OFFICIAL: PhII 920 and 940 overclocking results

[batmang]

So far my PII 920 is at 3.61Ghz on a K9A2 motherboard. I'm still trying to find the right voltage for the NB to make it 100% stable. 3.7 is bootable but eventually BSOD. 3.61 is fine so far. Haven't had time to test it.

 

[Markfw]

Originally posted by: batmang
So far my PII 920 is at 3.61Ghz on a K9A2 motherboard. I'm still trying to find the right voltage for the NB to make it 100% stable. 3.7 is bootable but eventually BSOD. 3.61 is fine so far. Haven't had time to test it.

Same motherboard as I have, but I have unlocked multi, making it way easier. I am now trying 233x16 and 900 DDR (I think that was it) to see if my PPD goes up. I tried 266 (something) that was still 3.7, but it bsod'ed right away.

 

[geokilla]

This isn't suppose to be a comment thread......but.

Maybe you just got a bad batch? You could always get another one when the AM3 Phenoms come out, and then test it again, so it's like letting the 45nm process mature or w/e. Or you can send me your current Phenom II =)

 

[BLaber]

Msi K9A2 has a FSB wall of 250ht , only few K9A2 mobos will get past that HT wall.

 

[BLaber]

Originally posted by: Markfw900
The latest. It seems stable @ 3.7, but its only doing as good as my Q8200 in F@H !!!!

Have you overclocked PII 's NB along with the core as well and if you have to what speed.

 

[Duvie]

With the threads recently showing Q9550's at 3.8ghz stock and 4ghz+ on air......I am thinking the phenom II is definitely underwhelming at this point.....

I will wait to put up Marks OC, since I know we have some tweaking to do thursday night before we call it....

Blaber give me some of your secrets if you have them when it comes to dealing with the NB.

I was thinking max OC stay away from the FSB/HTT.....but for Mark and performance in the folding world I think bandwidth is key and try to use a combination of things to balance OC and performance.

 

[batmang]

Originally posted by: BLaber
Msi K9A2 has a FSB wall of 250ht , only few K9A2 mobos will get past that HT wall.

I don't have that issue, thankfully. I've been up to 272 so far.

 

[SlyNine]

I've built a Core I7 920 at 3.8 ghz+ for my dad, But I used an old Alienware case ( used to have a P4 northwoodC 3.2ghz, then upgraded to a Athlon 4200X2) So when I put the side of the case back on then temps jump fast.

I was actually uping the QPI very easly, during a prime run at 3.8 ghz I decided to put the side back on and the temps shut up to 90C. Before the side of the case the temps were around 65C at the cores under full prime load small units.

Im sure it could have hit 4ghz easly. its right now at 3.5 ghz 24/7 prime ran for 40+ hours ( lol he leaves it on and I forgot it was running) came over to ask him if it ever crashed and he said its still running lol, I was like crap you can turn that off now. Thats also with turbo mode on so I believe one core is clocked at 3.8 during single threaded apps.

I have a Q6600 that no matter what I do cannot get it above 3.1ghz, Its at stock volts at 3ghz though.

 

[Markfw]

Well, I tired 18.5 * 200 (all stock except multi, works great) and then tried 266*14 and then now@233@16. 266 didn;t even get 5 seconds into the boot. 233 is fine so far.

I have PC-8500 memory, so there no problem there, and I have set the memory speed to always be under 1000 ddr.

And I installed windows@3.8, no problem, just F@H doesn;t like it.

 

[Cookie Monster]

Hmm, let me see.

First and foremost, overcloking the HT does next to nothing (and rather cause problems/instability) compared to Intel and raising the FSB like there's no tomorrow. With AMD, the bus speed at default is at 200MHz, with a default multiplier value of 10. So basically your sitting at 4000MT/s. The theroectical maximum for HT3.0 is 5200MT/s, and hence when you cranked the bus speed to 266MHz, it crashed since it went over the maximum limit (unless you mess around with HT multiplier). So basically decreasing the HT multi should impose no performance impact and allow you to raise the bus speed alot more. Most people at XS up the cpu multi instead however. Some also crank up the un-core speeds to over 2.6GHz. (People on water been hitting 3.0Ghz and up).

Motherboards based on SB600 seemed to have a really low FSB wall (~230 MHz), and the ones that show really good overclocking results are from SB750 based boards.

I think theres alot more complexity involved when it comes to overclocking AMD. You have alot of variables to test and record e.g NBv, CPUv, Memv, HTv, NB multi etc etc to find the best overclock. I guess its abit more challenging.

 

[Markfw]

That why I did the multiplier first. 3.7 is the wall, 3.8 won;t work using that. And temps @ 49c mean I don't have that problem.

Got any other ideas ? I won't go over 1.505 vcore.

 

[Cookie Monster]

I guess the SB600 is holding you back here. SB750 based motherboard might able to net you 200~300 more MHz, but with the added cost Im not so sure if its worth it.

 

[Markfw]

Considering how badly its getting beaten at F@H, 500 mhz more wouldn;t come close to my 3.5 ghz X3350 (Q9540) which gets 2400 ppd on the unit this get 1650.

Even 200-300 more wouldn;t let it beat most of my Q6600's !!!! (@3.3-3.4)

 

[Viditor]

Originally posted by: Markfw900
That why I did the multiplier first. 3.7 is the wall, 3.8 won;t work using that. And temps @ 49c mean I don't have that problem.

Got any other ideas ? I won't go over 1.505 vcore.

Most of the "4 GHz on air" overclocks were on 1.55 vcore...

Edit:BTW, I'm not saying you should do that (God knows I wouldn't), I'm saying that it appears that 1.55 is what's required...

 

[SunnyD]

Originally posted by: Markfw900
Considering how badly its getting beaten at F@H, 500 mhz more wouldn;t come close to my 3.5 ghz X3350 (Q9540) which gets 2400 ppd on the unit this get 1650.

Even 200-300 more wouldn;t let it beat most of my Q6600's !!!! (@3.3-3.4)

Mark, you have to also realize that AMD's SSE units are far weaker than Intel's. You can bet F@H is optimizing the code paths to take advantage of a better level of SSE support on the Intel CPU's than they do on AMD's.

 

[soccerballtux]

Originally posted by: Duvie
With the threads recently showing Q9550's at 3.8ghz stock and 4ghz+ on air......I am thinking the phenom II is definitely underwhelming at this point.....

I will wait to put up Marks OC, since I know we have some tweaking to do thursday night before we call it....

Blaber give me some of your secrets if you have them when it comes to dealing with the NB.

I was thinking max OC stay away from the FSB/HTT.....but for Mark and performance in the folding world I think bandwidth is key and try to use a combination of things to balance OC and performance.

Why isn't the anandtech OC article up there?

 

[Duvie]

Originally posted by: soccerballtux

Originally posted by: Duvie
With the threads recently showing Q9550's at 3.8ghz stock and 4ghz+ on air......I am thinking the phenom II is definitely underwhelming at this point.....

I will wait to put up Marks OC, since I know we have some tweaking to do thursday night before we call it....

Blaber give me some of your secrets if you have them when it comes to dealing with the NB.

I was thinking max OC stay away from the FSB/HTT.....but for Mark and performance in the folding world I think bandwidth is key and try to use a combination of things to balance OC and performance.

Why isn't the anandtech OC article up there?

I missed the OC results on page 10...thanks...appears AT has the best though I dont know what stability test they did at 3.9ghz other then running their test suite

AMD has proven in early demonstrations that the Phenom II x4 will offer overclocking headroom similar to the Penryn series. Early production sample processors have clocked anywhere from 3.9GHz on air to 4.4GHz on water

Appears to have been some cherry picked pieces...Most cannot reproduce AT results though they dont say if they stability tested it with prime95 or something..

 

[Duvie]

Gary Key's Blog on PhII 940 not reaching 4ghz

What is really strange is the behavior of the OS and Phenom II X4 940 at the 4GHz mark. We actually have 3.990GHz (19x210) stable for all tests except Crysis and the PCMark Vantage TV/Movies test suite. With that in mind, a simple change to 19x211 for a 4.009GHz clock speed results in the majority of our tests failing. We sometimes have trouble even entering Vista at 19x211, while 19x210 is about 97% benchmark stable. We have tried every possible combination (20x200, memory at DDR2-800, etc.) and even chilled the processor down to 16C and raised processor voltages above 1.7, nothing worked above 4GHz.

appears to be an issue with the OS and vista 64...

32bit ocers will be happy to know with the huge vcore AT tried they were able to get to 4.25...not sure if it was chilling the cpu to 16c (unlikely for most air coolers)....

Appears in their test the Yorkfield hit 4.2+ghz

 

[soccerballtux]

Hey Duvie would you be interested in updating what people here on AT get for their OC's? There was a thread that kept track of who had the highest OC's on the e21x0 Core 2 chips and it turned out to be quite popular. It was also nice to see the spread, median, and mean of the overclocks.

It would save us from having to read every post to get the useful info.

If not I'll start one and keep it updated.

edit: actually doesn't look like we have all that many Ph2 owners here. Nevermind.

 

[Idontcare]

Originally posted by: Duvie
Gary Key's Blog on PhII 940 not reaching 4ghz

What is really strange is the behavior of the OS and Phenom II X4 940 at the 4GHz mark. We actually have 3.990GHz (19x210) stable for all tests except Crysis and the PCMark Vantage TV/Movies test suite. With that in mind, a simple change to 19x211 for a 4.009GHz clock speed results in the majority of our tests failing. We sometimes have trouble even entering Vista at 19x211, while 19x210 is about 97% benchmark stable. We have tried every possible combination (20x200, memory at DDR2-800, etc.) and even chilled the processor down to 16C and raised processor voltages above 1.7, nothing worked above 4GHz.

appears to be an issue with the OS and vista 64...

32bit ocers will be happy to know with the huge vcore AT tried they were able to get to 4.25...not sure if it was chilling the cpu to 16c (unlikely for most air coolers)....

Appears in their test the Yorkfield hit 4.2+ghz

Wasn't there a prior situation with AMD chips very similar to this in which Windows 2000 (or was it XP?) at the time would not boot if the CPU was clocked faster than 1.2GHz (or 1.4GHz) because windows had a timing loop that was required to take more than a minimum number of seconds to process or else windows assumed the CPU was faulty and shut down?

 

[Flipped Gazelle]

Originally posted by: Idontcare

Wasn't there a prior situation with AMD chips very similar to this in which Windows 2000 (or was it XP?) at the time would not boot if the CPU was clocked faster than 1.2GHz (or 1.4GHz) because windows had a timing loop that was required to take more than a minimum number of seconds to process or else windows assumed the CPU was faulty and shut down?

Great memory, IDC! I recall the issue, but never encountered it myself.

 

[SlowSpyder]

This is not all out, I know there is more left in it, but I'm probably pretty close to the limits. I really think my SB600 is holding me back though.

But, right now I'm running at 3680MHz with a 2070MHz uncore. I'm running 1.505 volts to the core and 1.125 volts to the uncore. The chip will run at 3.7+ stable, 3.8GHz usually ends up with the PC locking up and rebooting in the Windows desktop at 1.505 volts. I just don't care to push more voltage as I'd like to find a good speed for the chip to run 24/7/365, and I think this is it. In summer this room gets very warm, 80F+.

Right now I'm running everything in my sig but the 900 watt power supply, I have it sitting here, just been too lazy to install it. Right now I'm using an Antec Neo Power 550 watt unit, otherwise the sig is correct.

So the short version... PhII 940 @ 3680MHz/2070MHz (1.505volts/1.125volts) on an MSI K9A2 Platinum. 8GB (4x2GB) of A-Data Vitesta 4-4-4-12 DDR800. Cooler is a Scythe Ninja Copper.

And I do have to be honest about my stability testing. I'm not too rigorous. I'll run OCCT for an hour or so, if that's fine than I just use the PC... game on it, browse, etc. If there are no problems, than I'm a happy camper. And so far, no problems at those speeds.

I paid $235 for the chip, and bought this motherboard a while ago for $150.

Idle is somewhere around 35C, usually below by a degree or two. Load temp is usually around 47C give or take a degree depending on the ambient temp.

I doubt I'll try for much more core speed, I'm at the point where it takes a lot more voltage for a little more MHz.

 

[lightstar]

Phenom 940
Asus M3A79-T
G.Skill DDR2-1066
Vendetta 2 air cooler

3.7ghz (18.5X200) @ 1.45v

running 2 FAH clients 24/7 for over a week now. . . .i had no resistance all the way up to 3.7 but anything close to 3.8 gives me trouble.