Jump to content
Corsair Community

i5 2500k very hot with Corsair A50


Hawken88

Recommended Posts

Hi guys,

 

Joined the forum in the hope of finding some help - tried another forum but the users were a little condescending

 

As you'll see from my spec, I have an i5 2500k. The processor is overclocked currently to 4.4GHz. Sitting on top is a Corsair A50. Within about 10 seconds of running Prime95, temperatures shoot up to 80 degrees Celsius. Leaving it another 5 minutes will see them hit 90 degrees Celsius.

 

So far, I've removed thermal paste using ArctiClean and applied Gelid GC-Extreme twice - I was advised that I may have used too much on the first application so did it again with less. I've also reseated the A50 an extra time to change the direction of the pipes as I wasn't sure which direction they should actually face. If anything, this made it worse.

 

So I was hoping that someone may be able to help. I'm entirely new to looking at cooling so please forgive me if I don't understand something fairly simple to you.

 

Thanks very much

Link to comment
Share on other sites

I'm not a fan of stress tests, Prime 95 is not a particularly good one, it's a bit harsh. What sort of temps does it get to in normal use? What voltage is it using?

 

I ran a 2500K (o/c to 4.4) for a couple of years, without reaching those sorts of temps, but I used either water cooling or a Noctua air cooler. How much TIM did you use? The recommended amount is a pea sized blob or a thin line. Too much can cause problems.

Link to comment
Share on other sites

The first time I applied the TIM I may have used too much as I did a small amount on the cpu and then on the cooler as well. So I removed that fully and then applied a small amount to the cpu, certainly not too much though.

 

In normal use, when it's idle it's around the 40 degrees mark. I'm not sure off the top of my head on the voltage, I will check that later. Would it help if I posted the BIOS settings I changed to overclock? I have them on email from a supplier...

 

Thank you for responding

Link to comment
Share on other sites

When you apply too much TIM, you usually get very erratic and "jumpy" temperatures at idle and load. Little air bubbles expand and pop causing the perceived heat spikes. It's not just a few degrees. It may move up and down as much as 10C, even at idle. Since you have already taken a second try and this does not sound like what you're describing, this is probably not the issue. By the way, with the GELID-X, don't bother with the spreader. It just makes things hard. Let the clamping force of cooler spread it out for you when you tighten it down.

 

With air coolers, the troubleshooting checklist is pretty short. If it's not TIM, the next place to look is the backplate and/or mounting system. Unfortunately, I have never seen an A50 and I don't know what kind of hardware is included for the mount. If it is tight to the board and feels sturdy, great. If not, then that is the place to start.

 

As Snapper mentioned, Prime 95 can be a killer and it's not the gold standard in testing it used to be. However, this is more of concern for more recent chips like Hasewell and Skylake and not so much for Sandy Bridge. Still, it's just good practice to run a second test of something else to make sure the test program is not responsible. There are free versions of AIDA64, OCCT, and Intel XTU that work just fine and generally do not have the same vampiric voltage drain of Prime95 on newer chips.

 

That said, 40C sounds a little high for idle unless it is very warm in your room (or case). So, if it's not the hardware than it has to be the voltage and BIOS settings. In this instance, I think the hardware stuff is easier to rule out before diving into BIOS. However, if by chance you are running 4.4GHz and the voltage is set to AUTO, we should set a specific voltage and try again before getting too carried away.

Link to comment
Share on other sites

Good to know regarding Prime95, thank you guys.

 

I'll try and get a picture of the setup tonight for you guys. I've got 2 fans coming today to help exhaust heat from the case - at the moment i'm only using the ones that came with the case so that may help. I'm also considering whether I get a high pressure fan too to blow air through the case. Currently not sure where the best position would be though....

 

So I'm not sure if it helps, but these are the settings that I've configured within my BIOS. My motherboard is a Gigabyte Z68AP-D3 and these were recommended by Overclockers.co.uk when I bought the motherboard from them.

 

 

Advanced Frequency Settings - 44x

CPU Ratio - 44x

Advanced CPU Core Features - Press Enter

Internal CPU PLL Overvoltage - Enabled

Real Time Ratio Change in O/S - Enabled

Press Esc

System Memory Multiplier - 16.00

Press Esc

 

Avanced Memory Settings - Press Enter

Performance Enhance - Standard

set the next line down to quick

Channel A - Press Enter

9-9-9-28 - then command rate 2n

Press Esc

 

Advanced Voltage Settings - Press Enter

Load-line Calibration - Enabled

Dynamic Vcore - +0.050v

Dram Voltage - 1560v

Link to comment
Share on other sites

I am not familiar with your board to specifically guide you, but the most important number is going to be the core voltage (Vcore). My best guess is it is under the "Advanced Voltage Settings" sub-menu. There does appear to be an offset of +.050v, but we need to know what that .05 is being added to. In general, the number we are looking for will be between 1.05 and 1.35v, however I suspect it is going to say AUTO when you find it.

 

 

Fans and case heat management can be important, but more so for quality of enjoyment (noise) or long use issues. They are not responsible for, nor can they prevent, Prime95 from hitting 90C in 10 seconds. Only the fans on the cooler will have a direct effect on the CPU temperatures in this setting.

Link to comment
Share on other sites

I'll get some more info on those bios settings tonight - I'm not too familiar so I'll make some notes and try and upload a couple of pics if necessary. Would it help to record the voltages shown in CPUID HWMonitor as well?

 

Ah ok, thank you for that information. I might see if I can try a different cooler on my processor - I have a spare PC in a cupboard that has a cooler. I just can't remember which cooler it is and if it will be compatible with my cousin. I'll also check that tonight too

 

Thanks again for the responses - really appreciate it

Link to comment
Share on other sites

HW Monitor (or other program) information could be helpful. It can tell us the actual voltage load at idle and 100% load, which is not always the same as what you set in the BIOS. It would tell you how much voltage is really being applied under the AUTO voltage table, something that otherwise would be a mystery.
Link to comment
Share on other sites

So, I've now got some exciting screenshots for you :)

 

5 min stress test was taken right at the end of the test - the CPU VCore seemed to stay around the 1.368V mark for most of it. I've then got a screenshot of HWMonitor once the stress test had finished so you can see the voltage there. When there isn't much activity, VCore seems to drop to around 0.970V

 

Does this help in terms of assessing the voltage?

1292127036_5minstresstest.thumb.png.d08b5b09b3c7675185fb00d21e743242.png

1164788871_Poststresstest.thumb.png.00e6701d46d02f26fe6288626ddfa865.png

Link to comment
Share on other sites

I am still curious about your Vcore value. Is it adding 0.05 to the Auto table? Or is it that in addition to a different value? The Vcore setting should be the second one down on under the Advanced Voltage Settings in the BIOS.

 

Either way, the effective value under load appears to be 1.368v with some normal overshoot to 1.38v. This is where we need someone who is or has been an i2500 owner. However, a little searching around leads me to believe you can run 4.4 or even 4.5 at 1.30v and I suspect the reduced voltage will take you back down under 80C. I am assuming this is your target/comfort zone. Any one with first hand knowledge is welcome to chime in, but I think you can stop taking the cooler on and off. Your temperatures appear to be voltage related.

Link to comment
Share on other sites

I will have a look at the BIOS as soon as I'm home - sorry, didn't get a chance last night. The target temperature is one that doesn't feel like my cpu is going to melt :) I'm not too specific really but I know that my temps are a lot higher than they should be. As long as they look relatively normal to more experienced guys like yourself then i'm happy

 

Just want to thank you for your help with this - very glad that I posted here

Link to comment
Share on other sites

I ran a 2500K at 4.4 with the Vcore set at 1.28 (with an H100i cooler and then a Noctua NH-U14S), but then again, not all cpus are the same ( I managed to get it to 4.8 but it wasn't stable and the temps were high). Do the cooler fan/fans ramp up as the temp goes up? You could also try knocking the o/c back down to 4.0 or even standard and see how that works.
Link to comment
Share on other sites

Hmm. The shot I was hoping for is not there. Is there a menu for "Advanced Voltage Settings"? The BIOS is unique to the board and I can only look at internet shots, so this is partial guesswork on this end. However, I think it is adding a +0.050 offset to the AUTO voltage table. Don't change it until it can be verified.
Link to comment
Share on other sites

Oh, you are quite right. I could not read it on this crappy laptop I am using. I was expecting to see a voltage number, but I only see the offset. How you overclock is unique to each BIOS/motherboard. I do believe it is adding +0.050 to the AUTO value, but I can't be sure. We need someone who knows that BIOS better than me.
Link to comment
Share on other sites

OK, well that brings me back to the inevitable moment where we discover a manufacturer has made a number of seemingly similar motherboards, but they have some significant differences in terms of actual use.

 

I do not believe you can set a specific voltage on the Gigabyte Z68-AP-UD3. Its specifications are listed as:

 

Overclocking: HTT frequency 80-200MHz, CPU multiplier: 16-59, max voltages, CPU +0.64V, QPI/VTT 1.7V, RAM 2.6V, PLL 2.52V, System Agent 1.35V, Graphics DVID +0.32V

 

This suggests to me the only way to configure voltage is to add (or maybe subtract???) an offset from the stock voltage table. This is an older style BIOS and an older method of overclocking.

 

From what I was able to find spread about, the voltage table looks something like this. Keep in mind every chip is unique.

 

Stock-4GHz on Stock VID

4.0-4.3GHz 1.300v-1.325v

4.3-4.5GHz 1.325v-1.375v

4.5-4.8GHz 1.375v-1.450v

 

 

So there are a few things you can try.

 

1) Turn that DVID (voltage offset) to 0. Hopefully that knocks 0.05v off your peak VID and would take us down to 1.32. This may get you under 80C on the stress test. Make sure you keep LLC enabled (it already is in your screen shot). I saw a few questions about voltage droop on this board. You want that on for stability.

 

2) Take your multiplier back down to 42 or 43 and see how that affects the VID in HWMonitor. This may also knock a few degrees of the max, but probably less appealing than option #1.

 

3) Leave it as is. Stress tests are one thing, but if you never break 60C in your normal use, there isn't a whole lot to worry about. You are not in the thermal shutdown range. So unless you have some really heavy and continual CPU loads to deal with, max temp testing may be irrelevant.

 

4) Get a heavier duty cooler. Bigger is usually better. There are thicker air towers and of course with water cooling 1.37v shouldn't be an issue. Perhaps not what you want to hear, but you may still have options open at the moment.\

 

I am fairly certain all of your temperature is voltage related, however the prudent thing to do (as suggested by someone else early on) is to put it back at the stock multiplier and run the stress test. A bad contact between the cooler and CPU lid will still give high temps regardless of voltage. When it comes back with lower numbers, you will know the lowest possible point for that cooler and CPU combination and can then properly evaluate the cost/benefits of higher clocks.

Link to comment
Share on other sites

Dropping that DVID to 0 has made quite big improvements to the temperature - I did that and ran a 30 minute stress test through Intel XTU - The highest temp was 75 degrees after that

 

Would you say that the 30 minute test is enough to say that it is stable?

927017119_HWMonitorvoltagedrop.thumb.png.a794522f792e03fbd2e025d74f06ee67.png

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...