2500K - do I really need aftermarket cooling? How much further could I get?

S

Spare-Flair

Supreme [H]ardness
Joined
Apr 4, 2003
Messages
7,471
It's been awhile since I've been in the overclocking game. I lived with my Q6600 B3 @ 3.4GHz for what seems like 3-4 years now. Amazing, the longevity of that thing.

I've just upgraded and now and I am completely out of the loop of Sandy Bridge overclocking and I'm just relying on my ASUS P8Z68-V Pro to handle all the automatic turbo mode overclocking. I've my got 2500K stable @ 4.43GHz just out of the box without any significant tweaking just running the ASUS auto tune utility and I am running on the stock Intel cooler.

I am thinking about upgrading it to a Corsair H70 or H80 but for the addition $90, would I really get any appreciable gains from the auto-turbo overclocking of this CPU? What are people's usual expectations and results?

I also have a second 2500K on a separate system I built on an H67 board so I can't test overclocking on that computer. Would it be worth it to go through the effort of swapping both CPUs around to see which overclocks the best or are 2500K results pretty normal across the board?
 
You can save a bundle by just getting something like the CoolerMaster Hyper 212+ - which is about as good as the H70 but costs less than $30 normally. And yes, if you are overclocking into the mid 4s, you'll want aftermarket cooling, the stock cooler isn't up to it at all.

Probably not worth fiddling with the other 2500K - they all tend to do about the same at moderate overclocking (into the 4.6 range), above that you might see a difference with different chips, but then you start reaching the point of diminishing returns.

Also check to see what the auto-tune is doing with the BCLK - sometimes those programs will up the BCLK for no good reason, it's better to just bump the multiplier a notch instead.
 
Switching to a good aftermarket cooler will not give you anything like an extra 1GHz. You should probably get to no more than 4.8-5.1GHz like most people. The best you will gain from a cooler at 4.4GHz is a significant lowering of your CPU's core's temperature and a probable drop in fan noise. Unless you are posting from a frozen wasteland your chip must be practically glowing when you stress test it at that speed with a stock cooler.
 
Spare-Flair said:
It's been awhile since I've been in the overclocking game. I lived with my Q6600 B3 @ 3.4GHz for what seems like 3-4 years now. Amazing, the longevity of that thing.

I've just upgraded and now and I am completely out of the loop of Sandy Bridge overclocking and I'm just relying on my ASUS P8Z68-V Pro to handle all the automatic turbo mode overclocking. I've my got 2500K stable @ 4.43GHz just out of the box without any significant tweaking just running the ASUS auto tune utility and I am running on the stock Intel cooler.

I am thinking about upgrading it to a Corsair H70 or H80 but for the addition $90, would I really get any appreciable gains from the auto-turbo overclocking of this CPU? What are people's usual expectations and results?

I also have a second 2500K on a separate system I built on an H67 board so I can't test overclocking on that computer. Would it be worth it to go through the effort of swapping both CPUs around to see which overclocks the best or are 2500K results pretty normal across the board?
Click to expand...

What temps does that run at if you run IntelBurnTest (run the test on High)? Those stock coolers are barely able to get the CPU from going into thermal protection at stock clocks once you get a 100% load...
 
my personal experience with 2500k

4200 mhz overclock
stock fan= 95c prime95 FAIL

4800mhz OC
30$ dc cooler = 69C prime95 WIN
 
Honestly the 2500K is already super fast at stock speeds. You really do not have to overclock a 2500K. I mean sure you can and there is NOTHING wrong with it but really they are already pound for pound so fast compared to anything else that is a true quad core.
 
tangoseal said:
Honestly the 2500K is already super fast at stock speeds. You really do not have to overclock a 2500K. I mean sure you can and there is NOTHING wrong with it but really they are already pound for pound so fast compared to anything else that is a true quad core.
Click to expand...

By the same token though, these chips are probably some of the easiest to OC in history and they all seem to have ample headroom, so why not take advantage of it? Personally I wouldn't bother testing two different 2500K unless I had time to waste or two mobos to test concurrently with (to save time), I also wouldn't spend more than $30-40 on a HSF for a run of the mill OC... I do think it's worth getting an aftermarket HSF tho, if for no other reason than having generally cooler temps and less noise (vs stock unit), Hyper 212+, Cogage TRUE Spirit, etc.
 
tangoseal said:
Honestly the 2500K is already super fast at stock speeds. You really do not have to overclock a 2500K. I mean sure you can and there is NOTHING wrong with it but really they are already pound for pound so fast compared to anything else that is a true quad core.
Click to expand...

Grab your pitchforks, men! :D
 
The stock cooler, when properly seated, will keep the CPU cool enough at stock speeds. If you start overclocking, your temps will rise quite a bit. At your speeds, I would not be surprised is you are running near shutdown temperatures. Get CPUID Hardware Monitor or similar program to check actual core temperatures. I know with my Gigabyte board, the software that came with that checks some other temperature or just checks the package temperature but is 10°C lower than Hardware monitor.

At stock speeds (3.5), the 2600K is at 70°C Package on stock cooler with the Large FFT Prime running. At 4.2GHz on stock voltages, it will jump up to around 90°C on the package temp..
This is with plenty of case cooling.
 
+1 on the Corsair H70 ;)

It's a good investment that'll only add to the longevity of your CPU.
 
I have a Corsair H50 and am stable at 4.8 and the max temp I've hit is about 55c
 
Paragon said:
At stock speeds (3.5), the 2600K is at 70°C Package on stock cooler with the Large FFT Prime running. At 4.2GHz on stock voltages, it will jump up to around 90°C on the package temp..
Click to expand...

Er, how does the temp jump if there's no voltage increase?

Wouldn't it be more likely that the OC will simply fail due to there simply not being ENOUGH voltage to sustain it?

In order to generate more heat, you need more voltage.
 
gweminence said:
Er, how does the temp jump if there's no voltage increase?

Wouldn't it be more likely that the OC will simply fail due to there simply not being ENOUGH voltage to sustain it?

In order to generate more heat, you need more voltage.
Click to expand...

No, the power usage is a function of both voltage and frequency (although it is voltage squared, so that's a much bigger factor). Power = heat, so there you go.
 
dude if your low on budget buy a 30$ 120mm direct contact cpu cooler (cm 212 xigmatek gaia etc)
having a k series cpu and not oc it is MURDER:mad:.
your cpu will run much cooler than crappy intel cooler and Dead silent if you dont feed it +1.35V
 
Forceman said:
No, the power usage is a function of both voltage and frequency (although it is voltage squared, so that's a much bigger factor). Power = heat, so there you go.
Click to expand...

Power = Voltage * Current OR
Power = Resistance * Current^2 OR
Power = Voltage^2 / Resistance

There is no frequency component in power useage, so I'm not sure what you mean.
 
gweminence said:
Power = Voltage * Current OR
Power = Resistance * Current^2 OR
Power = Voltage^2 / Resistance

There is no frequency component in power useage, so I'm not sure what you mean.
Click to expand...

CPU power usage, as opposed to say fan power use (which would folow the formulas you mention), is a function of both frequency and voltage. Or did you think that extra performance was free?

The actual formula, I believe, is: P = f*c*V^2 where c is the capacitance.

Edit: See more here:
http://en.wikipedia.org/wiki/CPU_power_dissipation
 
Need? No. I ran 4ghz+ on the stock cooler, no issues. However, I ended up buying a 212+ because it is so cheap and effective.
 
Forceman said:
CPU power usage, as opposed to say fan power use (which would folow the formulas you mention), is a function of both frequency and voltage. Or did you think that extra performance was free?

The actual formula, I believe, is: P = f*c*V^2 where c is the capacitance.

Edit: See more here:
http://en.wikipedia.org/wiki/CPU_power_dissipation
Click to expand...

The laws of physics don't change between fans and cpu's, I'm afraid.

It's not free. As frequency rises, more voltage is required. In the case of an auto setting, (or, on some boards, a vdroop setting), the board has latitude to draw more as needed -- ie, as frequency ramps up. So it does. The cpu draws more power, and heats up. Even in cases where you hard-set your vcore in bios to, say, 1.300, manufacturing inconsistencies and other random things will allow slight variations of voltage change as the freq goes up or down. Cases of OC failure are caused by the motherboard reaching a hard-limited setting in vcore, but requires more than it can get for a given frequency.

However, the issue remains the same: you can have any frequency you like at a given voltage, without an increase in power consumption. The setting might REQUIRE more consumption than you allow it, and so might fail. But frequency change isn't causing heat. Only voltage can do that.
 
gweminence said:
The laws of physics don't change between fans and cpu's, I'm afraid.
Click to expand...
True but irrelevant.

gweminence said:
Power = Voltage * Current
Click to expand...
More specifically the equation that always holds is:

"instantaneous power" = "instantaneous voltage" * "instantaneous current"

There are some equations you can apply to different types of average but they all rely on assumptions that may or may not hold in the case you are dealing with. In the case of digital electronics we can reasonably assume voltage is a constant giving us an equation of.

"mean power" = "constant voltage" * "mean current"

gweminence said:
But frequency change isn't causing heat. Only voltage can do that.
Click to expand...
You seem to be under the impression that for any device you can define a constant R which relates voltage and current. The simple fact is you can't. To predict the impact of different things on power consumption you have to understand the device in question.

A modern digital IC (be it a CPU, a GPU, a NB or whatever) is a collection of CMOS logic gates. Since CMOS logic gates have both N-Channel and P-Channel mosfets and do not need to use pull-ups or similar they draw little current when not switching.

However when a CMOS logic gate switches two things happen that cause a spike of current.
1: as the input voltage crosses the thresholds both the P-Channel and N-Channel mosfet are briefly partially on at the same time.
2: as the gate switches it must charge/discharge any capacitance on it's output. The current to do this must come from the power rails.

As clock frequency rises so does the switching rate of the gates. That means that the rate of current spikes increases and the average current increases.
 
gweminence said:
The laws of physics don't change between fans and cpu's, I'm afraid.

It's not free. As frequency rises, more voltage is required. In the case of an auto setting, (or, on some boards, a vdroop setting), the board has latitude to draw more as needed -- ie, as frequency ramps up. So it does. The cpu draws more power, and heats up. Even in cases where you hard-set your vcore in bios to, say, 1.300, manufacturing inconsistencies and other random things will allow slight variations of voltage change as the freq goes up or down. Cases of OC failure are caused by the motherboard reaching a hard-limited setting in vcore, but requires more than it can get for a given frequency.

However, the issue remains the same: you can have any frequency you like at a given voltage, without an increase in power consumption. The setting might REQUIRE more consumption than you allow it, and so might fail. But frequency change isn't causing heat. Only voltage can do that.
Click to expand...

You can define it any way you like, but the fact remains that increasing frequency increases the power draw. If you want to say that the increased frequency causes an increase in current draw (which is what I believe happens), fine. But greater frequency equals greater power. I didn't make up the formula.

Edit: What plugwash said.
 
Last edited:
212+ and don't look back. If you have no plan on overclocking to the limit then this fan is more than sufficient. AT stock and 100% load in prime 95 my max is 48c thats 28c above ambient. I mean for 25 dollars I was shocked at the performance of this cooler. I can not wait to get my new PSU in today So I can start Overclocking this bad boy. My max idle is at 28c So i am hoping I will be able to get a solid 4.5 with very little voltage increase.
 
It's a good idea... but if you dont plan to overclock and you won't put after market cooling... Go ahead and buy a non k version and save a few bucks
 
yes you really do need an aftermarket cooling. i wouldn't leave it at stock clocks even with the stock cooler.
 
naizarak said:
yes you really do need an aftermarket cooling. i wouldn't leave it at stock clocks even with the stock cooler.
Click to expand...

Now that's absurd :D

I'm debating between the H60 and H80 right now. I also have a Thermalright True 120 Black but I don't know if I have the mounting hardware for Sandy Bridge with it.

I'll run some benchmarks and burn tests @ 4.43GHz w/the stock cooler and see what temps it hits. Temperatures are not really a concern of mine, Intel CPUs are always quite robust. I'm more concerned about overclocking potential with improved cooling.
 
It'd probably be cheaper and more effective to buy the mounting kit for the TRUE... Not hard to find.
 
I'd recommend a cheap/good cooler like the coolermaster. But that said, I've been running the stock cooler on my 2500k at 4Ghz 24/7 at lower than stock voltage (1.25 I believe) and temps are no higher than at stock speeds/voltage. I haven't tried going above 4Ghz due to the stock cooler, so I can assume things change drastically beyond that point?
 
You must log in or register to reply here.
Back
Top