Overclock.net banner

1 - 20 of 20 Posts

·
Registered
Joined
·
733 Posts
Discussion Starter · #1 · (Edited)
Thought to share since this is overclock.net. Haven't ran it longer than just over 5min, don't know if I will cause not a realistic use case for me. Pain in the rear to lap a die, my first time, took .3mm off the top. Not a pro here by any means, just an enthusiast having fun.

5.4 Direct Die to Lapped Die .3mm off.jpg Cinebench at 5.4.jpg
 
  • Rep+
Reactions: mouacyk

·
Newb to Overclock.net
Joined
·
4,659 Posts
Looks cold. What cooler did you use and is the liquid chilled?
 

·
Registered
Joined
·
733 Posts
Discussion Starter · #3 ·
Looks cold. What cooler did you use and is the liquid chilled?
Liquid not chilled just ambient which is at 21C. 2 360mm radiators with gentle typhoons, ek res with d5 pump and Optimus Signature V2 Waterblock.

It's direct die to a lapped die, took .3mm of silicon off the top.

Lapping the cpu die made a world of difference.
 

·
Registered
Joined
·
2,553 Posts
Nice results. I’d be more impressed if it could run P95 small FFT AVX2 at 5.4 or Cinebench R20/R23 at 5.5.
 

·
Registered
Joined
·
733 Posts
Discussion Starter · #5 ·
Nice results. I’d be more impressed if it could run P95 small FFT AVX2 at 5.4 or Cinebench R20/R23 at 5.5.
5.5?
I wish! lol
May try it next.
Might need a chiller for that though
 

·
Just hangin' out
Joined
·
1,384 Posts
That CBR15 run at 5400MHz?
 

·
Registered
Joined
·
733 Posts
Discussion Starter · #7 · (Edited)
That CBR15 run at 5400MHz?
Yes, only did it for a bud, no idea how it compares or if that's what's expected of it.

Keep in mind hyperthreading is off.
 

·
Just hangin' out
Joined
·
1,384 Posts
Yes, only did it for a bud, no idea how it compares or if that's what's expected of it.

Keep in mind hyperthreading is off.
Thanks.

If it is not too much trouble, would you mind running that again with HWiNFO, like so:

2487764
 

·
Registered
Joined
·
733 Posts
Discussion Starter · #9 ·
Thanks.

If it is not too much trouble, would you mind running that again with HWiNFO, like so:

View attachment 2487764
How'd you get 1820? For some reason I got 1612 yesterday and only getting 1542 today. Is there something I'm missing, I just started Using C15 yesterday for the first time.
2487778
 

·
Registered
Joined
·
561 Posts
Even running HWiNFO at the same time will lower your score. Should be a bit better if you keep it minimized during the run. Windows is constantly doing whatever tf it wants in the background, which effects score. You can improve that by stripping Windows services and removing startup apps / tasks.

You can also set Cinebench to Real-time priority in task manager. That should give you pretty consistent scores because it puts all other tasks on hold. When you hit Run, the system will lock up and you won't see the render until it's completely finished.
 

·
Just hangin' out
Joined
·
1,384 Posts
How'd you get 1820? For some reason I got 1612 yesterday and only getting 1542 today. Is there something I'm missing, I just started Using C15 yesterday for the first time.
View attachment 2487778
As T.Sharp said, turn off background apps/services.

Never tried the real-time priority thing.
 
  • Rep+
Reactions: stryker7314

·
Registered
Joined
·
561 Posts
As T.Sharp said, turn off background apps/services.

Never tried the real-time priority thing.
If your OS is clean and stripped of bloat, the real-time score should be close to the normal score, but real-time will always be higher. Good way to test if your install is well optimized.
 
  • Rep+
Reactions: stryker7314

·
Just hangin' out
Joined
·
1,384 Posts
1820 = applications and services disabled
1842 = applications and services running, CBR15 real-time priority
1850 = applications and services disabled, CBR15 real-time priority.

2487857
 
  • Rep+
Reactions: T.Sharp and Nikado7

·
Crayon Evangelist
Joined
·
7,678 Posts
I've been trying to stabilize 5.4 on my 10900kf but AVX absolutely wrecks my hopes and dreams, and it absolutely is due to temperatures. Did the lap make a noticable difference? Currently running naked on a z490 RoG hero using thermal grizzly liquid metal, and being cooled by a MORA-3 in push pull.
I have it pretty solid on non AVX P95 maintaining under 80C but the instant I put AVX back into Prime and start a run, it shoots to 100C, throttles, them crashes.

What was your process for lapping the die? What was temp your drop? Did you also notice any voltage requirement drop?
 

·
Registered
Joined
·
446 Posts
Had someone Tried to lap a die and then glue IHS back to the pcb? Normally blocks are designed to cool down IHS as it has bigger surface then the silicon die, I wonder if there is actually gain in temps over direct die cooling.
 

·
Registered
Joined
·
235 Posts
Der8auer released just an video about 11900K direct-die water-cooling.

Temps was higher with direct-die than a Cpu + IHS.
Becouse core is so much smaller than heat spreader.
 

·
Registered
Joined
·
446 Posts
Der8auer released just an video about 11900K direct-die water-cooling.

Temps was higher with direct-die than a Cpu + IHS.
Becouse core is so much smaller than heat spreader.
yea exactly, optimal would be to slap golden or copper ( without nickel plating ) ihs on the die, glue the edges and call it a day.
 

·
Just hangin' out
Joined
·
1,384 Posts
Der8auer released just an video about 11900K direct-die water-cooling.

Temps was higher with direct-die than a Cpu + IHS.
Becouse core is so much smaller than heat spreader.
Interesting.

I would have believed multiple layers of thermal interfaces would be less effective at transferring heat than a single interface, i.e. waterblock to die.
 

·
Registered
Joined
·
561 Posts
Pretty sure those results only apply to the 11th gen chip. IIRC, the 9th and 10th gen performed best with direct die. Maybe he used a different block 🤷‍♂️
 
1 - 20 of 20 Posts
Top