Our Lapped CPU Heatsink Test Platform

Cooling | Reference|Recommended
Viewing page 4 of 4 pages. Previous 1 2 3 4

THE LAPPED CPU: ANY SKEWING OF TEST RESULTS?

The lapped CPU was installed back in the Asus P6T SE motherboard. Thereafter, heatsink testing proceeded as usual. That was back around Christmas 2012. So a year later, is there any indication that heatsink test results have been affected by the lapping? In other words, have there been any heatsinks that performed well that might not have done so well before the CPU was lapped? Or Those that did poorly which might have done better pre-lapping?

A quick look at the curvature of the base of heatsinks tested before January 2012 is one place to start. This table ignores other factors, such as size, and the quality of the mounting system; it's just a quick check for correlation between base profile and cooling performance. The ones with convex bases all tested well. The concave bases all gave poor results. The flat bases gave mixed results.

TESTED BEFORE LAPPING
Heatsink
Base
Cooling
Zalman CNPS9900DF
concave
poor
Noctua NH-L9i
flat?
OK
SilverStone HE02
flat/concave
mediocre
Prolimatech Panther
convex
good
Phanteks PH-TC14PE
convex
good
GELID GX-7
concave
mediocre
GELID Tranquilo
concave
mediocre
be quiet! Dark Rock 2
concave
poor
Enermax ETS-T4
flat (DT)
poor
Noctua NH-L12
convex
good
CoolerMaster GeminII M4
flat (DT)
mediocre
NZXT Havik 140
convex
good
Scythe Big Shuriken 2
flat
good
*DT = DirectTouch

Compare that with cooler tested after lapping:

TESTED AFTER LAPPING
Heatsink
Base
Cooling
Noctua NH-U14S
convex
good
Scythe Ashura
convex
good
Scythe Mugen 4
convex
good
SilverStone AR01, 02, 03
flat (DT)
good
NoFan CR-95C
flat
good
Noctua NH-U12S
convex
good
Phanteks PH-TC12DX
convex
good
Phanteks PH-TC90LS
concave
poor
*DT = DirectTouch

What is clear is that in general, convex base heatsinks fared well before and after the CPU was lapped. Concave base heatsinks also fared consistently: They performed poorly before and after the CPU was lapped. A flat base did not correlate strongly with good cooling pre-lapping but more so after the lapping.

Let's examine this further.

1. Flat Conventional Base Heatsinks

One heatsink retested right after the lapping was the Silverstone Heligon, which performed surprisingly poorly in the original review, hindered by less-than-ideal contact with the CPU. The base was checked with our straight edge and it appreared to be mostly flat, neither concave nor convex. It made better contact with the lapped IHS, and when retested with a single 14cm reference fan at various speeds, there was a 1-2°C improvement across the board. This result still did not make the Heligon significantly more competitive against smaller cheaper coolers, though its passive performance went another titch beyond the Thermalright Macho. So subtle improvements, not a dramatic one. Unfortunately, this is the only example of a flat solid base heatsink that we have in hand that was tested before the lapping.

2. Flat DirectTouch Heatsinks

DirectTouch heatpipe bases can only be flat, due to the nature of their construction. There are lots of questions around this type of base, including longevity and flatness, and a limit to how tightly it can be mounted on the CPU due to the fragility of the exposed heatpipes on the base. Morever, there is a sense among performance-oriented enthusiasts and traditional heatsink makers that DirectTouch represents something of a cheat, a cheap short-cut. But there's no denying its lower cost yet decent performance. A clear disadvantage is that most IHSs on most CPUs are a touch concave, and DirectTouch bases can only be flat, at best. They cannot be made with the subtle degree of convexity that the top "solid" base heatsinks tend to have. Our lapped CPU is now flat; hence, it stands to reason that DT heatsinks should now do better than they did on our test system before the lapping.

The Silverstone Argon 01, 02, and 03 coolers with direct touch bases, all reviewed this year on our lapped CPU, achieved great results, two of the three getting Editor's Choice awards and the third getting a Recommendation. Would they have done as well pre-lapping? We can't answer that directly, as there's no un-lapping our CPU, but we can take another look at the Enermax ETS-T40, a fairly large DT cooler similar in overall design which Larry Lee mentioned in the Silverstone Argon 03/01 review as having been "a dud". The ETS-T40 was reviewed on the pre-lapped test platform in mid-2012. We happen to still have this sample on our shelf, so it was brought out for a repeat test on the post-lapped CPU.

As Larry mentioned in his original review, "One look at the ETS-T40's mounting gear and we're optimistic. A simple metal frame is assembled around the socket and a thick bar goes over the base [which is used to clamp the heatsink to the frame & CPU], a system similar to that employed by Noctua, Prolimatech, and Thermalright on their high performance coolers." It is indeed a sturdy, secure and easy to use mounting system. I was able to mount the heatsink, and run Prime95 in just 10 minutes, with the stock fan dialed to 1060 RPM (7V). It was clear immediately that the results were going to be much better this time around. The full range of tests were done in two hours; the results are in the table below, compared to the ones obtained previously on the pre-lapped CPU.

Enermax ETS-T40
Fan Voltage
Fan Speed
SPL@1m
°C Rise
postlap
prelap
Stock Fan
12V
1930 RPM
35~36 dBA
35
42
9V
1520 RPM
28 dBA
37
43
7V
1060 RPM
20 dBA
43
49
6V
820 RPM
14 dBA
50
55
5V
580 RPM
11~12 dBA
61
64
Reference Nexus 120mm Fan
12V
1080 RPM
16~17 dBA
40
44
9V
880 RPM
13 dBA
42
46
7V
720 RPM
12 dBA
46
50
Reference Nexus 120mm Fan x 2
12V
1080 RPM
20 dBA
37
41
9V
880 RPM
15 dBA
39
43
7V
720 RPM
13 dBA
43
47
°C Rise: Temperature rise above ambient (20°C) at l oad.

With the lapped CPU, the performance of Enermax ETS-T40 improved dramatically across the board, by a minimum of 3°C to a maximum of 7°C. With a single reference fan, performance was 1-2°C better than the venerable Coolermaster Hyper 212 Plus, tested in 2010, soon after the i7-965 Extreme platform was rolled into service. It's possible that the Coolermaster 212 did as well as it did with its flat base because our CPU had not become as concave as it would be a couple of years after umpteen high pressure convex base heatsinks had been tested on it.

CAUTIONARY WORDS

Before you jump up and down clamoring for a retest of all DirectTouch coolers we've reviewed in the past four years, let's be clear about what these results mean.

1. DirectTouch base heatsinks will cool our our lapped CPU better than before it was lapped, when it was somewhat irregularly concave.

2. Such heatsinks will still need very good mounting mechanisms and good qualities in all other aspects of heatsink design and execution to compete with the heatsinks with solid bases that have reached the top rankings at SPCR.

3. A DirectTouch or other flat base heatsink that reviews well at SPCR now will cool as well for you if your CPU has a relatively flat IHS... or if you lap your CPU as we did ours.

4. With a CPU that has a concave profile, a flat DT or solid base heatsink will not cool as well; just how much worse will depend on the degree of concavity your CPU exhibits. The difference could be as little as a degree or two, but it could be as much as 5~7°C.

5. Slightly convex based heatsinks that have other critical qood qualities continue to review well on our lapped CPU. It is only one of a handful or critical features needed for high performance, so the simple fact of a convex base cannot assure good performance. But it is noteworthy that almost all heatsinks with convex bases have those other critical features for high performance.

6. Concavity of the IHS may be a less critical factor with smaller CPUs, like the socket 1150, 1155 or 1156 CPUs. The dimensions of the IHS surface being smaller, the same degree of curvature leads to a smaller gap between IHS and heatsink base.

7. Finally, to achieve high performance cooling, you must pay close attention to your gear, just as we do ours, and be prepared to make adjustments, change components, experiment, and monitor carefully. Just dropping in a highly rated cooler does not automatically give you great cooling and super low noise. It's less of an art than it used to be, but it still takes intelligent, informed care.

We will be making only one change to our heatsink rankings: The Enermax ETS-T40 will get a postscript and be introduced to the Recommended or Editor's Choice ranks. We may have to make some kind of handicap for flat base heatsinks to reflect that some users' CPUs are likely to have slight concavity rather than perfect flatness. On the other hand, it might be better to make note of this issue whenever we encounter a high performance flat base heatsink. Yes, it's another minor headache we'll deal with in the near future. Meanwhile, rest assured SPCR keeps trying our best to stay consistent and accurate in all of our real-world reviews.

LOOKING AHEAD

One comment made by the second of our two heatsink reps was that lapping makes the IHS thinner and weaker, so it could become subject to further deformation. I was aware of this opinion before lapping, but weighed the pros and cons and decided there was too much to be learned and gained. A year of testing after the lapping, our i7-965E CPU shows no sign of having become concave again. We're also not aware of any new records having been set for cooling in this past year, which might be indicated if the thinner IHS lead to improved cooling.

We're not certain what other factors will come into play for our heatsink testing program in the future. Everything wears, and certainly, the socket 1366 platform is already defunct, so it may be time to make another transition to a newer test platform before our current system breaks. Unfortunately, we do not have a new replacement board or CPU waiting in storage this time. A quick look through Newegg shows only server dual-socket boards available for LGA1366. So we are already in the hunt for a newer test platform, just in case. In the meanwhile, we will continue reviewing heatsinks as before, and add cautionary comments whenever we run into a heatsink with a flat base. This is unlikely to please everybody, but as the old song goes, you gotta please yourself.

* * *

Articles of Related Interest
SPCR's 2010 CPU Heatsink Test Platform
SPCR's Updated 2012 Small CPU Heatsink Test Platform

* * *

Discuss this article in the SPCR forums.



Previous 1 2 3 4

Reference|Recommended - Article Index
Help support this site, buy from one of our affiliate retailers!
Search: