DCS with i7-8700k

I'm probably one of the first here with an i7-8700k, so I thought I'd share performance and some settings. First off the specs:

CPU: i7-8700k

CPU Cooler: Arctic Liquid Freezer 120

GPU: EVGA GTX 1080 TI SC2 ICX

RAM: 16GB DDR4 @ 3200MHz

Mobo: ASRock Z370 Extreme4

Drive: Samsung 960 Evo 500GB

OS: Win10 Home

VR: Oculus Rift

i7-8700k overclock settings:

Delidded: No

Core Clock: 5.0GHz

AVX Offset: 0

Cache Clock: 4.2GHz

VCore: 1.36v fixed

LLC: 1 (lowest V-Droop on ASRock boards, opposite of most others)

MB voltages: default from XMP profile, mostly 1.15v or 1.25v

RAM: Straight from XMP profile, specific ram kit is this one: https://www.newegg.com/Product/Product.aspx?Item=N82E16820232476

Temperatures: Don't even think about using AIDA64 cache or FPU stress test without delidding, CPU test is fine though. Temps while running games never exceed low 70's for individual cores and averages are closer to 40-50C during gameplay. Idle temps around 30C with spikes to 50C for individual cores.

Passmark scores for CPU and memory:

Now for DCS results, settings for 1.5 and 2.x are the same except HDR gets replaced with the newer option (can't remember name), VR uses pixel density 1.4 in both versions. Test tracks attached at the end.

Settings:

Results:

Tracks here (requires F-5E):

DCS 1.5: https://drive.google.com/file/d/1RibThEokGC6Rnl0R01h3gEboQLlR__er/view

NTTR: https://drive.google.com/file/d/1PlpSZfNwLn5wJKGVWExUZD-8gTG8IRKo/view

Final thoughts: It cost an arm and a leg, but results are pretty good in 2.x. DCS 1.5 is still problematic in VR. The test mission is completely empty other than the player's F-5E. It's just a low flight over Batumi, NTTR is similar in that it's a low flight over Vegas in an otherwise empty mission. Multiplayer performance on 104th can generally be considered playable but not ideal. FPS is 90 over less populated areas but it runs at 45 pretty much anytime the coastal towns are in view. Given nearly all 104th missions are on the coast, it could be a lot better. The limiting factor is the CPU's single thread performance in these areas. GPU usage around Batumi is only around 30%! NTTR is well balanced though with both the GPU and CPU topping out at about the same time.

LoL, looks like it is overkill for NTTR and 1440p, it never drops below 150fps, despite HDR/DefSh ! damn !! Does that include the Vegas Strip ??

VR on the other hand could use some love. I'm glad I stayed with 1440p for now.

Yeah, the delid I guess is mandatory for 6-core Intels now, I got away on the 7700k without delid and 5-5.1GHz but for the 8700K I got the delid kit right with it. If you crank 50% more cores to the same or higher frequency, you will pump some serious juice through that little die.

I wouldnt be amazed if it is in excess of 130W at full load at 5.0G. They reduced the base clock from 4.2 to 3.7 to keep the TDP at 95W. Now we come and turn it back up. I wont even try p95avx, I stick with v26.6 and IBT for stability testing and Aida64 for quick testing..and I will use AVX Offset if the CPU is 5+G. That is just too much heat.

Have you tried any higher than 5 yet ?

You can look at the test track if you have the F-5, but I went for worst case scenario in both 1.5 and 2.x for an empty mission. Hence Batumi at about 100m heading north along the coast and flying from Nellis to the strip to McCarran Intl along the main highway just above the street lights/signs (between the buildings in the strip).

As for delidding and additional overclocking, I don't plan to do either. AIDA64 cache and FPU tests are massively artificial and far removed from anything I actually do. Temps in the regular CPU test are fine though, so I see no reason to delid. Additionally, gameplay temps are basically in the low 50's, which is surprisingly cool though they do spike to 70-ish for a second at a time (1 reporting cycle I think) according to HW monitor. Not sure what those are about but it's worth mentioning the same thing happens at stock settings too.

My spec

Windows 7 64 bit

8700k @ 4.9Ghz (on all cores)

NZXT Kraken x62 AIO cooler

Z370 Gigabyte Aouris Gaming ultra

32G Corsair 3866Mhz RAM

1080TI Zotac AMP

1TB M.2 samsung polaris

1TB Samsung Evo 850

2TB WB black stoarge

1TB Samsung storage

850W Corsair PSU

Corsair 750D Airflow case

CPU temp does not go above about 55 deg at the moment, did some rendering stress testing earlier to see max temps

If you guys want a quick stress test, try PUGB. That game will stress out your rig! LOL

How does it compare to your Ryzen system?

I wonder how much the CPU matters in 1.5.... I suspect it plays a far secondary role to the GPU.

I have the Ryzen 5 with a Geforce 1080 (not ti) and the GPU is what gets pinned AFAICT.

This totally bears out my experience between 1.5 and 2.1 in VR. Since I smoked the VR crack I don't even want to fly 1.5 anymore.

I'm all rotorcraft, so that probably pushes it harder than the F5 since I'm filling the viewframe with objects 100% of the time.

2.5 can't get here fast enough.

Here are the settings I use for 2.1.

Day:

https://i.imgur.com/8Vleo1E.jpg

Night (why render shadows at night?):

https://i.imgur.com/A7UauyY.jpg

Hi,

can you please perform the same test with a graph of Afterburner? This will show the CPU/GPU in DCS a lot more. Better to perform the track with the SU-25 which they all have.

Set in the bios Enhanced Turbo, which will ensure the turbo frequency for all cores the same.

I'm glad that you are sharing tests i7 8700k thank you very much. Pet

Afterburner is great, it can also log the graphs etc.. CTRL+F12 will let you enable/disable Overlay with the newest version 4.4.0.

+1 for the SU-25T as this is reproducible by everyone. :joystick:

+10 for your effort, I know how much time this all takes :thumbup:

I didn't record CPU/GPU usage but I did have it up in the background (task manager + resource monitor). For 1.5, it was all CPU limited. At no point did the GPU exceed 50% and it mostly hovered around 30-40%. The i7-8700k was pinned at 100% nearly the whole time on thread 11 even when doing 45 fps. The Batumi area is absolute CPU torture which is why I use it as my benchmark area in 1.5. There is no place in DCS 1.5 or 2.x that pushes the CPU harder that I'm aware of.

In 2.0, the CPU and GPU were almost perfectly matched. I never observed an instance where the GPU held back the CPU or the CPU held back the GPU. Sometimes one would be at 80% while the other was at 100%, but either way, they were matched given the 45 or 90 fps nature of the Rift. I did not look too much at monitor results as since 148 fps minimums in 2.x, so who cares? DCS 1.5 in monitor is CPU limited, just like VR. GPU usage does exceed 50% though, so... yay?

In regards to Ryzen's performance, I did do the exact same benches before cannibalizing it for the new build. Here are the results (1600x at 4.0GHz, RAM at 2800 MHz):

Edit: minor addendum. I was tinkering with some more modest overclocking on the 8700k for more of a daily use setup, using only 5.0 GHz for DCS and BoS where it is really needed. I found that 4.7 GHz all cores all the time can be done at 1.28v with LLC 3. This is actually below the default voltage of 1.31 when under load.

Please to create a reference track :) exclusively on the SU 25T and all available current maps thank you.

looks like Ryzen has nothing to fear with NTTR.

those Ryzen prices are free falling. 699$ for a 1950X, that's 43,75$/core !

If they had offered me this deal here in germany I would have bought that CPU instead of the 8700k. Our price here is still 899€.

Yeah, when I bought the 1600x I had no plans for VR but then Oculus cut the price in half. What I found is DCS was unplayable in 1.5 while NTTR ran at 45 fps all the time regardless of detail level. The 8700k still doesn't muster 90 fps consistently over the two areas tested but does hold 45 well. Other areas do get 90 fps most of the time. Of course if DCS used even 2 CPU threads for graphics rendering then it wouldn't be an issue for even something ancient like a phenom chip. But honestly, I don't have any expectation of ED doing that.

me neither, not in the forseeable future I am afraid.

That was one heavy factor for the 8700K over any 4G-Ryzen.

My rig is a DCS rig, not a Multimedia editing machine. I want SPEEEEEED and smooth fps gameplay.

Maybe in 1-2 years with my next one I will go AMD. I am actually a little sad that I dont have this AMD Ryzen hype at home.

I am tinkering with a miniITX server for my Living room to feed the TV and host all files 24/7 for every other PC. Havent had that in years. Maybe this will be my Ryzen project. This Ryzen hype really got me LoL

edit:

tbh, I am looking for an excuse hehe to build one. And I would now rather spend those 450€ for that than for VR. I can wait for VR till next year or so.

oh...Dropped to $350 now BitMaster. Last night, I was flying the A10C Normandy CAS mission. And I saw the SAMs streaking up at me and I was launching flares/chaff as I was jinking while looking at the SAM. It really is something else!

If you guys want a quick stress test, try PUGB. That game will stress out your rig! LOL

I'm playing that fine on my 2700k, never mind the rig though, it stresses my head!

I am tinkering with a miniITX server for my Living room to feed the TV and host all files 24/7 for every other PC.

I still prefer my Synology boxes for that purpose. I have two with wired connections to my router, and a third on the shelf as a disaster backup.

I'm playing that fine on my 2700k, never mind the rig though, it stresses my head!

I tell you what. I found out that my 1250W PSU was going bad. Firing up PUBG would kill the PC quite often. Once I replaced it, it was all good. But I remember all my cores going to 100% etc. Takes a lot of horespower to run it with high resolution, that's for sure.

I still prefer my Synology boxes for that purpose. I have two with wired connections to my router, and a third on the shelf as a disaster backup.

I love Linux too much to not do this myself, based on some nice SuSE and SAMBA.