Installation of the MCW60 was simple following the provided four page instructions which are clear to understand, as long as you pay attention to what you are doing. In these pictures below you’ll notice that I’ve put the block on the wrong way, it was the school holidays and I had four of my boys decided to have a fight over what channel to watch on the TV. Once I’d noticed the camera battery then decided to go flat and not wanting to waste time I carried on with the installation while the camera charged.

I used the supplied thermal paste that came with the MCW60 on the block installation.

Below I've included some quote form the install manual.

"Fittings compatibility notes:
The provided fittings thread is 1/4” NPSM. This thread is compatible with BSPP and G 1/4 threads.
G ¼, or BSPP fittings will fit, but may not necessarily seal properly; each must be checked prior to assuming that it will not leak just
because they fit together. Both NPSM and G ¼ (BSPP) are parallel thread and nominally the same size, the principal difference being
18 threads per inch for NPSM and 19 threads per inch for G ¼ (BSPP). Since most male end G ¼ fittings have a short thread length
they can generally be engaged in the NPSM threads without difficulty. The joint seal is effected with an o-ring which for the NPSM
barb is in a groove on the waterblock top and compressed by the flange nut barb. G ¼ fittings have the o-ring captured in a groove
under the fitting nut. G ¼ fittings will seal so long as there is a straight portion under the nut flats sufficient to bring the G ¼ fitting’s oring
into contact with the bottom of the waterblock o-ring groove, a depth of 0.080”."

Once the block was on it was time to put the MC14 memory blocks on.

The MC14 comes with thermal pads on but when I tried to put them on the underside of the card they would fall off once the card was turned over, SwifTech does state that it is better to use epoxy glue for better performance.

"Advanced Installation
For a superior mechanical joint and enhanced thermal conductivity, the MC14 Ramsinks may be permanently attached to the
memory modules using a thermally conductive epoxy glue such as Arctic Alumina or Arctic Ceramique Epoxy. Non electrically conductive
or non-capacitive glue should be used to prevent damage to the memory, thus precluding the use of Arctic Silver Epoxy
which is capacitive. Please refer to http://www.arcticsilver.com for installation guidelines."

So blowing caution to the wind I went a head and used epoxy to attach the heatsinks. The plus point of the MC14 is that you can trim them down, if need be using a pair of pliers.

Installation issues with 1/2” barbs and 3/4” or 5/8” OD tubing.
In some instances, there may not be sufficient clearance to install the Ramsinks on the memory modules located directly under
the water-block inlet and outlet when using the above mentioned tubing. This issue does not affect smaller diameter tubing such as
3/8” and 1/4” (1/2” OD and 3/8” OD). The pins of two of the memory Ramsinks can be easily shortened to provide clearance.
Simply use a small pair of pliers, and cut the pins individually as shown below:

I found that I only needed to trim one of the MC14s I had as it was under the inlet and outlet pipe, it does depend on your graphics cards lay out.

Once the MC14 memory heatsinks were on it was time to put the 7800GTX back into the system and refill, bleed and test for leaks. Something I have got down to a fine art over the last month. I then ran the system until I got stable temperature from the GPU, which meant that the thermal paste had settled down which would then give me fair results in testing.

I ran the same tests used in the NV-78 review and have put the standard HSF, NV-78 and the MCW60 in the charts so you can see the difference.

All tests were carried out with the room temperature being controlled as much as possible and NVtemp logger was used to record the temperatures.

The standard HSF idled between 41c to 42c, the NV-78 stayed at a steady at 39c both with an average room temperature of 22c. The MCW60 idled between 31c to 34c with an average room temperature of 19.1c.which is better than the stock HSF and NV-78.

Then I ran 15 Farcry benchmarks in a row, using HardwareOC benchmark again with the same settings, maxed out, used on the NV-78.

The MCW60 was run with the room temperature at 19.1c through out the Farcry test that’s 2.9c lower than the NV-78’s room temperature of 22c. Even if you add on 2.9c to the maximum temperature of the MCW60 it would still be lower than the NV-78.

I then decided to see if I could push the MCW60’s temperature higher and run a number of tests. First I ran Real time HDR test.

As you can see the temperature got to 39c once again.

Again the temperature didn’t raise that much from the idle temperature; the room temperature was an average of 21.8c through out the game of BF2. I then thought I’d overclock the card from 450MHz to 500MHz on the core and 1.25GHz to 1.28GHz on the memory which I was running on the NV-78 for every day gaming. Again I hit a maximum of 39c, so I went to 525MHz on the core and 1.33GHz on the memory and ran one cycle of 3dmark06.

This time it went a little higher, which showed me that there was any problem with the temperature readings. I hit a maximum of 41c, only 3c more than the other tests.