Etern4l

Many thanks for those insights. I can tell I will have a ton of fun with this stuff lol. So what happens when the pad melts, I presume the excess is permanently squeezed out of the contact area? It would follow that unlike graphite pads, this one is not reusable.

K, I'm sure you will enjoy the show then.

Did you play the game?

Thanks. Since I will receive the pad first, I wonder if it would be possible to hack it for application in a desktop CPU, e.g. heat up and stretch or roll to flatten before application.

Thanks a lot. BTW I do wonder what the behaviour of the pad is going to be under significant heatsink pressure and with high temps? Will some of the pad get pushed out from the gap between the IHS and the coldplate, resulting in a thinner effective layer? That would make sense.

Yes, you need 4K content, to appreciate a 4K TV. Upscaled content won't really look that much better. It's not DLSS.

That sounds great. Just to confirm - the method is: 1) delid 2) LM on die 3) LM on the IHS (can be an IHS from a different LGA1700 CPU). I guess I will burn some time and money trying to procrastinate on this for now, but the process would solve the cooling, if not open up some OC potential. One more noob Q: why would I bother looking for another IHS to use after delidding? Isn't the CPU warranty voided and CPU split open, so any future owners of the CPU would just have to accept this, or is it possible to reassemble a delidded CPU with the original IHS (for resale for instance)? Not clear why though. If a CPU is successfully delided, it has more value in a sense as it will run cooler? BTW Why does Intel sell CPUs which practically require this procedure, even when used in custom WC cooled PCs? I have a 360 AIO + 6 3000 RPM fans in push-pull config, still not enough to cool this down at stock settings with UV.

What's the curing process exactly? AFAIR it needs to be cold curing, which would be annoying? Which shipping method did you use? I can see quite a few options (plus the more familiar ones like FedEx) Yun Express (a few different options) E-packet or SF-Express-CD E-Express. Great tips as always! BTW What the bottom line delidding result for you again? AFAIR you weren't too impressed with the 12900KS initially? Then again, 13900K is a different beast (needs 50W more power for me). Did you delid the 13900K too?

Thanks for the LM tips, I have a feeling I will soon be in a position of having no choice but to put them into practice (nothing against LM in principle, but the setup is clearly more complicated and damages the CPU markings). Good point about the pads. There is also a 7950 paste, but laptop people have been complaining about it requiring 14h curing time. Let me actually order that too, as it could prove more suitable. Edit: no luck finding any apart from ebuy7. Oh well. It will get here by Easter or something.

Yep, the real reason why the new AW management slayed the AGA.

Yeah, the problem is that they are just 8.5W/mK, and most (the only?) great reviews I have seen are from laptop users. I have yet to watch the Linus video on it. Anyway, I ordered some for science, if anything. I've pretty much tried all the top pastes now and none are fully adequate. Will try this, then go to LM if needed. A lot of conflicting info on the benefits of LM when applied on the IHS though. It's simple: 1. Do you need/want to use a PC on the go? If yes, go to 2, otherwise go to 4. 2. Can you afford a desktop and a laptop (in terms of cost and space)? Yes, go to 5, otherwise 3. 3. Broken dreams, you must remain a sole laptop user. The end. 4. Build a desktop (sell your laptop). The end. 5. Build a desktop, and keep the laptop. The end.

Has anyone here tried the Honeywell PTM7950 phase-change pad? Apparently it's the new best thing. Just ordered some on ebay, as kind of disappointed with all the non-LM pastes I have tried. They may be great for laptops or lesser CPUs, but they are not really cutting it with a standard AIO and 300W workloads.

OK, so XTM70 wasn't bad - it kind of matched the performance of Alphacool Apex after it deteriorated a bit over the last two weeks. I used the suggested "easy" application method, but this almost surely resulted of overapplication of the paste. The paste is quite dry, almost clay like, so not sure how easy it would be to apply without the sticker mould. I would try a traditional application before dismissing this though. Not being quite happy with the result, I then tried Thermaltake TFX: instant fail - 2-3C worse than than XTM70, CB23 crashed, went in the bin. Sourced from a Chinese seller on Amazon so could have been fake. Lastly, used CoolerMaster CryoFuze (as recommended by @tps3443 if I recall correctly) and actually this improved noticeably over the XTM70 so running with it now - hopefully it lasts. Application method could matter as well, in the last one I used a card to ensure the thinnest and most even spread possible. Will see how it goes. I haven't ordered a PTM 7950 pad yet, kind of loathe to order from ebay or ebuy7, but I guess there is no other way.

For those who tried Corsair XTM70: did you use their applicator lattice sticker, and was there any curing time? I gave it a go, but I'm not too impressed so far. I used the applicator sticker on the coldplate as per instructions....

Mostly successful brining timings down, quite a bit - say from CL40 to CL34, with further reduction in secondary timings. I have not had much luck with raising voltages. Bear in mind I'm talking about Fury Beast desktop DDR5, and DDR5 is a different, well.. beast. I also only use HyperX RAM in laptops with no issues, albeit never had a chance to play around with tuning due to the usual BIOS limitations.

Amazing, but yeah - that's much easier said than done. Also your ambient water cooling is anything but ordinary with a gigantic radiator and dozens of fans lol

No worries. Also bear in mind that you may be able to run the Fury with tighter clocks. Works for me with Fury Beast on the desktop side.

It's like having birthday 4 times a year with those MSI BIOS people. +400MHz stable RAM speed (with a little contact frame tweak).

So you overclocked eCores from 4.3 to 4.8GHz and that's stable. Wow, congrats. LN2?

Yeah, I mean Linux is a bit more complicated for sure, and you need to learn a couple of things. In this case, what you were looking at (everything under /dev) is called a block device. This is basically a mechanism to read/write raw data on the drive, and clearly access to that should be limited for security's sake. What you need to do to make a file system out of this is mount the block device. You do this using mount command - in most cases you don't need to worry about complicated parameters, just need to know which device you want mounted where in the overall filesystem. Obviously, unlike Windows, there are no drive letters. In your cases all you needed to do, assuming you have a /mnt/bigreciever directory (in LInux you mount filesystems into directories, and those directories specify permissions of the root of the mounted filesystem) where you want to mount additional filesystems, is sudo mkdir /mnt/bigreciever chown reciever /mnt/bigreciever chmod u+rwx /mnt/bigreciever (as per the earlier post) and lastly sudo mount /dev/sdb1 /mnt/bigreciever Thus /mnt/bigreciever would become the mount point of the sub-filesystem contained on /dev/sdb1. Using /mnt as the location for mounted external drives is just one convention, you can mount the drive almost anywhere reasonable. Most linux distros have GUI tools that do all this, usually some file or disk manager. Some generic info links below, look reasonable: https://www.thegeeksearch.com/understanding-file-systems-and-mount-points-in-linux/ https://phoenixnap.com/kb/linux-mount-command Worth mentioning that if you just use mount, this is not persisent. To make it so, you need to use either a GUI tool to look up how it's done In most distros you still need to edit /etc/fstab https://en.wikipedia.org/wiki/Fstab but not all of them. BTW in many cases you can learn enough specific info using "man" e.g. man mount man /etc/fstab HTH

Well, there are viable technical solutions if they wanted to step out of the box a little and pursue them. A true 8 lane eGPU in a large enough enclosure would be one approach. If they integrated the GPU into the eGPU device, they could maintain the lovely "no upgrades mo********s!" paradigm (I think Asus has tried this).

ECC indeed seems to be of no use to most people, not only does it slow the GPU down, but perhaps a little surprisingly it also increases memory utilization, so same content will now require a bit more VRAM. Probably not hugely noticeable in games though. I ran it for a while (with a very conservative OC) and there were no errors ever detected and corrected so why bother.

I'm not saying a reason is strictly necessary, but a) Clearly, it's a question of probability that a need might arise, and this will vary by area, for example b) In places where most people don't carry arms, it might get taken the wrong way, or put people who don't carry or even own a gun off. Simple as that.

Why do you feel a need to attend those transactions armed? Is it a matter of principle, or do you feel like you could get robbed at gun point near a police station while selling a graphics card, in Mass. (of all places)? Of course, since you do carry, then it's nice of you to tell people, although many will understandably pass.

Yep, tech is going downhill and it's only going to get worse unfortunately.