• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Finally Direct Die my Garbage CPU

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Rainwater

Member
Joined
Nov 10, 2016
Location
Portland
I finally went Direct Die my 9900k. I ran Aida 64 Extreme last night for an hour 5.1 Ghz 1.405v with LLC Extreme and VRM LLC Extreme. Vcore stayed between 1.416v - 1.428v max temp of 101c with CPU FPU Cache and Memory stress. I lowered LLC and went to 5.2 Ghz and haven't passed 21 minutes without a crash. So I put LLC back to Extreme and it's at 1.41v in bios CPUID reading 1.416v - 1.428v. So far I'm at 16 min with a max temp of 102c. This chip is garbage it wants **** tons of juice. I feel like I need to remount the block. Well it just crashed again so I'm going back to 5.1 Ghz and lowereing Vcore to 1.395v see how stable I can get it at 5.1 Ghz

View attachment 213502
View attachment 213503
View attachment 213504
View attachment 213505
View attachment 213506
View attachment 213507
View attachment 213508
 
Ya IDK why it's taken me so long to do it. Well ya I do - LIFE -
But it seems pretty stable at 5.1 Ghz 1.39v in bios. I just noticed how cheap these 9900k are now I may just post this thing for $200 - $225 and grab another just so I can have some fun with it LOL.
Seriously anyone want a burgandy 9900k so I can try and find a golden one?
 
Last edited:
I have a question........I'm four hours into Aida 64 with CPU, FPU, CACHE and MEMORY being stressed. Max package temp 99c. One core has hit 100c. I'm at 5.1 Ghz 1.41v in bios with cpuid reading 1.416 - 1.428v. With speeds and vcore that high are my temps decent being as it is Direct Die with a 280mm aio? I know they are totally safe but should they be much lower?
 
Seems about right at that high voltage and clocks. It starts to throttle at 100C. I would keep things under 1.4V with stress test temps a few C away from the throttling point. ;)
 
........Ya....I figure worst case scenario I fry the SOB and get a new one :attn:
 
Last edited:
I did the same thing on my 7900x , I have a better block and far more rad's and it still runs hotter than I expected.
lapped the block etc its still quick to temp flux with tons of near ambient water flowing over it.
I always feel like I need to remount but it doesn't change anything lol.

The mod works it drops the temps but with these chip's on ambient its tough to not have high temps under lots of voltage sometimes less is more for the clocks unless its chilled or cold.

This is my 2nd 7900x and same results on 2 different mobo's.

Their may be better blocks out there but its a lot of $$ to just toss at it considering the one i have is solid copper with micro fins and a pressure plate they don't really improve much I've had a few different brands/types.
 
Why are you running such long artificial stress tests? Do you have an app that actually replicates this kind of load?

I ran out of cooler on my 9900k at 5.0ghz, 1.250V, 0AVX. It would run fine at 5.1Ghz @1.3V, but quickly overheated. 5.2ghz 1.35V -1AVX boots to windows fine and runs if you don't hammer it, but overheats so quickly it's not reasonable.

I run the H100i Pro cooler. I can't fit a bigger cooler in the case.

My artificial stress tests were always short, once it stabilizes, there's no point in doing more.

Also stress with something that replicates what I'm doing, I do CAD, Rendering, FEA and Video. The video rendering is probably the most stressful thing I have. It flatlines all cpu and the temps are hovering around 75°C. FEA is similar, though the vid render is uninterrupted. This is well under some of the artificial stress tests.
 
once it stabilizes, there's no point in doing more.
Quite the contrary. Many of these stress tests rotate through different types of loads (some don't). Any that do, you'd want to run through at least a few iterations of the loop. But even if it loops after say 30 mins, very few would call that 'stable'.

4 hours is about the median, I'd imagine. Much less than that and even more users fall off the 'yeah that's considered stable' bandwagon. There's others who I think are overkill running like multiple stress tests for several hours at a time... but if it's mission (or $) critical... makes sense. But some of these are just gamers... lol.

The short of it, is YMMV. What works for you, may not work for others. The best idea is to find a stress test app that jives with what you do. For example, I won't use Linpack/Linx, etc as, for MY uses, that type of testing is EXTREME overkill (versus overkill...what a stress test SHOULD do). I prefer AIDA64 with just CPU/FPU/Cache checked off. When I want to test memory I add the memory check box. This works well FOR ME and MY uses.
 
Dam 5.1Ghz is respectable. I'm sure I can get there with my 7700K cherry. Ran it since day one @ 5.0Ghz and 1.32v with lid intact right up till I upgraded to Ryzen. Figured once I got into the 5Ghz club, never tried any further. I'm currently playing with it @ 4.9Ghz and 1.4v.
 
I just got an 9900K, and with aida64 extreme I see a lot of avx in this which is rather fine with me as I render 3d in blender which is avx, I set my avx offset to -3, giving me 4.7.
The clock I'm sporting now is [email protected], 80c peak tempafter 20 mins, only stress cpu checked.
I run A 5x120, thick, rad so my block is always my limiting factor.



9900k aida.PNG
 
Last edited:
Back