Overclock.net banner
1 - 13 of 13 Posts

·
Registered
Joined
·
442 Posts
Discussion Starter · #1 ·
i just changed from the delid + relid
CLP between die and ihs and MX-4 between ihs and X63 cold plate.
results all the way to the right.

now im direct die mount.
CLP between die and X63 cold plate.
results are to the left.

settings the same in bios and in windows.
i might have actually increased my pump speeds (both kraken x63 (cpu and gpu) from 75% to 100%) on the direct die benchmarks.
i did lower it to 75% and it made cpu temps higher.

so what gives? why are my results about 6C higher then when i had ihs?
i dont think i had any tabs to file down on the corners of the socket.
felt pretty good and i used a 6" steel scale (rule) on top of the die.
i swept the area as much as i could (diagnol for sure) and i didnt see any light between it and the die. (no gap i could see).

here are the pics of results (before and after) and also the amount of CLP i used on die and x63 cold plate.
Font Screenshot Software Electronic device Technology
Font Electronic engineering Electronic component Hardware programmer Electronic device
Watch Analog watch Bicycle part Clock Rim

the CLP on the X63 cold plate dont look as good as it did with my eyes. maybe thats the problem?
 

·
Registered
Joined
·
388 Posts
I had to take a razor on mine and cut the corners of the socket down. They sit higher and seem to keep the cooler from reaching the cpu as good as it should. I've also had one of the 3 mounting screws of the direct die mount stick above the mount and ruin results as well. Its a pretty shawty thing direct die mounting it but the results are worth it. Just keep looking for something that's too high.
 

·
Registered
Joined
·
823 Posts
I had to take a razor on mine and cut the corners of the socket down. They sit higher and seem to keep the cooler from reaching the cpu as good as it should. I've also had one of the 3 mounting screws of the direct die mount stick above the mount and ruin results as well. Its a pretty shawty thing direct die mounting it but the results are worth it. Just keep looking for something that's too high.
This.
 
  • Rep+
Reactions: tdbone1

·
Registered
Joined
·
442 Posts
Discussion Starter · #4 ·
you know what i think i remember that bottom (3rd screw sticking up). i wonder if its hitting on the bottom of the plate.
crap.
the plastic socket the cpu sits in is pretty flush around the corners but it seems like the die mount might sit just a hair high.
i dont want to take the mb out to file stuff. crap. i will check with that screw first.
thanks for the good advice.
 

·
Registered
Joined
·
388 Posts
What I did was took a razor blade (red line) and went towards the green circle to shave that corner off, with it exactly as you have in the pic (direct die mount and everything, literally exactly as it is in the pic). Run that razor all over across the top of the direct die mount. The cpu die is bareeeeeeeeeeeely above the level of the direct die mount. Anythingggggg will mess with it.

Also remember the caps can be in the way. My cpu block will only work one way because it hits the caps any other way, and even then is a hair away from hitting them.

Circuit component Hardware programmer Font Audio equipment Electronic engineering


Here is a pic of my 8600k/z390 aorus. You can see how the edges were all razored off.

Product Circuit component Rectangle Line Gas


Heres my current one. You can see it barely misses the caps.

Building Automotive lighting City Neon Tower
 
  • Rep+
Reactions: tdbone1

·
Registered
Joined
·
442 Posts
Discussion Starter · #6 ·
just got done "shaving" some of that plastic off and came back here and seen your post.
did exactly as you said before you said it but thanks for the reassurance :)
btw nice water block. those hoses are huge and yea i can see how yours only fits one way.
my hoses are closest to the ram. when i have all 4 sticks in those elbows push on my 1st ram stick closest to the cpu.
it can run like that but i cant move the hoses up or down it pretty much has to stay and dont move them. i removed two of the stick so i dont have to worry about it anymore and also im going to give them to my son for his system that im slowly building :)
anyhow even after shaving some plastic off (mostly on the top two corners as it was not the same amount for all four corners) i wonder if the previous owner tried this.... there was not alot to shave off but there was some plastic that was raised a little bit. could barely catch it with my fingernail but the razer caught it right away.
still have about 7C difference then with heat spreader as pictured above.
the "after" picture above is pretty much identical to what i have now.
the "3rd" screw in the die bracket is raised but dont touch the cold plate.
Automotive lighting Blue Light Purple Automotive tire
 

·
Registered
Joined
·
388 Posts
I really don't see the temp going up without something still being wrong. It can't just sit with light or no pressure on the die it still has to smash it somewhat. If the cooler mount studs are too long or however it is designed, keeps it too high then that could be an issue. All of this stuff is designed for IHS stock heights so it really is luck of the draw on making it all work for direct die. I wish I remembered what I was getting with my 7700k when I direct die'd it, though it was with a crappy air cooler. But yeah, check those studs for mounting height etc. It really is a complicated process sometimes lol. A lot of coolers have mounts that set predetermined heights and that just doesn't work when you're trying to make it sit 1/8" lower.
 
  • Rep+
Reactions: tdbone1

·
Registered
Joined
·
442 Posts
Discussion Starter · #8 ·
im actually using the back plate and screws from a kraken m22 as the screws and backplate like you are saying were the wrong length,
the m22 stuff appears to work because the pump bracket (where the 4 thumbscrews tighten) flex down the 4 ears pretty good.
i pretty much tighten them down all the way until they cant go no more and then i start backing them out while watch my temps at same time.
i back them all out evenly about 1/2 turn at a time and it does seem like it likes it tighter then looser also like you said.
im not sure whats going on but i should get better temps without the ihs correct?
 

·
Registered
Joined
·
388 Posts
Should be way better. Especially if the ihs comparison was with factory thermal paste between the die and ihs.
 

·
Registered
Joined
·
442 Posts
Discussion Starter · #10 ·
comparison is with CLP between die and ihs and MX-4 between ihs and kraken x63 cold plate.
from the research i have read the temp should have went down like 3C to 4C more by doing direct die and leaving the ihs out.
 

·
Registered
Joined
·
442 Posts
Discussion Starter · #11 ·
ok i have some new info.
i made a video of the actual die mount bracket thingy.
i did wind up switching back to using the ihs with CLP on die side and on x63 cold plate side.
im 2C cooler then i was originally with the delid relid with clp and then used mx-4 between ihs and x63 cold plate.
i cant figure out why the direct die isnt cooler then with the ihs as i triple checked the cpu socket, the die, the direct die bracket (which had poor quality control imho (see in video) and the x63 cold plate and screws....
i used a razer blade on top of the die and had light behind it so i could see if there was anything higher then the die.
it appeared to me the die was the highest object.
maybe the kraken cold plate tightening was pushing the cpu pcb down into the socket away from the direct die bracket?
i did several adjustments on the 4 thumb screws and it just seems the looser the screws are the higher the temps go.
idk.
here is my current p95 with avx2 enabled and at 4.8ghz with avx offset 0 in bios.
Computer Font Screenshot Rectangle Electronic device
 

·
Registered
Joined
·
3,653 Posts
You have way too much LM from the looks of it. Paint it on very thin. It also looks like you are using CLU which is lumpy.
 

·
Registered
Joined
·
442 Posts
Discussion Starter · #13 ·
You have way too much LM from the looks of it. Paint it on very thin. It also looks like you are using CLU which is lumpy.
ive done it quite a few times and some times i do it thin and sometimes a lil heavy but just trying to figure out where it runs the best.
its CLP
i just cant understand why it runs cooler with the ihs.
and i have CLP on in two places.
 
1 - 13 of 13 Posts
Top