i7 8700K - Delidding Results

[mousegolf]

Hello Intel Fans,

Before and after purchasing components for my newest Custom Build since 2012 I did a lot of research on this site and several others, as well as YouTube tutorials. I hope anyone using any Intel CPU has learned about using the latest gadgets for delidding their CPU to add better thermal paste on the IHS (integrated heat spreader). Quite the fun experience these days with very little risk. The results of using liquid steel thermal paste dropped my overclocked 8700K running at 5.1GHz by 10 to 15 degrees Celsius under load. The Noctua D15 is a HUGE air cooler but it snuggly fit into my older tower with little effort.

Here are my results from benchmarking my system. [Average idle temp is 36c. Average gaming/load temp 50c to 67c.]

https://valid.x86.fr/1n7nb8
http://www.userbenchmark.com/UserRun/8714338
https://www.3dmark.com/fs/15578614

This is my new build:

ASUS ROG STRIX Z370-E GAMING
INTEL: i7 8700k
EVGA GeForce GTX 1070 Ti FTW ULTRA SILENT GAMING:
Noctua D15 Air Cooler
Corsair Vengeance LPX 16GB DDR4 DRAM 3000MHz

--------
ATX MID Tower: NZXT 410 Phantom
(purchased in 2014)

 

[trents]

Your temp drops after delidding are pretty typical.

 

[mousegolf]

Your temp drops after delidding are pretty typical.

Yes they are typical. We can only hope that those of us who use the current methods of our times will inspire others to do the same.

 

[maxfly]

what kind of tim did you use? ive never heard of liquid steel? good job!

 

[Woomack]

Yes they are typical. We can only hope that those of us who use the current methods of our times will inspire others to do the same.

Don't mean me wrong but looking at your temps and usage, delliding was pretty pointless in your case. I mean you see lower temps but it changes nothing as CPU was far from throttling point anyway.
I'm also not sure why anyone needs to "inspire" others to delid their processors and make them lose warranty. Of course, it helps when overclocking is limited by temps but most users won't ever need it.
I'm not against delidding as a way to lower CPU temps but recently I see on various forums that people try to convince others that delidding is required on new Intel chips, while it's not.

Regardless what I said above, I think you have a nice results and I hope you will have a lot of fun with this PC.

 

[EarthDog]

Don't mean me wrong but looking at your temps and usage, delliding was pretty pointless in your case. I mean you see lower temps but it changes nothing as CPU was far from throttling point anyway.
I'm also not sure why anyone needs to "inspire" others to delid their processors and make them lose warranty. Of course, it helps when overclocking is limited by temps but most users won't ever need it.
I'm not against delidding as a way to lower CPU temps but recently I see on various forums that people try to convince others that delidding is required on new Intel chips, while it's not.

Regardless what I said above, I think you have a nice results and I hope you will have a lot of fun with this PC.

QFT!

 

[Brando]

that's a good oc on a nice cpu. congrats and enjoy. i hope to do the same but i'm not sure i have the guts to lose my warranty at least until i win the mega millions lotto. good to know that air is enough to get a nice overclock with the right thermal interface.

 

[mousegolf]

Thanks for the input guys. There is a big hike in performance when the 8700K hits 5.2GHz. But. That much extra power use is just silly on a gaming computer IMHO. I have since toned it down to 5.04GHz which is about the same overall response as 4.9GHz. The delidding at default settings isn't necessary at all. At 4.5GHz an H60 Hydro would keep it cool or a $30.00ish AIR Tower cooler.

 

[EarthDog]

There is a big hike in performance when the 8700K hits 5.2GHz.

oh? What do you mean by this?

 

[mousegolf]

oh? What do you mean by this?

I'd say it shaves off about 300ms in response time when opening programs. It's already fast enough out of the box but you know how we enthusiasts are about pushing limits. Even though I can get a stable 5.2GHz it does spike into the 80c range when running a CPU stress. I know that's safe but I'm more comfortable seeing it spike into the 70c range at 5.04GHz.

 

[EarthDog]

And that '300ms jump you are talking about, that's more than other 100 MHz jumps.......or am I reading into that statement too deeply? I guess I understand you now to just say its a bit snappier compared to...........stock?

Yeah, 80C is fine. We tell people to keep it under 90C for stress testing and a long happy life. So, you are more than set temp wise. I am curious as to what was changed to have it suddenly stable, unless it was simply lowering the clock speed?

 

[mousegolf]

Lowered the clock speed to 5.1GHz made it stable. But still am keeping it at 5.04GHz.

 

[Woomack]

Personally I see no difference in games between ~4.2GHz and 5GHz+. All that matters to me is to have a least 6 cores at 4GHz+ as I can use it and I see the difference between 4 and 6+ cores. Recently I've switched 8600K to TR 1920X (long story and waste of money but still wanted it) and I see no difference. Before that I was playing on 7700K, 7900X and some others. Pretty much no difference.
If you say that couple of ms make the difference then you simply try to lie to yourself and explain why you made such choice ... and that's fine, my choices are even worse, like buying that TR 1920X or 7900X before that ... well, list is long but I constantly sell something to get new toys.

 

[mousegolf]

Personally I see no difference in games between ~4.2GHz and 5GHz+. All that matters to me is to have a least 6 cores at 4GHz+ as I can use it and I see the difference between 4 and 6+ cores. Recently I've switched 8600K to TR 1920X (long story and waste of money but still wanted it) and I see no difference. Before that I was playing on 7700K, 7900X and some others. Pretty much no difference.
If you say that couple of ms make the difference then you simply try to lie to yourself and explain why you made such choice ... and that's fine, my choices are even worse, like buying that TR 1920X or 7900X before that ... well, list is long but I constantly sell something to get new toys.

I see a difference in game load times at 5+ GHz. The speed increase is far more appreciated when I make System Image Backups a couple times a day. Or copy various GB/s of files from the Crive to my Interna Storage Drive.

Plus video editing. But overclocking is just another hobby (as you know) and those of us who enjoy doing it are always curious as to how far we can push a CPU and Memory until the system crashes. Just all part of the fun of owning a computer. Kind of like the Hot Rod guys in the 1950's.

 

[EarthDog]

Games load from the ssd... they can only go as fast as the drive will load it for the most part. Same with transferring from os drive to internal storage. The cpu gets tickled in those cases. Im surprised you are seeing improvements in activities that are drive limited.

Regardless, glad you seem to be enjoying your pc.

 

[mousegolf]

@ed,

You're close to understanding the process. Everything we interact with has to be transferred to RAM from the SSD/HDD so we we have an interface on the screen. The drives are storing that information until we call them with the mouse or a touch.

Regardless. We both enjoy our hobbies/work and that's what's most important.

 

[trents]

Loading times are affected by the power of the CPU because it does part of the work involved in the total process. Used to be years ago when each new generation of processor gave a quantum leap in performance you could really tell a significant difference in program loading times. Now all CPUs are relatively fast and so the part of the work load done by the CPU/memory in merely loading a program is done so quickly it's a pretty minor part of the total loading time. The vast majority of the time needed to load programs thus falls heavily on the shoulders of the storage device and it's data retrieving efficiency. This means that reving up the CPU has a very minor impact on app loading times since the storage device data retrieval efficiency is by far the weak link in the total process.

 

[EarthDog]

Well said (much better than what I tried to get across!).

I understand that data is passed through RAM, but, as was stated, its a minor impact in most cases. If a drive can only write at 300 MB/s it doesn't matter the CPU and ram can pump it over at 300 GB/s (random values, note). It takes a slow CPU these days to notably effect transfers, even with PCIe NVMe drives.

 

[Woomack]

If you check how CPU speed affects storage performance then CPU still helps but you can count it in single %. I mean when you use AHCI/RAID based on Intel integrated controller then CPU performance affects its speed (since there is no other dedicated processor). In other words when SSD can make 40MB/s random 4K read with 4GHz CPU (just an example) then 5GHz CPU will make it run at maybe 42MB/s. In typical daily work, random operations count much more than sequential, the same in games. It's also the reason why there is barely any difference between SATA and NVMe SSD in daily work/games. Random low queue operations on SATA are at about 30-100MB/s. The same on NVMe is between 40-120MB/s. The best NVMe drives will have that at 60-150MB/s ... high Optane drives maybe 200MB/s but their price/GB is way too high. Still it's speed of loading new, required data to RAM where it stays for longer so once it's loaded then should be used by the CPU without longer delays.

 

[mousegolf]

what kind of tim did you use? ive never heard of liquid steel? good job!

This is what I used.

http://www.thermal-grizzly.com/en/products/26-conductonaut-en