Haswell Stock Coolers - Good or Bad?

SpeedKing

Senior member
May 21, 2003
240
0
0
#1
I just upgraded my HTPC to an Intel i3 4130 running on a Gigabyte H81 board and decided to give the stock cooler a go, but I am pretty disappointed with the cooler's performance.

My desktop is an i7 4770K running at stock with a Coolermaster Seidon 240M liquid cooler and that idles at ambient (30C average in the Australian summer) and under load from Prime95 or Intel Burn Test it can reach around 65C.

The upgraded HTPC idles just shy of 40C and under load hits the mid 80s after running for 15 minutes or so. The heat sink is hot so I figure I attached it properly as it is transferring heat, however I am suspicious of the pre-applied thermal paste from Intel. It was applied in three strips; I've only ever used the pea drop in the centre and let that spread when fastening the cooler.

I also would have figured a i7 4770K running at 95W (Reported by HWMonitor under load) would pump out much more heat than a i3 4130 running at only 50W (Reported by HWMonitor under load).

I've just moved from an AMD X2 5200+ HTPC and C2Q Q9550 desktop, and I must say while Haswell is many times faster, it seems to run much, much hotter.

Any one else experience this with stock cooling performing so bad? Would third party thermal paste make the Intel cooler perform better or is it a lost cause?

I do have a Scythe Shuriken 100mm low profile cooler from my old AM2+ board which is Socket 115X compatible but I have lost the mounts for it for the Intel platform. Would this be a better cooler than the stock Intel one?
 

dma0991

Platinum Member
Mar 17, 2011
2,723
0
0
#2
The real question is that you're putting unrealistic loads on a HTPC that will never run at 100% for 15 minutes on a normal use case. Sure, the stock heatsink is borderline adequate, but it isn't apples to apples comparison either to compare it with another CPU with an expensive aftermarket cooler.

Plus there's other factors not taken into account such as whether the case temps are too high since HTPC cases would be smaller and have worse airflow compared to a full sized case.
 
Apr 22, 2012
20,395
0
106
#3
Tjmax is 100C. And Intel guarantees operation below that on stock settings with the stock cooler. Its not as hot as you think either. I bet your C2Q and X2 chipsets are in the 90-100C range. And the X2 you dont have any remotely accurate temperature measuring on the CPU.

SB/IB runs at the same temperatures too.
 

Essence_of_War

Platinum Member
Feb 21, 2013
2,649
0
76
#4
The response to the initial questions is "adequate" :p

Mid 80s under prime95 with stock cooler is not at all unreasonable. As ShintaiDK noted:
Tjmax is 100C. And Intel guarantees operation below that on stock settings with the stock cooler. Its not as hot as you think either.
and to add to that, I think throttling happens at 95C, so you're def within safe operating limits even under preposterously heavy load, as no HTPC stuff, indeed, probably no actual real-world workload in general, comes close to prime95 stress tests.

Would third party thermal paste make the Intel cooler perform better or is it a lost cause?
I would not apply third party thermal paste to the stock cooler. The stock cooler isn't actually a thermal paste, it's more of a thermal pad that's intended to act alone. Adding additional substances to the contact surface actually has a good chance of making the situation worse.

I do have a Scythe Shuriken 100mm low profile cooler from my old AM2+ board which is Socket 115X compatible but I have lost the mounts for it for the Intel platform. Would this be a better cooler than the stock Intel one?
It would probably be a better cooler, but honestly, unless you're going to BCLK OC the i3, it's probably not going to do much for you other than be a bit quieter under load if you set up a nice fan curve for it. I would go for it if the stock cooler is too noisy for your taste, but not if you're just worried about getting better temps on prime95. Does the noise bother you under HTPC loads?
 
Apr 22, 2012
20,395
0
106
#5
Throttle happens at 100C. Shutdown at 130C or so.
 

Essence_of_War

Platinum Member
Feb 21, 2013
2,649
0
76
#6
Throttle is at 100C? Hmm, I must have confused it with Ivy Bridge, then? :confused:
 
Apr 22, 2012
20,395
0
106
#9
I think Lynnfield was last to have 95C tjmax. SB was 98C if I recall right.
 
Jun 3, 2011
10,047
134
126
#10
You might have gotten one of the Horribad(tm) haswells, mine was decent, but you might have to delid. 65c load at stock with water seems way too high, i'd expect it to be around 45c.

Most people get 70c at 100% load when running their 4770k @ 4.5Ghz / 1.30~ volts (on H100i), or lower.

i assume you know everything about the botched haswell TIM job ?
 
Apr 22, 2012
20,395
0
106
#11
Never use the idle/low temperature as any relevance. Tj can be very inaccurate until it reaches tjmax, where its 100% accurate.

It looks something like this:
 

Khato

Golden Member
Jul 15, 2001
1,013
0
81
#12
About the only time I don't make use of the stock Intel cooler is if I intend to either overclock or frequently make use of AVX instructions, aka CPU based video transcoding. Outside of those two scenarios the stock cooler will easily keep the CPU cool at near silent noise levels.

Keep in mind that a number of the stress testing programs do make use of AVX instructions and are not at all representative of actual workloads. From what I've seen AVX instructions make a larger difference than overclocking - my i7-4770k running at 4.5 GHz runs cooler with non-AVX stress test than it does at stock running an AVX stress test.
 

SPBHM

Diamond Member
Sep 12, 2012
4,798
40
126
#13
85 under prime probably means under 75 for most CPU intensive tasks, so I wouldn't call it good or bad, just enough.
 

zir_blazer

Senior member
Jun 6, 2013
870
5
81
#14
The Retail Heatsink is really a piece of trash. It barely suffices, and depending on other factors (High ambient temperature, poor case airflow), you can get to thermal throttling scenarios during continuous full load scenarios quite easily. On a Core i7 4770/4770K or a Xeon E3-1245V3 and above it is even worse than yours because it pumps out quite a bit more heat (Hyper Threading specifically makes a difference), yet they still have the same crappy Heatsink than the model you have.
The Heatsink was included for Retail Ivy Bridge parts (Keep in mind those temperatures are in an open testbed, inside a case they will be worse), and the top model, Core i7 3770K, consumed slightly less than the 4770K. If the Heatsink was designed around Ivy Bridge TDP with maybe a small safety margin, it falls a bit short on Haswell.
 

dma0991

Platinum Member
Mar 17, 2011
2,723
0
0
#16
If the Heatsink was designed around Ivy Bridge TDP with maybe a small safety margin, it falls a bit short on Haswell.
The stock heatsink supplied is proportional to its rated TDP. I've had a side by side look at an Intel G620 (65W) and an Intel Core i7 2600K (95W) heatsink. They have different thickness and there's a copper base on the Core i7 2600K versus all aluminum for the G620.

Under stock settings, I wouldn't even worry about the temps and stability, even more so when a typical HTPC load isn't that heavy. Besides, thermal throttling will prevent it from permanent damage.
 

JDG1980

Golden Member
Jul 18, 2013
1,563
46
136
#17
The Haswell retail heatsink is pretty terrible. On low-TDP models (less than 65W) it will suffice if you aren't picky about low noise levels. For the K-series, I wouldn't even consider it for anything other than testing purposes. Regardless of what Intel says about TjMax, their warranty period is 3 years and it is in their financial interest to sell you a new CPU as soon as possible. IMO, this is why they switched from solder to paste. Intel wants to avoid the current situation where a lot of people keep their CPU for 7 years or more. Running closer to the thermal limits for long periods of time makes the processor go bad sooner.

You can get a decent aftermarket cooler for $20-$30. At that price, it's not worth taking the risk of going with stock.
 

zir_blazer

Senior member
Jun 6, 2013
870
5
81
#18
The stock heatsink supplied is proportional to its rated TDP. I've had a side by side look at an Intel G620 (65W) and an Intel Core i7 2600K (95W) heatsink. They have different thickness and there's a copper base on the Core i7 2600K versus all aluminum for the G620.

Under stock settings, I wouldn't even worry about the temps and stability, even more so when a typical HTPC load isn't that heavy. Besides, thermal throttling will prevent it from permanent damage.
I'm talking about Ivy Bridge (77W TDP) vs Haswell (84W TDP) Heatsinks which are identical according to what Legit Reviews said, at least for the mainstream models that have those TDPs. I'm sure that if you pick a Haswell Celeron with 53W TDP they will have a smaller Heatsink, but is not my point.
My point is that Intel picked a Heatsink intended for those 77W TDP Ivy Bridge and placed it on top of a 84W TDP Haswell, and even worse if you consider that real power consumption is slighty higher for Haswell according to the other link.

Thermal throttling will save it from permanent damage, but that also means that you can't enjoy the 100% of performance of your Processor when you need it the most (Else you wouldn't be Full Load!) if Intel underdimensioned the Heatsink and put it close to those scenarios.


IMO, this is why they switched from solder to paste. Intel wants to avoid the current situation where a lot of people keep their CPU for 7 years or more. Running closer to the thermal limits for long periods of time makes the processor go bad sooner.
I said something similar to that not long ago.
 
Apr 22, 2012
20,395
0
106
#19
My point is that Intel picked a Heatsink intended for those 77W TDP Ivy Bridge and placed it on top of a 84W TDP Haswell, and even worse if you consider that real power consumption is slighty higher for Haswell according to the other link.
You do know 95W SB, 77W IB and 84W HW uses the same heatsink, right?

So its actually designed for 95W TDP.
 

zir_blazer

Senior member
Jun 6, 2013
870
5
81
#20
Nope, I didn't know that. I through that Sandy Bridge had a better Retail Heatsink due to the 95W TDP (And also that you didn't hear people complaining about temperature with it as much as with Haswell), didn't knew that Intel actually reused that one for all three generations, just the last two.
 
Apr 22, 2012
20,395
0
106
#21
Nope, I didn't know that. I through that Sandy Bridge had a better Retail Heatsink due to the 95W TDP (And also that you didn't hear people complaining about temperature with it as much as with Haswell), didn't knew that Intel actually reused that one for all three generations, just the last two.
i7 2600K=216mm2. i7 3770K=160mm2. i7 4770K=177mm2.

Thats the reason. And the temperature is fine at stock.
 

zir_blazer

Senior member
Jun 6, 2013
870
5
81
#22
i7 2600K=216mm2. i7 3770K=160mm2. i7 4770K=177mm2.

Thats the reason. And the temperature is fine at stock.
You're forgetting the solder vs paste factor. If we go for die size, Ivy Bridge and Haswell should be extremely close due to the fact that Haswell die is a bit bigger, and it consumes a bit more power (Yet both less than Sandy Bridge), but actually Haswell seems to run quite a bit hotter regardless that.
 
Apr 22, 2012
20,395
0
106
#23
You're forgetting the solder vs paste factor. If we go for die size, Ivy Bridge and Haswell should be extremely close due to the fact that Haswell die is a bit bigger, and it consumes a bit more power, but actually Haswell seems to run quite a bit hotter regardless that.
Haswell runs colder than IB. I got 2 identical cases. Same coolers. And the i5 4670 is running at a lower temp then the i5 3570K.
 

zir_blazer

Senior member
Jun 6, 2013
870
5
81
#24
Haswell runs colder than IB. I got 2 identical cases. Same coolers. And the i5 4670 is running at a lower temp then the i5 3570K.
Both with Retail Heatsink and the gray strips of paste for the first application? What are the actual temperatures at Full Load?

After googling for an hour or so, I found some data. Some people say cooler @ stock, but that Haswell is much hotter than Ivy Bridge on overclocking conditions.
For actual test, what I found is that is rare to see results with the default heatsink, and of those that does, they're not done in the same circunstances, so you need to account for differences. The best one I found was this guy with a Core i5 3550 (100 MHz slowers than yours), reporting 83°C under Prime 95 inside a case. Hyper Threading should had around 5-7°C or so if it was a Core i7 3770K plus some more due to the extra Frequency, in what case it should be above LegitReviews Haswell, through that one is on an open testbed.
 
Feb 6, 2002
17,170
27
91
#25
Prime is a lot more intensive than watching video.

I was trying to test a hard drive on an i3 4330 and it starts cold at about 35c and after a while it ran at about 45-48c trying to run windows install. So I decided to turn the Power supply over and suck air from the case and out the back and see if it runs better.

I am not so sure prime is really a safe program to run.

So you plan on playing games or what?

I have mine on a Gigayte Z87N wifi in a Cooler Master 130 elite case. It does seem like the processor does warm up a bit. In an ITX case there is just not enough air flow really. I cant imaging a gaming video card in there heating up also.

I was wondering how to mount an extra fan in that case. I was thinking maybe you could use the extra Video Card IO Slot and put some kind of fan in that blowing out the side. It comes with a fan on the other side but it is blocked with all the wires. Since I am not using a video card maybe using the slot space for a low RPM Fan Might be a nice idea.
 
Last edited:

Similar threads



ASK THE COMMUNITY

TRENDING THREADS