How hot is too hot? cpu at 55c under no load

[FlippedBit]

I think this depends on how you define "idle" when measuring. I noticed with my own CPU that I get slightly different idle temps in the BIOS hardware monitor vs. a software monitor with Windows running. I think that using the BIOS hardware monitor is a true idle, whereas if the OS is running there will be some small amount of load.

My Core i7-920 was running a bit warm and I noticed a difference after blowing off the dust and reseating the heatsink and fan, as suggested. It now idles around 37C in BIOS and high 30's to low 40's in Windows *if* I have the fan on full speed. If I turn the fan down or use auto, it idles low to mid 40's.

That's with a Hyper 212 Evo heatsink+fan, but, my case is very small/cramped and I only have a single 80mm intake fan, and a single 120mm rear exhaust. I've been running that way since the Core i7-920 was released years ago (not overclocked) and never had a problem! Games run fine.

 

[Sonikku]

try hardware monitor

http://www.cpuid.com/softwares/hwmonitor.html

the cpu hovers around 70-80c now when web browsing, but much more worrisome now is that 123c system temperature. What the hell is going on now...

 

[escrow4]

What are your exact system specs, especially the case? 50 degrees for a HDD is rather warm.

 

[ShintaiDK]

By the looks of it its your case airflow its wrong with.

50C HD, 65C mobo.

 

[Sonikku]

What are your exact system specs, especially the case? 50 degrees for a HDD is rather warm.

Antec Sonata 2
Asrock H77M
4gb ddr3 ram
Intel Pentium g630 sandy bridge
Radeon 7770
Windows 7 64bit

 

[Ketchup]

By the looks of it its your case airflow its wrong with.

50C HD, 65C mobo.

I saw that too. And I agree.
What is your case fan configuration like?
What is your hard drive (SSD) configuration like?

 

[Sonikku]

Haven't had the chance to upgrade to a ssd yet.

 

[Sonikku]

Ok. I cracked her open, blew out a TON of dust with compressed air and got the heatsink remounted. Much to my disappointment, the heatsink was fastened very securely, ruling out that as an issue. After cleaning the crud off the cpu and and heatsink with the 90% alcohol stuff and Q tips I let it dry and used the drop method with arctic silver before getting all the points on the heatsink to double click.

After letting it set for a while I hooked everything back up again and turned it on. I immediately went into the bios and the temp settings and watched with shock as the cpu temp steadily crept up to 85c. The stock sink was moving at about 1900 rpm. I sighed and logged into windows anyway and launched speedfan and found it no longer seemed quite as glitched. Here is what it was before I started when both of the cores were running 70-90c from web browsing.

and here is what it is now.

A lot cooler, but strangely both of the individual cores seem to run 15c hotter than the generic cpu reading.

I still don't have an explanation as for why everything from software from the bios tells me the heatsink is spinning at 100%, but it was able to ramp up from 2000rpm to 2800rpm when gaming for extra cooling a week ago. Now I can't seem to get it to ramp up at all and it seems stuck at 2000rpm no matter what the temp. ugh.

 

[LTC8K6]

Is your BIOS current?

 

[sm625]

Just reach your finger in there and touch the cpu heatsink. If it is really idling at over 60C then it will feel hot as hell, too hot to keep your finger on it for more than a second. I'm betting that its actually not hot at all, and your readings are simply bunk. A G630 is a tiny chip, and at idle it is simply incapable of producing that much heat into a fully functional hsf.

 

[Ketchup]

Is hardware monitor showing you the same temperatures? Including hard drive?

I have used Speedfan a few times, but there always seemed to be a better option available, and I remember the older versions giving me more inaccuracies than other programs.

Also, can you give us an estimation on ambient temp?

 

[KingFatty]

If the CPU temp reading is unreliable, you might have a motherboard issue [confirm by touching base of heatsink with finger or thermometer to check against the value reported by the motherboard].

If the fan speed reading is unreliable, you might have a motherboard issue [confirm by downloading an app on your phone to accoustically check RPM against the value reported by the motherboard].

 

[Sonikku]

I just opened the case while a game was running and the apps were reporting a 90c core temp, but the heatsink felt cool. Is it possible I didn't use enough paste?

Is hardware monitor showing you the same temperatures? Including hard drive?

I have used Speedfan a few times, but there always seemed to be a better option available, and I remember the older versions giving me more inaccuracies than other programs.

Also, can you give us an estimation on ambient temp?

Yes and yes. The temps are coroberated by other programs. Even in the bios the cpu temp can reach up to 90c.

 

[Ketchup]

Bad contact, or the board's temp sensors/reporters have gone haywire. I lean more towards a mounting issue due to the heatsink being cool while running a game. You might be better off getting a cooler that mounts directly to the board, instead of the cooler mounts.

 

[ShintaiDK]

I just opened the case while a game was running and the apps were reporting a 90c core temp, but the heatsink felt cool. Is it possible I didn't use enough paste?

You only need to use a razorthin amount of paste. Too much paste and it works the other way around.

Did you try to reseat the cooler?

 

[Sonikku]

Yes I did. Watched all the youtube videos. Made sure all four plugs were turned the right way and pushed in until they double clicked. It was fastened very tightly by the time I was done.

 

[MongGrel]

Just from what I've read I'd slap an EVO on it and get some Noctua NT-H1 paste and stick a pea sized dab on the middle of the CPU, and re seat it.

That is really hot there.

Unless you want to get fancy IMHO.

AS5 is really out of date.

MX-2 still good these days I imagine.

 

[KingFatty]

If the heatsink is in good/proper thermal contact with the CPU, the base of the heatsink will remain approximately the same temperature as the CPU, that's inescapable physics.

If the heatsink's base is cool to the touch, then that means your CPU would also be cool to the touch. But if you are stressing the CPU, it cannot be cool to the touch.

So, how do you explain the heatsink base remaining cool? Maybe you'll need to give the heatsink base more time to warm up with the CPU? Run a benchmarking program to heat the CPU for an extended time like 5 minutes and periodically check the heatsink base to confirm it heats up.

Any chance you have a portion of the CPU exposed while the heatsink is installed? Feel around to look for any heat differences.

 

[LTC8K6]

I think the reported temps must be wrong.

 

[sm625]

I just opened the case while a game was running and the apps were reporting a 90c core temp, but the heatsink felt cool.

OK that's what I expected. G630 doesnt even have AVX so there is practically no way to get it to produce enough heat to ever truly heat a properly functioning stock HSF, especially if it thinks it needs to be pushing the fan to full speed. So now the only question is... is the cpu actually makign contact with the HSF? Here is how you test that: unplug the fan. Let it run under load without a fan. Try to use prime95 and make sure task manager shows cpu is at 100% load. With the fan unplugged the heatsink should get hot after a couple minutes. If this happens then everything is normal and you dont need to worry about the bogus temperature readings. That cpu will never overheat when the fan is running.

 

[EliteRetard]

Just from what I've read I'd slap an EVO on it and get some Noctua NT-H1 paste and stick a pea sized dab on the middle of the CPU, and re seat it.

That is really hot there.

Unless you want to get fancy IMHO.

AS5 is really out of date.

MX-2 still good these days I imagine.

Where do you get the idea that AS5 is "out of date"?

It's right in line with all the other name brand pastes. NT-H1 is not significantly better, at least not with stock CPU's and coolers, the gap may grow in extreme cases...but even then by only a few degrees. They did test an H80i AIO water cooler on an FX-8350 as well, and the performance differences even there were still less than 1c.

 

[EliteRetard]

And I'll agree with some of the others here, try running a stress test program while monitoring your CPU frequency and feel the heatsink once a minute or two up to 10 minutes or so. You should be able to notice the heatsink getting warmer. You can also unplug the fan as someone else suggested, that along with a full CPU load should get the heatsink notably warm/hot.

If the CPU frequency is dropping though with high temp readings then that would suggest it is indeed thermally throttling (stop testing if frequency drops a lot). I know one issue can prevent good contact with the cooler, even if it's properly clamped down. That would be the bracket that holds the CPU itself.

I have run into some that were made wrong and actually stuck up enough to interfere with CPU coolers. You can check by laying a straight edge (ruler, razor, torpedo level, etc) across the CPU and mounting bracket at various angles. The straight edge should always rest on the CPU, if it rubs against the bracket at all that's a problem.

 

[Sonikku]

Should I buy this? http://slickdeals.net/f/7833095-zal...ooler-cnps14x-9-99-after-rebate-free-shipping

Or wait to work out these other options first to see if it's something else? Money is kind of tight.

 

[LTC8K6]

If the heatsink is not warm with a reported high temp under load, then the CPU is not actually hot, or the heatsink is not seated properly.

If the CPU is actually hot, and the heatsink is doing it's job, then you should feel heat on the heatsink fins.

 

[MongGrel]

Where do you get the idea that AS5 is "out of date"?

It's right in line with all the other name brand pastes. NT-H1 is not significantly better, at least not with stock CPU's and coolers, the gap may grow in extreme cases...but even then by only a few degrees. They did test an H80i AIO water cooler on an FX-8350 as well, and the performance differences even there were still less than 1c.

If you like using conductive stuff, that takes awhile to cure, knock yourself out I guess.

I'd even use MX-2 over AS-5 any day, IMHO.