Help w/ 1055T OC on AsRock 880g Pro3

[Primate]

Hi,

This is my first build and I am new to OC so please bear with then inevitable noob topics I will ask about. I did extensive research here and in other forums trying to get my OC right and have been experimenting with what I learned for the past 2 days. The problem now though is that my temps seem to be going higher than they should and they got me worried enough to post this to ask for help.

The build is:
CPU: AMD 1055T, 125W
MOBO: Asrock 880G Pro3
RAM: G.SKILL Ripjaws Series 1x4Gb DDR3 1600 (9,9,9,24,2N; 1.5V)
PSU: Corsair CMPSU-650TX Enthusiast edition
Cooler: CoolerMaster V8
Case: CoolerMaster HAF 912
OS: Win 7 64bit

The system is stable and keeps low temps at stock settings, but I have had difficulties OCing. So far the only stable OC I have hit on is a 3.36 Ghz with a 240 FSB but by what I can tell my ram is running at 800mhz (CPUz) and more importantly, I am hitting close to 60 C under load (CPU temp from Speed Fan 4.43; AMD Overdrive Monitor reports 10 C less for CPU temp).

I am thinking I may have screwed the pooch on the mobo selection. I went with the 880G pro3 for the AMD+ support and the sata3, but the uefi is less than stellar. I have no options for HT or NB multipliers that I can find anywhere and I am thinking that is what's bottlenecking me since I hit a NB speed of 2400 with the default 10x multi.

I am trying to reach a 4Ghz 24/7 OC as I have read multiple posts and guides saying this is well within reach for a 1055t with just air cooling. Any ideas were to go from here?

Here are the screen captures in case I missed something (The AMD overdrive is just to show settings, I am doing all the tweaks from UEFI, since AMV OD does not let me change almost any settings ).

 

[Primate]

Found the HT and NB multipliers as soon as I posted here lol. It was expressed as speeds instead of X's. So now I dropped both to HT and NB to x8, increased the FSB to 265 and dropped the vcore to 1.45 (CPUz shows 1.48 for some reason) trying to get under the temps. I reached a "stable" 3.7 Ghz (15min under load for now) but the CPU temp still creeps to 60 C. See below.

 

[trents]

Don't worry about "CPU" temp. That is actually measured from the motherboard socket area. What is important is the "Core" temp which is running 46-47 C. That's fine. You just want to keep it to a max of the mid 50s C. range. In fact, with that amount of overclock your temps are doing great! Have you tested it with Prime95 blend for at least 2 hours for true stability? AMD Overdrive is not a very good stress tester. It's just not tough enough on the CPU. YOU will see your temps rise another few degrees with Prime95 blend but you have room for that.

Would also like to see CPU-z tabs: "Memory" and "SPD".

 

[Primate]

I am running prime95 as we speak (Blend at defaults). My temps went down actually, and Core temp is at around 42 C (I did drop the vCore a notch though). Here are the CPUz tabs.

I had read that PhII core temps are off by 10 C, that's why I was going by cpu temp. Do you have different info on that? (ref)

If you are right though, now I just need the sweet spot to get 4 Ghz .

 

[Primate]

Prime made the temps go way higher about 15-20 minutes into the test. Core temp went to 56-57 and the system did a reboot.

 

[trents]

Have you bumped up the CPU-NB voltage any? That usually helps with stability. Do you have Cool N Quiet and C1E disabled? Give you ram one or two small bumps over the stock 1.5 volts.

 

[QuietIce]

...

 

[Primate]

@ Trents: C&Q, C1E and Turbo are all disabled (CPU throttle also). The voltages are all manually set at stock (CPU-NB 1.175; HTT 1.20; SB 1.26; NB 1.1; CPU VDD 2.5) except for CPU (1.4). I will try raising the CPU-NB and the ram voltages a notch tomorrow and post the results. Thanks for the help!

@ QuietIce: I am aware of the 55 C being the max stable temp for PhII which is why I am concerned by my temps getting close to that. I have the stock HAF 912 fans, a 120 intake in front and a 120 exhaust, with cpu fan blowing towards the exhaust. Should I be looking at getting another exhaust fan above the CPU?
My question about which temp to go by is because most forums and articles on the 1055t mention it has core temp diodes off by about 10 C. But I see your point, since the 55C stability issue makes the max temp (whether it is 62 or 70) a moot point. Still, it would be nice to know my actual cpu temp! is it core temp, core temp +10, system temp, something else? to tell you the truth the pc does not feel hot at all, but my hand is hardly a good way to measure it lol.

Thanks for the help.

 

[Primate]

PS
I just noticed you were running a single stick of RAM? That will hurt your RAM performance and may affect your overclock as well.

damn... I did not know that. I went with 4g so I could max out latter on without having to replace modules.... Guess I'll just have to buy another 4g stick now

 

[QuietIce]

...

 

[trents]

Yes, you should put a top fan in. Good case ventilation is as important to keeping core temps down as is a good heatsink/fan on the processor. If you are using warmed air to try and cool the heatsink you are defeating the purpose.

 

[Primate]

Thanks for pointing that out. Yeah the case fans are 44.7 CFM each, so I have the cpu fan trying to push more air than the case fans are moving. So my shopping list now has 2 more fans over 40 CFM each (w/ leds while I am at it) for a total >90CFM intake/exhaust and another 4g RAM. I am going to concentrate on making the 3.7 OC completely stable until I make those upgrades and then go for the 4 Ghz. Google it question: how does 1 ram module affect OC? I know it has a negligible effect on performance (1x4 vs 2x2) due to the missing dual channel but I haven't seen any info on how it affects OC.

 

[Primate]

On a slightly different topic, what do you guys think of this particular mobo? for OC?

 

[Primate]

So I bumped the voltages for CPU-NB (1.185), RAM(1.51) and vCore(1.4125) a bit, ran prime95 and had the same result as before. All goes fine for 15 minutes then when it hits the 800k tests the core temperatures climb from 41 to 50-53 and crashes... so me thinks it is temp induced instability instead of RAM or NB undervolting.... should I try lowering the vcore until I hit low temp stability issues and then go a notch higher than that? ... yes I should, why am I asking.

 

[trents]

When running Prime95 over say, an hour's time, you will see fluctuations in temps but you shouldn't get as much fluctuation as you are getting. 2-3 degrees fluctuation is what you should be seeing. Do you have Turbo Boost enabled?

I hate to say it because it may add to your shopping list but the Coolermaster V8 is not that great a cooler. For the same money you could have done better. Still, if your core temps are maxing out at 53 C. you should not be encountering instability from temps yet.

Before you throw down bucks on that memory module, consider the fact that there is no guarantee you will be able to run in dual channel mode because you do not have a "matched pair". Even if you buy the same module as you are now using, at the factory dual channel memory kits come from modules produced in the same production run, from the same batch of silicon, etc. Trying to pair them later may not work. Dual channel memory modules need to be virtually identical. If it were me, I'd buy a dual channel kit and sell the one you have now. Also realize that 8 gigs may not overclock as well as 4 gigs because it puts more strain on the integrated memory controller. Unless you need the 8 gigs (and most people don't do computing tasks that need that much memory) I'd get a 4 gig dual channel kit and sell the 4 gig module you have now.

Try 1.2 for the CPU-NB voltage. Don't confuse the IGP/chipset NB with the CPU-NB. You probably also need a small bump in core voltage.

 

[Primate]

I droped the vCore to 1.35 (1.376 under load) with all other settings the same as above, and have been running Prime for an hour with no issue. The core temps peak at 43 when doing the 800k and 460k tests and drop to 38 during the other tests. 5/6 cores are maxed out but the last core runs at 11-20% load with prime95 alone it goes higher when I use other applications though, is this something I need to worry about? could it signal a drop in performance due to undervolting?

 

[Primate]

At around 2hrs of Prime95 it hit a rounding error and stopped running (didn't BSOD or reboot, just quited prime), I upped the vcore to 1.3625 (1.392 under load) I am seeing the same temps as before ~ 43 Core temp max. I already got another rounding error from Prime 15 minutes into it but it did not stop the program this time, just cancelled worker 6. Could this be a f-ed up core 6? the same one that is not getting full load? any ideas?

 

[trents]

Do you have Turbo boost disabled? What you are describing sounds like the Turbo boost effect which causes only some of the cores to dynamically overclock under load. If Turbo Boost is disabled all your cores should be running at 100% load all the time when running Prime95.

 

[Primate]

Yes, turbo boost is disabled. It was just prime95 getting a rounding error and canceling a thread. I can't seem to get rid of that problem though. At 1.35 vCore I get rounding errors and at higher vCores I hit 50C core temp and the system crashes. So I am fresh out of ideas...

 

[Primate]

I tried lowering the clock but got the same results. I also ran mem86 all night to make sure the rounding errors were not from bad memory and came back with no errors. So it is the cpu that's messing up. The situation right now is I get prime95 rounding errors within 10 minutes at 250-265 FSB with 1.35 vCore, but no heat issues. Any higher on the vCore and I hit 50ish on the core temp and the system crashes. There is no obvious instability in the system though, it's just the errors in prime95 but I know with errors that fast the system would corrupt quickly. This are the current settings:
FSB 250
vCore 1.35 (1.36 CPUz under load)
DRAM 1.5
NB 1.11
HT 1.205

HT link x8
NB x10
RAM at 9-9-9-24 1666 Mhz (no mem86 errors)

Any and all ideas are welcome