December 10, 2003 by Russ
Heatsink for Socket A CPUs
The MCX462-V is the Socket A counterpart to the MCX478-V recently
reviewed by Ralf Hutter. It’s a clear continuation of the classic Swiftech heatsink
design: Thick copper base, with helicoid aluminum pins press fitted into the
base. But the “V” does bring a new twist to the design. Well, a
new bend, actually. As with the 478 model, the pins are set into concentric
circles, and bent at multiple angels to increase the airspace between the
pins. According the Swiftech, “The spacing and angles between the concentric
rows has been precisely calculated to optimize cooling at air flow levels
as low as 22CFM and 23dbA”
Seen next to the previous version, the MCX462+, the new V model looks significantly
larger. The apparent size increase is deceptive, for the bases of both are
3″ squares, which keeps it within the AMD height restriction area of
the motherboard. The splaying of the fins does allow for an extra benefit,
the use of a 92mm fan without an increase in the base area of the heatsink.
As with the 478-V, brackets for both 80mm and 92mm fans are included with the kit.
- CNC machined, C110 copper base, 3″W x 3″L x .500″H, flatness
better than 0.0003″, micro surface finish 8 or better.
- 390 Helicoid Pins made of High Thermal Conductivity Aluminum Alloy, press
fitted in the base.
- Heatsink base dimensions 3″x3″x1.66″
- Weight: 23.2 oz (650g) without fan
NEW RETENTION MECHANISM
Well, after my review of the MCX462+,
which praised the classic Swiftech through-board retention mechanism for
being the most secure method of attaching a heatsink, and panned it for
being a giant PITA, Swiftech went and changed everything.
The new mechanism is a socket-only attachment, as opposed to the 4 bolt
of previous Swiftechs. It operates via a pair of 3 lug clips which snap
over the lugs as the heatsink is lowered vertically onto the socket. Then
the tension screw is loosened to gradually apply spring pressure to the
lugs, securing the heatsink at the proper tension.
This has the advantage of the old method by not requiring the motherboard
to be removed for mounting/unmounting. Not requiring the mounting holes also
expands the motherboard compatibility, as fewer and fewer socket A boards
have these holes. Also, opposed to traditional spring-clip lug mounts, this
system has the added safety of applying the pressure to the drop of the
After mounting and removing the V several times, I’m not completely sold
on the new mechanism. While more convenient than
through-board bolting, its not as simple a process as Swiftech suggests, or as smooth as other lug attachments. The problem lies in getting
the clips over the lugs. The clip doesn’t easily snap over
the top of the lugs as designed, and considerable fidgeting is required to
align it just right.
During the loosening of
the tension screws (which counter-intuitively tightens the heatsink onto
the CPU), I needed to squeeze the clips towards the socket to ensure that
all 6 lugs would catch. Without the squeeze, one side of the clip would twist
with the turning of the screw, resulting in only one or two of the lugs
being secured squarely. The same problem of clearing the lugs occurs when
removing the heatsink. After the springs are compressed the clips are still
aligned vertically under the lugs, so you can’t just lift the heatsink upwards
to remove it. Instead you need to slide and jiggle it around to get the
A simple design modification would improve the entire latching mechanism:
Shape the clip/spring/screw mechanism so that as the clip is loosened it
splays outwards, away from the lugs, and pulls in and upwards, towards the
lugs, as the tension is applied.
But once applied and tightened, the MCX462-V is very secure, with no wiggle
or sliding around possible. (Although at a weight of 650g without fan, Swiftech recommends that the heatsink be removed for shipping)
FIT AND FINISH
Swiftech has the finishing and polishing of heatsink bases
down to a science. There is simply no one doing it better. My comments from
the earlier review apply equally well here. So well in fact, that I’ll simply
report them again:
“As always, the finish on this swiftech is superb.
They really take pride in the finish on the mating surface, and it shows.
I can’t imagine that anything would be gained by trying to lap this thing
further, you’re more likely to end up making it worse rather than better.”
The Test Bed for MCX462-V
* ABIT NF7 motherboard, with latest BIOS revisions. Modified with passively
* XP2100 Thoroughbred B, at stock speed and voltage (1.65v, 62.1 Watts max)
* 1GB of PC2100 RAM.
* ATI 9500 graphics card, cooled with Arctic Cooling VGA Silencer.
* Seagate Barracuda IV hard drive.
* Enermax 365 PSU, modded with a Panaflo 80L1A fan @7 volts.
* Temperatures are read from the internal thermal diode of the CPU with
Motherboard Monitor 5.
* CPUBurn was used to achieve load temperatures.
* Ambient temperature during testing was 20°C.
To test the MCX462-V’s abilities with both 80 and 92mm fans the reference
Panaflo 80mm L1A was used in conjunction with a 92mm Papst 3412GL (36CFM,
MCX462-V with Papst 3412GL 92mm fan Fan voltage / CFM (calculated) 12V / 36cfm 7V / 21cfm 5V / 15cfm Idle temp 29°C 36°C 45°C Load temp 38°C 49°C 63°C Rise from ambient 18°C 29°C 43°C °C/W 0.29 0.47 0.69
MCX462-V with Panaflo 80L1A fan Fan voltage / CFM (calculated) 12V / 24cfm 7V / 14cfm 5V / 10cfm Idle temp 32°C 39°C 45°C Load temp 40°C 50°C 63°C Rise from ambient 20°C 30°C 43°C °C/W 0.32 0.48 0.69
MCX462+ with Panaflo 80L1A fan Fan voltage / CFM (calculated) 12V / 24cfm 7V / 14cfm 5V / 10cfm Idle temp 32°C 42°C 48°C Load temp 43°C 56°C 73°C Rise from ambient* 22°C 35°C 52°C °C/W 0.35 0.55 0.82
*Ambient temperature during MCX462+ testing was 21°
As Ralf discovered with the P4 version of the MCX462-V, a 92mm fan and an 80mm fan does not produce the results necessarily
expected. At its full 36CFM, the 92mm fan outperforms the 24CFM 80mm, but
not by much. It puts out 50% more air, yet results in only a 10% performance
gain. As the airflow drops, the performance difference narrows, until at
5v the extra 5CFM from the 92mm is of no help at all. The most logical explanation
for these numbers is the fact that the airflow from the 80mm fan is concentrated
closer to the center of the heatsink, where the greatest heat load is.
Compared to the previous edition,
the results are quite striking. Swiftech’s claim of
the MCX462-V being optimized for low airflow is not marketing hyperbole.
The temp reductions range from 2° at 24CFM, to an impressive drop of
10°at the 5v setting of 10CFM.
Overall, the performance of the MCX462-V moves Swiftech up closer to the top of the SPCR heatsink rankings, besting the perennial favorite Thermalright SLK-800
and falling behind only the heavyweights Thermalright
SLK-900A and Zalman 7000A. The MCX462-V is certainly a worthy improvement.
– Fit-and-Finish, top notch as always.
– Big boost in performance over previous version.
– New retention mechanism.
– Flexibility of 80 or 92mm fan sizes
– Improved motherboard compatibility
– Performance improved, but still bested by other heavyweights
– New retention mechanism
Much thanks to Swiftech for the review sample.
* * *
Discuss this article in the SPCR Forums.