Ryzen 5 1600x vs Intel i7 7700k

Hello to everyone,

Recently my dad got a new AMD Ryzen cpu. The 1600x to be exact. One of my good friends has an i7 7700k.
I wanted to know how the systems compare.

The Specs:
AMD Ryzen 5 1600x (Stock clock at 3.7 GHz)
16Gb G-Skill RAM running @ 2667mhz DDR4
EVGA GTX 980 Classified 4GB

-------

INTEL core i7 7700k ( Overclocked to 4.8 GHz)
16Gb Ram (2667mhz DDR4)
GIGABYTE GTX 970 Gaming G1 4GB
---------------------------------------------------------------

The Game setup:

Basically we have the same conditions:
Both games run on:

1920x1080 60hz Borderless
VSync off
FPS Limiter off
Overall quality: High
Dynamic reflections: Maxed out
Grid small pure
-----------------------------------------------------------
Intel i7 7700k
1 Cement mixer | 108FPS
2 cement mixer | 88FPS
3 cement mixer | 78FPS
4 cement mixer | 72FPS
5 cement mixer | 66FPS
6 cement mixer | 59FPS
7 cement mixer | 52FPS
8 cement mixer | 47FPS
9 cement mixer | 43FPS
10 cement mixer| 39FPS
11 cement mixer| 33FPS
12 cement mixer| 27FPS
------------------------------------------------------------
AMD Ryzen 1600x
1 Cement mixer | 118FPS
2 cement mixer | 97FPS
3 cement mixer | 88FPS
4 cement mixer | 81FPS
5 cement mixer | 73FPS
6 cement mixer | 67FPS
7 cement mixer | 60FPS
8 cement mixer | 54FPS
9 cement mixer | 48FPS
10 cement mixer| 43FPS
11 cement mixer| 38FPS
12 cement mixer| 28FPS
------------------------------------------------------------
AMD Ryzen 1700x @4 GHz
1 Cement mixer | 125FPS
2 cement mixer | 105FPS
3 cement mixer | 94FPS
4 cement mixer | 87FPS
5 cement mixer | 79FPS
6 cement mixer | 71FPS
7 cement mixer | 66FPS
8 cement mixer | 62FPS
9 cement mixer | 58FPS
10 cement mixer| 51FPS
11 cement mixer| 47FPS
12 cement mixer| 42FPS


So what do we see? Basically the AMD is a few FPS ahead of the i7 in the beginning but they come very close at the end .. it took each of them 12 T-series Cement mixers to get under 30FPS..
And you have to remember .. it´s not the best Ryzen .. it´s only a 1600x .. not an 1800x and it is running stock clock.

 

One has GTX980 while other has GTX970, there is no CPU load graphs or GPU load graphs which means we can't really say if difference is any way related to CPU or is it just GPU.

This is difficulty comparing different systems, there are lot of variables and while we can say system 1 and system 2 did perform like they did, there is no way to say if difference had anything to do with CPU from this data.

If you do same test again, but swap GPU's from system to system, then we start to see what is CPU's part in this and what is GPU's part. Alternatively, test setup could be set so that GPU will never max out during the test.

Of course we could try to do some weird math based on how much faster 980 is compared to 970 and try to estimate from that percentage how much difference there should be between the cards and see if AMD system got that much difference, but that would probably not result any better data.

So my point is that test was more about different GPU's than CPU's and while it does show that AMD is really good, especially with it's price tag (we already knew that), with better test setup we could see what is real difference between CPU's.

 

The 1700x even beats out my 14 core xeon in the Banana bench test, so yeah Ryzen does brilliantly with BeamNG because of the high core count and good single core speed. For the price its currently the best CPU for the game.

 

This looks like the graphics cards are the limiting factor since you have the graphics maxed out. Try it on low graphics and it should show the difference between the CPUs. With my i7 7700k I can run 5 Belaz at 60fps.

Edit: I managed 11 cement trucks at 60fps then beamng crashed. I had to lower the graphics as my 1080GTX was struggling with more than a handful.

 

i am waiting on my motherboard for my Ryzen 1700 which i will overclock

 

Sorry to bring this up, but I have Ryzen 5 1600x and I tested this;



With lowest settings 1080p I get ~15fps (12 cement mixer) (if I pause physics I get ~50fps)
With highest settings 1080p I get ~13fps (12 cement mixer) (if I pause physics I get ~26fps)


OP gets twice my fps with same cpu, almost 30 fps. How is that?

My specs:

AMD Ryzen 5 1600x
AMD R9 280x 3gb
8Gt RAM


Edit: Lowest settings; 9 trucks -> ~40fps, 10 trucks -> ~20fps

 

Mostly because the R9 280x is slower than the GTX 980 (https://gpu.userbenchmark.com/Compare/AMD-R9-280X-vs-Nvidia-GTX-980/2192vs2576), then he has double your RAM and maybe he has a faster ram.

 

Even though I went lowest settings? And if I pause physics I get 50fps. Doesn't that tell it's not about GPU? My RAM is F4-2400C16D-16GFT but only one stick installed atm.

Edit: Installed another stick, now having 16Gt Ram

12 cement mixers, lowest settings: ~18 fps (pause: 58 fps)
12 cement mixers, high settings: ~16 fps (pause: 28 fps)

 

You have 2400Mhz ram and what I have read about Ryzen is that it needs fast ram to work well, so you would need something like low latency 3200Mhz or something to get best out of it I guess?

If your GPU load stays under 90% or so you are not GPU limited, use MSI Afterburner to see your GPU load.


Today's version of the game on low graphics I get only 16FPS, it is not really any better on lowest, when paused FPS is bit higher up to 73fps, which could indicate that your GPU is also having it's limits there, AMD driver might also require more CPU power than Nvidia driver, but I have no idea of that.


With 12 cement mixers it seems that physics are loading CPU so much that graphics settings don't affect so much of the actual FPS anymore.

Then again I'm still stuck to I7-6700 until I get new motherboard and your Ryzen should have something like 40% better results with all threads utilized, so maybe it is once again because single core performance and ram speed's effect to that?

It is still bit odd though how there is no practical difference with physics running to my score. Not sure what to think about paused on low, difference can be GPU limit or SC limit, without GPU measurements it is bit hard to say.

Changing RAM speed and doing test might give some indication of how much effect that would have. 2666MHZ CL15 is what CPU-Z shows my memory running at the moment.

What is your banana bench result? 1st number is how well single core does physics, with 12 your score should be lot higher than mine, that 40% higher I think? If there is huge deviation from that to lower, then CPU would be under performing, changing ram speeds and running banana bench might give indication if it would be ram speed related, but still I think 40% would be pretty much to be just ram speed.


You can find banana bench from here:
"%yoursteamdir%\steamapps\common\BeamNG.drive\Bin64\banana-bench.x64.bat"

Also how with 8 cement mixers what kind of fps you are getting then? I'm getting 70fps physics running, gpu load on low graphics is then bit more than what 1050Ti can do, just by 1-2%, maybe there is some tipping point which after your FPS is not able to keep up from some reason?

Shadows and UI does 10 and 5fps drop with 8 cement mixers on low graphics here, so that is also variability that might be different.

 

I have to check that.

1197 MHz for me?

Here is my banana bench result:

60fps with 8 cement mixers and low graphics, physics running.

 

Hmm, here is one Ryzen 5 1600 result, which is lower than your results, but you have 1600X which is bit faster:
https://www.beamng.com/threads/banana-benchmark-results-comparison.8595/page-46#post-974600

To me it looks like that your CPU should be performing as excepted?

As DDR stands for Double Data Rate, 1197Mhz is doubled, so it is 2394Mhz which in practice is 2400Mhz.

Did you have shadows enabled? That might explain lower fps, but still it is bit odd that on low you are getting similar results to my system while your CPU should be performing bit better than mine (single core performance difference might explain some, but hardly all), even more odd is how OP results are so much higher.

On high details with dynamic reflections maxed out and shadows set to display all, I get 16fps, physics paused I get 45fps, single core performance is limiting that and it is highly dependent on if browser with lot of tabs is open etc.


It really does appear as if your GPU would be limiting factor, it is bit faster than 1050Ti, but I think it is pretty close to limit even at low details.

You can press ctrl-f twice to see amount of polygons on scene, here it is almost 7.3 million with shadows on and low details:


Also wait for GPU, if that is higher than 1 or so, then you are definitely GPU limited, but that works best when at full screen, in windowed modes it is 0.4-0.5 for me, with vsync limiting fps it can go over 4, so vsync off and fullscreen gives you best estimate.

1050Ti can do something like 6-6.4 million polygons at 100% load, depends of course amount of shader work etc. Your GPU maybe 7.1-7.5 million polygons, so it is quite there at the limit with low details, maybe, it really depends from many variables, but I think upgrading GPU is going to give you best improvements in performance.

Italy map on high details in main city can be around 7 million polygons with single car. Cement mixers are quite hard to run for CPU, but also for GPU it seems and as Vechiles don't really use LODs that I know of, it gets quite heavy quite fast.

Also you can open performance graphs in game, which can be bit confusing to interpret, but lot of yellow in upper graph is that CPU is waiting for GPU, as far as I know. This here is single core limited situation:


That is even task manager is not showing it as single core being maxed out and CPU only at 20%, single core (thread) on CPU is what is limiting FPS, increase single core speed and FPS is higher (that high peak was game running on foreground, other part of graph was game running on background):


My experience with 8086K is that in this game it's single core performance is limiting FPS more than gtx1080, also it has 12 threads, but those are not much use because single core speed is limiting first, that is case with really lot of games actually with gtx1080, in 99.6% of cases I would say 1070 does exactly same as 1080 when playing in 1080p.

First Project Cars game is perhaps only exception, but that is ridiculously GPU heavy mess that can drop frames even with 2080Ti if cranking every setting to ultra and it still is very hard on CPU too.

So while your system can benefit from faster GPU, unless you plan to go on 4K or play FPS games at over 100fps, there might not be so much benefit of going faster than gtx1070 performance level.