996601 on M4A79T Deluxe

[LoneWolf3574]

I'm just posting this as a kind of log as to what I'm doing so that others with similar setups may follow and hopefully not repeat any mistakes that I make. Any useful suggestions are always appreciated.

I'm running a AMD Phenom II x4 955BE on an ASUS M4A79T Deluxe (BIOS rev. 1103) with 8Gb of Mushkin 996601 for the last 3 weeks and, IMO, I've been running pretty stable at a CPU multiplier of x17.5 (3511MHz) and my RAM at 7-7-6-18-24-2T (CL-tRCD-tRP-tRAS-tRC-CR) voltage 1.86V (2 weeks at this voltage, 1.9V before) and DRAM frequency set at 1600MHz, all other settings are auto.

Now in the last week, I've set my multiplier back at stock (x16) due to heat issues, not that big of a deal atm, I've got some ideas on how to fix that (lapping the Xigmatek HDT-S1283 cooler). Now, today, I'm getting a BSOD every time I boot. I reset the CMOS since I can't even make it past the boot screen of Vista. I ran the MS memory tester from my Vista DVD and behold, I got a hardware error at about 39% on the first pass. So, I went ahead and set my 996601 sets to 1333 timings on the main screen and left everything else in auto (9-9-9-25 or so if I remember correctly, I'll check them again when I get home later). I re-ran the MS memory test and got a clean test after both passes and it seems stable for now.

I'm still new to the OC community, but it seems strange to me that after running stable for 3 weeks (admittedly a hot CPU ~42C @ idle per Core Temp) I should suddenly start getting the BSOD less than 5 minutes after startup. I'm going to try bumping the voltage to 1.8V, tightening the settings at 1333MHz, and playing with the NB frequency and voltage some per the suggestions that Mushkin Sean has been posting.

The only thing that really bothers me is that my overall CPU temp seems to be ~43-45C @ 26-28C ambient(Core 0-3 seem to avg. ~3C higher). The cooler has already been removed, cleaned, and had AS5 placed on it 1 time since my initial install since I thought that the problem may have been to much or incorrect application of the AS5. I've had prime95 running and got a failure after 2.25hrs and the overall CPU temp reached 62C. As I stated before, I'm going to try lapping the heatsink since numerous posts I've found on the internet suggest anywhere from a 6C to 12C drop in temperature. I'm tempted to lap the CPU, but I don't want to void the warranty yet, so I'll save that as a possibly for next spring.

[MushkinSean]

Hi,

Try the outer ram slots, 3 and 4 from the CPU, they work better for high speed.

Are you setting the voltage on the CPU, CPU/NB or are you leaving them at auto?

[LoneWolf3574]

Heh, all slots are filled w/ the 2x2Gb of 996601 (4 2Gb DIMMS), sorry I forgot to mention that.

Right now the NB frequency is set 2400MHz (2408.4MHz according to CPU-Z), 2600MHz = BSOD, and everything else is either in auto or its default. The current timings on at 1337.8MHz (hmm, interesting number) are 6-6-6-18-24-2T per CPU-Z, voltage is currently set at 1.86V. All these timings were set manually in the BIOS, with the exception of the CR. The tRC was originally 34, but I tightened it to 24 to see what would happen, so far so good. All other memory settings are left in Auto. The FSB is still at 200 since I've had a few issues in the past getting a BSOD on anything near 220, I'll play with it some more later.

Like I mentioned earlier, I'll keep this post updated as things progress. I've got my lapping kit on order from EasyPCKits.com and expect it to arrive early next week, I got the whole shebang, ArctiClean with Arctic Silver 5, Coarse Lapping Kit, Premium Lapping kit with glass, and the Final Polish Compound. After I lap the CPU cooler and imbibe in much beer , I'll run prime95 again to see where things go from there, hopefully I'll shed a minimum of 5C.

[LoneWolf3574]

Alrighty then, I completed the lapping last night (2 hours of work) and I could see my reflection in the now mirror like copper surfaces, especially after using the Final Polish Compound. I must say that I don't think I did that bad of a job considering it was my first time lapping a CPU cooler of any kind. Even though the coarse kit was included in my purchase, it wasn't need it since the surface wasn't that rough to start with.

Anyways, to the point, idle temperatures appear to have gone down from ~42C to around ~36-37C, a 5-6C drop . Initial startup temperature shows 38C with PCProbe II & Everest and actually dips down to 35C at times after everything has loaded. I'm currently avoiding using Core Temp (http://www.alcpu.com/CoreTemp/) since the only readout I can get in the current version is the hottest core temperature and not all four or the overall. Now for the Prime95 load information, I ran the torture test for 2 hours and for the first hour it seemed to hold at 53-55C (again a 7-9C drop ) but then things started to warm up from there to 60C (still a 2C drop) after 2 hours. After much anger , checking the fans to see they were running and their speeds, opening the case to verify I had indeed clamped the cooler down properly and centered it on the CPU, I realized that the ambient temperature near my tower had risen . Well, duh, hotter air pulled into the tower IS going to make it run hotter overall .

Overall, lapping the CPU cooler has accomplished its intended goal of reducing normal load temperatures by at least 5C, now I can start with the overclocking. So far I've got a CPU multiplier of x17.5 giving me a CPU OC of just over 300MHz for a frequency of 3511MHz. Naturally temperatures have risen, but I'm not as close to the 62C mark as before. Stability has me a little concerned since I had a total computer lockup in the middle of a game and Core Temp showed the hottest core at 55C (here I'll use it just to get an idea). The next step, I suppose, would be bumping the CPU voltage up from the default 1.350V and see if this minor stability issue goes away. While I'd like to see a 20% OC, I think a 10-12% OC is more realistic with air cooling* given how well my AMD Dragon system heats the air where it's located.

As I have mentioned before, I will continue to keep this updated as I change things with my system. It may be a couple of weeks at times as I experiment to see the stability before I post. If anybody has good ideas or suggestions on setpoints with this particular configuration (ASUS M4A79T Deluxe, AMD Phenom II x4 955 Black Edition, and Mushkin PC3-12800 (996601)) please leave a post and let me know. It may take a couple of days to respond, but I'll at least PM you and let you know I read your post, then I'll post the results later.

*Just so that you all know, air cooling/circulation is currently provided by the stock Cooler Master fans (intake) that came with the CoolerMaster RC-690-KKN1-GP tower (1 120x120x25mm TLF-S12-EB & 2 120x120x25mm R4-S2S-12AK-GP), 2 SilenX 120x120x38mm IXP-76-18 (exhaust), 1 SilenX 80x80x15mm IXP-52-11 (intake blowing the back of the motherboard), and the stock Xigmatek 120x120x25mm CPU fan.

Links:
CoolerMaster TLF-S12-EB: http://www.newegg.com/Product/Product.aspx?Item=N82E16811999061
CoolerMaster R4-S2S-12AK-GP: http://www.newegg.com/Product/Product.aspx?Item=N82E16835103024
SilenX IXP-76-18: http://www.newegg.com/Product/Product.aspx?Item=N82E16835226013
SilenX IXP-52-11: http://www.newegg.com/Product/Product.aspx?Item=N82E16835226001

[LoneWolf3574]

I've been reading more articles on various websites and came to the realization that perhaps the more important temperature to follow is the difference between the ambient and CPU temperatures verses the absolute CPU temperature. As of right now that difference is ~13-15C at idle, which appears to be the norm.

At the moment I am looking at purchasing 2 SilenX 120x120x25mm IPX-74-14 fans to replace the 2 CoolerMaster R4-S2S-12AK-GP fans, this would increase inlet airflow from approximately 149 cfm to 205 cfm. Although I'm not a big fan of LED fans, the front CoolerMaster stock blue-LED fan is tasteful and I think I'll leave it in place for now.

Links:
SilenX IPX-74-14: http://www.newegg.com/Product/Product.aspx?Item=N82E16835191013

[LoneWolf3574]

I got done installing the last WD hdd (planning on RAID-0 eventually), the last 2 SilenX 120mm fans, and a Zalman 6-fan controller yesterday, zero effect on the overall temperature. Oh well, plenty of air flowing now so I have little to worry about when it comes to cool air for my system. Best part is, even though my tower is louder, the only noise that I do hear isn't from the fans but from the air they're moving .

Today I downloaded and update the motherboard bios from 1103 to 1202, not my best move . While successfully updated, I ended up have to set everything to it's defaults (except Cool-N-Quiet is enabled) due to a BSOD and hard lock that kept coming up during the Vista load screen. I haven't exactly determined why since all the update did was add unleash mode to the ACC function (I left it disabled) and fixed the Intel 9300PT LAN card not waking up the system from S1/S3/S4 (I don't have one, so no big). I suspect that it may have something to do with the NB frequency being at 2400MHz since I had something similar happen when I tried to set that to 2600MHz with bios revision 1103. The only upside seems to be that I shed nearly 2C and I'm now about 11C above ambient (39C absolute). I gotta do some checking since I don't know what the default voltage is for the NB, but I seem to remember reading somewhere that increasing this voltage would help to stabilize things.

As always, any and all reasonable suggestions and ideas are welcome. By God, I'm gonna kick this pig until it works, .

EDIT: Sure enough the issue was the NB frequency being at 2400MHz, although I don't know why . I raised it to 2600MHz to see what would happen. I managed to get Vista loaded, but for some reason, according to CPU-Z, it's defaulting to 2000MHz and I can't get it to change. I double checked the BIOS and it's still set at 2600MHz, . I also raised the DDR voltage from 1.86V to 1.9V, that seemed to help with the benchmarks some, dropped the latency from ~57ns to ~49ns; I was much happier when I had ~47-48ns with the NB frequency at 2400MHz.

[LoneWolf3574]

I ended up reflashing my BIOS back to rev 1103, now everything appears to be stable and normal once again. With BIOS rev 1202, every time I did a cold boot, the POST screen would say that the OC had failed, press F1 to... and I would have to reload the settings I had last time I started up, TO MUCH EFFORT to do this EVERY TIME I start my computer.

Somebody suggested that I up the CPU voltage to see if that would stabilize things, of course I read this after I reflashed to 1103. At the moment, I'm just not wanting to deal with rev 1202. My guess, by the lack of posts, is that everybody else is having no problems with it and that my situation is unique. Oh well, life goes on and I'm really not willing to give up on this board yet, way to many excellent reviews on it and SO many tweaks for it.

I must say that I hope this is not an indicator of BIOS updates to come. I've got to include that I am rather disappointed with this being my first ASUS motherboard (Abit previously) that such a thing could cause such huge problems.

[LoneWolf3574]

Last night I spent about 3 hours or more rerouting all my cabling and wiring in my tower, most of that was spent trying to figure out how to do it. Needless to say, the interior is much cleaner now and has much better air flow.

I had an idea this morning, after having read several reviews regarding the testing of air cooled cpu heatsinks, I noticed they all had one thing in common, a controlled environment. It didn't matter what the ambient temperature was, most were ~24-28C, they were all controlled, this was the one thing I didn't have. I figured out that if I couldn't control the room temperature with the same precision as they did, I'd do the next best thing, turn a fan on high and direct it at my tower. Well what do you know, it worked, my average temperature after running a prime95 torture test was 54C after 3 hours. At the same time I was running a FurMark stability test on my GPU (71C average). I do believe I have this stock frequency temperature issue resolved now.

Ok, now with the ambient being 27C (80F), the cpu idling is running at 34C (93F), that's just 7C above ambient. The cpu at load is 54C (129F) and the maximum temperature for this cpu per AMD is 62C (144F). I would have to say LET THE OVERCLOCKING COMMENCE.

[MushkinSean]

I just went through the same thing building a rig for my Dad, it took forever to route all the power cables!

Soon you'll be like me, and have your fancy case in the corner while you run your mobo on the box top!

[LoneWolf3574]

LOL, at this rate, that just may come true.

I reapplied the version 1202 BIOS to my M4A79T Deluxe today figuring I'd try some tweaking to get past the BSOD that keeps popping up during the Vista boot screen. I bumped both the CPU and CPU/NB voltages up by 0.1V to 1.400V and 1.200V respectively and that seems to have solved the problem. Later I intend to go back and reduce these to the minimum need since I think only one may be needed. I came up with the 1.4V and 1.2V by reading on [url]Overclocking.net[/url] about other people OCing their 955BE cpu and found only one person with a similar air cooler (they have the Xigmatek Dark Knight, I have just the S1283) the same motherboard (didn't list their BIOS rev. but after the 1202 update) with a 18.0x multiplier, 1.4V CPU, 1.3V CPU/NB, and 2000MHz NB clock. I figured if they could get 3600 stable like that, what would happen if I tried something similar and worked my way up there. I haven't tested for stability as of yet, but my latency (according to Everest) for the RAM at 6-6-6-18 CR2 is 46.7ns (WOOT ), that's a reduction of 3.1ns from previously . And according to the chart given, a full 12.6ns faster than a Core i7 Extreme 965 on a Asus P6T Deluxe x58 platform with Triple DDR3-1333 (@ 9-9-9-24 CR1).

ATM, I must say that I am happy with the results. Idle is holding at ~41C or 14C above ambient.

EDIT: *sigh* Well the issues with the 1202 BIOS came back as a BSOD paging fault in a nonpaged area. I spent the last 3 hours trying various settings only to downgrade my BIOS back to 1103 again. I am aware that this usually is a RAM issue (ie bad), but the only way I could get my system stable was to set everything at it's default settings . I'm beginning to think I may have either a bad motherboard or cpu, either is possible at this point since I'm running stable with the 1103 BIOS and not with the 1202. Very annoying and no fun when trying to overclock.

[LoneWolf3574]

Been a bit but I've been busy.

Ok, here's the low down. I've upgraded the BIOS to 1303, amazingly, it's completely stable when compared to 1202 . I've wiped my HDD and taken the 2 320Gb WesternDigitals and put them into Raid-0, again, apparently stable. I've even managed to tighten my memory settings some more to 1333MHz at 6-6-6-16-24-2T @ 1.9V, although it doesn't seem to have made that big of a difference compared to the previous settings of 1333MHz at 6-6-6-18-24-2T @ 1.9V. I've OC'd the cpu to 3612.23MHz from 3212Mhz and I'm currently stable after running an 8 hour OCCT test. The video card, I've had to adjust AGAIN from 975/1155 @ 1.45V to 950/1150 @ 1.375V, but still have to run a stability test in Furmark (3 hours should be good). Most importantly I've upgraded from a Xigmatek HDT-S1283 to a Prolimatech Megahalems w/ a 1900rpm Scythe Slipstream. That shaved ~5C off my temperature, but the fan just is to weak (not enough static pressure) to push enough air through the heatsink fins .

I've heard that MemTest86+ is a good test for testing memory stability, if I can figure out how to use it, I'll give it a try.

[MushkinSean]

Awesome overclock!

It's amazing what a BIOS can do.

MemTest is a great program for testing your memory, just burn the image to a CD.
Then set the computer to boot to the CD.
If your OCCT stable for 8 hours I doubt you'll have any issues with your memory.
One thing about MemTest, it's a great program, for initial testing.
It does not load the ram the same as an intensive program in the O/S will, like OCCT for instance.