I've pondered de-lidding and replacing the TIM under the IHS on and off for a bit, but having trouble deciding whether it would be worthwhile. CPU: Xeon 1245v2 Idle temps: 47-51c each core Max load temps: 87-92c each core Measured with HWMonitor The temps don't bother me, though knocking 10c off the load temp could in turn mean reduced fan noise at load. Y'see, I have fans set to the lowest setting available, i.e. they only spin up just enough to prevent the CPU from throttling and/or cooking itself. I've seen some stellar de-lidding and TIM replacement results and some ho-hum results as well. I think what's preventing me from diving in is that all cores are showing similar temps, suggesting that the TIM application itself is just fine, even if the TIM is reportedly garbage. Another thing I've not been able to confirm is if the TIM is the same on the desktop parts as the Xeons. I don't see why it would be different, but I'd be a little miffed if I found myself £250-300 out of pocket because the Xeons were soldered or epoxied. Then there's a further issue that there's no IB Xeons on Amazon Prime, so I'm missing my main work/play PC until I can source a replacement. I've de-lidded successfully before, so familiar and comfortable with the process, but at the same time aware of the risk so don't want to bother for what could be no gains. What say ye?
I've done it, and seen huge differences. My 3770K now idles at around 30*C on the hottest core, but more impressively, won't go about about 50*C under full load and the rad fans turned right down. NB: Mine is in a full W/C loop. I reckon you'd see a noticeable drop in load temps, though. I also don't see why the Xeon's TIM would be different to the Ivy i-series, but then who knows... :O
That may just be the convincer then. Pre-de/re-lidding, did you have much of a spread of temps across the cores? I don't see any reason it would be any different either, but then it seems nobody knows, hence the issue :/ I suppose if it all goes wrong I can console myself with a tidy upgrade to a 1290v2!
I did - yes - about 5*C across the four cores, which used to irritate me considerably... Why would they make the chip in an entirely different fashion? It seems odd that they would, but then it is Intel. Still, as you say, if you break it, you can have a shiny new chip!
Like Sam, my 3770K is de-lidded and under an Enermax Liqtech 120X with the fans on their lowest (silent) setting I never see higher than 57°C under load. Three of the cores actually never go past 54 but there's always one I'm pro de-lidding now.
Another de-lidded 3770k here. However, I bought mine already de-lidded on the MP, because I'm lazy and CBA to lap the IHS and this one already was. A smidge of Coollaboratory liquid pro betwixt the die and IHS works wonders. It bimbles along in the high 20s and sits in the high 40s/low50s at full load (H100i).
Seems I shall be de-lidding then! I've got AS5 in stock, which I intend to use, unless there's any particularly compelling reasons not to? I didn't come across any, other than perhaps a couple degrees. Tis indeed.
To be honest, in terms of full disclosure, my 3770K is NAKED! I put MX-4 between it and the waterblock, and I can basically run it in a loop with a single thickness 240mm rad and fans running at 5V, keeping it under 50*C at load. It's totally worth it (and not at all scary seating a chip in the socket WITHOUT THE RETAINING STUFF!*) *That's a lie. My rear iris puckered considerably when I seated the block.
I shall definitely be re-lidding - the heatsink is extremely wide, and there's no way to be sure you've got it on level other than just going for it and hoping you don't hear a crunch. My first de-lidding experience was with an Opteron 165 and having decided to run it naked, I chipped a corner seating the block... s939 and it's bloody two hole mount! So, know the feeling! It was actually fine, but I did re-lid it.
I bought an angle grinder on impulse that I've not yet had an opportunity to use, seems as good a tool as any?
Sooooo... that was short lived. The IHS is impossibly closely attached to the package, to the point where I'm unable to get my thinnest blade (one I've previously used for de-lidding) under any part of it. There's no black rubbery glue visible anywhere around the edges like I would have expected, just IHS and PCB fused together, nothing in between. As a bonus though, I noted the heatsink wasn't making good contact (at all) when I popped it off, so there might be value in this little exercise yet.
Stop pissing about and buy a band saw. Seriously though when I get round to doing mine I think I will buy a small vice, seems the easiest and safest way I have seen.
In all seriousness though, what's second best when the razor blade doesn't work? I've been perusing, and have come across a de-lidded 1230v2, which suggests it should be fine, but the blade I'm using is barely able to hold itself straight under it's own weight it's so thin, so I can't go any thinner.