We're at PAX West today (8/31). Come say hi! Steve will be at the PC gaming panel at 7:30PM in the Sandworm theater. Article for those who prefer reading! www.gamersnexus.net/guides/3359-liquid-metal-aging-one-year-test-how-often-to-replace-liquid-metal Watch our latest hardware news video to catch-up on the industry: th-cam.com/video/ma1gh-21diQ/w-d-xo.html
I would test copper vs nickel plated copper contacts. IHS and or cooler coldplate. In terms of deterioration, I've read the nickel is far less prone to staining or just bad chemical reactions.
i had a no boot problem with my radeon R7 260x that has a cooler with a cooper base plate i had to put more liquid metal on it because the copper absorbed the liquid metal be aware this card was being used for basic internet tasks, gaming, and mining when the system was idle. after 3 - 4 months i got the card to refuse to work until i added more liquid metal
@GamersNexus I decided my 7700k with conductonaut about a year ago and haven't encountered any issues with it as far as I can tell. One thing I'd be interested in finding out is how the application with bare copper affects long term aging. I know that liquid metal doesn't react with copper like it does aluminium, but it still leaves a layer. I've heard stories of this mild interaction actually drying out the liquid metal itself over time unlike zinc covered surfaces and this accounts for the change in temperatures. Could you check out the longterm effects of this? I recall your video on the replacement pure copper IHS replacements by rockit and I suspect they would be susceptible to this kind of drying over a stock IHS.
I got a different question i could not find an answer to anywhere. What is better for heat transfer, a soldered CPU, or a CPU with Liquid Metal between it and the IHS?
I do have to say, Gamers Nexus and their test methodology is really refreshing. To hear the limitations of the study is something people kind of forget about at times and to hear it here is nice. You guys do a great job. Keep it up and nice new office!
I delided a core i7 7740x (I know... I didn't buy it, it was given to me) used it on my test bench for a solid year, never had issues, changed coolers multiple times, it allowed for a 5.3 ghz 24/7 overclock 72ish degrees under full load. I decided I wanted to seal the CPU up so I could change out the CPU, I didn't even reapply the liquid metal, I just put some silicon around the IHS, put it back in the socket to for pressure, booted it the next day, results were similar upon testing. maybe a degree off from my original results. Conductonaut is awesome, I've even had it on shunt resistors on my gtx 1070 for a year and it wiped off, no issues when I wanted to put it back to stock.
Everytime Steve says he doesn't seal the IHS because it negatively impacts performance I have to say don't put it between the IHS and substrate put it around the IHS like caulking a window or sink.
Jaggsta or if you knew anything liquid metal still yields better temps then soldering. Yes soldering is prefered for the rest of us. But deliding is best for the best overclocks and temps. But thats why im excited for the new i9. If its soldered and reviews show the OCs and temps are awesome, then im jumping there from my 6700k in a heart beat.
The issue is not really the lack of solder, its the amount of space between the CPU die and IHS. By deliding, you also remove the glue which makes it able for the IHS surface to be closer to the die, you also remove the excess amount of thermal compound and replace it with really thin spread of liquid metal. You will have a huge improvement in temperatures even if you used a normal paste during the delid its just that once you decide to delid, why not go all the way and slap LM on it as well as that will improve the temps even further.
TheWereCat yup this. I really want to delid my 6700k to do 4.9ghz. I can currently do 4.8ghz at a mere 1.325v fully stable. But temps hit 80 to 85c at full load. In normal gaming loads, it is kept well below 80c. Never hits 80 during gaming. But if i delid i can further drop temps and increase clockspeed. But im a coward and dont want to risk that 1% chance of failure.
Please dont call what they do science, it isnt. This isnt a bash of gamers nexus, its just that scientific method exists and GN doesnt follow it, I would think because its impractical to do so. For example for this test they would need a large number of identical setups, then publish their methodology and findings for peer review, then separate independent teams would need to do the same tests and find the same results, only then are we getting close to science. GN does great work, but they dont do science.
I have a 4770k and a 4790k running for 4 years now with LM without any noticeable degregation in thermal performance over the time. So I would say this stuff holds up pretty well.
@@tr0ublem4kerWZ switched to another platform a few years ago but until then still no problems with the original LM application. The temps did rise a bit after 5 years but I think a simple repaste with fresh LM would have solved that.
Had to re-apply my liquid metal today. I first applied it in 2019 achieving temps of 60c, and ~4 years have passed. My temps now before re-applying hit 100c on a 8700K , re-applying got my results of 60c max temps again.
8700k @4.8 GHz, Noctua aircooled, here. Max temp after about 7 months of use in Prime95 is ambient +60C (!), just like it was on day one. Go, Conductonaut!
Just wanted to say Thank you for all the research and benchmarks specifically on cases and cpu's. I learn a lot from yours, Linus, Jayz2cents, Science studio, Hardwarecanucks and the list goes on! Thank you again!
I absolutely LOVE your methodology. Usually, every "scientific" result that is presented to you in the media is done in such a way, that you don't know more than before or they reach unsupported conclusions. But not you. This was awesome. Thank you!
Great job as always! One suggestion/question: Testing with motherboard in vertical orientation would be interesting to see if gravity plays a part in liquid metal migration over time (as this is how most end users motherboards are mounted in their enclosures). But I totally get that you were trying to control for as many variables as possible and I appreciate your methods!
For guys who wants to check how often they should change it. I had it done 5 years ago on i7 8086k running 5.2GHz about 3 years. After that time I have noticed some blue screens and then decreased overclock to 4.4GHz (by changing offset and other settings in bios). After another 2 years I have finally decided to delid processor again and use new liquid metal. Now again I can overclock to 5.2 GHz stable running with my ddr4 3600 ram.(Max temperature 85C gaming) However I have decided to go down little bit for being safer with temperatures and voltage and running smoothly on 5.0 GHz. ( Max temperature gaming 70C). So in my example that was 3 years of stable running and I have extended to 5 years by decreasing overclock drastically.
Got a Xeon E3-1230 v2 running on liquid for the 3rd year 24/7. It's running good and cool as day 1! Just checked it, the liquid metal is the same as day 1.
PSA: liquid metal 1.5 to 2 years and went bad. I delidded my cpu and used liquid metal and thermal grizzly. Noticed when i woke up one day my fans were spinning faster then normal. I replaced the thermal grizzly no change. P95 was shooting my cpu to 100+ c instantly. So i checked the liquid metal, and it was dried up and basically non existant. Just a bit of hard residue. Had to use a razor to scrape it off the IHS, and it just rubbed off the cpu. Re applied fresh stuff. P95 now shoots to 58c. Just a heads up my pc ran 24/7. 50, 60 70 days uptime. Only time it was off was hard crash, or power outage. Pc was only used for gaming. Cpu i7 8700k 5ghz locked. Corsair AIO
Same. Just replaced the liquid metal on my 8700k after 3 years. Temps were steadily going up after 2.5 years. Started at about 25c idle and went to about 50c idle. Back to normal now. Longer term testing would definitely be beneficial for viewers.
So, I used LM on a 4790K in a laptop, then brought the LM out into weather with an ambient of -5 to -10C (around 12-24F). Either from low mounting pressure or from the cold exceeding this LM (at the time, CLU) (there was QC issues on the heatsinks in the Clevo P770ZM which were the start of doing unified heat sinks for Clevo in this fashion), or air oxidizing from contact issues, it had become hard and crusty. Only time I saw it become hard where it needed sanded (not talking about just staining of the copper IHS or copper heatsink, both of which were lapped to mirror finishes). Never seen that since then, and do not know if it was from the air or from the cold compromising the paste. If you could check those scenarios, that would be pretty cool. And thank you for the case study. I do not blame Coollabs for the incident as it seems like either the contact issue or the cold (which limits on temps for use are on the packages, I think 10C for CLU and 8C for TG Conductonaut). But, the limit on cold would be worth testing when that temp causes the alloy to go to a solid, then when it compromises it where it doesn't come back correctly to liquid. Best regards and keep up the good work.
March 2017: i7-7700k delidded and Conductonaut on the DIE. Resealed with hightemp Silicone (with little gap of course). Tested March 2017 with Bequiet 280mm AIO. 10°C less on 1.25v 4,5GHz all Cores, compared to no LM (from 65°C ->55°C). 1.4v 5GHz all Cores 25°C less with LM from ~90°C ->65°C. In 2017 i had to remount 3 Coolers. Since 3 AIO died all within 3 Months each...(all same model from RMA). Since then i use a 140mm Single tower Cooler from Noctua. In 2019 i switched Cases for USB-C Support. Since the delidding 2017 I never delidded again or looked under the IHS. Temps are still the same with 65°C-66°C max. Running 1.4v 4.5GHz all Cores. So 3 Years without fail. Conductonaut is KING. German Engineering BABY.
It seems like 4 years might be where it's at. (judging that most "success stories" came from people < 4 years and judging by the @SpesDesperatio comment here. His "case study" is quite convincing. What is your experience with liquid metal after 4 years? Anything change?
After about 4 years, the first but significant deteriorations in the temperatures of my 8700K are witnessed. The normal thermal paste between the CPU cooler and the headspreader has been replaced again and again in the meantime and then made small but insignificant differences. Now after 4 years some cores are significantly hotter than usual. Otherwise the difference was always 3-5 degrees between the cores at maximum load, meanwhile it's already a clear 11-15 degrees.
I have a 4790k running a naked delid with a Supremacy Evo (block on die, no IHS). I used Coollaboratory Liquid Ultra and after 2 years of not touching it my core temps are still within 4C. I delidded using a Rockit 88, painted all the capacitors with liquid electrical tape, barely tinned the die and middle of the block with liquid metal using a little paint brush, and mounted it using the EK Naked Ivy kit. Temps went from 85C to 60C.
Liquid metal doesn't dry. In order to dry, it would first have to be wet. Liquid metal isn't wet. It can harden, but hardening isn't the same as drying.
I've been delidded and had LM for about 3 Years and over the last 6 Months temps have slowly creeped up about a degree a Month. My temp under load is about 7-10c hotter [under load] than when I applied it. That said, my temps are still very safe but I'll be reapplying here soon.
LM never degraded for me noticeably (outside of XOC usage). So unless you try to hit highscores just leave it until you notice temps after x years. (You probably change gpus before)
I used coollaboratory liquid metal on my old C2Q6600 between lapped IHS and copper base waterblock, meaning both copper surfaces, for about 3 years and there were significant changes in thermals I would have noticed. Now running ThermalGrizzly, same as used in the video for 10 months on my delidded 8700K and it is consistent with results of case study in the video
Holy Shit! I can't imagine the electric bill for 1 year of testing just for this video. Love GN, this is where I always come for my deep dive info. Amazing work as always!
I've been using Coollaboratory Liquid Ultra on my Arctic Accelero Xtreme IV 280(X), installed on my MSI 280X Gaming 3GB OC, for over 3 years(constant use, single application). There's been no thermal performance degradation whatsoever with it in that period of time. I also recently disassembled my delidded 3570K that had CLU applied under the IHS at the same time as the 280X(over 3 years ago). I'd seen no thermal performance degradation with it either in that same period of time. When I pulled the IHS off of it, the CLU looked exactly the same as it did the day it was applied several years prior. So far as I could tell, it had not "dried out" at all. Conclusion: In some cases a single application of CLU will last at least 3 years(likely longer) without impacting thermal performance in any noticeable way. EDIT: I've witnessed Conductonaut "dry out" when applied to bare copper. After only 1 day's use. For the record. Funny thing is the Accelero Xtreme IV has a bare copper base. And, like I said, the CLU on it hasn't stopped working just as good as the day I applied it over 3 years ago. Is it "dried out"? Maybe. Maybe not. Don't know. Don't really care. If it ain't broke...
Yeah so i've been running Conductonaut for a little over 2 years on my CPU and even my GPU and they still run so cool without a single issue. It's incredible.
In that time period or from going from the stock cooler to the water block? Hard to say if the temperature actually fluctuated in those 2 year because i never wrote down my initial results, i also modded the card so that the power limit is open. At an ambient temp of 26°C my idle gpu temp in the warmed up loop is around 28-29°C and in furmark i reach around 45-49°C. Note that my GPU reached top 1 of the GPUPI 32B benchmark scores on HWBOT in its class so when i say its modded i really mean it. :)
Instructions not clear, my liquid metal formed into a man with a cold star that asked where it can find Sara Connor. I'm not sure if it will impact my cooling efficiency in the second year of use.
4 ปีที่แล้ว
@Legiro The T-800 is on its way a year late because of time travel weather. It will optimize your system.
Thanks Steve and all at GN! Just transferred my machine into a new case today, and got some Kryonaught, and reduced my 6600K temps by 8C today. I have it on a Asus Z170I ITX motherboard and a Corsair H80i AIO. Thanks for the recommendation!!!
I delid my i7 6700 non-K on July 6th and I got really impressed by the results, before the delid I was able to hit 4.3 Ghz with 1.35v (34x127) using a HyperT4, the temps would go as high as 80c sometime.. After my delid I was able to set it to 4.5 Ghz with 1.39v (34x133) and the temps rarely go higher than 70c with the same cooler! I'm loving it and I hope it keeps this way, as I don't have the conductonaut with me to replace it hahahaha Ps: Yeah! 6th gen overclock well via BCLK, and YES, bclk OC is safe on 6th gen and YES, bclk OC demands more voltage.
@Gamers Nexus 6700k delidded in june 2016 and resealed with conductionaut and a tiny bit silicone adhesive. Been running approx 1-2 hours pr. day average, at 4.7 GHz @ 1.35vcore. Till this day I have seen no degradation what so ever - temps are still max 55’ish depending on ambient (running with large custom cooling). Used PK-1 thermalpaste on top of the IHS, which however has been chanced a couple of times - no difference there either 🙂 Hope it helps, keep up the good work!
I just recently had an issue with my CPU overheating. I had applied Thermal Grizzly Conductonaut in May 2020. Now, 4 years later I noticed an overheating issue and decided it was time to replace my application of liquid metal. When I removed the heat-sink, the liquid metal had completely dried. In the area where the CPU contacted the heat-sink it looked like fine grit sand-paper. around the edges where some had squeezed out, it was in small shiny blobs that looked like liquid metal, but was rock hard. Did my best with rubbing alcohol to clean it for about half an hour, but made very little progress on restoring the original mirror finish on my Noctua NH-D15 heat-sink. After reapplying more Conductonaut the overheating issue went away. I have an i5-9600K all-core overclock to 4.7 at 1.26v. It typically idles around 30-35c. It's been basically on 24-7 for the past 4 years. It's only under significant load for gaming or encoding videos. Spends most of it's time as a home-theater PC but with an RTX 3070 it get's pressed into gaming service when needed.
Thank you for taking the time to explain the methodology, variables and what the results mean. A lot of people don't take the time to understand testing results and GN does a great job explaining how and why tests and results are the way they are and what they represent.
With regard to your experience of having to re-apply LM several times to get optimal results, I had similar results. What I found was that it is VERY detrimental to use too much, as I did in my first attempt. This is to say that ONLY BARELY enough should be used to serve as a transfer medium. Any more than that will retain heat on the CPU die, increasing temps dramatically. To get it right, I would apply very little LM to the CPU die and the IHS inner surface to match the die. I would do a test fit in the CPU socket to clamp the IHS to the die together without any adhesive. I would then remove the CPU, separate the IHS from the die and inspect for any overage that may have been squeezed out. I would use a dry Q-tip to carefully remove any excess, then re-seat the CPU and IHS in the socket. I would repeat this process a few times until I was sure that no excess LM was beading out from between the CPU die and IHS when compressed by the socket clamp. This technique always provided the best temps in subsequent testing, and I ended up getting a 30C drop in temp from the 7700K stock TIM. As others have stated, using conformal coating over the peripheral components around the CPU die is a wise precaution. My daily driver is a 7700K delidded with LM @ 5GHz, and it has been performing at the same temps for 20 months. My plan is to run it for three years before inspecting the LM application, unless I notice degradation in performance sooner.
I have been using lm in my laptop since around the start of the year. At one point I took it apart for inspection, ~2 months after application. I didn't see anything particularly wrong with the application and so only touched up a couple tiny places where the lm seemed to have not contacted well. However a couple weeks after this I noticed my temps rising as high as they ever did with normal TIM. I cleaned the cooler and cpu and gpu dies completely and reapplied the lm. I found that there was a dark residue on the copper which was very hard to remove that I suspect is the culprit. It's been over 4 months since then and I have not noticed some small decrease in thermal performance. Right after the reapplication I recorded some truly miraculously low temps. The laptop cpu was idling at ~ambient temperature. The lm I used was Conductonaut.
I was looking into cooling of FPGAs using liquid metal solutions. However, I haven't found any long-term tests of performance degradation until I came up with the idea to check Steve's channel now. This is ultimate dude! You're doing very very good work with all the benchmarking and real-life testing. Keep up the good work, thumbs up 👍👍👍
I have an i7-6700K delidded and overclocked to 4.6 GHz on all cores. I used CoolLab Liquid Pro and my temperatures over ambient are the same compared to when the CoolLab Liquid Pro was applied in November 2017. I have the same cooling and neither my cooler or thermal paste were removed. I'm using an H100i V2 for cooling. I intend to keep the setup the same and use it for long term testing to see how the CoolLab Liquid Pro holds up.
One more case study, my rig, 9900k Watercooled, delidded direct die cooling, temps haven’t changed since application in August of 2020. Now Jan 2022. I think the aging is defiantly not as much of an issue as application. 9900K @4.8 Temps idle 28*C Load Max 70*C
on the plus side, the 2020 refresh of the Asus ROG laptops will help with the big picture of this subject: 1. they have an automated LM application method 2. there's gonna be a boatload of laptops with the factory LM now, bumping up the overall sample size too
Running my i7-7700k@4,9GHz during load since March of 2017, so close to 6 years as of now (Jan.2023). I applied Thermal Grizzly Conductonaut only between the die and the ihs. Between ihs and cpu cooler used conventional paste. Now its time to renew the liquid metal, prime95 gets the cpu to 100°c instantly. The cooler at fullspeed doesnt change this.
I think such kind of case study should clearly distinguish/separate in liquid metal applied on 1) copper (as in notebooks, graphic cards or on top of a CPU IHS) and 2) nickel plated copper (as inside delidded CPUs, between the CPU die and the nickel plated IHS) The majority of comments indicates in case 2) no or very rare change of liquid metal is required during the life time of the CPU, while in case 1) it is advisable to change liquid metal after 1/2 year, because after 1 or 2 years the liquid metal has become hard/very difficult to remove and the temperatures are strongly increased, in worst case thermal throttling occurs (the higher the operating temperature the faster the liquid metal seems to get hard when applied on copper). In my understanding copper is absorbing steadily gallium from the liquid metal to form the mechanically stable and thermally quite conductive CuGa2 and this absorption is resulting in a different percentage mix of the different metals remaining in the TIM gap (mainly the gallium percentage is permanently reducing) with the ultimate result, the liquid metal gets hard at room temperature, because the gallium percentage in the liquid metal becomes too low. I think many people understand liquid metal is not allowed to come in contact with aluminum, but they see no great difference between copper and nickel plated copper. But when clearly distinguishing/separating into copper and nickel plated copper the contradicting messages, some persons saying they have not exchanged liquid metal for up to 10 years and experienced no problem, while other persons are claiming strong increase in temperature within 1 or 2 years (or even faster) for which the reason was found in liquid metal being completely or partly hard and extremely difficult to remove (the silver CuGa2 layer on the copper is only possible to remove by grinding) may be explained by nickel coating results in a large difference in how often liquid metal should/must be changed.
What was done in this case study was along the lines of "Accelerated Life Cycle Testing". This is one of (major) testing methodology used to determine the usable life of a design (you can read more online). Since I do not have the data from this test (temp ambient, high offset, low offset, cycle time, etc) thus I cannot say for sure, but if the test was conducted for an year, their test results may be representative of 5 - 20 years. I would not need to include the fresh TIM in the study, as you already have the baseline from the start of the test. If the data was recorded over the test duration, would could plot the highest temps per cycle and see the trend to be more conclusive.
The reason to re-test with fresh liquid metal and thermal compound is to show that any degradation (and there was some) is due to the cooling system itself becoming less effective.
Yes, but, a computer already has "thermal cycle" constantly. I mean temps jump up and cool down pretty much any time you do any kind of work, like watch a video. However, there it runs 24/7. So I would say this is closer to 3-5 years testing (at most), than 6-20.
Would have loved to see this test continue on for 3+ years, as I have a feeling most people never change their TIM again after installing a cpu or GPU, however I know thats unreasonable as it would have introduced more variables (multiple systems) or prevented this one year video from happening.
6yo 3570k, oc'd at 4.3 Ghz, now it's my dad's pc but still works like a champ. We're talking 2012. I'm kinda curious to have a look at what happened inside but I'll have to remove the paste if I do and tbh I have no intention to do so as long as it runs. Coollaboratory paste btw.
Would love to see you guys test something like the Graphite Thermal Pads that Innovation Cooling makes...I know Linus did some testing on it, but their testing is kinda...loose in its methodology a lot of the time, would be nice to see a more controlled test of that product (plus, if it works as well as advertised, it might make test bench swaps faster for you since there's less to clean up).
I use an IC Graphite thermal pad on my Ryzen 1700 and preforms on par with the old liquid metal I was using. Was not expecting graphite to work as good as it does. I love it because I can reuse the same pad unlimited amount of times which is great for me since I take apart my custom loop and clean it on a regular basis.
I asked about this a few months back and he said in an ask GN episode that they'd try to do it. I don't expect it to be any time soon now that RTX has released though. I'm also super interested to get tech Jesus' take on them so hopefully they eventually get around to it.
I tried Panasonic soft pgs with the wraith prism cooler and results were similar to thermal paste (1 or so worse degree worse). The cooler applies a lot of pressure, but the surface of the cooler didn't look very smooth. There were some small marks on the cooler and gaps around the heat pipes that are not ideal for graphite.
I have just done my third application in three years on my 2015 Macbook Pro (thermals are amazing). The heatsink keeps absorbing the liquid metal and it's drying up, this time I added twice as much as I did before and am sure the absorption will stop now, it can only absorb so much. I will do one more application in a year's time if needed, in 2 years if it's still drying up I will use Kryonaut and set/forget. I delidded an old 4690k and the temps didn't change for 3 years. I was able to hit 4.8GHz within the safe voltage range. I don't think there is any need for maintenance on desktop CPUs if your first application is done well.
Changing out 2 year old liquid metal today on my 8700k with copper IHS, never did have any performance drops, temps were consistent the entire time from what I could see watching hardwareinfo every day.
I've been using conductonaut on a gl502vs-ds71 laptop for almost 3 years already and just replaced with a new one. The aging impact on the temperature was negligible but I did noticed some erosion on the copper (about .01 mm deep) when cleaning.
I've had LM on my delidded 7700k for a couple of years now with no adverse effects or increases in temps yet I haven't even refreshed the MX4 paste between it and the cooler ...I do keep the system as dust free as possible ....so far so good..
I applied Liquid Metal to my 3770K back in 2014, I had to redo it recently. When I first applied it the temperatures ran at 53C. Just before I replaced it it ran at 84C. After replacing it the temperatures are back stable at 60C. Something to note is that after reapplying it I set a different fan curve causing the fan to spin up at a much higher temperature. I used liquid metal in between the die and the IHS and the IHS and the cooler.
On which of both interfaces did the liquid metal become hard? I assume no problem on "die - IHS", but on "IHS - cooler" the liquid metal got hard. Furthermore I assume the cooler is copper only (not nickel plated like the IHS)?
My liquid metal application was absorbed into the IHS. I noticed when my system became unstable and overheating and took the IHS off and saw it was BONE DRY. All that was left was some crusty oxides. My application was applied around the time of this video.
How come this video is lacking on what should be the proper way to apply LM? You guys talked so much about how crucial the application is but then failed to mention a single thing about the process. Am I the only one? Now all I came out with from these 12 minutes is that LM's performance = application. And onto search for another video on how to do it properly. Awesome.
As long as the temps stays low (or at least well below stock), no need to redo the job in a timeframe. Eventually will likely have to, assuming one holds onto the build long enough. I'd still try repasting the CPU & using better quality next time, my first was done with AC5 between IHS & cooler, should had used a better paste. On the other hand, my i7-4790K temps are still better than out of the box (22-25C idle, 54-58C load), so am not concerned. Idle was 10C higher & under load, was hitting 80C or higher quite often during gaming/benching. So for a first time job, am satisfied. The only reason why one may need to redo a liquid metal job within the first year or two would be if improperly performed. This is why I chose to remove all traces of sealant possible, this may not seem to me much by some, yet that little bit of extra space means the IHS is closer to the die. Which is what we want for best temps. So after applying CLA, gently lowered the die into the socket, placed the IHS on, held to prevent movement (but not too hard) & locked the lever down. No glue. Temps has been well under out of box, both at idle & load, however the largest difference was under load. Rarely will hit 60C & this is only when running benchmarks or long Malware scan. The Noctua NH-D15 was the used cooler before & after delid, my best cooler to date.
About 3 months ago, I put LM (Thermal Grizzly Conductonaut) in my Gigabyte Aero 15 laptop, both on the i7 CPU and the GTX 1060. Unfortunately, temps under full load remaiend the same (83C on the GPU, with OC), no improvement over stock, which seems not to be uncommon for this model. As suggested by others, I used only a very thin layer of LM. I recently noticed that my temps went up to 91°C and the GPU was throttling. I opened up the laptop again and found that ALL the LM was absorbed into the Copper heat spreader. There was no brittle residue or anything left, the DIE was almost completely dry while the copper now has changed form bronze to a silver color. I reapplied the LM (a bit more this time) and temps are back to 83°C. I Hope the copper heat spreader absorbed all the LM it could and that it won't react anymore with the LM, but I'll keep checking the temps. TL;DR: My Liquid Metal was fully absorbed into the Copper heat spreader after only 3 months of low to moderate use. Obviously , this can happen A LOT faster than the one year span some people report. YMMV
Even if the results don't look very interesting, I'm very happy that someone took the time to actually test this. Now we at least have a base line where we can say that liquid metal doesn't just magically disintegrate over time, and it's possible to apply it in a way where performance stays consistent.
I have no experience with that PSU, but I do have experience with other multi rail PSUs and high power draw 12v, and mixing and matching rails can usually avoid OCP. You just have to detect which cable belongs to which rail. I used to run two Titans running at 425 watts each off of two 8 pins each on a multi rail PSU. Used to trip OCP until I determined which cable was from which rail.
I delidded my i7 3770k in April of 2012 (was among the first in forums doing it with a razor blade). I have very thin layers of liquid metal on the die and on the inside the ihs plus on the surface of the ihs and the cooler (nickelplated copper of the Thermalright Silver Arrow SB-E). The liquid metal was from a syringe that I bought over forums in 2004 or 2005 from the guy that later sold liquid metal with the Coollaboratory brand. This combination is running for over 8 years now and I noticed no change in temperatures at all. I found screenshots of 2012 of 12h Prime95 and my delta T is 2 degrees lower after 30 minutes today (42 vs. 40 at same oc settings). But I honestly don't know if I have the same fan settings now (but I still use the same case and fans and the CPU fans ran way below 12V) and the old GPU may have produced a bit more heat (but brobably not in idle).
What temps do you get? Also, do you get really big temp spikes when near idle? I occasionally see a 10*C spike on my 7700k, but obviously much better temps compared to Intel's compund. I used Coollaboratory Liquid Pro
I honestly would like you to try this for laptops. If the transfer from DIE is bad would make a significant difference. In this experiment there are too many factors like what do you consider “cool” when it’s idle? Or when system is off during a winter.
@GamersNexus I have a 4770k with a 3 year old Liquid Ultra application... Still holds. Can't do meaningful testing as I don't have any old data but as far as I remember it's still reasonably close to what it was back then. Again, I cannot provide any data but in the current state, but what I can say is that it could still go on.
The slight changes from original 2017 application and 2018 new application is probably the aging of the AIO. Slight drop in fan RPM as they wear, aging of the coolant as you alluded to. Definitely a good video for those that de-lidded their pricey CPUs.
I wonder if similar testing could be done in an experiment with LM between compressed sheets of glass being heated and cooled, simulating a similar situation. This could also easily be set as an autonomous experiment to lessen error, and could be used to compare aging of different compounds.
Just got my CPU out after 2,5 years of use after applying liquid metal (conductonaut)... Did so because one core spiked suddenly to 100°C. after getting the cpu out i found out the metal had dried up nearly completely, so can confirm it does need reapplying over time (at least, in this case). CPU: i7 8700K @5.1GHz (daily use as gamer) cooler: corsair H115i (or predecessor, not 100% sure anymore)
Applied conductonaut about 1 year ago on 7700k. Overclocked to 4.9 at 1.32 vcore temps still about the same as day 1. Using an aio cooler h115 i to be exact and swapped fans out for Corsair ml140’s.
Maybe I missed it in the video but did you see any signs of corrosion? Also I've had two systems running LM for a while now, one running Coollaboratory Liquid Pro for 2.5 years and one running Conductonaut for 1.5 years and neither have any signs of deteriorated thermal performance.
Probably on a laptop would deteriorate due to movement and after a year you should change it, meanwhile on a setup which is sitting all the time that time could be longer...
Steve I de-lidded my 3770K using CLU. I don't think Thermal Grizzly was around at that time. **April 10th 2015** CPU overclock: 4.7Ghz, 1.150v bios, Ultra High LLC, 1.6PLL (I won the silicon lottery with this bad boy) Ambient 21.5 degrees c Corsair H100 AIO with Scyth AP15 fans Gaming Temps 42, 47, 51, 43. **September 2nd 2018** Same CPU and Overclock Ambient today is 25.6 degrees c Corsair H100i V2 (Upgrade) with Scyth AP15 fans Gaming Temps 47, 50, 53, 42. I did change GPUs during that period also.
I run liquid metal on my 4770k. Coming up on a year and haven't seen any noticeable decay in cooling, but also haven't done any actual testing to confirm. Just going by what I see when I look at hwinfo after gaming for a while or encoding or whatever. It works well, it seems to age well, and when done right is safe.
Amazing content! I realize that I started to look for the same type of scientific backed data on other channels as well, and get disappointed when it's not there.
We're at PAX West today (8/31). Come say hi! Steve will be at the PC gaming panel at 7:30PM in the Sandworm theater.
Article for those who prefer reading! www.gamersnexus.net/guides/3359-liquid-metal-aging-one-year-test-how-often-to-replace-liquid-metal
Watch our latest hardware news video to catch-up on the industry: th-cam.com/video/ma1gh-21diQ/w-d-xo.html
I would test copper vs nickel plated copper contacts. IHS and or cooler coldplate. In terms of deterioration, I've read the nickel is far less prone to staining or just bad chemical reactions.
i had a no boot problem with my radeon R7 260x that has a cooler with a cooper base plate i had to put more liquid metal on it because the copper absorbed the liquid metal be aware this card was being used for basic internet tasks, gaming, and mining when the system was idle. after 3 - 4 months i got the card to refuse to work until i added more liquid metal
@GamersNexus I decided my 7700k with conductonaut about a year ago and haven't encountered any issues with it as far as I can tell. One thing I'd be interested in finding out is how the application with bare copper affects long term aging. I know that liquid metal doesn't react with copper like it does aluminium, but it still leaves a layer. I've heard stories of this mild interaction actually drying out the liquid metal itself over time unlike zinc covered surfaces and this accounts for the change in temperatures. Could you check out the longterm effects of this? I recall your video on the replacement pure copper IHS replacements by rockit and I suspect they would be susceptible to this kind of drying over a stock IHS.
Hey, wo ist den Deutsches Interview, Alter? Dein Deutsch war doch ganz gut!
I got a different question i could not find an answer to anywhere. What is better for heat transfer, a soldered CPU, or a CPU with Liquid Metal between it and the IHS?
I do have to say, Gamers Nexus and their test methodology is really refreshing. To hear the limitations of the study is something people kind of forget about at times and to hear it here is nice. You guys do a great job. Keep it up and nice new office!
It's quite crazy that it's been 3 years since this video came out. An update may be interesting to see.
I 2nd that!
when ? its a long wait...
4th
@@gamingknight8514 5th
1 year isn't long enough :-(
I delided a core i7 7740x (I know... I didn't buy it, it was given to me) used it on my test bench for a solid year, never had issues, changed coolers multiple times, it allowed for a 5.3 ghz 24/7 overclock 72ish degrees under full load. I decided I wanted to seal the CPU up so I could change out the CPU, I didn't even reapply the liquid metal, I just put some silicon around the IHS, put it back in the socket to for pressure, booted it the next day, results were similar upon testing. maybe a degree off from my original results. Conductonaut is awesome, I've even had it on shunt resistors on my gtx 1070 for a year and it wiped off, no issues when I wanted to put it back to stock.
Timmy Joe!
I love you Timmy Joe!!!
Hey Timmy Joe! Love your channel! Hope your sub box blows up soon!
Everytime Steve says he doesn't seal the IHS because it negatively impacts performance I have to say don't put it between the IHS and substrate put it around the IHS like caulking a window or sink.
timmy joe reviews your wife
No capacity for 100 systems running thermal stress tests year long? Say it ain't so! Cool video.
2y delided 4770k, used Conductonaut. Temps havent changed since day one. Still 23°C down as when freshly delided
TheWereCat deliding is the gift of the pc gaming gods. Its crazy what dropping temps do to performance.
or if intel did it right from start and soldered them wouldn't need to do this.
Jaggsta or if you knew anything liquid metal still yields better temps then soldering. Yes soldering is prefered for the rest of us. But deliding is best for the best overclocks and temps. But thats why im excited for the new i9. If its soldered and reviews show the OCs and temps are awesome, then im jumping there from my 6700k in a heart beat.
The issue is not really the lack of solder, its the amount of space between the CPU die and IHS. By deliding, you also remove the glue which makes it able for the IHS surface to be closer to the die, you also remove the excess amount of thermal compound and replace it with really thin spread of liquid metal.
You will have a huge improvement in temperatures even if you used a normal paste during the delid its just that once you decide to delid, why not go all the way and slap LM on it as well as that will improve the temps even further.
TheWereCat yup this. I really want to delid my 6700k to do 4.9ghz. I can currently do 4.8ghz at a mere 1.325v fully stable. But temps hit 80 to 85c at full load. In normal gaming loads, it is kept well below 80c. Never hits 80 during gaming. But if i delid i can further drop temps and increase clockspeed. But im a coward and dont want to risk that 1% chance of failure.
Congrats on 300k subs, hope to say congrats on 500k by Christmas. Best science based channel for PC tech.
Please dont call what they do science, it isnt. This isnt a bash of gamers nexus, its just that scientific method exists and GN doesnt follow it, I would think because its impractical to do so. For example for this test they would need a large number of identical setups, then publish their methodology and findings for peer review, then separate independent teams would need to do the same tests and find the same results, only then are we getting close to science. GN does great work, but they dont do science.
@@kidShibuya nice gatekeeping!
Thanks, taipei 101!
+Nathan Brown: You are confusing scientific method with academic publishing. Related, but not the same thing.
@@GamersNexus thanks, I love your work, just hate what has happened to the name of science in the media due to its loose usage. It's my trigger lol.
Delid 3770k OC 4.8 for approx 5 years. Never reapplied liquid metal. Still have extremely low temps under load.
This is the kind of comment I like to see, nice!
I have a 4770k and a 4790k running for 4 years now with LM without any noticeable degregation in thermal performance over the time. So I would say this stuff holds up pretty well.
Whats the temperature on those cpus?
@@famousfighter2310 I wolud have to dig up old screenshots but from what I remember in the mid to high 60s in gaming and workloads without avx.
@@d3nso_ any repasting over 4 years?
@@tr0ublem4kerWZ switched to another platform a few years ago but until then still no problems with the original LM application. The temps did rise a bit after 5 years but I think a simple repaste with fresh LM would have solved that.
The way to put liquid metal is to put it in q tip first then rub using the q tip.
Hugely reduces spilling risk
I see black q-tips , they are specialized one for that task ??
@@danbigboss6484 I believe so
I love how you put the stuffed animal in the background
edit: I also like this angle a lot more for filming
That's from our Patreon backers!
they have a good taste!
Snowflake!
that's what I was thinking
lol I thought you meant stuffed as in taxidermy, glad to see it's only a stuffie
Had to re-apply my liquid metal today. I first applied it in 2019 achieving temps of 60c, and ~4 years have passed. My temps now before re-applying hit 100c on a 8700K , re-applying got my results of 60c max temps again.
Why no footage of old liquid metal after a year? Would have liked to see how the cpu and liquid metal looked after a year.
That would have been interesting to see, but I don't particularly care what it looks like if the performance is the same as new.
They neclver show that shit
I assume because theres probably not much to show. I would think that it doesn't look very.
@@dustinandtarynwolfe5540 Not good enough. This GN not LTT he should show the full details.
@@therealb888 what does appearance have to do with performance
8700k @4.8 GHz, Noctua aircooled, here. Max temp after about 7 months of use in Prime95 is ambient +60C (!), just like it was on day one. Go, Conductonaut!
Go 5.0GHZ! Noctua ND15?
At first I thought you ran Prime for 7 months, I was like "what madlad does that"
Just wanted to say Thank you for all the research and benchmarks specifically on cases and cpu's. I learn a lot from yours, Linus, Jayz2cents, Science studio, Hardwarecanucks and the list goes on! Thank you again!
I absolutely LOVE your methodology. Usually, every "scientific" result that is presented to you in the media is done in such a way, that you don't know more than before or they reach unsupported conclusions. But not you. This was awesome. Thank you!
Great job as always! One suggestion/question: Testing with motherboard in vertical orientation would be interesting to see if gravity plays a part in liquid metal migration over time (as this is how most end users motherboards are mounted in their enclosures). But I totally get that you were trying to control for as many variables as possible and I appreciate your methods!
For guys who wants to check how often they should change it. I had it done 5 years ago on i7 8086k running 5.2GHz about 3 years. After that time I have noticed some blue screens and then decreased overclock to 4.4GHz (by changing offset and other settings in bios). After another 2 years I have finally decided to delid processor again and use new liquid metal. Now again I can overclock to 5.2 GHz stable running with my ddr4 3600 ram.(Max temperature
85C gaming) However I have decided to go down little bit for being safer with temperatures and voltage and running smoothly on 5.0 GHz. ( Max temperature gaming 70C). So in my example that was 3 years of stable running and I have extended to 5 years by decreasing overclock drastically.
Got a Xeon E3-1230 v2 running on liquid for the 3rd year 24/7. It's running good and cool as day 1! Just checked it, the liquid metal is the same as day 1.
man bat what motherboard are you using?
I really find it cool "Brought You By Us"
Dope...
PSA: liquid metal 1.5 to 2 years and went bad. I delidded my cpu and used liquid metal and thermal grizzly. Noticed when i woke up one day my fans were spinning faster then normal. I replaced the thermal grizzly no change. P95 was shooting my cpu to 100+ c instantly. So i checked the liquid metal, and it was dried up and basically non existant. Just a bit of hard residue. Had to use a razor to scrape it off the IHS, and it just rubbed off the cpu. Re applied fresh stuff. P95 now shoots to 58c. Just a heads up my pc ran 24/7. 50, 60 70 days uptime. Only time it was off was hard crash, or power outage. Pc was only used for gaming. Cpu i7 8700k 5ghz locked. Corsair AIO
Same. Just replaced the liquid metal on my 8700k after 3 years. Temps were steadily going up after 2.5 years. Started at about 25c idle and went to about 50c idle. Back to normal now. Longer term testing would definitely be beneficial for viewers.
@@chrisbryan89 hey, was there any kind of corrosion?
So, I used LM on a 4790K in a laptop, then brought the LM out into weather with an ambient of -5 to -10C (around 12-24F). Either from low mounting pressure or from the cold exceeding this LM (at the time, CLU) (there was QC issues on the heatsinks in the Clevo P770ZM which were the start of doing unified heat sinks for Clevo in this fashion), or air oxidizing from contact issues, it had become hard and crusty. Only time I saw it become hard where it needed sanded (not talking about just staining of the copper IHS or copper heatsink, both of which were lapped to mirror finishes). Never seen that since then, and do not know if it was from the air or from the cold compromising the paste. If you could check those scenarios, that would be pretty cool. And thank you for the case study.
I do not blame Coollabs for the incident as it seems like either the contact issue or the cold (which limits on temps for use are on the packages, I think 10C for CLU and 8C for TG Conductonaut). But, the limit on cold would be worth testing when that temp causes the alloy to go to a solid, then when it compromises it where it doesn't come back correctly to liquid.
Best regards and keep up the good work.
March 2017: i7-7700k delidded and Conductonaut on the DIE. Resealed with hightemp Silicone (with little gap of course).
Tested March 2017 with Bequiet 280mm AIO. 10°C less on 1.25v 4,5GHz all Cores, compared to no LM (from 65°C ->55°C). 1.4v 5GHz all Cores 25°C less with LM from ~90°C ->65°C.
In 2017 i had to remount 3 Coolers. Since 3 AIO died all within 3 Months each...(all same model from RMA). Since then i use a 140mm Single tower Cooler from Noctua.
In 2019 i switched Cases for USB-C Support.
Since the delidding 2017 I never delidded again or looked under the IHS. Temps are still the same with 65°C-66°C max. Running 1.4v 4.5GHz all Cores.
So 3 Years without fail. Conductonaut is KING. German Engineering BABY.
It seems like 4 years might be where it's at. (judging that most "success stories" came from people < 4 years and judging by the @SpesDesperatio comment here. His "case study" is quite convincing.
What is your experience with liquid metal after 4 years? Anything change?
Ahhh, thank you for moving the hair light out of frame.
Yes! But, still completely killing contrast in the image. Get it higher up and add some better barn doors for the camera and light!
I agree, but didn't want Steve to yell at me. :-)
We can't get it higher up and we are still working on the shot. We just moved.
Sorry, I was just nitpicking about the hair light contrast! The office and everything else looks amazing. Congrats on the upgrade!
Yep, all meant to be helpful. Not hacking on GN.
After about 4 years, the first but significant deteriorations in the temperatures of my 8700K are witnessed. The normal thermal paste between the CPU cooler and the headspreader has been replaced again and again in the meantime and then made small but insignificant differences. Now after 4 years some cores are significantly hotter than usual. Otherwise the difference was always 3-5 degrees between the cores at maximum load, meanwhile it's already a clear 11-15 degrees.
So 4 years about is where it's at, huh?
“This video is brought to you by us!”. Great work! Well done. This makes me happy 👍
PS5 is gonna be probably the most wide spread example, but from what I’ve seen it’s caked on there
I have a 4790k running a naked delid with a Supremacy Evo (block on die, no IHS). I used Coollaboratory Liquid Ultra and after 2 years of not touching it my core temps are still within 4C. I delidded using a Rockit 88, painted all the capacitors with liquid electrical tape, barely tinned the die and middle of the block with liquid metal using a little paint brush, and mounted it using the EK Naked Ivy kit. Temps went from 85C to 60C.
Liquid metal doesn't dry. In order to dry, it would first have to be wet. Liquid metal isn't wet.
It can harden, but hardening isn't the same as drying.
I've been delidded and had LM for about 3 Years and over the last 6 Months temps have slowly creeped up about a degree a Month. My temp under load is about 7-10c hotter [under load] than when I applied it. That said, my temps are still very safe but I'll be reapplying here soon.
LM never degraded for me noticeably (outside of XOC usage). So unless you try to hit highscores just leave it until you notice temps after x years. (You probably change gpus before)
I used coollaboratory liquid metal on my old C2Q6600 between lapped IHS and copper base waterblock, meaning both copper surfaces, for about 3 years and there were significant changes in thermals I would have noticed.
Now running ThermalGrizzly, same as used in the video for 10 months on my delidded 8700K and it is consistent with results of case study in the video
Holy Shit! I can't imagine the electric bill for 1 year of testing just for this video. Love GN, this is where I always come for my deep dive info. Amazing work as always!
I've been using Coollaboratory Liquid Ultra on my Arctic Accelero Xtreme IV 280(X), installed on my MSI 280X Gaming 3GB OC, for over 3 years(constant use, single application). There's been no thermal performance degradation whatsoever with it in that period of time. I also recently disassembled my delidded 3570K that had CLU applied under the IHS at the same time as the 280X(over 3 years ago). I'd seen no thermal performance degradation with it either in that same period of time. When I pulled the IHS off of it, the CLU looked exactly the same as it did the day it was applied several years prior. So far as I could tell, it had not "dried out" at all. Conclusion: In some cases a single application of CLU will last at least 3 years(likely longer) without impacting thermal performance in any noticeable way.
EDIT: I've witnessed Conductonaut "dry out" when applied to bare copper. After only 1 day's use. For the record. Funny thing is the Accelero Xtreme IV has a bare copper base. And, like I said, the CLU on it hasn't stopped working just as good as the day I applied it over 3 years ago. Is it "dried out"? Maybe. Maybe not. Don't know. Don't really care. If it ain't broke...
Yeah so i've been running Conductonaut for a little over 2 years on my CPU and even my GPU and they still run so cool without a single issue. It's incredible.
What difference in temps did you see on the GPU?
In that time period or from going from the stock cooler to the water block?
Hard to say if the temperature actually fluctuated in those 2 year because i never wrote down my initial results, i also modded the card so that the power limit is open. At an ambient temp of 26°C my idle gpu temp in the warmed up loop is around 28-29°C and in furmark i reach around 45-49°C.
Note that my GPU reached top 1 of the GPUPI 32B benchmark scores on HWBOT in its class so when i say its modded i really mean it. :)
I use thermal grizzly's liquid metal on everything. No problems so far
Have you experienced any temperature changes from your Conductonaut since 6 years when you wrote this comment? Let's make this open and shut case. :)
Instructions not clear, my liquid metal formed into a man with a cold star that asked where it can find Sara Connor. I'm not sure if it will impact my cooling efficiency in the second year of use.
@Legiro
The T-800 is on its way a year late because of time travel weather. It will optimize your system.
The T-1000 (Terminator 2) wanted to kill John Connor. The T-800 from Terminator 1 wanted to kill Sarah Connor.
@@angolin9352 I bet you're a riot at parties
Why would a man be carrying a red dwarf with him?
Thanks Steve and all at GN!
Just transferred my machine into a new case today, and got some Kryonaught, and reduced my 6600K temps by 8C today. I have it on a Asus Z170I ITX motherboard and a Corsair H80i AIO.
Thanks for the recommendation!!!
i first found this channel from its first delidding video and been watching ever since!!
For me, it was looking for hype-free information about the difference between the 4GB and 2GB GTX 960 cards.
I delid my i7 6700 non-K on July 6th and I got really impressed by the results, before the delid I was able to hit 4.3 Ghz with 1.35v (34x127) using a HyperT4, the temps would go as high as 80c sometime.. After my delid I was able to set it to 4.5 Ghz with 1.39v (34x133) and the temps rarely go higher than 70c with the same cooler! I'm loving it and I hope it keeps this way, as I don't have the conductonaut with me to replace it hahahaha
Ps: Yeah! 6th gen overclock well via BCLK, and YES, bclk OC is safe on 6th gen and YES, bclk OC demands more voltage.
@Gamers Nexus
6700k delidded in june 2016 and resealed with conductionaut and a tiny bit silicone adhesive. Been running approx 1-2 hours pr. day average, at 4.7 GHz @ 1.35vcore.
Till this day I have seen no degradation what so ever - temps are still max 55’ish depending on ambient (running with large custom cooling).
Used PK-1 thermalpaste on top of the IHS, which however has been chanced a couple of times - no difference there either 🙂
Hope it helps, keep up the good work!
What about the findings that liquid metal travels and starts eating away at nearby components? Serious issue in laptops.
how about that graphite thermal pads?
I just recently had an issue with my CPU overheating. I had applied Thermal Grizzly Conductonaut in May 2020. Now, 4 years later I noticed an overheating issue and decided it was time to replace my application of liquid metal. When I removed the heat-sink, the liquid metal had completely dried. In the area where the CPU contacted the heat-sink it looked like fine grit sand-paper. around the edges where some had squeezed out, it was in small shiny blobs that looked like liquid metal, but was rock hard. Did my best with rubbing alcohol to clean it for about half an hour, but made very little progress on restoring the original mirror finish on my Noctua NH-D15 heat-sink. After reapplying more Conductonaut the overheating issue went away.
I have an i5-9600K all-core overclock to 4.7 at 1.26v. It typically idles around 30-35c. It's been basically on 24-7 for the past 4 years. It's only under significant load for gaming or encoding videos. Spends most of it's time as a home-theater PC but with an RTX 3070 it get's pressed into gaming service when needed.
Thank you. It reinforces the findings I've found in other places here, when it degrades in a period of half a year, after 3.5-4 years.
Thank you for taking the time to explain the methodology, variables and what the results mean. A lot of people don't take the time to understand testing results and GN does a great job explaining how and why tests and results are the way they are and what they represent.
Pause at 10:16: when you can't believe your GPU for a reasonable price really just got delivered to you.
Keep up the good work =)
test direct die cooling with lm
We want to!
Speaking of they sell die guards on eBay to hold your processor in socket without the IHS
pretty sure De8auer has been working on a direct die guard, probably what Steve is waiting for since he knows the quality will be top notch
@@smokeyninja9920 I'm sure he has one for socket 2066 but I'd like one of his for my 8700k
saultube44 CPU’s didn’t always have heat spreaders, and there were no issues I am aware of. What are you basing this statement on?
With regard to your experience of having to re-apply LM several times to get optimal results, I had similar results. What I found was that it is VERY detrimental to use too much, as I did in my first attempt. This is to say that ONLY BARELY enough should be used to serve as a transfer medium. Any more than that will retain heat on the CPU die, increasing temps dramatically. To get it right, I would apply very little LM to the CPU die and the IHS inner surface to match the die. I would do a test fit in the CPU socket to clamp the IHS to the die together without any adhesive. I would then remove the CPU, separate the IHS from the die and inspect for any overage that may have been squeezed out. I would use a dry Q-tip to carefully remove any excess, then re-seat the CPU and IHS in the socket. I would repeat this process a few times until I was sure that no excess LM was beading out from between the CPU die and IHS when compressed by the socket clamp. This technique always provided the best temps in subsequent testing, and I ended up getting a 30C drop in temp from the 7700K stock TIM.
As others have stated, using conformal coating over the peripheral components around the CPU die is a wise precaution. My daily driver is a 7700K delidded with LM @ 5GHz, and it has been performing at the same temps for 20 months. My plan is to run it for three years before inspecting the LM application, unless I notice degradation in performance sooner.
I have been using lm in my laptop since around the start of the year. At one point I took it apart for inspection, ~2 months after application. I didn't see anything particularly wrong with the application and so only touched up a couple tiny places where the lm seemed to have not contacted well. However a couple weeks after this I noticed my temps rising as high as they ever did with normal TIM. I cleaned the cooler and cpu and gpu dies completely and reapplied the lm. I found that there was a dark residue on the copper which was very hard to remove that I suspect is the culprit. It's been over 4 months since then and I have not noticed some small decrease in thermal performance. Right after the reapplication I recorded some truly miraculously low temps. The laptop cpu was idling at ~ambient temperature. The lm I used was Conductonaut.
I was looking into cooling of FPGAs using liquid metal solutions. However, I haven't found any long-term tests of performance degradation until I came up with the idea to check Steve's channel now. This is ultimate dude! You're doing very very good work with all the benchmarking and real-life testing. Keep up the good work, thumbs up 👍👍👍
I have an i7-6700K delidded and overclocked to 4.6 GHz on all cores. I used CoolLab Liquid Pro and my temperatures over ambient are the same compared to when the CoolLab Liquid Pro was applied in November 2017. I have the same cooling and neither my cooler or thermal paste were removed. I'm using an H100i V2 for cooling. I intend to keep the setup the same and use it for long term testing to see how the CoolLab Liquid Pro holds up.
One more case study, my rig, 9900k Watercooled, delidded direct die cooling, temps haven’t changed since application in August of 2020. Now Jan 2022. I think the aging is defiantly not as much of an issue as application.
9900K @4.8
Temps idle 28*C
Load Max 70*C
This is one of the most thorough channels I've found when it comes to computer testing and information. Thank you.
on the plus side, the 2020 refresh of the Asus ROG laptops will help with the big picture of this subject:
1. they have an automated LM application method
2. there's gonna be a boatload of laptops with the factory LM now, bumping up the overall sample size too
Running my i7-7700k@4,9GHz during load since March of 2017, so close to 6 years as of now (Jan.2023). I applied Thermal Grizzly Conductonaut only between the die and the ihs. Between ihs and cpu cooler used conventional paste. Now its time to renew the liquid metal, prime95 gets the cpu to 100°c instantly. The cooler at fullspeed doesnt change this.
I highly recommend picking up Micro application brushes in 2.0mm, search them on Amazon, love doing liquid metal applications with them.
I didn't consider thermal contraction/expansion relationship to wear. You guys are legit.
I think such kind of case study should clearly distinguish/separate in liquid metal applied on
1) copper (as in notebooks, graphic cards or on top of a CPU IHS) and
2) nickel plated copper (as inside delidded CPUs, between the CPU die and the nickel plated IHS)
The majority of comments indicates in case 2) no or very rare change of liquid metal is required during the life time of the CPU, while in case 1) it is advisable to change liquid metal after 1/2 year, because after 1 or 2 years the liquid metal has become hard/very difficult to remove and the temperatures are strongly increased, in worst case thermal throttling occurs (the higher the operating temperature the faster the liquid metal seems to get hard when applied on copper).
In my understanding copper is absorbing steadily gallium from the liquid metal to form the mechanically stable and thermally quite conductive CuGa2 and this absorption is resulting in a different percentage mix of the different metals remaining in the TIM gap (mainly the gallium percentage is permanently reducing) with the ultimate result, the liquid metal gets hard at room temperature, because the gallium percentage in the liquid metal becomes too low.
I think many people understand liquid metal is not allowed to come in contact with aluminum, but they see no great difference between copper and nickel plated copper. But when clearly distinguishing/separating into copper and nickel plated copper the contradicting messages, some persons saying they have not exchanged liquid metal for up to 10 years and experienced no problem, while other persons are claiming strong increase in temperature within 1 or 2 years (or even faster) for which the reason was found in liquid metal being completely or partly hard and extremely difficult to remove (the silver CuGa2 layer on the copper is only possible to remove by grinding) may be explained by nickel coating results in a large difference in how often liquid metal should/must be changed.
What was done in this case study was along the lines of "Accelerated Life Cycle Testing".
This is one of (major) testing methodology used to determine the usable life of a design (you can read more online).
Since I do not have the data from this test (temp ambient, high offset, low offset, cycle time, etc) thus I cannot say for sure, but if the test was conducted for an year, their test results may be representative of 5 - 20 years. I would not need to include the fresh TIM in the study, as you already have the baseline from the start of the test.
If the data was recorded over the test duration, would could plot the highest temps per cycle and see the trend to be more conclusive.
The reason to re-test with fresh liquid metal and thermal compound is to show that any degradation (and there was some) is due to the cooling system itself becoming less effective.
Yes, but, a computer already has "thermal cycle" constantly. I mean temps jump up and cool down pretty much any time you do any kind of work, like watch a video. However, there it runs 24/7. So I would say this is closer to 3-5 years testing (at most), than 6-20.
Would have loved to see this test continue on for 3+ years, as I have a feeling most people never change their TIM again after installing a cpu or GPU, however I know thats unreasonable as it would have introduced more variables (multiple systems) or prevented this one year video from happening.
Yeah, would be nice to watch that rn
6yo 3570k, oc'd at 4.3 Ghz, now it's my dad's pc but still works like a champ. We're talking 2012.
I'm kinda curious to have a look at what happened inside but I'll have to remove the paste if I do and tbh I have no intention to do so as long as it runs. Coollaboratory paste btw.
Thank you for listening to our comments about the hair light...looks much better...Also, thank you for this video...great content!!
Would love to see you guys test something like the Graphite Thermal Pads that Innovation Cooling makes...I know Linus did some testing on it, but their testing is kinda...loose in its methodology a lot of the time, would be nice to see a more controlled test of that product (plus, if it works as well as advertised, it might make test bench swaps faster for you since there's less to clean up).
Still would like to see a video on the IC Graphite thermal pad, but this vid is still informative thumbs up. 👍
I use an IC Graphite thermal pad on my Ryzen 1700 and preforms on par with the old liquid metal I was using. Was not expecting graphite to work as good as it does. I love it because I can reuse the same pad unlimited amount of times which is great for me since I take apart my custom loop and clean it on a regular basis.
I'd like to see the graphite thermal pad compared to things like regular
I asked about this a few months back and he said in an ask GN episode that they'd try to do it. I don't expect it to be any time soon now that RTX has released though. I'm also super interested to get tech Jesus' take on them so hopefully they eventually get around to it.
I tried Panasonic soft pgs with the wraith prism cooler and results were similar to thermal paste (1 or so worse degree worse). The cooler applies a lot of pressure, but the surface of the cooler didn't look very smooth. There were some small marks on the cooler and gaps around the heat pipes that are not ideal for graphite.
I have just done my third application in three years on my 2015 Macbook Pro (thermals are amazing).
The heatsink keeps absorbing the liquid metal and it's drying up, this time I added twice as much as I did before and am sure the absorption will stop now, it can only absorb so much.
I will do one more application in a year's time if needed, in 2 years if it's still drying up I will use Kryonaut and set/forget.
I delidded an old 4690k and the temps didn't change for 3 years. I was able to hit 4.8GHz within the safe voltage range. I don't think there is any need for maintenance on desktop CPUs if your first application is done well.
Changing out 2 year old liquid metal today on my 8700k with copper IHS, never did have any performance drops, temps were consistent the entire time from what I could see watching hardwareinfo every day.
I've been using conductonaut on a gl502vs-ds71 laptop for almost 3 years already and just replaced with a new one.
The aging impact on the temperature was negligible but I did noticed some erosion on the copper (about .01 mm deep) when cleaning.
Erosion on copper? That's strange
Still feels like this was only a few years ago. Wild. I'd love to see some new looks at the Thermal Grizzly pastes they currently sell!
I've had LM on my delidded 7700k for a couple of years now with no adverse effects or increases in temps yet I haven't even refreshed the MX4 paste between it and the cooler ...I do keep the system as dust free as possible ....so far so good..
I applied Liquid Metal to my 3770K back in 2014, I had to redo it recently.
When I first applied it the temperatures ran at 53C. Just before I replaced it it ran at 84C.
After replacing it the temperatures are back stable at 60C.
Something to note is that after reapplying it I set a different fan curve causing the fan to spin up at a much higher temperature.
I used liquid metal in between the die and the IHS and the IHS and the cooler.
On which of both interfaces did the liquid metal become hard?
I assume no problem on "die - IHS", but on "IHS - cooler" the liquid metal got hard.
Furthermore I assume the cooler is copper only (not nickel plated like the IHS)?
My liquid metal application was absorbed into the IHS. I noticed when my system became unstable and overheating and took the IHS off and saw it was BONE DRY. All that was left was some crusty oxides. My application was applied around the time of this video.
bro, your content is on another level! tests like this? no one is on this level. Linus Tech Tips got nothing on you! Mad props!
How come this video is lacking on what should be the proper way to apply LM? You guys talked so much about how crucial the application is but then failed to mention a single thing about the process. Am I the only one? Now all I came out with from these 12 minutes is that LM's performance = application. And onto search for another video on how to do it properly. Awesome.
The way he spread it in another video was nothing to brag about.
As long as the temps stays low (or at least well below stock), no need to redo the job in a timeframe. Eventually will likely have to, assuming one holds onto the build long enough. I'd still try repasting the CPU & using better quality next time, my first was done with AC5 between IHS & cooler, should had used a better paste. On the other hand, my i7-4790K temps are still better than out of the box (22-25C idle, 54-58C load), so am not concerned. Idle was 10C higher & under load, was hitting 80C or higher quite often during gaming/benching. So for a first time job, am satisfied.
The only reason why one may need to redo a liquid metal job within the first year or two would be if improperly performed. This is why I chose to remove all traces of sealant possible, this may not seem to me much by some, yet that little bit of extra space means the IHS is closer to the die. Which is what we want for best temps. So after applying CLA, gently lowered the die into the socket, placed the IHS on, held to prevent movement (but not too hard) & locked the lever down. No glue.
Temps has been well under out of box, both at idle & load, however the largest difference was under load. Rarely will hit 60C & this is only when running benchmarks or long Malware scan. The Noctua NH-D15 was the used cooler before & after delid, my best cooler to date.
I have had my 3770k delided for 4 years and it is still giving me good temps using coollaboratories LM. Never taken the IHS back off. 4.3ghz 1.2v.
About 3 months ago, I put LM (Thermal Grizzly Conductonaut) in my Gigabyte Aero 15 laptop, both on the i7 CPU and the GTX 1060. Unfortunately, temps under full load remaiend the same (83C on the GPU, with OC), no improvement over stock, which seems not to be uncommon for this model.
As suggested by others, I used only a very thin layer of LM. I recently noticed that my temps went up to 91°C and the GPU was throttling. I opened up the laptop again and found that ALL the LM was absorbed into the Copper heat spreader. There was no brittle residue or anything left, the DIE was almost completely dry while the copper now has changed form bronze to a silver color. I reapplied the LM (a bit more this time) and temps are back to 83°C. I Hope the copper heat spreader absorbed all the LM it could and that it won't react anymore with the LM, but I'll keep checking the temps.
TL;DR: My Liquid Metal was fully absorbed into the Copper heat spreader after only 3 months of low to moderate use. Obviously , this can happen A LOT faster than the one year span some people report. YMMV
how did it go after second re-application?
Even if the results don't look very interesting, I'm very happy that someone took the time to actually test this. Now we at least have a base line where we can say that liquid metal doesn't just magically disintegrate over time, and it's possible to apply it in a way where performance stays consistent.
This test is gold. Thank you for this!
I have no experience with that PSU, but I do have experience with other multi rail PSUs and high power draw 12v, and mixing and matching rails can usually avoid OCP. You just have to detect which cable belongs to which rail. I used to run two Titans running at 425 watts each off of two 8 pins each on a multi rail PSU. Used to trip OCP until I determined which cable was from which rail.
Floodlight almost gone - improvements all around :)
This angle feels a lot closer to what you had in the old studio, it provides more "depth" compared to the full frontal one too, I prefer this one.
I delidded my i7 3770k in April of 2012 (was among the first in forums doing it with a razor blade). I have very thin layers of liquid metal on the die and on the inside the ihs plus on the surface of the ihs and the cooler (nickelplated copper of the Thermalright Silver Arrow SB-E).
The liquid metal was from a syringe that I bought over forums in 2004 or 2005 from the guy that later sold liquid metal with the Coollaboratory brand.
This combination is running for over 8 years now and I noticed no change in temperatures at all.
I found screenshots of 2012 of 12h Prime95 and my delta T is 2 degrees lower after 30 minutes today (42 vs. 40 at same oc settings). But I honestly don't know if I have the same fan settings now (but I still use the same case and fans and the CPU fans ran way below 12V) and the old GPU may have produced a bit more heat (but brobably not in idle).
My 7700k performs the same and I've been using it with liquid metal for 19 months.
What temps do you get? Also, do you get really big temp spikes when near idle? I occasionally see a 10*C spike on my 7700k, but obviously much better temps compared to Intel's compund. I used Coollaboratory Liquid Pro
@@fieldcar sits around 43c in games. I've got a corsair water AIO on it. It's been mining at night. Gets closer to 50c when mining.
Fantastic vid !!! First one with this info. Thanks so much
I honestly would like you to try this for laptops. If the transfer from DIE is bad would make a significant difference.
In this experiment there are too many factors like what do you consider “cool” when it’s idle? Or when system is off during a winter.
@GamersNexus I have a 4770k with a 3 year old Liquid Ultra application... Still holds. Can't do meaningful testing as I don't have any old data but as far as I remember it's still reasonably close to what it was back then. Again, I cannot provide any data but in the current state, but what I can say is that it could still go on.
You are doing the science. Love it.
Does this apply to the new laptops that come with factory applied liquid metal?
The slight changes from original 2017 application and 2018 new application is probably the aging of the AIO. Slight drop in fan RPM as they wear, aging of the coolant as you alluded to. Definitely a good video for those that de-lidded their pricey CPUs.
I wonder if similar testing could be done in an experiment with LM between compressed sheets of glass being heated and cooled, simulating a similar situation. This could also easily be set as an autonomous experiment to lessen error, and could be used to compare aging of different compounds.
Forgot to add that it would allow for visual feedback and could potentially be used with a microscope if desired
did the test keep going? 5 years later..
Good info, thanks guys, good job
Use a heat gun to warm up the liquid metal prior to application to see if that improves the "wetting" action.
Just got my CPU out after 2,5 years of use after applying liquid metal (conductonaut)... Did so because one core spiked suddenly to 100°C. after getting the cpu out i found out the metal had dried up nearly completely, so can confirm it does need reapplying over time (at least, in this case).
CPU: i7 8700K @5.1GHz (daily use as gamer)
cooler: corsair H115i (or predecessor, not 100% sure anymore)
Applied conductonaut about 1 year ago on 7700k. Overclocked to 4.9 at 1.32 vcore temps still about the same as day 1. Using an aio cooler h115 i to be exact and swapped fans out for Corsair ml140’s.
Maybe I missed it in the video but did you see any signs of corrosion? Also I've had two systems running LM for a while now, one running Coollaboratory Liquid Pro for 2.5 years and one running Conductonaut for 1.5 years and neither have any signs of deteriorated thermal performance.
Probably on a laptop would deteriorate due to movement and after a year you should change it, meanwhile on a setup which is sitting all the time that time could be longer...
Steve I de-lidded my 3770K using CLU. I don't think Thermal Grizzly was around at that time.
**April 10th 2015**
CPU overclock: 4.7Ghz, 1.150v bios, Ultra High LLC, 1.6PLL (I won the silicon lottery with this bad boy)
Ambient 21.5 degrees c
Corsair H100 AIO with Scyth AP15 fans
Gaming Temps 42, 47, 51, 43.
**September 2nd 2018**
Same CPU and Overclock
Ambient today is 25.6 degrees c
Corsair H100i V2 (Upgrade) with Scyth AP15 fans
Gaming Temps 47, 50, 53, 42.
I did change GPUs during that period also.
I run liquid metal on my 4770k. Coming up on a year and haven't seen any noticeable decay in cooling, but also haven't done any actual testing to confirm. Just going by what I see when I look at hwinfo after gaming for a while or encoding or whatever. It works well, it seems to age well, and when done right is safe.
Amazing content! I realize that I started to look for the same type of scientific backed data on other channels as well, and get disappointed when it's not there.
Really appreciate your appreciation of the scientific method.