This is my attempt at a review of the Akasa Freedom Force cooler through my experience fitting it to my GTX260.
(Note: all pictures should be clickable for larger version. Some are also less cropped and extend up to 2.5MB in size.)
Let's start with the box you get:
Nothing too fancy, some info on the front about it (including "GTX285 ready" more on this later) and the back also has a clear cut-out allowing you to see the direct-touch heatpipes.
In the box you get:
The main heatsink / fan, mounting nuts/bolts, RAM sinks, thermal paste, fan adapter and user guide.
The two 80mm fans are ~15mm thick and have white LEDs in them, they're not excessively bright and, in my opinion, quite nice with out being over the top or tacky. They are wired together to a standard 3pin motherboard plug that can be connected to the PSU with the included adaptor.
Four direct-touch heatpipes contact the GPU, small GPUs or ones without a heat spreader might not make good contact with all four and thus not get full benefit of the coolers potential.
The mounting system has three positions to cover most mid-range and high end cards. The largest ones are further apart than most other coolers and fit Nvidia's GTX2xx range. The brackets have two, surprisingly and reassuringly long, screws to hold them to the base. Once removed you can adjust the hole position of the mounting bolts to suite your card. The hole for the screw isn't round so it's held captive and the rubber washer / spacer is a friction fit so the bolt is held in place. Once the bolts are in the bracket is screwed back on the base. As the bolts are held in and captive it's not fiddly and works well, it gives the impression of being well thought out.
Front side naked:
There are ~40 fins on each side that the heatpipes connect to, as the total spread is ~92mm I estimate the fin spacing to be ~2mm. This is quite reasonable and bodes well for low noise cooling, there are around 20 aluminium fins on the top of the GPU and these are also widely spaced.
The test subject:
XFX GTX260 "XT" This had slightly higher GPU / shader clocks than "standard" card. It uses Nvidia's reference design and cooler, it's 216 core, 55nm model.
The stock cooler uses a single "blower" fan, with 4 pin PWM control, to blow through an aluminium fin array that has three heatpipes in it to distribute the heat from the GPU core. The cooler also makes contact, through grey thermal pads, with the NVIO chip, RAM chips and various VRM parts.
Parts to be cooled:
Everything with a red dot is contacted by the stock cooler; that is The GPU core, the NVIO chip (left most), 14 RAM chips, 12 VRM components (right)(I believe, technically, these are FETs that form part of the VRM circuitry) and 3 other items.
[The blue dot items were all treated to lots of "super-glue" to fix them firmly to the board to significantly reduce the "coil whine" this card suffered from - whole nuther story there]
The Freedom Force came with 9 RAM sinks, not quite enough I think
New version is supposed to have 22 heatsinks with it, better but
Maybe by "GTX Ready" they mean it's ready when you're ready to buy a load more heatsinks
Which I did
The top 2 were some I had left over, the bottom 4 packs were purchased for this project. (At significant expense, in fact slightly more than the Freedom Force cost
) You could of course use much cheaper aluminium ones.
On the left we have the black Akasa RAM sink, pretty small in my opinion, then blue Zalman RAM sink, Enzotech BMR-C1L, BMR-C1, MOS-C1 and CNB-S1. The BMR-C1L is 14x14x9mm, BMR-C1 is 14x14x14mm, MOS-C1 is 6.5x6.5x12mm and CNB-S1 is 36x36x27mm
First I tackled the NVIO chip
NVIO chip and heatsink:
The Enzotech CNB-S1 was much too tall to fit under the edge of the Freedom Force so I trimmed the fins that wouldn't fit and bent down the remaining ones. I also had to elongate the mounting slots as they weren't close enough together. I cut down the supplied foam shim to make that fit too.
NVIO Heatsink fitted thus:
I cleaned up the chip first with ArctiClean and used Arctic Silver 5 thermal paste, it's held on with the supplied Enzotech spring-loaded mounting pins.
It took a lot of work to determine how much needed to be trimmed off each BMR-C1L RAM sink as they vary. The top two only need one row of pins trimming, some of the right hand ones that end up under the heatpipes needed all the pins trimming. With a lot of back and to I ended up with all the heatsinks cut to fit under the Freedom Force while leaving the maximum amount on each one. To fit them I cleaned the RAM chips with ArctiClean and then pressed each chip on firmly and held it for a good 10 seconds. This seems to work well as they are all well stuck on, hopefully helps the heat transfer too.
I fitted the MOS-C1 to the VRMs using the same method as the RAM sinks. They didn't need to be cut down which was a blessing, although the small size and height made pressing them on firmly a little tricky as they tended to want to fall over. I'm not at all sure the contact is great of if they provide enough cooling. Enzotech do make MOS-C10 heatsink that is 10x10x14mm, it has 16 pins vs 4 on the MOS-C1s I used. I'm not sure you could use on all the VRMs and some might have one sink on two, If I was doing it again I'd try a mixture I think. I'd also get the CNB-S1L low profile heatsink for the NVIO chip to save cutting that down too.
All heatsinks fitted:
With all the Enzotech heatsinks fitted and after a test fitting of the Freedom Force. Subsequently I put a little more Arctic Silver 5 thermal paste on too for full coverage on the left side.
Fitted without fans:
More shots showing the fit without the fans. (and after some testing hence the dust
Ready to go:
Slotted in to test system in Mini P180
Unfortunately I never did any good testing and recording of results with the stock cooler
Looking on the net shows temperatures of the GPU of 60 - 80Â°C+ depending on load and fan speed. My recollection is at idle it was pretty quiet but slightly above the noise floor of my PC. Under Folding@home load it was fairly quiet but not really quiet, don't remember if I even tested under Furmark. SPCR's review of a GTX260 put the idle fan speed at 40%, which mine was, and 700rpm. I don't remember mine being that low, more like 1000rpm but I could be wrong, think they tested 192 core 65nm where mine is newer 216 core 55nm and has a revised board layout and, presumably, cooler. SPCR's load test showed 50% and 880rpm but significantly more noise, more noise sounds right, again not sure of the fan speeds. They also showed a PCB temp around 10Â°C lower than the GPU core.
I have a screen shot of Furmark run with the Case and GPU fans running moderate speed and the core hitting 79Â°C with the PCB in low 60s. The twin rear 80mm case fans are ~1400rpm with the Freedom Force fans ~1300rpm so fairly quiet.
The same run also had some "full cooling" with the case fans ~2000rpm and the Freedom Force fans ~2200rpm. This reduced the GPU to 67Â°C and the PCB to 53Â°C. This, I believe, is a pretty good result for Furmark on GTX260 although it's not "SPCR quiet!" The Freedom Force fans seemed to be good acoustically, I didn't notice any particular untoward rumbling or buzzing beyond what you'd expect of an 80mm fan for a given speed. Probably the best 80mm fans would be quieter in dB(A)/RPM/CFM comparisons but not drastically so. I think few people would feel the need to change the fans on acoustic grounds.
In the original configuration the GPU fans were plugged in to a motherboard case fan header so the speed was controlled at the same rate as the case fans. Not altogether unreasonable as when need to get more heat off the GPU also need to get it out the case but lacked total flexibility. Also the GPU fans were only tied to the GPU temp by SpeedFan. It can read the GPU temp and control the motherboard fan speeds, if I'd set it up right and left it on auto and not over ridden it!
Since then I've replaced the fans on the Freedom Force with a pair of Arctic Cooling F9 PWM fans. These are both run off the cards fan header and are controlled by the cards BIOS. I thought changing the 80mm fans for 92mm would mean they would overhang the heatsink and better cool the surrounding components.
Freedom Force with 92mm fans:
They don't, this means the Freedom Force is larger than I thought it was and comparable to the Thermalright T-Rad2 GTX that also uses 2x 92mm fans. The fans are zip-tied to each other in the middle and to the heatsink at each corner. Both fans run off the cards header using Arctic Cooling's "PWM Sharing" system. One fan's speed signal is connected to the card header and the other fan is connected to a motherboard header. This means the fan speed can be seen by motherboard monitoring software (SpeedFan) and graphics monitoring software (GPU-Z). Presumably the fans run very close to the same speed as each other and the two report similar numbers, I think the card is more accurate and finer grained than the motherboard.
I've now modified the BIOS with NiBitor to set about the highest stable clock speeds and also reduced the idle fan speeds and tweaked the temp/speed profile to aim for slightly lower temp. I couldn't find any good guide / explanation of these settings. The min Duty Cycle is clear, TOperating seems to be the temp the fans speed up beyond but how TMin, TLow and THigh affect it I've no idea!
BIOS settings compared: (Original - New)
New testing with 92mm fans etc. I will write up a General Gallery Thread for my PC for full details of it. In brief it's Asus P5B-E Plus with Q9400 @ 3GHz (8x375), 4x2GB DDR2 750, 1TB Samsung F1 in SQD, the GTX260 and Antec Signature 650w PSU. The case is Cooler Master ATCS 201 with much improved breathing. Now consists of mesh grills on the Blow Hole (80x80mm) as intake, mesh on front intake (80x160mm) and now a bunch of 10mm holes in the floor with taller feet. Exhaust is 80mm fan behind/below GPU and two more 80mm fans behind CPU, all 2000rpm rated Arctic Cooling F8s.
My PC as tested:
Testing was done with room ~21Â°C. First up we have 100% case and GPU fan speeds. Furmark run in default benchmark mode 1920x1200 for 10 minutes.
This gives GPU temp peek of 66Â°C and PCB of 56Â°C, for an over-clocked GTX260 running heavy Furmark I think this is quite acceptable. The case fans are much louder than the GPU fans, moderate level of drone and air whoosh, no particular unpleasant tones just plenty of volume, not SPCR quiet at all!
Next I left the case fans on 100% and allowed the BIOS to regulate the GPU fan speed:
The BIOS stabilised the GPU temp at 75Â°C with the PCB barely hotter than before at 57Â°C. The GPU fans reached 36% / 917rpm, the noise level is as before as the case fans already drowned out the GPU fans when they were at 100%. What this test tells us is the GPU fan speed required under ideal conditions is pretty low.
Finally we have a test with the case fans at a much quieter 50% / ~1200rpm
The GPU temp reached 78Â°C with the fans at 41% / 1050rpm. The PCB hit 59Â°C although I measured the physical temp of the board on the back of the VRMs with Scythe Karma Thermo to be, in places, 85Â°C
I've no idea how bad that is, on the plus side Furmark doesn't crash and the thing's been happily running Folding@home for months.
Not that it gets anything like as hot with Folding@home or "normal" usage so I guess it's OK
Noise wise, in this configuration, it's pretty quiet, not inaudible but not really intrusive and unnoticeable when playing a game or with any appreciable ambient noise. I like to think pretty darn good for a potentially potent gaming PC
When Idle the fans can be slowed right down:
Bear in mind only the CPU fan is 120mm, the case fans are 80mm and the GPUs 92mm. At these speeds the PC is barely audible, the Samsung F1 HDD is the most noticeable component.
Summary? I've made it work and work well and very quietly for me. The effort has been huge and the additional expense significant but then all bleeding edge PC silencing is
Would I choose it again? Now maybe not as there are other choices, at the time the only other option I could find for sale was Thermalright HR-03+ that won't fit in my case (not wide enough) as it extends too far up from the motherboard. The T-Rad2 GTX wasn't yet available, and I believe with it too VRM cooling isn't straight forward. Doing it again I would buy cheaper aluminium RAM and VRM heatsinks, but then the Enzotech copper ones are quite pretty
Do I recommend it? Yes but... On a mid level card with no VRM or RAM chips to worry about then it should be great. On an ATi 4850/70/90 paired with a good Zalman or Thermalright VRM heatsink it should also perform well with minimal hassle. The same could be said of the Accelero S1/S2 that are much cheaper... for me that they don't fit my case or graphics card ruled them out.
For a GTX2xx owner I would suggest reading my story first! Paired with Thermalright GTX VRM heatsink pack it might work well with minimal effort.
At around Â£35 it's priced like a premium graphics cooler, but the performance potential is there too.