Did my first ever lapping of an IHS today.... using TG AM5 holder mechanism... I ran out of sand paper ( TG supplied + sand paper in the house for other sanding jobs) so just did what I could with what I had for the time being. Hoping the 7600X will get better thermals under load, although undervolting is a thing too but that 3. something mm IHS is a too thick for my liking.
Also scored an Asrock X670E Steel Legend mobo as well & wanted to use my old AM4 Cryorig R1 Ultimate cooler but removing the SAM mechanism from AM5 socket & trying to put the TG alternative bracket on the back did not go well.... the screws that hold the SAM on are too small in length to properly hold the TG bracket on.... not to worry, still have the Deep Cool AK620 to fall back on for the time being. Will apply TG Kryonaut TIM. Had experience with this TIM before & I'm happy with it.
I'll be interested to see how much the lapping helps.
Make sure you cooling method is able to tighten the block down with adequate pressure. Some of these will only clamp down so far and if you take a mm off. May nkt tighten as much.
I delidded one of my R5 7600 and temps are maybe better, but it changes nothing. The CPU boosts the same, and max OC is the same too. In the best case, I can run fans slower, but this CPU is already ~80W, so it also doesn't change much.
The main problem with delidding is that the best is to use a CPU without IHS, and then you may have problems finding a cooler that will fit. Most mounting kits are for perfect stock CPU height including IHS, so most coolers will have a 1mm+ gap between the cold plate and CPU dies. Noctua offers additional kits for delidded AM5 CPUs, but except that, you may only install some AIOs or custom liquid coolers, or try to modify the mounting.
Many (or maybe most) tower coolers are designed for perfect fit. It means the plastic spacers won't let you go any lower than the CPU IHS height. It's required to modify spacers or use many washers, and it wouldn't look good. Later screws can be too long, which also depends on the mounting kit (backplate mainly, as not all have go-through holes).
On some motherboards, if you lower coolers by 2mm, it will cause the cooler to lack space around the socket. It sounds funny as it's only 2mm, but somehow it's happening, and wide coolers with many heat pipes touch VRM heatsinks. Some very small, non-tower coolers can't be installed at all. Like you can't install Noctua NH-L9x series coolers on delidded AMD CPUs using the AMD motherboard bracket/backplate. With the backplate, there will be 1mm+ space between the cold plate and the CPU dies.
I thought it would be easy to do that until I delidded the CPU, and then I couldn't find any cooler at home that would fit. I only had an Alphacool CPU block that had perfect mounting without any modifications.
Lapping seems to cause no problems with coolers, but the effects are not as good as expected. In both cases, you lose the warranty, so why not remove IHS at all? There are also offset mounting kits that help to drop temps by 2-3°C more. I didn't like the Thermal Grizzly one, as it was a pain for installation and looked bad. Noctua kits for their coolers are simply bar replacements with moved mounting holes by a couple of mm, so it's as simple as a standard mounting kit ... but it requires a Noctua cooler which isn't cheap.
I'll be interested to see how much the lapping helps.
Make sure you cooling method is able to tighten the block down with adequate pressure. Some of these will only clamp down so far and if you take a mm off. May nkt tighten as much.
Thanks, yes I could see that happening with my AK620 cooler & its mounting system limitations. I only took off about a mm of the IHS. I'll give it a few days of testing & come back with more OC results.
Many (or maybe most) tower coolers are designed for perfect fit. It means the plastic spacers won't let you go any lower than the CPU IHS height. It's required to modify spacers or use many washers, and it wouldn't look good. Later screws can be too long, which also depends on the mounting kit (backplate mainly, as not all have go-through holes).
On some motherboards, if you lower coolers by 2mm, it will cause the cooler to lack space around the socket. It sounds funny as it's only 2mm, but somehow it's happening, and wide coolers with many heat pipes touch VRM heatsinks. Some very small, non-tower coolers can't be installed at all. Like you can't install Noctua NH-L9x series coolers on delidded AMD CPUs using the AMD motherboard bracket/backplate. With the backplate, there will be 1mm+ space between the cold plate and the CPU dies.
I thought it would be easy to do that until I delidded the CPU, and then I couldn't find any cooler at home that would fit. I only had an Alphacool CPU block that had perfect mounting without any modifications.
Lapping seems to cause no problems with coolers, but the effects are not as good as expected. In both cases, you lose the warranty, so why not remove IHS at all? There are also offset mounting kits that help to drop temps by 2-3°C more. I didn't like the Thermal Grizzly one, as it was a pain for installation and looked bad. Noctua kits for their coolers are simply bar replacements with moved mounting holes by a couple of mm, so it's as simple as a standard mounting kit ... but it requires a Noctua cooler which isn't cheap.
Yes, lapping can open up a so called can of worms with mounting of 3rd party tower coolers. It was probably a good thing that I ran out of sandpaper when I did it & only took off about a mm of the IHS. Although, I can say at this point with very similar ambient temps in the room with the system before the lapping, the idle temp is about 1C less. It's not much, but its very early days still because I have swapped motherboards as well so there maybe other contributing factors at play here too.
This is an area I'll have to spend more time researching....
I can only compare if I duplicate my customised PBO settings to this Asrock X670E board like I did with the MSI B650 board running p95 small ffts for 2 hrs - that was the last OC test I did on that board before any lapping of the cpu. So transferring those bios settings to this Asrock board, then I'm in a position to do a like for like with the only variation being how effective the VRMs are on each board as the varying factor with the thermals (ambient temps being the same) It's Impossible to do a like for like in the strictest sense without doing them on the same board.
Too much bother to test the lapping effectiveness again without installing everything back on that MSI board. I'm sure you've seen mobo reviews where they test a board's VRM capability & it's heat output - you'll know what I mean.
Darn. Would love to have seen how much of a difference that made on these chips. 1C at idle is quite literally within error.
Well aware what it takes to replicate . That said......the vrm thing between these two boards (many boards) is negligible, so it is just a matter of putting the same settings in.
Seems to me that lapping was more beneficial in earlier iterations of CPUs; tolerances were adequate but shaving a few microns off could easily make a small (1-2°C) difference. Along with a better interface (goo, gel, pads). But the combo with water made the biggest difference. I'd make a "ham sammich" bet that with as many air and water cooling options available with modern manufacturing techniques and tolerances that out-of-the-box is close to good enough.
Depending on the mounting mechanism, washers between backplate and holes could provide a benefit. At least that is how the AQ Kryos Cuplex works.
I actually added and old Coolit universal mounting bracket on the back of mine to make a fan mount for the back of the socket, which added at least 2mm maybe more to the through bolts that the spring loaded threaded thumbscrews go into. This is AM3+ though.
If you'd like a pic LMK.
Darn. Would love to have seen how much of a difference that made on these chips. 1C at idle is quite literally within error.
Well aware what it takes to replicate . That said......the vrm thing between these two boards (many boards) is negligible, so it is just a matter of putting the same settings in.
At the time I thought, if I'm upgrading the board - then it will be a good time to embark on this lapping adventure cause' you know... all the parts are out on the bench so...
Is there a difference in the board's chipset temps between X670E & a B650 board? In any case, I'm betting it is very minuscule to say the least. However I can honestly say the whole of system idle temps using HWiNFO between before & after lapping have dropped 3 - 4 C, so I think that's pretty good. But under load with p95 small FFTs, the boost is a little lower & temps about 5C higher - keeping in mind, this is comparing with different motherboards but identical Cooler & TIM.
Couple screenshots of the 2 different systems. Tried to mimic identical settings in each bios before running the tests. Ambient temps were within about 1 C of each run - different days.
First up the MSI B650 board...
Next, the current motherboard, obviously after the lapping ...
So, at this stage, I will remount the cpu but have to do a little more lapping to be 100% sure of ultra flat surface for top of IHS & bottom of cooler. As I was alluding to before in my above post on this topic, I ran out the grit grades of sandpaper I needed on that day when I was lapping but have more on order so waiting for delivery to completely be satisfied with the job. This current mounting of the cooler was only temporary anyway just see the effect, if any & if it is negative or positive in its outcome.
Seems to me that lapping was more beneficial in earlier iterations of CPUs; tolerances were adequate but shaving a few microns off could easily make a small (1-2°C) difference. Along with a better interface (goo, gel, pads). But the combo with water made the biggest difference. I'd make a "ham sammich" bet that with as many air and water cooling options available with modern manufacturing techniques and tolerances that out-of-the-box is close to good enough.
The problem is the thickness of Zen 4 IHS - its a record for thickness compared to other generations of cpu from either intel or amd for desktop usage. The smaller node manufacturing process has delivered impressively low power consumption figures at stock & boots compared to Intel atm, but the density of transistors is now facilitating extra heat production - laws of physics.
Depending on the mounting mechanism, washers between backplate and holes could provide a benefit. At least that is how the AQ Kryos Cuplex works.
I actually added and old Coolit universal mounting bracket on the back of mine to make a fan mount for the back of the socket, which added at least 2mm maybe more to the through bolts that the spring loaded threaded thumbscrews go into. This is AM3+ though.
If you'd like a pic LMK.
Yes, I have been thinking of a way around this because I have several after market cooler kits here with an assortment of bolts, nuts, washers... you name it accumulated over the years already. So I'll have to get a bit inventive in how the cooler is lowered in proportion to the IHS top being removed. I'm taking in the ball park of a couple or less of millimetres here.
Likely... X670E has two PROM chips. Not entirely sure how that's relevant here though. I doubt it would make much of a difference temps-wise, even if it were in the loop.
Any chance you can point to a guide or thread that shows the difference between stock and lapped AM5 results? You touched on why today's silicon runs hotter (density and removal through silicon, not increased production)... and while I'm sure it will improve some, I'm really curious what it actually does... I haven't seen a lot of people do this and wondering why. Idle temps are still largely irrelevant. As far as the 5C INCREASE, I'm guessing that has to do with your mount. I don't see temps going up from this adventure, lol.
Did my first ever lapping of an IHS today.... using TG AM5 holder mechanism... I ran out of sand paper ( TG supplied + sand paper in the house for other sanding jobs) so just did what I could with what I had for the time being. Hoping the 7600X will get better thermals under load, although undervolting is a thing too but that 3. something mm IHS is a too thick for my liking.
Also scored an Asrock X670E Steel Legend mobo as well & wanted to use my old AM4 Cryorig R1 Ultimate cooler but removing the SAM mechanism from AM5 socket & trying to put the TG alternative bracket on the back did not go well.... the screws that hold the SAM on are too small in length to properly hold the TG bracket on.... not to worry, still have the Deep Cool AK620 to fall back on for the time being. Will apply TG Kryonaut TIM. Had experience with this TIM before & I'm happy with it.
You're having heat issues on a 7600X? Where is it going? mine just runs into the mhz limits at like 70C under full load testing.
Can push it a bit more in other methods but it's more like a chip limitation I see not thermals.
I have the same Mobo its been good on the newer bios updates I haven't tried the newest one yet I heard its more for Samsung memory compatibility anyways.
Likely... X670E has two PROM chips. Not entirely sure how that's relevant here though. I doubt it would make much of a difference temps-wise, even if it were in the loop.
Any chance you can point to a guide or thread that shows the difference between stock and lapped AM5 results? You touched on why today's silicon runs hotter (density and removal through silicon, not increased production)... and while I'm sure it will improve some, I'm really curious what it actually does... I haven't seen a lot of people do this and wondering why. Idle temps are still largely irrelevant. As far as the 5C INCREASE, I'm guessing that has to do with your mount. I don't see temps going up from this adventure, lol.
Hardly any chance of finding info or threads about diff between stock & lapped AM5 atm. I could be wrong though of course, most enthusiasts seem to go for water cooling & all that involves or even delidding. Been there before with other platforms with H2O but air cooling is for me. So lapping is something I'm obviously not afraid to do.
You're having heat issues on a 7600X? Where is it going? mine just runs into the mhz limits at like 70C under full load testing.
Can push it a bit more in other methods but it's more like a chip limitation I see not thermals.
I have the same Mobo its been good on the newer bios updates I haven't tried the newest one yet I heard its more for Samsung memory compatibility anyways.
Nah, just don't like 95C then it throttles. Of course one can change that in bios to lower but their are consequences. Going by the info in your signature, your using a comprehensive H2O solution so of course you'll get better temps than me sticking to air cooling. The newer bios then should work well with my Samsung IC DIMMs. The PBO offset is listed in my screenshots above but its negative 30. It's boosting to 5.5GHz but throttling down cause' of the mount issue atm.
I do have a big water loop, but it's just overkill also, I have a much smaller loop on my spare PC and same block just a move no remount the new smaller loop is similar in temps.
Good air is just a bit behind water imho its pretty good right now for cheap, the Thermal right Dual coolers are amazing for the price my friend got one for a 5800X and it was surprisingly good so I'm a bit shocked the chip you have is doing poorly with -30 offset too.
is it a mount issue or some kind of defect between CPU runs. I pre ordered this cpu and board from a main distributor in Canada before they were on the market. Wonder if there was a chip to chip or socket to socket issue, we have the same cpu and mobo Afterall.
Just adding to this, if you compare the results of others with you do you think you should maybe RMA if you're having major differences between others, it's hard to cure a bad CPU that's what warrantee is for, now only when I think I have a good chip do I start modifying anything to make it faster.
Thats just my mindset like if the chip is getting much worse performance, I try RMA and if it doesn't change. it's definitely mount.
It Costs AMD a bit of money to test and send that chip out but it's their platform I bought into anyway.
I don't like wasting my efforts I want it to be a good start.
Hardly any chance of finding info or threads about diff between stock & lapped AM5 atm. I could be wrong though of course, most enthusiasts seem to go for water cooling & all that involves or even delidding. Been there before with other platforms with H2O but air cooling is for me. So lapping is something I'm obviously not afraid to do.
Nah, just don't like 95C then it throttles. Of course one can change that in bios to lower but their are consequences. Going by the info in your signature, your using a comprehensive H2O solution so of course you'll get better temps than me sticking to air cooling. The newer bios then should work well with my Samsung IC DIMMs. The PBO offset is listed in my screenshots above but its negative 30. It's boosting to 5.5GHz but throttling down cause' of the mount issue atm.
Jayztwocents did this and had to mod the mount to make it work but the cpu is made to to keep boosting till it hits 95C. I run a 7900X with a 480mm x 60mm Rad with push/pull fans and it still will go to 95C under full load. They made the IHS thicker so it would work with AM4 coolers.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
. That said......the vrm thing between these two boards (many boards) is negligible, so it is just a matter of putting the same settings in.
![[url=https://valid.x86.fr/ca6lrl]](https://www.overclockers.com/forums/[url=https%3A//valid.x86.fr/ca6lrl][img]https%3A//valid.x86.fr/cache/banner/ca6lrl-6.png[/img][/url]){kind=link}