Haswell Stock Coolers - Good or Bad?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

nwo

Platinum Member
Jun 21, 2005
2,309
0
71
#26
I see no difference between my i7 4770k Haswell stock HSF and my Ivy Celeron G1620.
 

SPBHM

Diamond Member
Sep 12, 2012
4,856
81
126
#27
I see no difference between my i7 4770k Haswell stock HSF and my Ivy Celeron G1620.
isn't the 4770K the same as the 2600K (95W), it makes no sense to use the same cooler for the Celeron?

just checked some google images and the 2600K looks a little taller than the G1620 one, also, are you sure the fan can't go faster on the i7 cooler? and what about the base? intel used to have all aluminum for cheaper CPUs and copper for the higher TDP ones,
 
Apr 22, 2012
20,395
0
106
#28
There is 3 coolers if I recall right. 95W(copper), 65W(alu base) 45W for T models that is used for the 3 generations.
 

SpeedKing

Senior member
May 21, 2003
240
0
0
#29
Thanks everyone,

I have read up on the Haswell TIM issue, and understand that most of the time my CPU will be almost running at idle when decoding 1080p and below content.

I do a bit of video encoding using the HTPC, which can run it at load for 30 minutes or more at a time, so the use case of 100% load is not unreasonable from that point of view. The old Athlon X2 could cope with this fine even with it's stock cooler.

Maybe I will shift my encoding to my desktop which can cope temperature wise where in the past I had it batch scheduled on my HTPC so it could be done while I wasn't home.
 
Jul 12, 2000
27,225
23
81
#30
It just...is. I haven't installed my new cooler yet so am still on stock. I did back it down to 4ghz until it is set up, but it idles fine and unless you are stress testing or have a poor ambient you can get away with it temporarily. Edit: Added two minutes of 100% load since someone mentioned idle usage is BS.




Load

 
Last edited:

Essence_of_War

Platinum Member
Feb 21, 2013
2,649
0
76
#31
I do a bit of video encoding using the HTPC, which can run it at load for 30 minutes or more at a time, so the use case of 100% load is not unreasonable from that point of view. The old Athlon X2 could cope with this fine even with it's stock cooler.
100% load of video encode != 100% load prime 95.

Prime95 stress tests are a simulation of basically a power virus. You should see markedly higher temperatures than during a video encode session.
 
Feb 10, 2012
60
0
66
#32
I never put the retail HSF on my 4770k as the build was planned in advance. But I did pick up a 3770k on the cheap a month ago and, before I put into its final resting home, I slapped the retail hsf on it just to make sure the cpu and mobo were in good working order. Not satisfied with that, I decided to overclock a little. Then a little more. At 4.4 GHz I figured I'd see just what the stock hsf could handle, so I ran Prime95. For over an hour. It never went over 93 degrees. I was fairly shocked by this, and I'm sure it would be different for Haswell, but that little heatsink was holding its own.
 

dma0991

Platinum Member
Mar 17, 2011
2,723
0
0
#33
I do a bit of video encoding using the HTPC, which can run it at load for 30 minutes or more at a time, so the use case of 100% load is not unreasonable from that point of view.
Then clear up the clutter in that cramped case and improve its airflow. Getting an aftermarket ultra low profile heatsink is not going to help much.

Mid 80s for 30 minutes shouldn't be too detrimental. I've ran an overheating mobile Intel C2E >100C with ThrottleStop for hours it still survives after I've fixed the overheating months later.
 
Oct 27, 2006
19,794
302
126
#34
As always, no two CPUs will be truly identical even at stock. Much the same way you cannot buy two hard drives and have them be truly identical either, even if they are the precise same model and the serial numbers are sequential.

With SB and previous, things were much more predictable at least on the stock side of things. With Ivy and HW, things can vary quite a bit more from chip to chip. One of the best ways to check, outside of voiding the warranty, is by testing with the stock HSF and then putting a known great HSF like a Noctua or even a 212+ on it, and comparing baselines of stock settings and identical application loads.

So let's count the lotteries.

(1)- Silicon lottery.

This is the actual die of the CPU. Intel guarantees that this will work effectively at stock speeds in all supported extensions with the stock cooler on compatible motherboards/memory/yadda. This can be skirted a bit by finding out batch codes that indicate more robust overclocking batches. Ask any old timer about which 300A to get, or even G0 vs. B3. In newer cases, we've seen generally massive improvements with delids of seemingly 'bum' Ivys and Haswells, but occasionally one will still be terrible for OC, and that's the silicon lottery in effect (assuming the lack of user error, and I've seen enough experienced users simply find a gimpy K chip regardless).

(2)- TIM Lottery.

This is new to Ivy and Haswell. Of course, Intel has changed over from soldered to a simple squirt of proprietary paste to bridge the gap between the top of the die and the integrated heat spreader. This has been shown to be pretty wildly variable, so this is a big one to 'win' if you want to OC and/or have low temps even on stock.

(3)- IHS lottery.

This is the quality of the IHS itself. On both the inside cap, the outside mounting rim, and the top surface of course, there can be small imperfections and variations that will affect heat transfer. Dedicated souls will sand/polish this (lapping), in order to improve this (with varying degrees of success) the actual contact area that gets heat off of the thing and into the heatsink for dissipation. If you've ever looked at something under very high magnification, it can be shocking how something that appears smooth can actually be pretty gnarly in actuality. Of course, modding the IHS in this way voids warranty and means you'll want to mark the CPU in some way for visual ID.

These are far from the only factors, as you have local AC line quality, other components that may be contributing to stability, ambient and case temperature/humidity/airflow/fluctuation, it really gets quite granular to the obscure the more you think about it.

I've built and worked on tens of thousands of systems for portions of four decades now, and can say confidently that (A)- The stock Intel HSF is more or less adequate for stock use and potentially mild OC, and (B)- I absolutely hate using them and feel that they are of low quality, and obviously designed to be as cheap as possible to make while still hitting their targets (namely, cooling the chip enough for average use to make it through warranty).

The pushpin design annoys me to no end. Take any normal board, even a moderately high end one, mount the chip and attach the Intel (or 3rd party model even) push-pin HSF and it WILL flex the mobo a bit. Of course this is all more or less fine technically, but I hate it even more so for that effect.



Ugh. I vastly prefer using a quality aftermarket unit of not too high an expense, as I like to really take care of my stuff. Yes, it will be of little value in a short amount of time either way, and yes the stock stuff works all right, but I just feel better knowing I'm lowering the operating temperate substantially and not putting undue stress on anything. With very low power processors and dirt cheap builds I will forgo this, but anything that will see heavy use and any high-clock/K series/etc will get at minimum a 212+ Evo.
 

nwo

Platinum Member
Jun 21, 2005
2,309
0
71
#35
isn't the 4770K the same as the 2600K (95W), it makes no sense to use the same cooler for the Celeron?

just checked some google images and the 2600K looks a little taller than the G1620 one, also, are you sure the fan can't go faster on the i7 cooler? and what about the base? intel used to have all aluminum for cheaper CPUs and copper for the higher TDP ones,
Ah yes, you are right. The 4770k cooler does have a copper base while the Celeron cooler is all aluminum. Everything else looks the same though. I haven't actually used the 4770k cooler, I just took a quick peek to compare it to the G1620 which I will use in my new office build.
 
Sep 1, 2013
28
0
0
#36
A friend had very high temp with a FX-6300 recently. We removed the stock heatsink and there was barely any left from the stock paste. New TIM fixed the issue of course. Might be worth checking
 

sequoia464

Senior member
Feb 12, 2003
870
0
71
#37
I asked this same question a couple of weeks ago - after considering all of the input I decided to try the stock cooler.

I have an I5 -4440 - right now I'm running prime95 and my temps haven't been over 69°C - usually idles upper 20's to low 30's C. (ambient is right around 68°F currently)

The cooler that came with the 4440 has the copper insert.

Summers can get pretty warm where I am, so I plan on monitoring temps when summer approaches and then - if necessary - look into a better cooler.

Right now I'm enjoying not hearing anything from the cooler.
 
Last edited:

Morbus

Senior member
Apr 10, 2009
998
0
0
#38
That whole Fahrenheit and Celsius thing you going on there, sequoia, got me a bit confused. Ambient is around 68º F, right? Care to translate that to human?

Actually, never mind. GIMF.
 

Tristor

Senior member
Jul 25, 2007
314
0
71
#39
Sorry for the slight hijack, but it seems I have a question that's close to on-topic here. Anyone have any recommendations for a replacement for the stock HSF that fits within a similar form factor? My HTPC is a low-watt SB box (i5-2400S) where the stock HSF performance has been adequate however the noise level of it has increased over time due to what I think is a bad bearing in the fan.

I cleaned it all out the other day (it needed it) and that did seem to help some, but I still get pretty bad fan whine. I'd be very interested to replace it with something better (lower temps are a bonus). The problem is my HTPC case has VERY limited space, so I can't do anything much taller or much wider than the stock HSF (height is less limited than width oddly in this case).

Other than some questionable coolers from Zalman I don't see many options out there, but I am open to some suggestions. I see lots of folks arguing in favor of the stock HSF here, but nobody throwing out suggestions for a replacement for OP and at least some suggestions would help me out ;)
 

sequoia464

Senior member
Feb 12, 2003
870
0
71
#40
That whole Fahrenheit and Celsius thing you going on there, sequoia, got me a bit confused. Ambient is around 68º F, right? Care to translate that to human?

Actually, never mind. GIMF.
Thanks, it was confusing. Fixed the post - added the pertinent Cs's and F's.
 

Insert_Nickname

Diamond Member
May 6, 2012
3,609
141
126
#41
Other than some questionable coolers from Zalman I don't see many options out there, but I am open to some suggestions. I see lots of folks arguing in favor of the stock HSF here, but nobody throwing out suggestions for a replacement for OP and at least some suggestions would help me out ;)
If you're just looking for an inexpensive and quieter replacement for the stock Intel cooler, there is the Arctic Alpine 11 PLUS. If that is too big, you might look at the Freezer 11 LP. If it needs to be as small as possible, there is the Alpine 11 Rev. 2, but it has a slightly lower cooling capacity.

If you can get hold of one of the copper core stock coolers (bundled with 77-95W rated CPUs), its really good too. I think your 2400S only has the standard aluminium heatsink bundled.

For the record: 35-45W low profile heatsink. 50-65W standard aluminium heatsink. 77-95W copper core heatsink.
 


ASK THE COMMUNITY