Question / Help Streaming with Intel i7-5960X

[Hrobnicek]

Hi there,

i saw a post back in days in Q&A section, but seems to me that thread is dead since person who started it no more updating status or something.
So, i was able to buy this monster CPU, and first of all, no, for me its not wasting of money.
Iam playing/streaming games, yes, also iam working as IT administrator, doing some rendering from time to time etc. so i can use it for all kind of work. Also, i wanted something which i can hold for atleast 3-5 years w/o upgrading.
I would like to give you all some feedback on this CPU if you are interested and mainly for gaming/streaming purposes.

Lets start with my setup:

CPU: Intel Core i7-5960X
MOBO: Asus X99-A
RAM: Corsair 16GB DDR4 2666MHz CL15
GCARD: MSI N770 TF 2GD5/OC Gaming
COOLING: Corsair Cooling Hydro H105

I think thats all we need, but if you want more details, just ask.

So far i tried to stream 3 games, all with same setting.

resolution: 1280x720
fps: 60
bitrate: 2500
cbr: ON
cfr: ON
x264 CPU Present: Slow

Streamed games:

World of Warcraft
Heartstone
Path of Exile

All games without problems, no frames dropped, CPU around 50% - 60% WITHOUT any kind of OC. Temps of CPU around 50-60 degrees Celsius.

If you want more information or anything, just let me know and i will try to do my best.

 

[Hrobnicek]

Iam adding log file from 45minutes of streaming Heartstone. I know its not the most heavy task game, but i will add more in next days.
Dont mind that error msg for slow server, sincce iam getting that error for every twitch server no matter what.
Server iam using is around 30km away from me, and still iam getting "slow server" issue :)

 

[dodgepong]

The "slow server" warning isn't important if you're streaming fine.

Are you having any issues, or are you just reporting that it is working well?

 

[Hrobnicek]

Hi there Dodge,

just reporting, i dont have any issues so far or i was able to solve them because somebody allrdy asked before me :)))
Like i sad, someone promised information and nothing hapened, so i decided to report some info by myself.

 

[yanis31]

hi and thanks for the post!

but what about preset "slower" and games that actually use alot of cpu? <- this is the elephant in the room here
(im not too familiar with wow tho but i can't imagine it's a tough one to stream)
would love to see this on maxed out battlefield4 full-ish 64player servers (paracel storm map during storm could be just the cake!) / tomb raider 2013 without game fps limiting (i remember it's cpu usage spiking alot during cutscenes) / crysis3 and other cpu multit-threading intensive games that i can't think of at the moment...
games that can't utilize 4 cores properly stream pretty decent already on amd's... that's why i believe it's a really bad way to test this 16thread cpu.

and to make this even more complicated (and i realise this is alot to ask just for some info)
would there be some way to get videosamples? let's say 720p on "slow" or "slower" on different bitrates from let's say 2000 to 3500 ?

 

[Hrobnicek]

Hi yanis31,

well, thats a lot of questions :)
But sure, i will try my best.
Iam not sure what games i should use, sincce i dont have BF4.
But what i know, GW2 was also eating i7 like hell. So i want to try that out.
I will prepare video samples for 720@60 and 1080@60 at slow and slower for 2500kbps,3000kbps and 3500 kbps.
It will take some time, but i think at the end of the week it can be done.

Peace

 

[yanis31]

Hi! And thanks!
i am not familiar with GW though...
BF4 is the go-to game because it uses alot of cores... (i know this is a really silly way to put it but i just had to)
- more than 4 to begin with. And i have experienced the worst cpu stream performance with this game out of any game i have ever attempted to stream by a long shot. so i have a feeling the GW will not represent this quite as well - some games that do not utilize multicores well and have a limited performance because of it - often turn out to be some of the best streamers with the amd 8 cores just for the reason there's more resources left for stream encoding...
(also i threw in the amd just because of it's architecture - please no Intel vs AMD wars, let's keep this place mature)

 

[Hrobnicek]

Hi,

i am not FUNboy of Intel neither AMD. I have few computors with Intel and AMD processors, so dont worry.
Well, i own BF3, if its somehow "same" in way of CPU usage.
Another game i can think of is DayZ Standalone or Arma 2 with DayZ mode. I believe they are both pretty heavy on CPU.
Also, i will be doing my testing probably during the next week. Some work come into my way :(

Peaccce

 

[yanis31]

cool dude! im not anywhere near being able to afford a 5960x so not much will change in my life even if you do it later :)
also wasn't aiming the "wars" against you - just every other thread even if i mention my cpu in one sentence - i get attacked! ... gosh i almost feel like buying you BF4 for all the effort haha! just alot of financial stuff in my way at the moment :) - because im not sure about ARMA since i don't have it ... but dayZ standalone i stream 60fps without a hitch... so that one atleast will not represent this situation at all... those BF4 64player servers where i have to stay on 30fps in order not to make my gameplay choppy if i stream using cpu...

 

[Boildown]

Thanks for posting about it. The question (what's the best computer and how good will it make my stream?) definitely comes up from time to time. I'd like to see some more log files from your tests, particularly looking at the duplicated frames numbers for the preset, framerate, resolution, and game.

Your one log file so far shows "07:36:22: Total frames encoded: 157529, total frames duplicated: 136 (0.09%)" which is fantastic at slow preset. There's an excellent chance it could achieve the "slower" preset and still keep the duplicated frames well less than 1%.

If you want to know a CPU-bound game that I would appreciate you trying, do Planetside 2. Its a free-to-play shooter that is very CPU-heavy in large firefights, it would be fascinating to know if your setup can do slow or slower while keep your FPS around 60 in that game.

 

[Hrobnicek]

Hi Boildown,

sounds like a deal to me. Planerside 2, i will test it.
During this week i will try do some tests.

Also, keep at mind, iam running everything at stock freq. So there is probably whole another level after i OC this CPU.
I will report soon with more results, tests samples and logs.

Peace

 

[Cryonic]

i can run x264 1080p 60FPS with "fast" preset with max 55% CPU-load. But CPU-heavy games like DayZ SA forced about 97% CPU-load overall. Anything was working fine, no frames dropped on the stream, but i dont sure that DayZ was running at full speed.
CPU used:
i7 5820K @ 4,5GHz with HT on. Once you overclock your beast to 4,2-4,5GHz, you can run any game and stream 1080p 60FPS with the slow preset. 2 extra cores help a lot^^

 

[yanis31]

i read somwhere that planetside can't utilise more than 2 cores properly because it's a DX9 title ... which would render testing it completely useless...
there's a big difference in having bad performance because the game can't utilize a multicore cpu,
and just using alot of cpu, because for the former - OBS will gladly step in and utilize the left over resources (of which there will be a whole lot on a 5960x) and render a smooth stream.

back to battlefield 4 64player servers... until i see this one i can't take the data seriously unfortunately... :)

i already mentioned maxed out DayZ - it doesnt use your cpu at all - that's the reason it runs bad, and thats why i can stream it 720_60 "fast" like it ain't no thing... but on the BF4 full64servers i better not go above 30fps with the same setting or the gameplay will start to stutter (probably can go a bit higher than 30 but if i dont feel smooth @ 45 i just use 30)

i can run x264 1080p 60FPS with "fast" preset with max 55% CPU-load. But CPU-heavy games like DayZ SA forced about 97% CPU-load overall. Anything was working fine, no frames dropped on the stream, but i dont sure that DayZ was running at full speed.
CPU used:
i7 5820K @ 4,5GHz with HT on. Once you overclock your beast to 4,2-4,5GHz, you can run any game and stream 1080p 60FPS with the slow preset. 2 extra cores help a lot^^

-therefore massive doubts about this one^^ (unless of course any game doesn't include well multithreaded ones)

also - don't take windows task manager too seriously.. it often shows load across all 8 of my cores when it clearly isn't running on them...

with my first fx6350 i had a motherboard driver bug that made 1 core run 100% while pc was idle -
guess what task manager showed? - all 6 cores around 16-17% load
-process explorer on the other hand - clearly showed 1 core 100% and other 5 idle - just the load was jumpind to different cores from time to time...

i hope you guys see why i mention the same thing over and over - because you seem to be going in completely the opposite direction from where you should be....

anyway - hope all goes well and the 1080p60fps data "for other games" sounds good already.

 

[Hrobnicek]

Hi again,

well i will definitaly do some research before i start testing.
Because today, find any game which is realy properly uti for more than 2 cores is realy hard.
Maybe i will buy BF4 just to test it. Its not too much money to spend. Also it will serve later on for other testing after i will buy new gcard.

So far i tried 1080p@60fps with slower present. Only game i think strugled was World of Warcraft.
Medium is totaly fine and what i remember slow was ok too.
But it was just 1minute test because i had spare time :)
So expect more and proper testing, hopefully in next few days.

Peace

 

[Cryonic]

Well i can clearly stream anything with 1080p 60fps with the 5820K @ 4,5Ghz.
I hit the 95-97% load, but the newer games like Evolve dont drop below 50-60fps and OBS doesnt drop frames, so i keep the settings. If i hit the CPU-limit, i will use a faster preset. And if you have over 90% load, it doesnt matter what you use to monitor it. My main tool is MSI afterburner to monitor anything at the same time and get peaks. I can see the load on all 12 threads.

 

[Boildown]

Planetside 2 used to be like that, but they had to make it run on the Playstation 4 using the same codebase, so now its a lot more multi-threaded than it used to be. Either case, its still worth testing, as whatever CPU is unused by the game, the encoding process can use, and the main thing I'm interested in is what preset you can work down to and the duplicated frames number you get at it.

 

[yanis31]

Well i can clearly stream anything with 1080p 60fps with the 5820K @ 4,5Ghz.
I hit the 95-97% load, but the newer games like Evolve dont drop below 50-60fps and OBS doesnt drop frames, so i keep the settings. If i hit the CPU-limit, i will use a faster preset. And if you have over 90% load, it doesnt matter what you use to monitor it. My main tool is MSI afterburner to monitor anything at the same time and get peaks. I can see the load on all 12 threads.

cool :) however you should try Battlefield 4 full 64 player servers :)

 

[Cryonic]

Well BF4 with 64 people is almost a torture test for your CPU :D

Anyway i`m really happy with the X99 plattform, its a huge step over the 4770K that i had before. I know some people would be happy about the 4770K, but if you want maximum quality out of 1 rig, you have to step it up^^

 

[yanis31]

i tried to gather some benchmark results about how the 5820k with an overclock compares to my cpu...
but admittedly after about 1 hour of messing with this i got a bit tired and checking cinebench and x264 hd benchmark results both on my pc and online concluded that 5820k@ 4.5-4.6ghz could be looking like 60-80% higher total multithreaded performance than my 8370e @ 4.7ghz (fsb230 x 20.5multi)

and so i just tested the 1080_60 performance of win8 desktop capture and a 1080p youtube video with some dynamic blinking lights, colors n whatnot in fullscreen...
(youtube is not quite ideal since it eats "some" cpu but i guess its small enough not to matter)

all tests ran at 3000kbps
cpu usage:

1080_60 veryfast
static desktop ~24%
video ~50%

1080_60 fast
static desktop ~36%
video ~80-90% with some jumps to near 100% - high cpu encoding message popping up from time to time (obs reports 0% dropped frames like it always does on my pc even if stream starts to freeze from time to time - so don't even look at that)

864_60 fast
static desktop ~22-24%
video ~40-65%

(the slower presets seem to be more "jumpy" on the cpu usage)

The above maths suggest to me that you could still run into some trouble if you attempted this with BF4 64 - That's why i want to see it... because if it actually can handle it - im sold.

Well BF4 with 64 people is almost a torture test for your CPU :D

Anyway i`m really happy with the X99 plattform, its a huge step over the 4770K that i had before. I know some people would be happy about the 4770K, but if you want maximum quality out of 1 rig, you have to step it up^^

exactly .. that is why i'm thinking to either step up to one of them when actually possible or just build a cheap 2nd rig with a haswell i5 or another AMD and put this one to encoding! still need to research the pluses and minuses of using a capture card/transferring the audio to 2nd pc and all that stuff... im sure it's not too bad though.

Also - which sane person actually needs 1080p60fps? sure it looks good but half of your audience won't be able to watch it unless you are partnered.
i have a i3-380m laptop with 4gb ram and intelhd graphics (integrated in the processor)
and it cannot handle my 720p60fps streams using browser... it is "almost" fine in the window but drops a bit of frames but fullscreen it can barely show 40fps... flash is really bad with twitch streams
it only works ok if i use VLC player to watch the stream...

 

[Boildown]

Measuring CPU usage is the wrong way to benchmark this. Run your game and then run OBS at various presets with the same resolution, downscale, FPS, etc. And then look at the log file and measure the duplicated frames. If duplicate frames are less than 1%, it passed, if they're more than that, it failed. Make sure to give it at least five minutes of encoding time to make it statistically valid. And try to run a timedemo or something that is a consistent CPU effort on the game-side.