- Joined
- Feb 16, 2013
- Messages
- 16
- Motherboard
- Supermicro X9SAE, Asus Sabertooth Z97, RIVE
- CPU
- E3-1275v2, 4790k, 4930k
- Graphics
- nVidia GT 610, GTX 650 Ti, 650
I recently built a hackintosh using GA-Z68MA-D2H-B3 and i7-2700K. I wasn't planning on overclocking it and only got the 2700k because I needed an i7 and HD3000, and I was able to get it on ebay cheaper than a 2600k. I ran it on Intel's stock heatsink/fan and was generally happy with it's performance and temperature; it idles in the low 30s and goes up to the high 60s or low 70s under load (Handbraking/QuickTime conversion). For comparison, my i3-2120 (also on stock heatsink/fan) idles in the 20s and core temp goes up to the low 50s under load. I wasn't at all concerned because the core temperature rarely goes above 65, and it nearly doubles the performance of i3 on video editing/encoding.
A few weeks into it, I was bitten by the OC bug: what's the point of having a Z68 & 2700k if I don't OC it a little? Right around the same time, I was reading many threads from various sources denouncing the usefulness of Intel stock heatsink, and how it shouldn't even be used to run at stock speed. Since I was hitting the low 70s occasionally, it would be smart to use an after market heatsink/fan. Cooler Master Hyper 212 Evo was highly recommended as one of the best performance/price option, I thought it should help me to lower the temperature by 10 or more degrees.
I won't get into the installation except that it could be terrible frustrating the first couple of times. What I want to share with you is this: the unit I have is either defective or this heatsink has serious design flaws. With the CM, the idle temp is slightly higher, not much but noticeable. The problem is under load, it quickly hits high 80s and could stay in the 90s for the entire handbrake process.
Below are a few pictures and screen shots to compare the 2 heatsinks. The Cooler Master is always on the left and Intel on the right.
Firstly, the CM is significantly larger and taller:
At idle, both performed about the same, but Intel is slightly better. I want to state that the screen shots were taken about 30 minutes apart (the time it took me to replace the heatsink), and the temperature in the room didn't change much during that time, and I don't think the ambient temp is correct.
To simulate load, I ran Handbrake to convert a test avi file to mp4. In addition to the obvious temperature difference, with the CM setup running at a dangerously high temp of near 100 degrees, the high temp also triggered the CPU voltage reduction, down-clocking (2.77GHz shown here, but to as low as 2.38GHz to cool down the CPU). This translates into performance loss, as shown in Handbrake performance (760fps on stock vs. 620fps on CM, about a 20% drop).
Here is another demonstration of the performance loss: I ran 3 consecutive Geekbench tests, and on the last run, the temperature was high enough that the under-clocking would kick in occasionally resulting in a drop of score from it's normal 12100 range, a drop of about 5%.
Here is the same using Intel stock heatsink, the screen shot was taken during and right after 5 consecutive Geekbench tests, and temperature was stable during all 5 tests:
The Cooler Master Hyper 212 Evo looks good on paper: big fan, high air flow, big fins and large surface area. In theory it should have performed much better than the Intel stock. However, the heat needs to be transferred over distance (about 5cm) from the CPU through heat-pipes to the fins and fan, and that process may not be efficient. That said, I don't think that's the primary reason for it's poor performance. There are many good reviews out there, and there are many satisfied customers; they can't all be wrong.
I can think of 3 reasons for its poor performance. Firstly, its performance is greatly dependent on how it's mounted. In the last 4 days, I must have remounted it more than 15 times, and the temperature differences could be as great as 20 degrees idle (at worst it idled in the 50s) and 10 degrees under load (low 90s, but eventually temp would reach 98 during a longer stress test and triggered down-clocking). To compound this problem, it is a significantly heavier heatsink (about 500g), so when the case is up right and it is turned on its side, there is significant amount of torque that I wouldn't be surprised that it is only making contact on one side of the CPU. This may indeed be the case because when removing the heatsink to reseat, the thermal paste is usually thicker on one side. Some one actually explained this problem much better on Amazon, and I would recommend looking it up (the most helpful 1-star review on Amazon).
Secondly, the heat-pipe is hollow. Copper is a good heat conductor, but for heat transfer both the distance and cross sectional area are important. Although this alone shouldn't be a problem since laptops and all-in-one computers such as iMacs use this method to cool their CPUs.
Thirdly, the design might be difficult to quality control during manufacturing. The surface contacting the CPU is has more than 4 individual pieces
that it might be difficult to make it perfectly flat. If any stress were applied on any of the heat-pipes, it could make the surface uneven and degrade its ability to transfer heat away from the CPU.
During one of the stress tests (temp reaching 98deg), I put my finger on the fins, lower heat-pipe and the base of the heatsink to assess its temperature. Both the fins and the lower heat-pipe were cool (ambient temp), only the base got a little warmer, but not higher than body temperature. In hind sight, I should have remove the fan and see if the entire heatsink would get warm. I bet that it wouldn't.
To wrap it up, I just want to make 2 points:
#1, Intel's stock heatsink is a decent performer (unlike what you might have read elsewhere). Many bad opinions were given without actual evidence. As long as you don't overclock, I think it's going to serve you pretty well. As a matter of fact, the readings I got were similar to i3.
#2, Despite good reviews, CM Hyper 212 Evo may not work well for you. It could just be sample variation, and I just got a lemon. Somehow I suspect the problem is more common than it appears.
A few weeks into it, I was bitten by the OC bug: what's the point of having a Z68 & 2700k if I don't OC it a little? Right around the same time, I was reading many threads from various sources denouncing the usefulness of Intel stock heatsink, and how it shouldn't even be used to run at stock speed. Since I was hitting the low 70s occasionally, it would be smart to use an after market heatsink/fan. Cooler Master Hyper 212 Evo was highly recommended as one of the best performance/price option, I thought it should help me to lower the temperature by 10 or more degrees.
I won't get into the installation except that it could be terrible frustrating the first couple of times. What I want to share with you is this: the unit I have is either defective or this heatsink has serious design flaws. With the CM, the idle temp is slightly higher, not much but noticeable. The problem is under load, it quickly hits high 80s and could stay in the 90s for the entire handbrake process.
Below are a few pictures and screen shots to compare the 2 heatsinks. The Cooler Master is always on the left and Intel on the right.
Firstly, the CM is significantly larger and taller:
At idle, both performed about the same, but Intel is slightly better. I want to state that the screen shots were taken about 30 minutes apart (the time it took me to replace the heatsink), and the temperature in the room didn't change much during that time, and I don't think the ambient temp is correct.
To simulate load, I ran Handbrake to convert a test avi file to mp4. In addition to the obvious temperature difference, with the CM setup running at a dangerously high temp of near 100 degrees, the high temp also triggered the CPU voltage reduction, down-clocking (2.77GHz shown here, but to as low as 2.38GHz to cool down the CPU). This translates into performance loss, as shown in Handbrake performance (760fps on stock vs. 620fps on CM, about a 20% drop).
Here is another demonstration of the performance loss: I ran 3 consecutive Geekbench tests, and on the last run, the temperature was high enough that the under-clocking would kick in occasionally resulting in a drop of score from it's normal 12100 range, a drop of about 5%.
Here is the same using Intel stock heatsink, the screen shot was taken during and right after 5 consecutive Geekbench tests, and temperature was stable during all 5 tests:
The Cooler Master Hyper 212 Evo looks good on paper: big fan, high air flow, big fins and large surface area. In theory it should have performed much better than the Intel stock. However, the heat needs to be transferred over distance (about 5cm) from the CPU through heat-pipes to the fins and fan, and that process may not be efficient. That said, I don't think that's the primary reason for it's poor performance. There are many good reviews out there, and there are many satisfied customers; they can't all be wrong.
I can think of 3 reasons for its poor performance. Firstly, its performance is greatly dependent on how it's mounted. In the last 4 days, I must have remounted it more than 15 times, and the temperature differences could be as great as 20 degrees idle (at worst it idled in the 50s) and 10 degrees under load (low 90s, but eventually temp would reach 98 during a longer stress test and triggered down-clocking). To compound this problem, it is a significantly heavier heatsink (about 500g), so when the case is up right and it is turned on its side, there is significant amount of torque that I wouldn't be surprised that it is only making contact on one side of the CPU. This may indeed be the case because when removing the heatsink to reseat, the thermal paste is usually thicker on one side. Some one actually explained this problem much better on Amazon, and I would recommend looking it up (the most helpful 1-star review on Amazon).
Secondly, the heat-pipe is hollow. Copper is a good heat conductor, but for heat transfer both the distance and cross sectional area are important. Although this alone shouldn't be a problem since laptops and all-in-one computers such as iMacs use this method to cool their CPUs.
Thirdly, the design might be difficult to quality control during manufacturing. The surface contacting the CPU is has more than 4 individual pieces
that it might be difficult to make it perfectly flat. If any stress were applied on any of the heat-pipes, it could make the surface uneven and degrade its ability to transfer heat away from the CPU.
During one of the stress tests (temp reaching 98deg), I put my finger on the fins, lower heat-pipe and the base of the heatsink to assess its temperature. Both the fins and the lower heat-pipe were cool (ambient temp), only the base got a little warmer, but not higher than body temperature. In hind sight, I should have remove the fan and see if the entire heatsink would get warm. I bet that it wouldn't.
To wrap it up, I just want to make 2 points:
#1, Intel's stock heatsink is a decent performer (unlike what you might have read elsewhere). Many bad opinions were given without actual evidence. As long as you don't overclock, I think it's going to serve you pretty well. As a matter of fact, the readings I got were similar to i3.
#2, Despite good reviews, CM Hyper 212 Evo may not work well for you. It could just be sample variation, and I just got a lemon. Somehow I suspect the problem is more common than it appears.