It’s that time… I’ve been faithfully using my 2600K for the past 5+ years. And now, with the launch of Coffee Lake CPUs I decided it was time for an Upgrade, Mostly due to CPU bottlenecks I have in DCS in VR . The Internet is full of Benchmarks and reviews. However, No one does flight sims Benchmarking. So I hope I’ll be able to provide some light on this niche games of ours. And help you my fellow simmers do make a more educated decision.
System Specs
The new rig is a direct upgrade. All that changed is the CPU, the Motherboard and Memory.
Common Hardware
Graphic Cards |
(System has Dual Graphics card for the multiple screens cockpit requires. Only the 1080 is used for graphics in game. second card only does 2d stuff and video encoding.) |
Storage |
|
Displays |
(System has 5 screens connected on all tests. Only difference is the virtual location of those screens.) |
Old Rig
CPU | Intel I7-2600K@4.4Ghz (OCed) |
RAM | 24GB or DDR3 @ 1600Mhz (2*4GB + 2*8GB) |
Motherboard | Gigabyte P67A-UD7-B3 |
New Rig
CPU | Intel I7-8700K |
RAM | 32GB of DDR4 rated at 3200Mhz (4*8GB) |
Motherboard | Asus Maximus X Hero |
Methodology
Prior to the Hardware upgrade I’ve done several baseline tests.
Post upgrade (and updates for drivers and such), the same tests are done on the 8700K. However there are several tests done in 3 distinct hardware configurations.
- CPU in stock speeds, Memory at base speed 2133Mhz
- CPU in stock speeds, Memory’s XMP profile enabled, running at their rated 3200Mhz.
- CPU is running on extended boost (Asus MCE – allowing 4.7Ghz boost on all cores, up from stock 4.3). Memory @3200Mhz.
Synthetic Benchmarks
First set of tests are CPU synthetic Benchmarks using Cinebench R15 and CPU-Z both single and multi-core tests.
Simulator gameplay Benchmarks
I’ve used FRAPS to record avarage FPS per second for 15 minutes. as well as Min/Max/Avg. these Numbers will be crunched down to graphs. However, the Raw data was trimmed a bit during the data processing so that the graphs will line up properly. which included trimming the leading second or two from some of the data streams, and then the ends were trimmed so that they will end together.
BMS
BMS is a bit tricky to test, as it does not have any way to reliably do a benchmark. I’ve opted to do the following. I’ve started a new Campaign in KTO with BMS 4.33.3. I’ve ran the time for a few hours on Day one and selected a squadron from Osan which is pretty close to the FLOT. I’ve started the FPS monitoring before loading on a runway start. I’ve set Viewpoint to be stationary (no TIR).
After takeoff, I’ll wait on 350kts till No.2 declares airborn, then start a turn and flight along the route with TGP in AG mode and tracking the waypoints climbing @ 400 kts up to FL250.
Starting from around 11 minutes in the test, there is uncontrolled variation due to the FLOT and engagement with enemy fighters (that happens differently everytime due to BMS dynamic nature). and the FPS data does not track perfectly.
DCS
DCS allow us to use Replay of tracks and consistently do the exact same flight.
I’ve recorded a track and then ran it twice in each of the test scenario, once in full screen with settings on HIGH preset. the second run was with a VR headset with a custom Graphics settings and PD set to 1.5.
I’ve also measured FPS during the recording session, frame rate is lower then in the playback, but not in any meaningful way (less the 1% difference), this low discrepancy allows the DCS replays to be used, which allows DCS data to be very consistent.
Please note, test sequence does not include any weapons exploding, and no ground units (and no Airports). So a more intense mission will result in lower frameRates then recorded in this benchmarks.
All DCS tests were done with DCS 1.5.7.10872 Update 4 from Oct. 20th, 2017