Aurora R8, i9-9900K performance issues

Thanks guys ( GTS81, Dell630i, rz72019, and others) for the help on this.   Here are some new benchmarks that I just ran:

New Idle temp:

https://drive.google.com/open?id=12QLllGRCvvvyy5BH8AkEzDD6F6xL5iNQ

New R20:

https://drive.google.com/open?id=15l4Wzuwsec958u75WvFLAi8wRxnwoLU9

New TimeSpy:

https://drive.google.com/open?id=1ggImGXR57nzc0jcxkA5u0HDTnzuaDICV

Awesome, congrats! I can't believe they told you that it was performing as expected before this!  

---

@RedGreenApple wrote:

 

So as of this morning, took out the pump, cleaned the CPU, re-applied the the paste.  The temps have dropped about 10 degrees.  I'll post some screenshots later.  I am able to get about mid 4000 in R20.  

I can't explain how poor of a job Dell did with the paste (or whatever they used).  Also the contact wasn't 100% correct either, one corner was tilted a bit. 

I don't think the CPU would last more than 3 years if I hadn't done the work. 

---

I am trying to report this to Dell, but it needs to be just technical facts (like above). All these other comments and opinions go to a different department.

Basically, it sounds like the Liquid-Cooler was not installed properly. For one, the heat-sink (and plate) was not screwed-down perfectly-flat against Intel processor. IMO, that is a critical error. But can you elaborate about the thermal-paste?.    

This was troubleshot and repaired by user with PC-hardware and CPU-install experience.

At this point, @RedGreenApple has done the re-paste and we will need to rely on his memory of how the pump was seated, and the state of the paste when he first removed the pump. Wish there were some pics taken to show how the pump wasn't seated properly, and whether the paste looked pristine (bad), semi pristine (still bad), or overflowing outside the heat spreader (also bad because so much was applied that even a badly seated plate could mash it down).

If you can remember how the pump was seated, based on this pic below (sorry, gotta wait for mod to approve), was it one of the screws that weren't tightened correctly? Or was the "teeth" of the mount not sitting inside the grooves in the pump head properly?

That's right GTS81.  I actually took the whole computer to a shop just to make sure I don't screw things up.  I wish I took a few pics during the whole process but at the time, I just wanted to get this done.  

Yes, you are also correct about the issue with the seating of the pump.  One of the screw wasn't screwed in at the right angle, as a result, the contact wasn't 100% there.  As for the paste, I got the Noctua NT-H1.  As for the previous paste, it wasn't applied evenly (not sure what brand Dell used).  

The bottom line is that this should never happen to any pre-assembled PC.. let alone a $3000 Alienware.    

@GTS81 @r72019 @Willk. @RedGreenApple 

I think a Cinebench R20 score of mid-4000 is still a bit low for a stock i9-9900K. I just got an i9-9900 (non-K). After using Throttlestop to lift the power limit and VR current limit, I got 4600-4700 in Cinebench R20. i9-9900 can only run at 4.6 GHz all-core while a stock i9-9900K can do 4.7 GHz all core. This difference should be reflected in the benchmark results. There are online screenshots showing that a stock i9-9900K can score somewhere around 5000. 

What is worth mentioning here is that I now do believe that Aurora may have a minor VRM overheating issue that caused some throttling during multi-score tests. I noticed that when I ran the test right after a cold boot, my i9-9900 can stay at 4.6 GHz for almost the entire duration of the test with a real-time power consumption of 140-145W. It was in those tests when I can get over 4700 in R20. But most of the time the processor cannot stay steady at 4.6 and would get lowered to 4.5 very early on during the test with the power consumption dropped to around 135W. 

That was something we discussed here: https://www.dell.com/community/Alienware-Desktops/Aurora-R8-Newer-Bios-Seems-to-Limit-CPU-Performance/m-p/7625656#M27955