperature. Usually anything above 55C is already pushing it.

Get better cooling will let you clock higher than 2.9GHZ stable.

Yeah, what OC101 said.

I would dang near put money on the fact that you are limited by your cooling on this one. It won't be stable bumping the volts up more because you are just pumping out more heat and it is already pretty high.

What sort of cooling are you running? Have you tried reseating your HS?

I would agree with your suspicions, but I get the same temps no matter if I use 1.34V or 1.51V. Either way I'm ~61-65C. It's just a rather hot room and the air around the mosfets and CPU is stagnant and hot as well. I use an Arctic Cooling 64 Pro and IC Diamond paste.

Particle said:

I would agree with your suspicions, but I get the same temps no matter if I use 1.34V or 1.51V. Either way I'm ~61-65C. It's just a rather hot room and the air around the mosfets and CPU is stagnant and hot as well. I use an Arctic Cooling 64 Pro and IC Diamond paste.

Click to expand...

It is possible that your chip just doesn't clock well. However that does not mean that your cooling is working all that well either. To put it simply 61-65c is just to hot. Most of the air cooling setups that people have given temps from so far are usually in the low 50s at full load and idle in the 40s (not a lot of info to go on here but that is what I have seen while looking around).

I am honestly not sure if the AC64pro is a sufficient cooler for the 9850, at stock maybe but when OCing I just don't think that it would hold up. So far most of the air cooling setups that I have seen being talked about are the Tuniq Tower and the U120E, even on the stock cooler at stock volts I believe there was a reported temp of 55c with an OC to 2.8.

My advice is to get a better cooler to start with, also work on increasing the airflow through the case (the air should not be stagnant), you really shouldn't be running at temps that are that high.

Well, after sweating through several hours of M3A32 BIOS and countless reboots I've finally arrived at what I think is a stable OC. I checked it for about an hour w/4xPrime95 then switched over to SETI. While I believe SETI to be a better stability test, it also takes time for the work units to be validated by another machine.

My approach was just the opposite of "normal". Since so many people were having problems with the NB at speed I decided to down-clock and crank up the CPU multiplier instead. Things are getting a little hot with the air cooler, though, and something else is getting very hot as well but I need to find out what "TmpIn3" on AOD refers to on this system.

Here's a wild suicide run that managed 3.2 GHz - on air:
http://valid.x86-secret.com/show_oc.php?id=352629

Water is due in sometime next week so maybe next weekend I can crank it up some more ...

Here is what I ended up with since I can only make 2.8GHz while synchroclocked. Considering this, I think it is worth while for people even with "lemon" chips to investigate what their individual cores can reach. You might only have one gimpy core.

I know core 0 is the worst performer followed by core 1. Cores 2 & 3 seem to be limited by CPU temps. I'll leave it as is for now and play with the non-sync after I get WC next weekend (I hope). If it is a temp issue the Fuzion will take care of it ...!