Review: i7-8700k vs R5 1600x in BoS

[BeastyBaiter]

About 6 months ago I grabbed an R5 1600x to replace my ever problematic and lackluster i5-4690 + Biostar Z87 mobo. At the time I had no intention of getting VR and the 1600x seemed like a great option without breaking the bank. However, after moving to VR I found the single thread performance in both BoX and DCS fell a bit short. After much consideration, I ended up grabbing an i7-8700k. This review will attempt to cover all aspects of the two chips in BoS and DCS. So....

 

First off let's deal with the elephant in the room, the price difference:

 

AMD Ryzen 5 1600X:

CPU: $240              https://www.newegg.com/Product/Product.aspx?Item=N82E16819113434

Mobo: $100            https://www.newegg.com/Product/Product.aspx?Item=N82E16813144018

Cooler: $24            https://www.newegg.com/Product/Product.aspx?Item=N82E16835856005

Total price: $364 USD

 

Intel i7-8700K:

CPU: $420            https://www.newegg.com/Product/Product.aspx?Item=N82E16819117827

Mobo: $165          https://www.newegg.com/Product/Product.aspx?Item=N82E16813157789

Cooler: $68          https://www.newegg.com/Product/Product.aspx?Item=N82E16835186152

Total price: $653 USD

 

So yeah, the i7 costs nearly twice as much. This is not a fair comparison but that isn't the point of it. The point is to give our community an idea of what to expect out of either system. Here's the additional list of parts that were shared between the two builds:

 

RAM: 16GB TridentZ RGB 3200          https://www.newegg.com/Product/Product.aspx?Item=N82E16820232476

C Drive: Samsung 960 Evo 500GB     https://www.newegg.com/Product/Product.aspx?Item=N82E16820147594

GPU: EVGA GTX 1080TI SC2 ICX       https://www.newegg.com/Product/Product.aspx?Item=N82E16814487337

 

Now onto performance, here are synthetic benches for the 1600x system, I included CPU and memory as these are the two interesting bits. The Ryzen system is at 4.0 GHz and 2800MHz RAM (memory in general is a problem for Ryzen):

 

 

 

Next is the balapan bench we've been using for a while with Ryzen:

 

And finally a look at DCS at high-ish settings (NTTR is top pic, 1.5 is second):

 

 

As can be seen, the monitor results are generally very good. There really isn't anything to complain about. Unfortunately the VR results are problematic due to the single thread performance. It is my opinion that neither engine was actually intended to run above 30 fps given their age and their heavily single threaded nature. Higher IPC and clock speeds of modern CPU's allow them to reliably hit 60 fps or more, but they weren't built for VR or high refresh rate gaming. Regardless, let's look at the i7-8700k's performance, starting with synthetics:

 

 

Aside from the massive single thread difference thanks to the 25% clock speed increase and minor IPC advantage, we can also see a huge difference in memory latency. This is the exact same memory kit but Ryzen simply does not like it. The Ryzen system used that kit at 2800 MHz while the i7-8700k used it at 3200 MHz, but even at 2133 MHz, the i7 only had 19 latency. I never noticed this as an issue in gaming, but it is certainly worth mentioning.

 

Moving on to BoS, we have the balapan test again:

 

 

This offers a near solid 90 fps in VR, but the balapan test is basically medium detail (high setting with nearly everything switched off or to minimum). Though the track is busy, it does show that not even the most powerful gaming CPU when overclocked can actually run BoS at max detail at 90 fps. Still, it is a fair bit better than the 1600X at 4.0 GHz.

 

Moving to DCS, we have this (same settings and track):

 

 

 

Despite the huge difference in performance in these legacy engines, both are actually good CPU's. The 1600x performs admirably for a monitor even at the highest detail settings. But it does fall just ever so slightly short of the single thread performance to reliably get 90-ish fps in VR. The i7-8700k fairs better in VR, but at a hefty premium and it too falls short since it can't run ultra detail, which is what one would expect from such a ludicrously powerful system.

 

Here are the settings to run the R5 1600X at 4.0 GHz:

Voltage: 1.375v in Ryzen Master

Clock: 4.00 GHz

Yes, it is that simple.

 

i7-8700k 5.0 GHz, 0 AVX offset settings (this chip is NOT delidded):

Voltage: 1.36v in bios

Clock Multiplier: 50

LLC: 1 (max setting on ASRock boards)

assorted MB voltages: 1.25v instead of 0.9v

Disable anything and everything applying offsets or autovoltage to CPU.

Try 500 different combinations of this, pray to the new gods and sacrifice a lamb to the old gods.

 

Links to the two benchmark tracks for DCS (requires F-5E and NTTR):

https://drive.google.com/file/d/1PlpSZfNwLn5wJKGVWExUZD-8gTG8IRKo/view?usp=sharing

https://drive.google.com/file/d/1RibThEokGC6Rnl0R01h3gEboQLlR__er/view?usp=sharing

Edited November 19, 2017 by BeastyBaiter

[SCG_Fenris_Wolf]

I call these heritage engines.. single thread is still so important, and will probably remain so, I give it another 5 years.

 

Good to see you found a very good new CPU you're happy with! Godspeed comrade! S!

[Mitthrawnuruodo]

Very nice comparison. 

 

It will be interesting to see how things change with future CPU products. Hopefully single-thread performance keeps creeping upward. 

[skline00]

Thank you for your effort. Very concise analysis.

[=TBAS=Sshadow14]

Whats the core loading for i7 compared ryzen in IL2

ryzen still only loading up core 0 mostly in IL2 and failing to use the others?
(eg FX8350 in IL2 uses 4 cores pretty evenly under Win7, but my mates Ryzen in IL2 uses mostly 1 core under windows 10)

did AMD fix this issue or a windows 10 fix

 

Edited November 24, 2017 by =TBAS=Sshadow14

[BeastyBaiter]

It isn't an AMD issue, it's a BoS issue. BoS uses 1 core for graphics rendering which is the vast majority of BoS's CPU usage. This is a pic from balapan track on the 8700k at 4.7GHz all cores all the time:

 

[=TBAS=Sshadow14]

hmmm thats interesting it does that...

Heres my IL2 core loadings on the 8350
(Nothing else in background running i even have all aero and advanced windows themes disabled and all auto updates on my system disabled)

FX8350 @ 4.2Ghz
GTX1060 G1
Windows 7 Home premium X64
Watching a replay with big dogfight lots of explosions and effect over that graph time period
Core 0 = 40% | Core 2 =17% | Core 4 = 22% | Core 6 20% |

Edited November 25, 2017 by =TBAS=Sshadow14

[BeastyBaiter]

BoS is bouncing around cores in that pic. How do I know? It shows 18% CPU usage but those 4 cores add up to 99%. What's really going on is 1 core is running at 100% (12.5% total) and the remaining 7.5% is being used by the other constantly changing 3. I had the same thing with my old i5-4690, it never showed more than 25% usage for any individual core but was definitely running 100% on a core all the time. Task manager simply didn't show it, it showed the average over small period of time which was 25% cause it was constantly bouncing between cores. Both the R5 1600X and i7-8700k look like my previous pic though, with fixed core usage in any given instance of the game. So it isn't an AMD or Intel thing but might be an age thing. Maybe Windows is using fixed thread assignments on modern CPU's but using the old jumping system for older ones. I know Ryzen CPU's initially jumped too but Windows was patched about a week after release to stop that since it murdered Ryzen's performance.

[Fishbed64]

BoS is bouncing around cores in that pic. How do I know? It shows 18% CPU usage but those 4 cores add up to 99%. What's really going on is 1 core is running at 100% (12.5% total) and the remaining 7.5% is being used by the other constantly changing 3. I had the same thing with my old i5-4690, it never showed more than 25% usage for any individual core but was definitely running 100% on a core all the time. Task manager simply didn't show it, it showed the average over small period of time which was 25% cause it was constantly bouncing between cores. Both the R5 1600X and i7-8700k look like my previous pic though, with fixed core usage in any given instance of the game. So it isn't an AMD or Intel thing but might be an age thing. Maybe Windows is using fixed thread assignments on modern CPU's but using the old jumping system for older ones. I know Ryzen CPU's initially jumped too but Windows was patched about a week after release to stop that since it murdered Ryzen's performance.

I thing its much less complicated ,FX cpu is much slower(IPC mainly) than Ryzen or any modern intel i5 or i7 so the calculations on FX is running on more cores... 

[SCG_Fenris_Wolf]

Not really. Beastybaiter is right, even in the details.