[Gamegpu.ru]Call of Duty Modern Warfare Remastered GPU Perf

[psolord]

Hey Head1985, my friend can you do me a favor?

Could you please do a 60sec run, of the beginning of the charlie don't surf mission and upload the csvs in mega or any server you like? Start the benchmark when the scene starts right away.

Make sure you have enabled FPS,Frametimes and MinMaxAvg in fraps "FPS" tab.

I will need the three CSV files that will be produced. No overclock needed.

Thanks.

 

[Head1985]

Hey Head1985, my friend can you do me a favor?

Could you please do a 60sec run, of the beginning of the charlie don't surf mission and upload the csvs in mega or any server you like? Start the benchmark when the scene starts right away.

Make sure you have enabled FPS,Frametimes and MinMaxAvg in fraps "FPS" tab.

I will need the three CSV files that will be produced. No overclock needed.

Thanks.

http://www.edisk.eu/download/14233/benchmarks.rar_52.77kb.html/
1440p same setting we used in my test

Looking at results you are pretty bad CPU/platform bottleneck.I have 50%(58%more in comparable fps) more Fps and also i running in higher resolution.I told you 2500k is bottleneck in this game.

 

[psolord]

Thank you very much my friend.

Here is how your results look.

Spoiler: 1070,1440p,6700k,charlie don't surf,60secs

You have the same frametime variance as I do, but your performance is a lot better overall.

I don't have a cpu limit, gpu usage is maxed out, let's not go there again, haha.

The gtx 1070 is a bomb, that's all. If I had a 1070 I would probably not get the same performance as you, but that's a different thing altogether.

 

[Head1985]

you really believe GTX1070 is 100% faster than GTX970?Cmon...
I cant have 60% more performance in 1440p than you in 1080P.Just accept it.I upgraded from 2500k 5Ghz to 6700k 4.5ghz and its another level.You cant compare it.
I have now 3Ghz DDR4 ram and that help alot in games.
If not CPU then your slow DDR3 hold it back.

 

[96Firebird]

GTX 1070 sure is awesome, a great upgrade from a 970. I'm still waiting to finish my home office, but once that is done the 1070 will be ordered and the 970 will go into my 4K HTPC.

 

[Head1985]

Here is my old test 2500k vs 6700k.Its already more than year old when i upgraded, so games are outdated by now.2500k is massive bottleneck for GTX970.Psolord is 100% bottleneck by his CPU.GTX1070 cant be 60% faster in 1440p than GTX970 in 1080p.

Spoiler: Test

Test system:
2500K 4.5Ghz
8GB 1600 8-9-8 1T
GTX970 1500/8000
win7 64

6700k 4.5Ghz
16GB 2937 CL15 1T
GTX970 1500/8000
win7 64
Also use Frametimes and comparable FPS as primary.Fraps is here just for min FPS.


Watchdogs-1920x1080 all max SMAA temporal.Testing method is driving in cityhttps://www.youtube.com/watch?v=48Os2_Ul8YU
2500k fraps log-Avg: 58.942 - Min: 40 - Max: 87
Frametimes and comparable FPS
http://abload.de/img/watchdogs60ss0.jpg
6700k fraps log Avg: 82.677 - Min: 73 - Max: 91
frametimes and comparable FPS
http://abload.de/img/watchdogspcs0r.jpg

Skyrim+Mods
2500k 38.2fps http://abload.de/img/tesv_2015_08_25_13_17tmsse.png
6700k 56.3fps http://abload.de/img/tesv_2015_08_25_19_150asse.png

World of tanks-all max 1920x1080 + FXAA.Kharkov map replay
2500k fraps log Avg: 83.136 - Min: 62 - Max: 101
frametimes and comparable FPS
http://abload.de/img/wotimsfx.jpg
6700k fraps log Avg: 112.393 - Min: 87 - Max: 126
Frametimes and comparable FPS
http://abload.de/img/wotuwsr4.jpg

Crysis3-all max 1920x1080 + SMAA.Test method https://www.youtube.com/watch?v=dSRPOp3ss5Y
2500k fraps log Avg: 50.839 - Min: 38 - Max: 78
Screenshot at same place http://abload.de/img/crysis3_2015_08_19_04wus7z.png
Frametimes and comparable FPS
http://abload.de/img/crysis3ymsyk.jpg
6700k fraps log Avg: 77.112 - Min: 72 - Max: 82
screenshot at same place http://abload.de/img/crysis3_2015_08_25_214vsvw.png
frametimes and comparable FPS
http://abload.de/img/crysis3a6shb.jpg

GTA5-all max 1920x1080 + 2xMSAA.Advanced settings all max.Test method is driving in city https://www.youtube.com/watch?v=G4aEWP2pAEE
2500k fraps log Avg: 59.276 - Min: 48 - Max: 68
frametimes and comparable FPS
http://abload.de/img/gta5vesy4.jpg
6700k fraps log Avg: 65.094 - Min: 51 - Max: 77
Frametimes and comparable FPS
http://abload.de/img/gta5twssx.jpg

Far cry4- 1920x1080 Ultra.Test method is is when you leave pagan ming palace until car crash https://www.youtube.com/watch?v=LJBDENSqLSo
2500k fraps log Avg: 81.496 - Min: 55 - Max: 99
Frametimes and comparable FPS
http://abload.de/img/farcry47vspp.jpg
6700k fraps log Avg: 86.963 - Min: 70 - Max: 106
Frametimes and comparable FPS
http://abload.de/img/far4gvstl.jpg

Witcher3- 1920x1080 all max.Hairworks off,DOF off.Test methodhttps://www.youtube.com/watch?v=Eiw3nqsxLmA
2500k fraps log Avg: 58.565 - Min: 46 - Max: 71
Frametimes and comparable FPS
http://abload.de/img/witcher306sa8.jpg
6700k fraps log Avg: 67.163 - Min: 58 - Max: 77
Frametimes and comparable FPS
http://abload.de/img/witcher3knsip.jpg

Dying light-1920x1080 all max.Test method same like here to 1:50 timehttps://www.youtube.com/watch?v=Tci8A8Sylqk
2500k fraps log Avg: 72.068 - Min: 53 - Max: 87
Frametimes and comparable FPS
http://abload.de/img/dyinglightivsdi.jpg
6700k fraps log Avg: 74.457 - Min: 57 - Max: 89
Frametimes and comparable FPS
http://abload.de/img/dyinglightnns7w.jpg

If GTX970 dont bottleneck there is masive gain in games like crysis3-more than twice faster than 2500k
watchdogs 60%
skyrim 47%
WOT 37%
Witcher3 37%

 

[psolord]

My friend Head1985, thank you for these comparisons, but I believe we have a major misunderstanding here.

In all the 2500K screenshots and videos you posted, the GPU usage is not maxed out. Far from it.

What I have said many times, is that if the GPU usage is 100% on the 2500k, you will NOT get better performance on the 6700K.

The actual reason you are getting better performance on the 6700k, is because your GPU usage was not maxed out on the 2500k. This is what a cpu limit is. The GPU not being able to max out.

Now coming to where all this discussion started, my GPU usage, in the charlie don't surf 60sec benchmark I asked you to run, is maxed out.

See the gpu usage in 1080p and 1440p. Cpu is at 4.3Ghz not even 4.8Ghz and the GTX 970@1500



I know the cpu usage is also very high, but it is not giving any cpu limit. It is just very high. Probably if I install a better graphics card, I will not be able to max it out. But it is OK for the 970.

And here is the benchmark result you see in the picture in text.

2016-11-23 18:07:49 - h1_sp64_ship 1080p
Frames: 6712 - Time: 60000ms - Avg: 111.867 - Min: 86 - Max: 171

2016-11-23 18:09:15 - h1_sp64_ship 1440p
Frames: 4644 - Time: 60000ms - Avg: 77.400 - Min: 62 - Max: 114

The reason you are getting 160fps average on 1440p, is because the 1070 is indeed THAT much faster in that specific scene. That's all there is to it. No cpu limit from my 2500k in this specific test.

Again, I am not saying the 6700k can offer no benefit. It offers no benefit when the GPU usage was maxed on the previous processor. In your examples it was NOT!

 

[Head1985]

Test scenes videos was made with 6700k at 2.5ghz with HT off (except crysis3 i was done that at 4.5Ghz) when i was testing HT.It is only there for test scene location.
BTW nope GTX1070 is not that faster than GTX970.Its average 50-55% faster than GTX970 stock vs stock.
GTX970 1500/8000=stock GTX980
GTX1070 2100/9400=10-15% behind stock GTX1080.
How much is stock GTX1080 faster than GTX980? 65% average?Then how can GTX1070OC be around 100% faster than GTX970 when its even slower than stock GTX1080?

Btw how can you say 2500k its ok for GTX970 when from my testing 6700k is far faster than 2500k(both at 4.5Ghz)

EDIT:
You can test it yourself.Test it with watchdogs in that test scene(with same settings) even if you have 99% GPU usage you will have like 60% less fps than me with 6700k and GTX970.

 

[96Firebird]

Huh, I thought this thread was about COD:MW Remastered...

 

[Dave2150]

Huh, I thought this thread was about COD:MW Remastered...

So did I.......

 

[psolord]

Head, as you can see, people are getting annoyed if you take a Modern Warfare discussion and start shoving Crysis, Witcher and whatever in there.

I am not saying you are lying and I am not disputing your findings. Actually I am glad that we have a fellow benchmarker doing tests of his own. I am not even disputing that the 6700k is a far better processor. Yes in some specific cpu limited games, it will be insanely faster. I never denied that.

This whole discussion started because you thought that in this specific game, Modern Warfare Remastered, and for the locations we studied, the 2500k OCed is a limit, which is not. I showed you the gpu usage graphs. You cannot go above maxed out no matter what cpu you have. As a hobbyist benchmarker it's very important to understand that.

All the games you mentioned, are known to be cpu hogs at many locations. I have them, I have tested them, I have included them in my benchmarks and I know. The reason you got such better performance from the 6700k in these games, is exactly because your gpu was not maxed out with the 2500k. In some games it happens. Your 6700k however, maxed it out and hence the difference.

As for the performance delta of the 1070 and the 970, I believe we have established that it was around 68% in all resolutions, in the crew expendable mission. The charlie don't surf mission starts at sea and is generally lighter, so maybe that's why the 1070 flexes its muscles even longer.

 

[Head1985]

We should make new thread and compare GTX970 vs 1070.But only after you upgrade to 6700k hehe

 

[psolord]

For now I have ordered a GTX 1070 to play with, until Kabylake arrives, hehe.

I know I will eat some cpu limits in my face, but that will only make the new cpu sweeter!

 

[Head1985]

Ok i hope you will test same test scene with your gtx1070.

 

[Head1985]

Hey psolord i know you already testing GTX1070.I tested that charlie dont surf mission 60s in 1080p.From avg and comparable FPS its 2x faster than your gtx 970.Can you test that with your GTX1070?We will see how much cpu bottleneck you have.
http://www.edisk.eu/download/60648/benchmarks.rar_68.42kb.html/

btw i am pretty bad cpu bottleneck in 1080p with 6700k 4.5Ghz.I watched your video and i have 30-80 more fps in that scene so you must be even more cpu bottleneck.

Edit my own video recorded with shadowplay(so 10%performance hit) as you can see in some scenes i have 50% more fps.And we both have GTX1070.We should have same performance.

 

[psolord]

Hello my friend.

Yeap my GTX 1070 arrived and I am hot on the heels testing a quadrillion things.

Now to recap things, these are the two benchmarks of the GTX 1070 vs the 970.

Cod Modern Warfare Remastered 1920x1080 v.High GTX 970 @1.5Ghz Core i5 2500k @4.8GH - 129fps

Cod Modern Warfare Remastered 1920x1080 v.High GTX 1070 @2Ghz Core i5 2500k @4.8GHz - 181fps

As you can see I did get a solid 40% performance increase. If the system was so much cpu limited, it would have landed on its face and I would not be getting a 40% performance increase.

Your system is getting better performance for sure. There's not denying that. Here are two screenshots from our videos at the same places for easier comparison.










Now as you can see, especially for the screenshot where the soldier lands, we can clearly see that we have the same gpu load at 97%, yet the performance on the 6700K is considerably better. In my book that's just wrong. This is not supposed to happen.

A clear cpu limit makes the gpu usage having lots of dips and being very jagged, like so (from my Dying Light Q9550+970 benchmark, which was severely cpu limited).



Now in both my Modern Warfare Remastered runs, the gpu usage was like so.



I know it's not optimal, but if both our systems show 97% gpu usage, that does not explain the 25-30% framerate difference.

I will send Unwinder a question as to why this happens and hopefully he can explain it somehow.

I do fear however that my bottleneck may not be the cpu itself, but the PCI Express, since I only have PCI express 2.0.

Still I will stop worrying about it too much. Talking about cpu limits at 100 and 200 fps is kinda moot.

What really matters for me, going from the 970 to 1070, is that I am now getting around 70-80fps in Forza Horizon 3 and Deus EX : MD, while I was getting around 40-45 before. And that exactly is the 1070's worth. Not getting 300fps in charlie-dont-surfa-metron worries me just a tiny bit!

 

[Head1985]

Dont forget i am using shadowplay so -10% performance hit in that video..Can you post fraps 60s benchmark so we can compare fps?
This is my run
http://www.edisk.eu/download/60648/benchmarks.rar_68.42kb.html/

btw if you want we can compare more games.I also have Infinity warfare and rise of tomb raider and deus ex.

 

[psolord]

I did some tests a couple days ago

2500k 4.8ghz

2016-11-24 22:52:33 - h1_sp64_ship charlie 1080
Frames: 8756 - Time: 60000ms - Avg: 145.933 - Min: 102 - Max: 230

2016-11-24 22:54:04 - h1_sp64_ship charlie 1440
Frames: 6752 - Time: 60000ms - Avg: 112.533 - Min: 88 - Max: 160

2016-11-24 22:55:28 - h1_sp64_ship charlie 4k
Frames: 4294 - Time: 60000ms - Avg: 71.567 - Min: 53 - Max: 92

2016-11-24 22:57:59 - h1_sp64_ship charlie 720p
Frames: 10395 - Time: 60000ms - Avg: 173.250 - Min: 102 - Max: 310

860 4ghz

2016-11-24 20:06:40 - h1_sp64_ship charlie 1080
Frames: 7568 - Time: 60000ms - Avg: 126.133 - Min: 76 - Max: 217

2016-11-24 20:08:10 - h1_sp64_ship charlie 1440
Frames: 6230 - Time: 60000ms - Avg: 103.833 - Min: 77 - Max: 154

2016-11-24 20:09:35 - h1_sp64_ship charlie 4k
Frames: 4054 - Time: 60000ms - Avg: 67.567 - Min: 51 - Max: 89

2016-11-24 20:28:19 - h1_sp64_ship charlie 720p
Frames: 8745 - Time: 60000ms - Avg: 145.750 - Min: 78 - Max: 274

All with 1070 2000/9000.

Huge differences.

can you post 1080p with 6700k @ 3ghz ht on and off?

 

[Head1985]

You will not like this...Skylake at 3Ghz is still faster than sandy bridge at 4.8Ghz...
Ok i retested it with 1070 at 2000/9000
first 6700k at 4.5Ghz

2016-11-29 22:14:47 - h1_sp64_ship 1080p
Frames: 12063 - Time: 60000ms - Avg: 201.050 - Min: 133 - Max: 290

2016-11-29 22:17:34 - h1_sp64_ship 1440p
Frames: 9199 - Time: 60000ms - Avg: 153.317 - Min: 116 - Max: 198

2016-11-29 22:19:17 - h1_sp64_ship 4k
Frames: 5231 - Time: 60000ms - Avg: 87.183 - Min: 63 - Max: 105

Then 6700k at 3Ghz(3Ghz cache and 3Ghz ram) HT ON

2016-11-29 22:26:14 - h1_sp64_ship 1080p
Frames: 10102 - Time: 60000ms - Avg: 168.367 - Min: 96 - Max: 281

2016-11-29 22:29:40 - h1_sp64_ship 1440p
Frames: 8514 - Time: 60000ms - Avg: 141.900 - Min: 98 - Max: 195

2016-11-29 22:35:22 - h1_sp64_ship 4k
Frames: 5194 - Time: 60000ms - Avg: 86.567 - Min: 65 - Max: 105

And 6700k at 3ghz ht off

2016-11-29 22:40:09 - h1_sp64_ship 1080p
Frames: 10382 - Time: 60000ms - Avg: 173.033 - Min: 95 - Max: 279

2016-11-29 22:44:43 - h1_sp64_ship 1440p
Frames: 8500 - Time: 60000ms - Avg: 141.667 - Min: 96 - Max: 196

2016-11-29 22:46:30 - h1_sp64_ship 4k
Frames: 5126 - Time: 60000ms - Avg: 85.433 - Min: 61 - Max: 105

 

[psolord]

That is really good and actually I do like it, because I am gathering more data!

Now time for some more testing, since I need to align all this data.

Can you post CPUz, Cinebench R15 and Geekbench 4 results with the 6700k@3Ghz with no HT? At your convenience, no harry.

And yes guys, this has to do with Modern Warfare remastered!

 

[Head1985]

geekbench4
https://browser.geekbench.com/v4/cpu/1200893
CPUz

CB

 

[psolord]

Hello again my friends. Head1985 thanks for your quick reply and sorry for the late reply of mine.

Here are my respective tests of the 2500k @ 4.8Ghz.

https://browser.geekbench.com/v4/cpu/1258531



To recap, my GTX 1070 and 2500k @ 4.8Ghz gave us

2016-11-24 22:52:33 - h1_sp64_ship charlie 1080
Frames: 8756 - Time: 60000ms - Avg: 145.933 - Min: 102 - Max: 230

while your 1070 and 6700k at 3GHz no HT gave us

2016-11-29 22:40:09 - h1_sp64_ship 1080p
Frames: 10382 - Time: 60000ms - Avg: 173.033 - Min: 95 - Max: 279

So, although in terms of raw cpu power a 2500k@4.8Ghz is faster than a 6700k@3Ghz without HT, the 6700k was still faster in the charlie-dont-surf-ometron.

This alone, makes me believe that the 2500k is not cpu limited with the classic meaning of the expression, but there is something deeper at play.

Without knowing any better, I still believe that it's my PCI Express 2.0 that is the culprit here and that, not in terms of raw bandwidth, but maybe in terms of IOPS or overhead or something.

My chipset being a really old P67 may not be helping either, but the PCI express controller responsible for the graphics cards, is embedded into the cpu since the Sandy Bridge, so this may be irrelevant.

I now have enough data to take the discussion to more knowledgeable people than me, so thank you for your feedback mate.

 

[psolord]

@all

If anyone has an ivy bridge with PCI express 3.0 mobo and a GTX 1070, his input running some charlie-dont-surf-ometron benchmarks would be invaluable. You know, for science!

 

[Head1985]

Fast Ram help alot skylake in games, but dont do anything in programs like cinebench...
https://www.purepc.pl/pamieci_ram/t...pamieci_ram_wybrac_do_intel_skylake?page=0,13
I have 3Ghz cl14-15-15 1T

 

[psolord]

Well I am running DDR3-2133 @ 10-12-12 2T.

I know that Skylake can benefit from fast ram, but still the fastest cpu from the page you posted is a Broadwell with DDR3-2400 11-13-14.

The second one, faster than all skylakes, is a Broadwell with DDR3-1600.

Even if you get your ram to DDR4-2133 it will be faster. It's not (just) the ram.