Gadgetory

after our launch day investigation it's deleting the 99 hundred K and finding its shortcomings we've been working on a follow-up involving lapping the inside of the IHS and applying liquid metal to close the story on improvement potential with the deal in process we're also returning to bring everyone back to reality on deleting the 99 hundred K because it's not quite as easy as it may look from what you're seeing online before that this video is brought to you by the be quiet dark rock 4 and dark rock pro for CPU coolers these high-end coolers to focus on a smarter approach to air cooling by adding a mini fin stack on top of the direct contact cold plate adding small bumps to the fins for increased service area and by using silent wanes 135 millimeter fans custom-built for high performance cooling without too much noise the pro is a dual Tower cooler rated for 250 watt TDP while the dark rock 4 is built for 200 watt TDP s learn more at the link of the description below just a quick reminder on this we already kind of went over this process in our review of the 9900 K but now we're adding on in a bit so we've lapped the inside of the IHS basically sanded it down and we've added liquid metal and this is something that we knew before launch Rowan was working on as well and we followed it up a bit after him so our initial testing focused on thermal paste high-end paste versus solder and you can see those results in the previous in the review or later on in this content but now we're looking at liquid metal and seeing how much further we can improve it we'll also be talking about just is it even worth deleting and kind of addressed that already so the task of lapping the inside of the IHS arose out of necessity for this one it wasn't because we necessarily wanted to the truth of the matter is that no matter what all of the tech reviewers including us make it look like on camera it's actually a higher level of difficulty to do this deal it than all the previous deal id's so there's something we tried to touch on in the review because it's easy to look at everyone deleting the processors and think this is a good idea I should do it but it's not quite that you're seeing edited down versions of what everyone's doing again including us so it might take a couple hours whereas on camera it looks like maybe 5 or 10 minutes so it's worth talking about that and making sure everyone's clear on if you should actually do this or not and the 9900 Kay to really get it working well it's not exceedingly difficult with addy lid but it is definitely more work than popping the CPU into addy litter turning a screw and putting some liquid metal on it it's not quite that easy so for this job of course you'll need to remove the silicon adhesive that's not difficult then you'll also need to carefully remove all the excess indium solder from the die and the IHS it is principally important that both the surfaces are smooth when the deal is done if it's not smooth and there are chunks of indium still remaining the CPU will almost instantly throttle and used especially with liquid metal liquid metal surface tension requires it to have an opposing surface that is also coated in liquid metal and depth differentials could throw off the contact and increase the likelihood of air gaps between the liquid metal this isn't like a paste or it'll just squish out and fill in any of the gouges or bumps and make up for mistakes although we haven't tried it one could theoretically use a liquid metal to help dissolve some of the indium we know that one is a gallon stand and the other ones just indium but again we haven't tried that we took the more barbaric approach of a knife but this requires a lot of patience slipping or getting too aggressive will very likely result in a dead CPU and it's the difficulties a bit higher here as for internal lapping of ice we have some footage of that process as well there are better ways to do it but it worked out decently so the way we did this was we wanted to help deal with gouges that were left over from indium removal where I kind of use the knife and scrape it up and you might end up with a slight cut mark or slight difference in depth between where the indium was and where it wasn't of course and so what we did we used grit 600 sandpaper then 1,200 and 2,000 then 2,000 wet and 3,000 3,000 wet ended up with a very polished service so it worked well there's more you can do of course there always is with this stuff but it was good enough there's good enough to see an improvement and we'll talk more about this process towards the end but let's go over some of the new results first and get those charts on the screen so we're not going to reread and re-explained everything from the you can check the review the first quarter of it if you want to see those results initially so if you haven't checked that you have questions about testing or what-have-you then check that content out as well when testing versus a high end thermal grizzly cryo not paste as we said in the review we found that the differences didn't amount to much Intel solder was only 4 to 5 degrees better than good paste on average leaving plenty of room to improve after our liquid metal application we saw a temperature reduction from about sixty four point four degrees Celsius delta T over ambient to about sixty degrees over ambient that's an additional four degree drop in our blender AVX workload it is possible to get even bigger temperature reductions of course but that might be something we explore separately with dice and E and this is where it's kind of interesting the reason we didn't see better performance here with liquid metal or well we did but it's only four degrees is because we have not yet sanded down the die to be smoother the IHS could be improved as well and so there are some marks on the die or on the surface of it from removing the solder previously and those marks are where you have issues with the surface tension of liquid metal as opposed to paste which can fill even fractions of a millimeter of a dip where liquid metal struggles and that's the point of this content it's to illustrate how much extra effort is required to really get good performance we spoke with their bearer separately about his improvement and he spent an entire week getting his dewitt dialed surface smoothness is now the biggest limiter to our performance results as a reminder this is on our Kraken X 62 standardized test bench cooler not the crazy setup we had for the stream a four degree drop isn't bad considering there's still so much more we could do and it's what helped contribute to our higher overclocks went on stream a few days ago speaking of the stream let's look at some of our overclocking results our original overclock was confined to 5.2 gigahertz at about 1.3 five volts after accounting for V droop as we began running up against thermal limitations with the solder and X 62 cooler although the extra nine degrees will help with overclocking on the X 62 we switched to a 540 millimeter se radiator from ek for our live stream we also attached four of the Vardar furious bands to it producing enough airflow to create hair flow as you can see in the livestream footage and it was very loud as well with all these things combined we stopped our overclocked up to five point three gigahertz stable or five point four with pseudo stability and we could definitely eventually achieve stability with more work we also increased the cash ratio to 50x eventually later on and we still have a lot more work to do on memory configurations but we saw increases in Cinebench and times by performance alike I will post some of those numbers on the screen now for Cinebench as reference just changing memory increased the score from about 2018 to about 2060 for multi-threaded the point of showing that is to just show the performance difference from memory change as we used a slightly different build for the live stream here's our scaling chart with the memory we measured 2153 CB marks when at 5.0 gigahertz all core a climb of 4.3 percent over the 2064 score going to 5.2 gigahertz all core pushed us to 22 65 CB marks or another 5% climb over the previous score and a climb of nine point seven percent over the baseline 2064 marks at five point three gigahertz we measured 22 92 cv marks and we're now hitting diminishing returns about a 1.1 percent improvement for the extra 100 megahertz all court Cinebench did not finish at five point four gigahertz with this set up but times five physics did complete four times by we measured eleven thousand eight thirty three marks at 5.0 all core twelve thousand 21 at five point one or a 1.6 percent increase then twelve thousand one 47 at five point two another one percent climb and then twelve thousand four hundred nineteen at five point three and in at twelve thousand five twenty one at five point four gigahertz we measured a score of twelve thousand six thirty five at five point three with a 50 X cash ratio and four thousand megahertz ram but this is not comparable to the previous results this is where we bring it back to reality then deleting has never been a hard requirement with intel's processors never so it's easy to get that confused out of the box it works it does okay it's never really been commendable thermally but it's fine there's a reason they can sell it like that and not get left out of the market it has become a serious requirement for actual real overclocking efforts as we've Illustrated 20 degree reductions on a 79 ad XE 18 core part our very easily done with a deal it in liquid metal so absolutely it helps this can help outside of overclocking as well like if you want to just drive down your noise levels by reducing your fan speed because the processor doesn't run as hot anymore so there are there have been good uses for d-lighting outside of the livestream type stuff we do where its competitive overclocking kind of exits practicality with the 9900 K though that benefit of the the practical benefits start to fade and you start to look more at primarily competitive overclocking benefits which are realistically very impractical for most users so with the soldered CPU it's still not a requirement to deal with it the solder is absolutely better than the original Dow Corning thermal paste I didn't tell you so we give them credit for that improvement Intel started this one out of necessity the company could not have achieved the clocks that it did with thermal paste at least not as easily and not across all eight cores not without some form of increased cooling efficiency and that was through a better interface so even still it's clearly possible to improve performance you could go direct I to we've got to cut it off somewhere though and as for whether it's worth it to deal it realistically the answer is no more so than previously previously the answer has been maybe just it's really not difficult you pop it in 2d later you turn a screw you put a new interface on it and you're done not that hard but now and it's still not that hard but now you remove the solder too and there's a higher risk of Killian or damaging the CPU there is more trial and error testing involved where you might think it's pretty good put it into a system hopefully in test bench setup not an actual computer and then maybe it's overheating instantly which sometimes happens with the previous deal it attempts to but not nearly as often so if you're going to deal it just be aware that it might be a couple of hours of trial and error if you're really an experienced with this stuff and you haven't done it before and it might be maybe an extra thirty minutes to an hour if you have done this before because there is still going to be more testing involved generally the answer though is no it's not worth the elating Upton extreme scenarios were you doing competitive overclocking something like that because an extra couple degrees nine even 15 as de Barras on some of his tests degrees off of the top yeah it can help with reducing noise it can help with other things like overclocking but it's a lot more work involved and so it's harder to recommend that route so games require more attempts to get it right there's more trial and error testing if you start grinding down the die lapping the IHS it's more likely you'll see worthwhile improvements and it's just it's not like the 798 exe where you instantly see 20 degree drops across all cores with less than an hour of work involved so we'd only recommend deleting at this point for people who are pushing high clocks who are competitively overclocking or who are so neurotic about noise that a 100 200 rpm reduction in fan speed will make your day and there's nothing wrong with that but that's kind of where we see it being worth d-lighting at this point anyway those are the results with liquid metal we know it's possible to do better and we know you can lap the die or send it down I know you can grind down the iatest some more but this I think is a pretty good representation of something that most people could accomplish in less than an afternoon a couple hours of work anything more than that you start questioning the the time being worth it or not more so than it is he questioned here so that's it for this one as always subscribe for more thanks for watching check out our our initial review for the foliar coverage of thermals and Romans video of course is still worth checking out if you want to see his more extreme approach where he grinded down the die as well and go to store documents access net to pick our new limited edition Graf logo shirt that's quad foil now it's got four-way SLI foil going on check that on the store thank you for watching I'll see you all next time