What's new
Carbonite

South Africa's Top Online Tech Classifieds!
Register a free account today to become a member! (No Under 18's)
Home of C.U.D.

TheOverclocker Presents - Intel Core i9 10980XE

Gouhan

Forum Addict
TheOverClocker.com
Rating - 100%
96   0   0
Joined
Apr 22, 2010
Messages
3,311
Reaction score
977
Points
5,265
Core-i9-10980XE-main-scaled.jpg
Intel is back again, this time with their most powerful HEDT CPU yet, the Core i9 10980XE. 18 cores running at up to 4.8GHz. Intel is leveraging it’s improved 14nm manufacturing process and extracting as much as possible from it to good measure. The 10980XE may not be the best CPU at any single task, but it does well enough in them all that it makes a great case for itself. With a variety of configuration options, this is easily the most flexible CPU Intel has to offer. Find out more in TheOverclocker Presents – Intel Core i9 10980XE

Read More
 
This is one big monster...but I assume this will mostly benefit people doing editing and video rendering etc and not much for the gaming aspect.

I mean look at the 9900K vs the 9900KS,
Some people claim that the KS can be as high as 6-8% faster than the K series. I have seen some guys doing tests and found that yes and no depending on the use for the CPU. So basically my understanding to the difference between the K and the KS is that the KS has a factory stable all core overclock as to the K series. I believe this might not be as of a huge difference in games because most developers still make use of 4 cores (some stretching to 6 cores) at the most from what I have seen. Yes the KS has a base of 4.0GHz, not that I really will feel the difference in the base clock, because the boost takes over once you start to give the CPU the greater demand. Also, from some tests I saw in the gaming benchmarks, overclocking both the K and the KS to 5.0GHz there was about a 1% gain in the favor of the KS, so why then buy the KS? Also when it comes to Blender benchmark the Ryzen 7 3700X is still ahead of the Core i9-9900KS. Besides the only other key difference between the Core i9-9900K and the Core i9-9900KS is (understandably) the thermal design power (TDP) rating of the chip, up to 127 watts in the Core i9-9900KS, from 95 watts in the Core i9-9900K.

My question is, is it really worth paying more for such little difference in gaming?
 
Well believe it or not, game performance isn't just limited to 4 cores or 4 threads anymore.
You can easily see this when you monitor core clock load during gameplay. Check to see the frequency that each core is running at. In red Dead Redemption 2 for instance, over 10 threads/cores were loaded (exact number is in the video towards the end you can see in RDR2)

As for TDP
Well yes they had to increase that for the base 4GHz clock and 5GHz Turbo frequency. However the operating voltage on the KS is lower than the average voltage on 9900K. I can't imagine there's a 9900K that can't do 5GHz all core, but the worst one will need 1.35v for this, whereas the worst 9900KS needs on 1.25v for this. This directly affects TDP. Same frequency and performance but higher power draw on 9900K in this example.

What Intel is doing now is tuning process as that's all they can actually do until next gen is ready. Scaling frequency has a predictable performance gain and they attempt to offset that with core count ( 10900K has 10 cores and a bpost of 5.3Ghz same 127W TDP as 9900KS). So there are clear optimizations for the node. However any more optimization to bring the power down, will have to come from key architectural changes and that isn't feasible at present.

Finally for performance.
In all the games I tested, 8 core 8 thread 9900KS at 5.2GHz beat 5GHz all core 9900KS and everything else. Believe it or not games still scale in performance past 5GHz.
1581613823270.png
 
Well believe it or not, game performance isn't just limited to 4 cores or 4 threads anymore.
You can easily see this when you monitor core clock load during gameplay. Check to see the frequency that each core is running at. In red Dead Redemption 2 for instance, over 10 threads/cores were loaded (exact number is in the video towards the end you can see in RDR2)

As for TDP
Well yes they had to increase that for the base 4GHz clock and 5GHz Turbo frequency. However the operating voltage on the KS is lower than the average voltage on 9900K. I can't imagine there's a 9900K that can't do 5GHz all core, but the worst one will need 1.35v for this, whereas the worst 9900KS needs on 1.25v for this. This directly affects TDP. Same frequency and performance but higher power draw on 9900K in this example.

What Intel is doing now is tuning process as that's all they can actually do until next gen is ready. Scaling frequency has a predictable performance gain and they attempt to offset that with core count ( 10900K has 10 cores and a bpost of 5.3Ghz same 127W TDP as 9900KS). So there are clear optimizations for the node. However any more optimization to bring the power down, will have to come from key architectural changes and that isn't feasible at present.

Finally for performance.
In all the games I tested, 8 core 8 thread 9900KS at 5.2GHz beat 5GHz all core 9900KS and everything else. Believe it or not games still scale in performance past 5GHz.
View attachment 59638

I agree with you 100% and yes I also found that in all games if you run a core monitor all cores are showing to be fully utilized as per the overclock settings, because the cpu is set to make use of all cores, BUT how does this relate to when you take away 2 or 4 cores and compare FPS and performance in the same game itself? The big debate that has been going on for long now is the fact that single higher core speed is delivering better performance than lower core speed with more cores. This can relate to your finding that's 5.2GHz is still giving better performance as my understanding is that developers still do not focus much on the multiple process availability from more cores so much but still on the speed hungry single core performance enhancement. I have seen guys overclocking the 9900K to 5.3GHz and getting amazing performance out of that because of the single core speed hungry game build. I unfortunately can't go any higher than 4.9GHz on all cores due to my aircooler and ambient temperature I have in my room at home. If maybe you could do an interesting test to see the findings what you will get when you only run for example 4 or 6 cores compared to 8 and compare the difference in the FPS in the same game. I get what you are saying and I initially also argued that for long until I was sadly proven wrong by a group of guys that did testing on amount of cores needed vs utilized comparison in game compared to relying on running monitor software and evaluate performance while bench marking games. Hence this is sometimes quite a debate but I would love to know what the actual difference is when testing core for core on the actual cores itself in a game and when does the FPS actually don't go any higher regardless of core count or core speed.

What is your thought on this?
 
This is probably not THE best example and excuse the guy's ridiculous choice of music LINK, but this is a good comparison in showing the CPU utilization from the 9900K and the 8700K at 5.0GHZ. I guess this also gives some proof towards the argument that having more cores is not the better option for gaming being less utilized than having less cores and give it the higher clock speed the games wants from the cores. Hence I would like to see a test done on for ex. the 9900K with 8 cores at the highest stable overclock and then maybe the i7 8700K at 6 cores in the same game with the highest stable overclock and what effect this will have on the utilization, performance, FPS etc. Also it could help in temperature drop that could potentially give room for a slightly higher core clock and this giving the single core for core power the game might want?
I had a look last night, when I do Prime95/cinebench or Aida64, it really utilize the crap out of my CPU and thus I can only do 4.9GHz overclock, BUT going into games I don't even have more than 40% average of CPU package utilization which means I could potentially run a 5.0GHz overclock on my Aircooler in games, but the cpu is becoming a Spur sparerib grill when I do 5.0GHz on the benchmarks hence I prefer to stay @ 4.9GHz as per the stress results. I don not regret buying the 9900K at all, but many people argue I should've gone with either the i7 8700 (which is claimed to still be the best overclocker today) or even the i5 9600 because I could get a slight higher core speed due to the i9 Hyper threading relating to higher temps and the more cores needing more power resulting in more heat. So the whole argument even to day still is do people really need more cores or just higher core speed with less cores/threads. Maybe developers is now in favor of making use of more than 6 cores I don't know and I am open to any opinions, suggestions, thoughts of anyone on their findings regarding the cores, core speeds and performance in games.
 
Great thread....glad i discovered it guys XD
So im also a avid overclocker, and have had the 9900k, settled on running it without hyperthreadig to reduce temps, but at a higher clock. Even with reduced threads the increased(0.1ghz) core clock haad slightly(like 2%) gain in cpu limited games, like Far Cry 5-uses more single core speed. But in sottr wich is in my opinion greatly optimized for multi threads, no gain or loss was seen.
Have since moved to 10980xe, to try it out.
On both these chips im running a custom closed EK-loop, with decent enough airflow, maybe a bit little in the rad department, but decent nontheless.
9900k i saw pull 180-190w on a cinebench run, maxing out at, or just below 80°, with a oc applied to it of abiut 5ghz on 16 threads,
And 150-160w with hyperthreading disabled, minus 5° on load in that case.
Power draw measured through msi afterburner(pacage), and corsair link(load minus idle)
On 10980xe i didnt get that big an overclock, i had temp issues on 4.6ghz.
4.5ghz all thread is where i settled on, reaching 82° on a cinebench run, and pulling around 350w
Since then i have discovered the 'per individual core' overclocking mode, where you can set each individial core's max boost and voltage.
After 3 straight days of overclocking the 18 individual cpu's, clocks i settled on:
4 cores on 4.8ghz
6 cores on 4.7ghz
5 cores on 4.6ghz
3 cores on 4.5ghz

I found that some cores need quite a bit of power to be stable, no matter the multiplier, but at that point they can be overclocked as well
Some cores just are hot, no matter the voltage(probably uneven solder or something),
Most cores just refuse to overclock further than a certain freq, hard limits to the silicon. And yet, on a all-core oc i can do 5ghz on 10 cores(core #0-#9, wich included some bum cores)
Thus, i will be doing my next round of overclocking without hyperthreading for 18 cores only,
And after that,
Disable some bum cores and do an all-core oc as well.
Look forward to sharing the experience if you guys want me to
Cheers
 
Thanks for sharing this @Jaco52222 this is definitely some awesome feedback and info to know.
I have to admit that is a crap load of power to run at 350W and you will never be able to really run that on air cooler so this is good to know that if a person would like to own the new intel cpu's and overclock them they will need most likely custom loop water cooling or a dam good AIO.

Needless to say you did what I still wanted to attempt with the 9900K and that is to see what the difference will be in turning the hyper threading off. Now that you have done the tests to find that there is a bare minimum gain in performance I won't even bother.

The fact that you mention about the different amount of cores in groups assigned to different voltages I can vouch for as Jayztwocents also did this I believe and he also had to assign like this in groups to get the best stable benchmark, I just can't recall but I think this was for the latest AMD threadripper cpu when he switched from intel to amd for his video guy a few months ago.

Glad to know somebody did this comparison between the 9900 and the 10980 series cpu. I believe after your results we can agree that the 9900 is still THE cpu to use for gaming due to the higher single core speed it can achieve and the 10 gen cpu is much better at multi processing...you can advise me if I am wrong to argu this way. Also for the money, I don't have them deep pockets for Intel's 10 gen so my 9900K will have to do for the next 5 years or so.

Brilliant feedback and looking forward to your next findings to share with us. :cool:
 

Users who are viewing this thread

Back
Top Bottom