OC'ng GA-990FXA / AMD 965BE

[CyberNerdz]

I just upgraded from my 5yr. old system. Mobo blew up on me and had an AMD Athlon X2 64 6000+ and 2x 4GB DDR2. Since I was due for an upgrade anyway, I was able to budget purchasing a new mobo, CPU and memory for less than $240

Anyway, after installing Corsair's H80 last night I started experimenting with different settings. I was surprise it was as easy as just adjusting the multiplier to x20 that brought me to 4Ghz and then CPU Freq. to 205 and at 4.1Ghz now ( it was not booting at 210+). I ran Prime95 overnight for about 12 hrs stable at highest CPU temp at 39°C with room temp of 23.3°C

I'm not comfortable increasing the voltage of the CPU core nor the memory. I also would like to keep my CPU temp below 55°C, hoping this CPU can last me as long as the last AMD I had

Question: Is there any other recommended OC settings I can do with this conditions I have?

Thanks in advance.

Ps. My system specs is on my signature

 

[trents]

Unfortunately, you pretty much tie our hands when you say you are uncomfortable with increasing voltages, and I take that to mean unwilling. If you got the CPU to 4.0+ ghz on the stock voltage that's pretty remarkable and in large part it must be your good cooling.

I'm not sure why you are so hesitant to increase voltages as when it is done incrementally and responsibly while keeping an eye on temps there is very little chance of damaging anything.

What about memory frequency and timings as an option for a little more performance? If you could attach pics of CPU-z tabs: CPU, Memory and SPD we'll take a look at that with you. Do you know how to attach pics using the built-in forum tool? Click on Go Advanced at the bottom of any new post window.

 

[swampdonkey]

I just upgraded from my 5yr. old system. Mobo blew up on me and had an AMD Athlon X2 64 6000+ and 2x 4GB DDR2. Since I was due for an upgrade anyway, I was able to budget purchasing a new mobo, CPU and memory for less than $240

Anyway, after installing Corsair's H80 last night I started experimenting with different settings. I was surprise it was as easy as just adjusting the multiplier to x20 that brought me to 4Ghz and then CPU Freq. to 205 and at 4.1Ghz now ( it was not booting at 210+). I ran Prime95 overnight for about 12 hrs stable at highest CPU temp at 39°C with room temp of 23.3°C

I'm not comfortable increasing the voltage of the CPU core nor the memory. I also would like to keep my CPU temp below 55°C, hoping this CPU can last me as long as the last AMD I had

Question: Is there any other recommended OC settings I can do with this conditions I have?

Thanks in advance.

Ps. My system specs is on my signature

Your temps wont go above 55 with the h80 unless you have a serious overclocker. I am curious to how you got to 4.1ghz with no voltage increase.

 

[CyberNerdz]

Thanks for the responses. Probably with guidance I can incrementally adjust the voltage. I just had a bad experience before when I was not monitoring my temps, since system was already running stable for months after OC'ng, that I burned out my CPU. Apparently when I installed a new video card, I had moved my CPU's HSF and misaligned its surface contact and caused my CPU to burn out. Oh well we learn from out mistakes ...

Here's the CPU-Z's ss...again appreciate the help.

 

[trents]

Unfortunately, you left out the most important part of the HWMonitor interface, the part that shows core temps. They are the critical ones if they seem accurate and on the Deneb CPUs they generally are. Also, we need to see "load temps" not idle temps. Could you redo that. Have HWMonitor open while you run a 20 minute Prime95 blend test. Then upload the pic of HWMonitor again with the core temp section in view.

 

[CyberNerdz]

Thanks but there was no mention about getting Core temps. I do not see any accurate readings from HWInfo even after running Prime95 for 30 minutes. I also tried latest HWiNFO64 and each core temps are grayed out. What am I missing?

 

[trents]

Now, go into bios and disable the core unlocking tool. You have no cores to unlock anyway with that CPU and with it enabled the core temp sensor reading process is disrupted. That's why the core temps show 0c. Can we try that again?

 

[CyberNerdz]

Wow that simple BIOS change fixed the core temp, thanks Ran Prime95 for another 30 minutes and here is the SS of the temps. I also observe that the TMPIN2 has jumped to 64°C, which I think is more realistic than the 12°C it keeps reporting. Thanks again for your patience.

 

[CyberNerdz]

Is this true?

TMPIN0 = Mobo temp
TMPIN1 = CPU temp (mobo sensor)
TMPIN2 = NB temp

If this is the case, it seems like I have to put some attention as well to getting an after market NB heat sink. Suggestions?

 

[trents]

You know it's really hard to say which of those TMPINx lines represents which sensor temp and in the end it may not matter because they are all really close together in the same area of the motherboard. Bu if you have some extra case fans you might try laying them in around the VRM/CPU socket area and the NB to see what effect that has on those unknown temps. One of the drawbacks of water cooling is that you eliminate the air turbulence in the socket area that a traditional air cooler can provide. Lately, I've started keeping a stock of 50mm fans on hand for this purpose. They are small and fit into tight places. What I do is stick them to the motherboard heat sinks with double-sided ahesive strips. I cut a small square of tape with scissors, stick one side to the fan hub (after I remove the paper label from the hub because it will come loose eventually) and then stick the fan to the top of the heat sink with the other sticky side of the tape. It's also handy to have some fan cable "Y" splitters on hand so you can run multiple small fans off one motherboard fan header. Keeps those motherboard heat sinks nice and cool. http://www.newegg.com/Product/Product.aspx?Item=N82E16835104008

 

[CyberNerdz]

Now, go into bios and disable the core unlocking tool. You have no cores to unlock anyway with that CPU and with it enabled the core temp sensor reading process is disrupted. That's why the core temps show 0c. Can we try that again?

Thanks for the advise on the temp question. I do have half a dozen fans left from my old build that I took off when I replaced my mobo/cpu and have been doing the same thing attaching them

I was wondering about your thoughts though on the original issue and did post the temp of the cores above, as requested.

 

[trents]

You mean overclocking/performance measures that don't involve increasing voltages?

 

[CyberNerdz]

I was just not really sure why you wanted to see the core temps but now that I see that the core temps were already at 52°C on 100% load then maybe I should just stick with what I have now (4.140 Ghz at x20, 207Mhz w/o voltage increase). I do however have a chipset heat sink I should probably install now. I bought this for my old board when I was also getting high temps on its chipst.

 

[trents]

Yeah, the chipset heat sink would be a good idea but I would also suggest blowing some air on the mosfet/VRM heat sink between the CPU socket and the I/O stuff.

You know I don't think you should be getting 52c core temps with an H80. Something's wrong there. How do you have the H80 fan(s) oriented? Are you bringing cool air into and across the radiator or are you moving it from the case through the radiator to the outside?

 

[CyberNerdz]

I'm "moving it from the case through the radiator to the outside." The case is designed to exhaust the air out to the top. I'm not sure if your familiar with Raven's case designs but it has 3 intake fans at the bottom of the case that brings in cold air then flows to the top. The top in this case is the back panel of the mobo. I really can not bring air into the radiator because of this. Here is a YouTube video of the design.

Anyway, I'm pulling the board to install the heat sink for the chipset and might as well check the H80 mounting I did. Thanks again.

 

[trents]

I think you might do well to reseat the water block and redo the TIM.

 

[CyberNerdz]

Just finished running Prime95 for 15 minutes just to see how the new chipset heatsink was performing. It obviously helped drop it 8° (66 to 58) but I will install a fan to it to even cool it down some more.

Reseating the H80 and applying Antec's Formula 5 thermal compound also helped dropped my core temps with full load at 47°C, from earlier reading of 52°C. I'm also running now at 4.2GHz (20.5 x 205) and still with stock voltages. I will let it run Prime95 overnight and hopefully runs stable

 

[trents]

Hey, that's good news! I noticed that the stock voltage on your CPU was quite on the high side so I suspected two things: One, that was partly to blame for your higher than expected temps, and two, you may have had plenty of overclocking headroom without adding anymore volts to the CPU. Seems like I was correct, at least on my second suspicion. 4.2 ghz on a Deneb 965 is pretty darn good.

 

[CyberNerdz]

Stopped Prime95 after 2.5hrs. and everything is stable Peaked CPU temp was at highest at 48°C with case temp at 35°C. Summer heat wave is a different story so I just have to wait and see. I don't have much high CPU demand apps anyway and most of my games put more stress on my GPU, which is ok for now.

Question now is, do I need to tweak my memory timing? Although memory is rated at 1866Mhz and is only running at 1656Mhz (207x8) but I feel the timing delay is a bit high 11-11-11-30. This is set AUTO in the BIOS but wondering if there are any noticeable difference if I drop it to maybe 9-9-9-24 and 1333Mhz.

Time to update my sig

 

[trents]

Yes, it might be worth experimenting with quicker timings and lower frequencies. Most people don't or can't run the ram at 1600 mhz on the Deneb core CPUs as the IMC is a little weak. You are only able to run the ram at your current speed because you significantly relaxed the timings but as you know that offsets some of the performance gain of the higher frequency. Which is best may also be very application dependent.