Zero Infinity - Phantom Knight analysis Introduction I just recently recevied this new heatsink named the "Phantom Knight" by Zero Infinity. Zero Infinity are new kids on the block in the cooling world and this is their first attempt at a high end CPU cooler. First thing I notice with this cooler is its unusual shape. It almost takes on a figure 8 shape and is a bit different than anything we have seen before. Perhaps this is for cooling efficiency purposes, someone has worked out this particular design is going to provide more effiency than a straight up box style cooler such as the Noctua NH-U12P SE. Next I notice a middle mounted fan, this design traditionally has not always been great, not only because we can't mount fans on either side easily, but performance has not been on par with the top coolers that feature externally mounted fans. The arguement on the other side is that with this design there is some sort of pull/push going on and perhaps again that is a benefit in terms of cooling efficiency. This is something we will find out in due course. Looking at the base we find a DHT or direct heat pipe style cooler. The DHT can give benefits of immediately delivering heat from the CPU into the fins of the cooler, instead of the heat first being drawn into a base, then up into the heat pipes and to the fins. This can be difficult to pull off well, as some coolers in the past have had bases that are anything but flat. On the Phantom Knight it does seem very flat, but there are some small notches I can notice if I drag my finger nail across the surface. I would suggest they are so small they wouldn't have any detremental impact on performance, that being said we will find out later! The base is not highly polished, but polished well enough that I get a good reflection of my USB stick when I hold it up close. The 88mm copper heatpipes then extend up to the 45 layers of aluminium fins. There is an additional smaller, approx 1CM set of thick fins just above the base of the heatsink. Zero Infinity have decided to throw in a couple neat touches with the Phantom Knight. There is four different color tops to choose from with red, green, orange and purple available with the retail kit. I have an earlier sample here and I have yellow instead of red, thats cool, i LIKE yellow, infact I am going to use yellow on mine. The mounting system on this pot is solid, it features a flexible system that mounts to AM3, LGA775, 1156 and 1366. The system works via 4 threaded poles come up from the mount place, secured by a nut, the heatsink then sits directly over the top of these poles and is secured via thumb screws. This is a slightly different approach than the Noctua heatsink I have been using on my test platform. It has a mounting system that is first secured to the platform, then the heatsink is secured to that. When mounting a number of times I found the Phantom Knight extremely easy to mount on both X58 and AM3, it is quick to mount also and does not need any screwdriver (thumbscrews), which would be of great benefit if you are using this heatsink on a number of platforms, as many overclockers would. The metal backplate is coated in a plastic to ensure there is no problems with contact on the back of the motherboard. I did have one small gripe with the AM3 mounting, it requires the heatsink to sit sideways, this means that ram clearance on some boards would be very tight. I have spoken to Zero Infinity about this and they said the mounting system will be revised in the future but will be as currently is for retail release. I think this issue would only affect a very small number of motherboards though. There is a "Zero Infinity" 120mm fan mounted inside the unit, the specifications on the back of the fan read 12V, 0.29A and 2000 RPM. 2000 RPM is relatively slow for a performance fan but this could be for noise reasons and perhaps the Phantom Knight is designed to perform best at low RPM. The RRP of this unit is $70 and it is currently only available in Asia. It should be in the Australian market later this year with hopes of moving into US and Europe in the near future. Testing So how do you test the quality of a heatsink? I think the only way to do it is in a comparitive way. Comparing the heatsink directly against another unit, so I am going to put the Phantom Knight up against the Noctua NH-U12P. The Noctua NH-U12P has done very well in reviews against the Megahalems and Thermalright Ultra Extreme 120 and is a safe benchmark of a good CPU cooler. My platform will be as follows: * Gigabyte EX58-UD4P * 6GB of GSKILL Perfect Storm BBSE 2000 9-9-9-24 * Intel Bloomsfield 960 @ 1.35v 4000 MHz cpu, 3600 MHz uncore, 200 BCLK, 1600 8-9-8-24 rams, 1.35 vcore, 1.45 vtt and 1.7vdimm I will be doing the following tests: * Temperature @ bios after 1 minute idle * Temperature @ windows after 1 minute idle * Temperature after 5 minutes LinX load * Temperature after 10 minutes LinX load * Maximum temperatures reached during LinX load * Temperature after 10 minutes LinX load has finished, waiting 3 minutes to cool down I will perform 2 seperate rounds of testing, the first round will be with a 59 CFM noctua fan that is very quiet to use. The second with a loud, performance Delta fan that pushes 105 CFM. This is to simulate the 2 different setups, an air cooled 24/7 platform and benching platform. I will conduct each test 3 times. dismounting and remounting after each complete round of benchmarks. Running each set of tests 3 times per heatsink will ensure there is no bad results due to mount, I will then average out the 3 sets of results to give an overall picture of the heatsink performance. It is worth noting that the 105CFM Delta fan mounts much better to the Noctua than the Phantom Knight due to the unusual shape of the Phantom. When the fan is running you can feel air being sprayed either side of the heatsink, this would indicate not all air is being pushed through the heatsink and gives the Phantom a disadvantage in this test. I still feel this test is required as overclockers looking to buy this product will more than likely want to use high CFM fans that don't fit inside the Phantom. I will be using Realtemp to monitor my temperatures so I can get differences in cores. I have attached a temperature probe directly to the CPU but found it was just as good to use realtemp because we are actually after delta temperatures for comparitive reasons. It does not matter if I not getting the precise temperature to 1 millionth of a degree as long as we can compare the two heatsinks and see which one comes out on top. LinX has been choosen as the tool to generate heat. I am using Ceramique thermal paste which has become the defacto standard thermal paste in overclocking circles. Below is an example (with another thermalpaste because I ran out of ceramique after the review) of how I am spreading the thermal paste, then I am mounting the heatsink. The ambient temperature in the room was a solid 22 degrees celcius for the duration of the testing. Results First the results with the 59CFM Noctua fan. Now the results with the 105CFM Delta fan. Performance Analysis Some quite interesting results to be seen there. First lets look at the 59 CFM fan. Bios idle temp is the same and windows idle temperatures are almost idential, but when we move into LinX loaded temperatures we start to see a couple degrees different in the favor of the Phantom Knight. To be honest this is not what I was expecting, entering the market with a new product, target RRP of $70, so it should go on sale for $55 or $60 and it is beating the established Noctua, yes but only by a small margin. Looking down through maximum temperature during LinX load and recovery temps they both look very similar. On to the heavier testing, with our 105 CFM Delta fan attached! Starting with the bios idle temp immediately we see the Nocuta is 2 degrees in front, but when we move to windows idle temperature they are basically on level playing field, very similar results. The results do start to change a little during our LinX testing, the Noctua does take a lead of 1-2 degrees across all the cores and continues right through the 10 minutes. Recovery temperatures do come back similar again like our initial windows idle temperatures were. So what does those mixed results tell us? Looks like Phantom Knights design plays a big part here and its inability to "accept" all of the air from a Delta straped to the front. As I mentioned earlier the different shape of the Phantom means some of the air hits the side of the heatsink and redirects off, instead of being forced through the heatsink like it is on the Noctua. This does cost a few degrees performance when you are mounting fans on the outside, but when you play the fan inside it is a different story. We can see the Phantom showing a small advantage through all the 59 CFM fan testing which is great news for end users considering the Phantom for their 24/7 workstation or gaming system. Conclusion These results are so close they are almost impossible to call, but it certainly does show that the Phantom Knight is class act and capable of keeping up with the better heatsinks in the market place. I am looking forward to seeing what Zero Infinity can do in the future and through a few revisions this heatsink should be more efficiency with the few small bugs it has totally ironed out. I do though find the stock fan a little noisy, it does a nice job of keeping the CPU cool but for my 24/7 rig I would be considering replacing the 120MM fan that comes with the product with another. I have no hesitation recommending this heatsink to users for their platforms. I have been using it for a good two weeks now and find it a pleasure to remount onto other boards, it is very quiet and does keep the CPU at a great temperature. After all, it did just beat the Noctua with mid-range CFM testing, and that is impressive from a new player on the market.