Motherboards Memory Storage Cases/Cooling/PSUs IT Computing Displays Mobile Mac CPUs & Chipsets Video Digital Cameras Linux Gadgets Systems Trade Shows Guides Home Increase Font Size Decrease Font Size Change Page Size
Stuck at 4GHz with the Phenom II X4 940...
Stuck at 4GHz with the Phenom II X4 940...
Date: January 29th, 2009
Author: Gary Key
 
 

We have been testing the overclocking capabilities of the Phenom II X4 940 (seriously, who names these products?) for an AMD roundup and have hit the proverbial brick wall. It has not been for the lack of trying or even using a stellar motherboard to test these processors. The motherboard choices have ranged from some wicked little 790GX overclockers from ASRock, Gigabyte, and DFI to the flagship 790FX products from ASUS, MSI, and Foxconn. We even tried a few NVIDIA 780a based motherboards along with a slew of newly arrived AM3 compatible boards. We changed cooling, processors, video cards, disk drives, memory, and tried every available voltage setting. It did not matter. We could not break the 4GHz barrier and still complete our benchmark test suite.

The only commonality between our 30 different setups is the operating system. We recently standardized on Vista Ultimate 64 SP1 for testing. Granted, we had this same problem when our 940 engineering samples first arrived and we asked AMD about it. However, AMD never did get to the bottom of it before the launch date. We thought our results might change with retail processors. Alas, they did not. We have four retail CPUs, three with 0850 lot codes and one from the 0849 batch that all behave the same way under Vista 64.

Our final benchmark stable clock speed is 3.955GHz reached via a 17.5 core multiplier and a 226 HTT setting on our . This required a 1.6V VCore setting (with droop, real voltage is around 1.585V) on our DFI DK 790FXB-M2RSH motherboard. Memory speed is DDR2-1205 at 5-5-5-18 timings with VDimm at 2.060V. This is the setting we will utilize in our upcoming roundups. We mention it now as our Core i7 920 will operate at 4GHz and Q9550 at 4.25GHz, not exactly fair, but we are looking at platform capabilities on air cooling in the overclocked sections. As one would say, it what it is.

 

 


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, 1GHz NB speed, etc.) and even chilled the processor down to 16C and raised processor voltages above 1.7, nothing worked above 4GHz.

At least under Vista 64 as Vista 32 was much different. The settings mentioned earlier allowed us to reach a stable 4.275GHz (19x225) with the same components. However, we are not utilizing Vista 32 in testing anymore, especially considering our standard benchmarks are completed with 4GB and 8GB configurations on the DDR2 platforms. As such, it appears at this time that any overclocking comparisons will be limited to under 4GHz on the AM2+ and AM3 platforms. We have addressed this problem with AMD again and hopefully an answer is forthcoming. In the meantime, we figured out a way to get a screenshot above 4GHz before we received the standard BSOD routine. At least it is a consolation prize at this point.



 


49 Comments
Username:
Password:
HT Link Frequency by Ryun, 376 days ago
I heard that taking down HT 3.0 down to 1.0 helped with overclocking. Maybe give that a try if you haven't already?

Reply
PAE by punjabiplaya, 376 days ago
Maybe some limit to the PAE or something? Maybe a quick 32bit vs 64bit at maximum common stable OC could point out something? Or I could just be blabbering.

Reply
NB speed by duploxxx, 376 days ago
Don't tell me you just tested the systems against each other with a NB speed of 1.8-2.0 ????

Reply
Nothing wrong with 4GHz by simontay1984, 376 days ago
Its not as if you'd need to overclock more than 4GHz anyway...
I think 4GHz is plenty fast enough! Its a vast improvement over the original 65nm Phenom. There's no way you'd have been able to reach anywhere near a 4GHz clock with the orignal Phenom!

Reply
Stuck at 4GHz?!? by Zak, 376 days ago
Stuck at 4GHz?!? Man, that truly sucks, LOL :)

Reply
RE: Stuck at 4GHz?!? by Griswold, 375 days ago
Indeed, AMD and AT should immediately face the consquences and close shop!

Reply
NB speed (Uncore part) by BLaber, 375 days ago
You need to get that NB speed higher bcz PII's can easily do 2.8Ghz of NB speed, if you push youll get 3.0ghz stable as well.

Reply
Hmmm. by yyrkoon, 375 days ago
Sounds like AMD either did something non standard with their implementation, or somehow Microsoft missed something in their code, or just flat out limited AMD64 somehow. Whatever it is, it sure sounds odd that *only* AMD64 would be affected, in a 64bit environment and I64 was not.

Reply
RE: Hmmm. by Penti, 375 days ago
Theres plenty of PHIIs benching over 4GHz.

Reply
Other 64-bit Operating Systems by TA152H, 375 days ago
Have you guys tried it on Windows 7, 64-bit, or Windows XP 64-bit? It could be that Vista is doing something to cause it as is being suggested, or it could be that there are one or more 64-bit instructions that simply can't execute at as high a clock speed, or even a 64-bit register that can't respond quickly enough when overclocked.

It is strange that it happens at the 4 GHz threshold, for sure, but it still could be a red herring. It's hard to say without more information, but trying Windows 7, and Windows XP 64-bit would give a little more data to help the analysis.

Reply
RE: Other 64-bit Operating Systems by Griswold, 375 days ago
Eh, W7 is essentially Vista under the hood. There wont be any differences, I'm willing to bet real money on that. And XP-64 is just a piece of shit to begin with. If you want to test something truly different that makes sense, do it with a Linux flavour.

Reply
RE: Other 64-bit Operating Systems by TA152H, 374 days ago
Essentially is an ambiguous term, and there are always changes to "what's under the hood" even from service pack to service pack.

I'm not here to judge XP-64 or any operating system, I don't care whether you like it, don't like it, or whatever, it's more about does the processor run on it at higher speeds, regardless of its merits.

Most people buying a processor aren't going to shackle it with Linux, although, for the purposes of identifying where the problem lies, even it could prove useful. The most important consideration is Windows 7, from a practical standpoint, since XP is dying, Linux never lived, and Vista should never have been born. There's no doubt in the near future, most new computers will be running Windows 7, like it or not.

Reply
RE: Other 64-bit Operating Systems by WillR, 373 days ago
I just ordered a Phenom II 940 yesterday. It's somewhere between NewEgg and me this moment. Oh, and I mention this because I intend to run Linux on it. I also don't intend to overclock it. Just as few people overclock their procs as few run Linux on them.

Reply
RE: Other 64-bit Operating Systems by Donkey2008, 373 days ago
There are thousands of data centers that would probably disagree with your Linux statement.

Vista runs perfect on my system. It didn't always do so, but neither did Windows XP for the first 18 months.

Windows 7 = Mohave Experiment. Microsoft changes the interface a little and everybody goes bananas at how great it is.

Reply
Cooling by Penti, 375 days ago
How about using Phase-change cooling? Not that it wouldn't still get beaten.

Reply
64bit by Holly, 375 days ago
Sounds to me like Phenom 2 is having problems running multimedia extension instructions in 64bit mode. Just to note 64bit can quite well use very different instructions (=different parts of cpu). That's about it.

Reply
samples by Visual, 375 days ago
Overreacting much? Jow many cpu samples have you tried with the same result?
It seems to me like you just got a bad overclocker or two and are crying foul... but I may be wrong.
If you tried multiple cpus (though I really don't know what number would be ok) from different batches, then I retract my statement. But if you did all this swapping of motherboards, settings and whatnot with the same one or two cpu samples, I think your whole post is completely pointless.

Reply
RE: samples by Calin, 375 days ago
There were 4 CPUs, three from a batch and one from another, as stated in the article. While 4 CPUs limited at under 4GB in Vista64 and working at 4.25 in Vista32 is nowhere near statistical significant, I think I'll take his opinion as true until/unless I hear otherwise.

Reply
Maybe 64-bit software is at issue by Calin, 375 days ago
As the 64-bit software exercises more of the internal structure of the processor (extra registers and so on), there might be a local hot spot or maybe some critical path only under 64-bit software.
Could you please try with another 64-bit operating system that limit in overclocking? It shouldn't take long to confirm/infirm this using either a version of Linux, FreeBSD, OpenSolaris, and you could test stability with Prime95 (mPrime) or something else. Even a live CD might be of use for a bit of testing

Thank you

P.S. As the scope of the articles grew and grew (compared to the years past), they appear less and less often. These blogs are very nice in keeping the front page lively

Reply
why 9550 by duploxxx, 375 days ago
why do you use a Q9550 in this test against ph2 940, the direct price competitor is Q9400 ......... compare what you need to compare also ddr2 against ddr2. If not there is no added value to this review.

Reply
It's fair by ssj4Gogeta, 375 days ago
"We mention it now as our Core i7 920 will operate at 4GHz and Q9550 at 4.25GHz, not exactly fair..."
It IS fair. It's not Intel's fault that AMD can't make better overclockers.

Reply
RE: It's fair by duploxxx, 375 days ago
again, not in same price range, why don't you take a compare of q82-q8300 then against p2 940 afterall that is what intel is doing lately, schrinking the caches

Reply
formatting toolbar by ssj4Gogeta, 375 days ago
How do I use the formatting toolbar while posting a comment? When I click on bold, italics, or underline, nothing happens. And when I click on the link button, it asks me for url and text, but when I click on OK, nothing happens. Same for Quote button. I have javascript turned on. Please help. :)

Reply
RE: formatting toolbar by Calin, 375 days ago
bold
italic
underlined

I can't either - but you can manually edit the posts with [ B ] ... [ / B ] and so on

Reply
There is more work to be done & AMD needs to give us an answer by East17, 375 days ago
That's all ... There is need to test the OC capabilities in WinXP 64 Bit & 32 Bit and then in Windows 7 32 Bit & 64 Bit and then we will have the whole picture.

Right now there is not much that we can conclude from these tests besides the obvious : "There is something strange here"

Some tests in Win2003 Server R2 with all the workstation features enabled would be also usefull .

Reply
Experiencing the same problem by MagnumMan, 375 days ago
I am experiencing the same kind of wall, at a lower frequency. I have an 0850DPCW Phenom II X4 940 under water cooling with 4x2gb memory running at DDR2-960, FSB 240, 5-5-5-12-2T @ 2.2V. I have the CPU running at 3.6GHz @ 1.4125V, HT is 2.4GHz @ 1.325V. I have tried pushing the voltage as high as 1.65V on my ASUS M3A79-T Deluxe board (both CPU and NB..) with the reference voltage as high as 2.7V (currently 2.6V) but I cannot get the CPU stable even at 3.66GHz by dropping the FSB and raising the multiplier. I tried for many days to get past 3.6GHz and I cannot. I also have 3x 3870 overvolted and cap-modded, running 965/1250 in 3-way crossfirex and I can play Dead Space until I'm blue in the face at 3.6GHz, or run prime95 for 24 hours. But I go up to just 3.66GHz and prime95 dies in a matter of minutes. Lastly, I am using Vista 64-bit... I do wonder if this is due to the 4x2gb memory... when testing Vista 32-bit were 4 memory sticks used or not? That could be the difference. I expected a lot more out of the CPU... mind you it beats the Phenom 9850 BE which I got to 3GHz on this platform so I'm not complaining much, I expected to get to 4GHz or more.

Reply
Wow...Inteltech is having problems with an AMD processor. by helldrell666, 375 days ago
And what else Inteltech...doesn't like about the new AMD processor?
I got the processor myself two days ago and it's now clocked at 3.62GHz on stock voltage operating like a champion.

I think that You, Inteltech, should not be allowed to review AMD products.Stick to your Intel cpus and great gpus and leave AMD alone.the same goes for those N-vidia sponsored sites.




Reply
RE: Wow...Inteltech is having problems with an AMD processor. by Adul, 375 days ago
Considering that this site was favorable to AMD for a long time I find it odd you call them bias. Fanboi much.

Reply
How about the X4 920??? by Ratman6161, 375 days ago
I know its easier to focus on the 940 BE with its unlocked multipliers. But for a lot of us the point of overclocking is to buy a cheaper part and make it as fast or faster than the more expensive top of the line part. So I think a lot of us would be more interested in what could be done with an x4 920.

Reply
Other OS confirmation by RobberBaron, 375 days ago
Definitely pop in another 64 bit OS to verify that this isnt a hardware limitation instead of an ugly Vista 64 problem.

Reply
4Ghz wall?? by ghostfaceChillah, 375 days ago
May I point you to some results that differ from your conclusion?

http://www.overclock.net/amd-cpus/44249...enom-ii-steppings-max-oc-database.html

Reply
So what wrong are you doin? by 0vercl0ck, 374 days ago
Isn't it really strange?

Team Finland (SF3D & Sampsa) reached 6435.62 MHz with an AMD Phenom II ES (45 nm).

AMD Phenom II
Windows Vista Ultimate Edition SP1 (Build 6001)
CPU Arch : 1 CPU - 4 Cores - 4 Threads
CPU PSN : AMD Engineering Sample (ES)
CPU EXT : MMX(+) 3DNow!(+) SSE SSE2 SSE3 SSE4A x86-64
CPUID : F.4.2 / Extended : 10.4
CPU Cache : L1 : 4 x 64 / 4 x 64 KB - L2 : 4 x 512 KB
CPU Cache : L3 : 6144 KB
Core : Deneb (45 nm) / Stepping : RB-C2
Freq : 6435.62 MHz (286.03 * 22.5)
MB Brand : DFI
MB Model : LP DK 790FXB-M2RS
NB : AMD 790FX rev 00
SB : ATI SB750 rev 00
GPU Type : ATI Radeon HD 4870 X2
GPU Clocks : Core 507 MHz / RAM 500 MHz
DirectX Version : 10.0
RAM : 2048 MB DDR2 Dual Channel
RAM Speed : 572.1 MHz (1:2) @ 5-5-5-18
Slot 1 : 1024MB (PC2-5300)
Slot 1 Manufacturer : OCZ
Slot 2 : 1024MB (PC2-5300)
Slot 2 Manufacturer : OCZ



Reply
So what wrong are you doin? by 0vercl0ck, 374 days ago
EGzperts? by 0vercl0ck, 374 days ago
Aftere reading all the posts above it is clear that the great majority of the posts, except 2, is made by egzperts!

so what is the definition of an expert?

An asshole out of town?

Iyt is clear that thge attempt of this article more likely was to prove that AMD can not handle more than 4.o giigs.

The truth however is that your overclockers know shit!

http://valid.canardpc.com/show_oc.php?id=481726

So what is wrong with you guys?

Shitheads not overclockers!





Reply
The standard BSOD routine. by 0vercl0ck, 374 days ago
The standard BSOD routine is unrelated to CPU overclocking!

Set your RAM to normal stable performance, than overclock CPU!

Once you got crashes as the result of overclocking CPU, lower the setting to previos stable, and ONLY than overclock the RAM!

Your standard BSOD routine is related to wrong memory setting and have nothing to do with CPU!

"Memory speed is DDR2-1205 at 5-5-5-18 timings with VDimm at 2.060V"
When overclocking first set memory to standard setting, overclock CPU, than play with memory timing!

"Set the memory to 5-5-5-12, at optimal voltage, than overclock.66

"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, 1GHz NB speed, etc.) and even chilled the processor down to 16C and raised processor voltages above 1.7, nothing worked above 4GHz."

Reply
RE: So what wrong are you doin? by BlueBlazer, 374 days ago
Those samples were given to them by AMD directly, in other words...

THEY WERE CHERRY PICKED

Go browse XS and check how retail PhII actually overclock ON AIR is like...

MANY CAN'T GET PASS 3.9GHz FULLY STABLE... DESPITE EXTREMELY HIGH VCORE...




Reply
Unbiased? by Murphy9090, 374 days ago
Hey this is coming from the same guys that used the excuse of using a pretty crappy Heatsink on the comparison between the i7 and the Phenom 2. Their excuse: It was the only one available that was Phenom 2 ready. Amazing considering its an AM2+ part and the only thing that changed was a die shrink and some streamlining.

One other thing, the AMD 940 was never designed to go up against the i7, period, end of story.

Glad to see another Anandtech article whining about how bad there overclocks are when they could only pull 2.9 ghz on an original Phenom, w/ the new 750SB motherboards. Check out some forums to learn how to overclock. And try using the same coolers to compare products so your biased doesnt shine through yalls blue tinted lenses

Reply
RFor sure unbiased! by 0vercl0ck, 374 days ago
Murphy 9090,

You got that one right baby!
;)

They are overclockers want to be!

Once one try AMD, the real man will never go back to InHell in sLide.

i7 is a microwave oven, not a CPU!

Reply
RE: RFor sure unbiased! by BlueBlazer, 374 days ago
Retail PhII have problems getting pass 3.9GHz FULLY STABLE on air cooling..

Go back to your FantasyZone...

Reply
RE: RFor sure unbiased! by WillR, 373 days ago
Could you please not screw the English language so much?

Also, i7 is a more energy efficient proc than Phenom II or Core 2. Unless all you do with your computer is "oh, l33t 1 has mor fps!" then you have to take processing time into account as well as power draw. Core i7 runs hot, but who cares if it's stable at higher temps? My only problem with i7 right now is the related hardware's price. It's why I just purchased a 940 yesterday. Price/performance of C2Q and PII are similar, Socket T doesn't appear to have much life left in it, and we all benefit from AMD staying in business.

Damn, fanboys on either side are annoying.

Reply
RE: Unbiased? by Murphy9090, 374 days ago
Correction: My apologies Anandtech didnt do the article w/ the crappy cooler, that would be another AMD bashing site"TH". But I will stick by my arguement of the AMD Phenom 2 940 never being developed to compete with a brand new product like the i7. Also the guys saying AMD needs to shut there doors, roflol. The only reason you can get your intel chips at the prices your paying is because of AMD's competition in the market. Wish for AMD's demise and watch Intel run away w/ your wallet.

As others have stated, please try w/ a different X64 operating system and get back to us. AS most of us have come to realize that Vista has some strange tendencies.

Reply
Dear AMD fanboys, by geekfool, 374 days ago
The Phenom II blows chunks, go troll somewhere else now.

@ghostfaceChillah: apparently you didn't actually read the content of the link you posted, because none of the 940s there managed to get over 4GHz on air. Take your foot out of your mouth before you post, mkay.

@0vercl0ck: if you think that 6.4GHz with a cherry-picked engineering sample is an achievement, you're an idot... but I think we already knew that from the style and content of your posts. Let me guess, you're 15 years old?

@Murphy9090: "the AMD 940 was never designed to go up against the i7". That may be true, but AMD's marketing department certainly made it seem that way... until the Phenom II launched, and everyone realised it's a POS.

Reply
RE: Dear AMD fanboys, by Murphy9090, 373 days ago
Well glad to see a good argument? Blows chunks really, then the Q9400 Q9550 must blow chunks by your so eloquent statement. Considering they are similar in price and speed.

I was trying to clear some images up about the Phenom 2 and we get "blows chunks." Great choice or words, and your bashing the other idiot on these comments. I'm not a fan boy. by no means. and have no problem switching from one company to the other. Instead of watching apples to oranges comparisons, lets compare apples to say um maybe apples. Or even get close with applesauce

Reply
RE: Dear AMD fanboys, by haukionkannel, 373 days ago
Well you are right that in that list none of them got over 4GHz using air. On the other hant PhenomII seems to like cold, so estimations about some hotspots in 64 code execution can be real. In that list there was 4.3 with 64 bit with TEC cooling. So there is more potential. We need new stepping to achieve it with air though.

Reply
Not Unusual by Goty, 372 days ago
It's well documented that Vista is much tougher on CPU overclocks than XP was (same with going from 98 to XP), and it's no secret that moving to a 64-bit operating system has much the same effect.

I don't see why you're so confused that your overclocking headroom goes down when going to Vista x64. When you start getting into those extra registers, you've got more points of possible failure, so naturally it's going to be harder to get an overclock stable.

Reply
AMD K10 has enough memory bandwith...latency should be lower (improved) by jmurbank, 371 days ago
Let us get some things straight. One AMD K8 and K10 processors have enough memory bandwidth to start out with. K8 and K10 have issues with latency, so it is better to lower the timings for better performance. Over clocking the memory gains nothing for these processors. When over clocking processors, the path ways and transistors starts oscillating. Over clocking requires higher quality power to go further. Probably the best way to do a good over clock is use a oscilloscope and check the quality of both the power supply unit and power supply in the motherboard.

If the Phenom II processor can not handle past 4 GHz, then it could either be its built-in power circuitry that may not be enough to power the processor at this speed or the quality of the transistors and pathways are not at a high enough quality. If it is possible, turning off one or two cores to gain back the ability to over clock the processor further. If it can the power grid is a little under designed or the transistors can not handle very hot temperatures at a high speeds. If want all four cores to be over clocked, super cooling may have to be done.

I prefer AMD because a total system build is cheaper compared to Intel. Intel is OK and costly if you want a pristine 80x86 processor straight from a company that invented this type of processor. I can care less about over clocking.

Windows users crying over Vista that they can not over clock their processor as far any more need to think of something else. Over clocking and using GNU/Linux will not provide a reliable setup and it never has. GNU/Linux uses system resources more efficiently or uses the system to near 99% to do tasks and it finishes faster. Windows Vista does a lot of stuff and it goes slow, so you think over clocking might compensate for the lost of performance.

Reply
amd overdrive by stainremover, 371 days ago
just a suggestion, but have you tried using amd overdrive to overclock? all those videos of the 3dmark world record @ 6.5ghz involve using amd overdrive...

Reply
Hmmm... by Jayw, 369 days ago
Gosh Almighty...It's the memory controler. How a 64 bit OS and a 32 bit OS deal with mem..verses the mem controler on the processor..

Thats just my humble opinion. Iam still weighting the benifits of 64 bit and speed is always the prime consideration.

Reply
Phemon 940 BE works without any problem at 4GHz by DavidMMM, 366 days ago
the proof is here:

http://forum.computergames.ro/137-overc...g/445255-ce-poate-phenom-ii-940-a.html

probably with a better mobo and better air cooling, he can pass the 4GHz barrier air cooled.

Reply
Comments Page 1 of 1





AnandTech.com Blog Categories
All categories
Anand's Macdates
Anand's Theater Construction
Anand's Updates
Cases and Power Supplies
CeBIT 2008
CES 2008
Computex 2009
Derek Decanted
Eddie's Got Game
Gary's First Looks
IT Computing general
Jarred's Musings
Kris's Corner
Raja's Ramblings
Rob's Experiences...
Ryan's Ramblings
Virtualization
What's New with Wes
Blank
Blank

Blank

Latest news by
DailyTech

 February 9, 2010

Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank

 February 8, 2010

Blank


more Blogs Discussions



pipeboost
Copyright © 1997-2010 AnandTech, Inc. All rights reserved. Terms, Conditions and Privacy Information.
Click Here for Advertising Information