Gadgetory

the semi-custom loop ice wolf cooler landed in our shop after one of our viewers generously offered to loan us the unit straight from the German manufacturer alpha cool the ice wolf GP X 1080 deploys a quick-release solution for integration into a semi custom liquid cooling loop and differentiates itself from pre-built hybrid products like the EVGA hybrid by providing full coverage of the PCB not a full coverage water block but full coverage aia saw uses aluminum fins and a base plate that contacts the vram and the vrm directly using a built-in pump to circulate liquid through whatever radiator is attached before getting to our review of the ice wolf this content is brought to you by AMD freesync devices like the affordable LG 29 um 67p 2560 by 1080 ultra wide which is currently $300 check the link below for more information on that and this is the $130 ice wolf liquid cooler from alpha cool the unit here is one we're looking at the rest of this is part of their Ice Bear line and you'll need something like this to hook into because these quick releases are that basically screws that tension in you need a radiator on there obviously to use it so it's this isn't something you can buy for 130 bucks and use out a box you do have to buy another part but that's what we're looking at today we've mounted it to our test subject the 1080 founders edition which is undergone quite a bit of abuse in the last six months of its life but it's back and it's going through some more tear downs and build ups with the Alpha coil unit so we're mounting the block to the GPU which contains a pump pre filled with liquid and large heatsink for the full PCB and back side of the PCB out of the pump block protrudes two tubes and in and an out valve and those tubes terminate with locking valves these can be connected to compatible products again like the ice bear which is 280 millimeter cooler that we bought for around hundred and forty five dollars and that also cools the CPU by the way EK WB's quick release valves will not work with the ice wolf and the same is true for a couple other competitors but we do have the e KW v x LC predator and will be soon pitting it against the ice wolf check back this week for that review alva cools tubes use an 11 millimeter outer outer and our rubberized encased in coils to help prevent bends that would kink flow and G quarter inch threads allow for relatively universal fitment to other liquid cooling products you could even avoid using the locking and quick releases if desired though the value kind of plummets at that point since you may as well go for an open loop setup the base plate provides full coverage of the PCB but again it is not a full coverage water block the liquid lives only in the pump and this means that future GPU upgrades won't require repurchasing the pump part of the solution you'd only have to pay for a new aluminum plate and cooling block and back plate and things like that and then reattach the pump to the new card so in theory it's a bit cheaper that way but that's assuming that the mounting holes are the same spacing for future video cards the entire unit is comprised of a base plate a back plate and integrated pump which is this part right here that has the tubes coming out of it and then thermal pads that you have to cut the size a bit annoyingly we'll talk about that more later and all this means that you can actually run this isolated from a CPU cooler if you wanted to which again the ice bear is a CPU cooler that's got its own disconnect valve up here you can unscrew it and connect it to whatever you want or you can connect it into a loop if you run it alone it makes for kind of weird value because the block is basically just feeding straight into a radiator and the whole cool part of this the entire reason you'd buy something like this is because it hooks into a CPU cooler and makes for a poor man's open-loop cooling solution or lazy man's effing loop cooling solution basically you don't have to deal with the whole separate reservoir separate pump all the stuff that you do with open-loop cooling but again it's not quite as effective as an open-loop setup so we'll look at all the thermal performance today the main things to look at we've already talked about installation in the previous video but we'll be talking about that again today because the previous video is a blind installation so now I've got some thoughts on it compiled based on our experiences with that installing the IU's 12 isn't particularly challenging but it can be pretty annoying at times the Theroux had for instance need to be cut to size by the user and about half the pads are pre-cut like the vram but the rest is up to you to do a thoughtfully supplied template by alpha cool does make this easier to their credit but it just seems like a silly place for the cooling company to save money when they spent so much on the massive aluminum heatsink regardless not the biggest deal in the world the quick-release system however is a little more obnoxious its finicky the threads will sort of lock but if you apply extra pressure for reinsurance because obviously you don't want leakage they'll slip and need to be retighten this is part of the spring tensioning system that they have it is a weak point in the spring tensioning system and this is a casualty of weak plastics in the valve among other things Alvah cool could improve this by adding some feedback to the user for fully secured and sealed connections similar to ek water blocks which have a clear pop upon mocking note though that you should also expect a few drops of liquid to leak out during connection this is normal we'd recommend putting down paper towels as safety and then running a leak test prior to fall operation speaking of leaks during the connection to our I spare 280 unit also from alpha coal that we purchased we had an issue where the cheap plastic locking mechanism and it's spring popped out of the valve resultant of the aforementioned slipping threads when you tighten them just to be sure fortunately we hold the tubes up when installing these types of things so no fluid was really lost at least none to the board where it would matter just be careful when you're connecting everything get into the testing our full testing methodology as always is defined in the article in the description below you can click that link for the full review some extra tests and the methodology the main things to talk about here we've connected more than just the normal software to read the gpio temperature attached to this video card are two thermocouple probes these were used for our EVGA vrm testing and they basically one goes to the MOSFETs it goes the third MOSFET up from the bottom and the other one goes to the back side of the PCB where there's a bit of a hot spot starting with a baseline thermal analysis this simple test runs fairmark for just 25 minutes and it's all automated with scripting we've written in-house so it provides an understanding of performance prior to endurance testing for baseline GPU temperatures we're seeing the alpha cool GP X Pro idling at 5.4 Celsius delta T prior to any real testing this compares to about two salsas delta T on the G on a hybrid that we built which is just a founders Edition card with an EVGA CLC attached to it and that's something ages ago before the EVGA hybrids were available if you're wondering how that makes sense regarding the lower idle temperature on the EVGA solution that we built here's why there are two things that cause the increased GPU diode temperature on the ice Wolff cooled GTX 10 a DfE it's the same video card the same clocks but this one as you saw with idle runs a couple degrees warmer the first reason is because this base plate is sinking for the vram and the vrm and any other hot components on the front side of the PCB and even sinking the backside of the PCB so it's pulling more heat in to that central area where the pump is responsible for obviously sending it out through the tube system to get dissipated once it hits the radiator so that increases the GPU diode temperature because you're just pulling all this other heat in from the card to one central area now also well and because the liquid is then working harder and running hotter now also same idea liquid running hotter because we've got it attached to this this is the CPU cooler it's attached to the CPU even though the CPU is not really doing anything it's still generating Heat even when it's idle it generates at least some level of load and so that warms up the liquid as well and that's why you get those differences in results when it's idling when it's under load it's a bit of a different story though as for low temperatures we're looking at the GPX Pro at about fifteen point five Celsius on the GN high rate at about fifteen point three Celsius no real difference here but we do see that the alpha cool unit is more efficient at keeping the GPU temperatures under control when it's under load because we're also cooling everything else on the card let's move on to endurance testing this first chart shows the GPU MOSFET number three and PCB backside temperatures for the devices the probes remained in the same place for each test but I'll talk about some potential flaws here in a moment we're seeing load temperatures are about the same when tested over a one-hour firm mark vrm burnin and the fats are significantly different in temperature though the Alpha cooled GP x pro is keeping its mosfet number three around six celsius cooler than the GN hybrid solution which relies entirely on a base plate and vrm blower fan for vrm cooling and because the blower fan ties the GPU temperature the GPU temperature is low because it liquid cooler it doesn't run that fast as for the PCB backside temperature is my primary theory for the increase seen in the GP x pro PCB backside is that the massive backplate and thermal pads which connect the front side just through the top of the card are helping to better spread heat across the entire unit this can cause a thermocouple to look like it's experiencing a higher temperature than is perhaps realistic and that's also because it's indirectly touching the back plate so it's sort of the center of the transfer on the backside of the card endurance tests with GPU over clocks manually applied position the GPX pro and GN hybrid about 0.3 Celsius apart in GPU temperature with a GPX Pro again advantaged in MOSFET temperatures the probe on the PCB is reporting higher temperatures but that's because we've got a thermocouple sandwiched between the thermal pad of the backplate all the heat is transferring through that area as explained given that this thermocouple is positioned on a hot spot the 42c backplate temperature or back side of the PCB temperature I should say isn't that s'what it's attached to is still way more than acceptable it's perfectly fine it's not like we're hitting 130 Celsius on this PCB and just as a reminder vrm temperatures can go higher than what you might be used to seeing with something like a GPU diode it's not the same we're 100 Celsius is really really bad on a vrm not such a big deal a lot of them have a t.j.maxx of 150 C so a bit of a different temperature scale therefore what's acceptable one final test to go over before sending you to the article for the rest these are in more of a vacuum we're looking at only the GPX Pro for this set of tests this benchmark is with the CPU being loaded with prime95 LF of T is maxing out load at 100 percent while the GPU is being tortured with fairmark also at 100 percent load and fir mark is pretty stressful on the V RMS and power design too this results in high temperatures in both the CPU and the GPU across the entire board and since we're using a shared loop that's an important thing to look at the results position the CPU at about 52 C delta T under load or about 70 to 73 when accounting for our ambient and this is with the GPU at around 19 Celsius delta T or roughly 4 Celsius higher than when the CPU was under minimal load in the previous test sepia cooling isn't great to be honest but it is cooling two hot devices at 100% load which is pretty uncommon even in gaming especially in gaming regardless we'll talk about that more in our future ice bear review for low rpm testing if you're curious about the noise and performance trade-off we've got that as well it's in the article in description below and we've also got the noise testing in that article as for the impact of cooling on clock rate stability and maximum clocks we'll break this down in two different ways the first is a straight chart the next is an overtime line graph in this table we're looking at the average clock and maximum clock of both solutions when playing GTA 5 and dirt rally alpha cool manages to sustain a higher average clock even with the stock configuration that's had no overclock applied with GTA 5 we're seeing a difference of about 14 megahertz between the GN hybrid an alpha cool 1080 Fe at stock clocks the Alpha cool variant at 1866 megahertz to the GN hybrid variant at 1852 megahertz the maximum clock is about the same for this setup dirt rally also posts an identical maximum at clock rate between the non overclocked cards which makes sense but we're seeing a boosted average clock rate with the alpha cool GP x pro and that gain is about 45 to 50 megahertz though recurrent results are similar though we're seeing the higher maximum clocks on the Alpha cool card the result is a 15 megahertz higher average clock rate with dirt rally with a GP x pro at 20 86 megahertz to the GN hybrid 20 70 megahertz it's actually somewhat substantial since we're already pushing the limits of what Pascal can do GTA 5 gives us a 26 mega Hertz increase in maximum clock rate with sustained averages about 40 to 45 mega Hertz higher than the GN hybrid card just for another way to visualize that data here's that line graph with an over time plot of frequency on the charts between the hybrid DIY solution we built and the Alpha cool DIY solution we built the question is whether or not this makes any real difference in gaming in the real world so our temperatures are pretty good and actually we've got a slightly higher clock rate with the Alva cool solution over the other solution the GN hybrid built with an EVGA cooler and a vrm blower fan but does it matter that's where FPS testing comes in for dirt rally at 4k Ultra we're seeing stock clocked GT X 1080 hybrid performance at around 90 1.7 FPS average 82 ish on 1% low and 76 ish on the point one percent lows looking next to the alpha cool version also at stock clocks before overclocking is effectively identical there's no noteworthy change when looking at overclocks it's clear that the Alpha cool card has a very slight lead about three to four FPS average and that's a result of a slightly higher overclock capable because of the solution even though these differences are significant insofar as the larger gaps in 1% 0.1% lows with these two cards the performance is basically the same it's not detectable to the user at least any reasonable user who actually is living in reality FPS is high enough on each device that performance looks about the same still this shows that there's at least some benefit from the increased clock rate stability of the alpha cool card is just irrelevant moving on to GTA 5 at 1440p with very high settings and ultra settings we're seeing pretty minimal performance scale in between the devices they're effectively identical performance on the card with stock clocks as equal and there's no significant difference between the GN hybrid and the alpha cool overclocked GTX 1080s there is no perceptible performance difference between them at all so is this worth it well the end result is pretty similar when we look at FPS to other liquid cooling solutions versus a stock 1080 like the EFI card stock with its original cooler you can see about a 5% FPS increase just from that higher sustained clock because thermals are no longer concerned because the EFI cooler will run into thermal limitations at some point it can handle more heat but GPU boost basically tells the the GTX 1080 is saying I really want to stay at or below 81 Celsius so we're going to modify the clocks make sure that happens these eliminate that problem so you do gain about 5% FPS there we've talked about that in the past in previous reviews of the 1080 and of our build of the 1080 hybrid originally but does it matter when you move from one a liquid-cooled solution to another well not so much the FPS difference is basically none it's almost no FPS difference a couple frames sometimes but even when the lows are higher on this thing it's still not a big deal so the question then becomes one of value particularly why buying one of these would make sense there are three possible answers to that first it's a fun project and being an enthusiast is part way about building things that aren't necessarily needed but cool to do in that regard the Alpha cool ice wolf is a fun teardown and build up project for a Saturday and the psychological or uniqueness benefit of that may be great enough to justify the purchase it's just kind of fun to do the second answer is that because we're getting a semi custom loop out of this we're also getting the benefit of a multi pump system that has some redundancy in the case of failure so we're consolidating to a single radiator in the loop this increases the cooling efficiency sort of per square area used in the case but it also provides a peace of mind in the event of one failing pump that you'd obviously want to replace it pretty immediately the final possible answer is related to noise the noise argument here is that you could run lower fan RPMs through something like a 280 millimeter radiator if connected to the CPU block and completely eliminate the GPU fan merge the CPU and GPU cooling into one or two fans as it were and then you end up with lower rpms across those fans because it's all liquid so it performs pretty efficiently as is this makes for lower system noise levels overall while retaining cooling that is still better than an air-cooled card and that's really the big one and the one we've talked about before in terms of advantages if none of those three things are interesting to you it's probably not worth the purchase the quality of materials is high enough the modularity is interesting the efficiency is good but if you don't care about those three factors the fun project factor the noise factor or you think you get better noise from an alternative solution and you just don't care about the other benefits of this then don't buy it it doesn't make sense for those use cases the next thing these quick release valves are awful they're probably the worst ones I've seen these from ek far and away better and we'll be talking about these shortly these are simple button press pull it out and when you reconnect it there's a really definitive click so you know it's connected with these when you're screwing them back together the threads if you go just one tick too far will loosen and you'll have to retighten it there's also the risk where we had the whole kind of casing explode and I had to reassemble it while holding the tube up in the air so one week so thanks again to Eric for loaning us this unit to work on we'll be sending it back to him with some extras maybe a shirt or something so if you have something that you think we don't have access to like one of these let us know might look at it and work something out but otherwise as always pay traveling at the post roll video how is that directly subscribe for more we've got some interesting content coming up this week including a look at our new solution for robocopy and our files on the GN servers and moving them to our Nass and then from the dash to the web it's really cool setup I'm excited to talk about it so subscribe for that links in the description below I'll see you all next time you