CPU Cores for Gaming, Streaming, & Rendering Simultaneously - Is More Actually Better?
Linus Tech Tips
·Linus Tech Tips
·2015-05-07
·
1,317 words · ~6 min read
0:00
so we reviewed the core i7 5960x here an
0:04
eight core hyper-threaded ddr4 fueled
0:07
monster of a computer processor that seems to be positioned as the elite tier
0:12
CPU in gaming systems among boutique system builders and yet
0:16
doesn't appear at first glance at least to offer a compelling improvement over a
0:21
more mainstream and much less expensive
0:24
ddr3 equipped quad core processor like
0:28
the core i7 4790k perhaps there is more to this than meets
0:33
the eye though like how i've been using an iphone 6 for a couple weeks i mean
0:37
actually that one's pretty easy i have a review coming subscribe so you don't miss it anyway back to CPU performance
0:42
for now though
0:50
Corsair gaming RGB keyboards feature precision cherry mx RGB key switches for
0:55
16.8 million color per key backlighting for virtually unlimited customization
1:00
click now to learn more so here's the problem with the way that benchmarks are
1:05
typically run in a CPU review in order to isolate as many variables as possible
1:11
and keep the results as reproducible as possible this is the correct way to do
1:16
things scientifically speaking by the way background processes on the system are
1:20
kept in check so only the applications being benchmarked at a given time are
1:25
run and fresh system images are loaded very regularly to ensure random extra
1:31
junk isn't interfering with getting the best results possible however in the
1:36
real world a more cluttered environment the kind where you've got you know a
1:41
thousand sometimes poorly coded random apps in the system tray like bloated
1:46
gaming peripheral software anti-virus game streaming voip clients printer auto
1:50
updaters etc is oftentimes more realistic
1:54
and beyond that power users and prosumers may be interested to know how
1:59
their system will perform in games when they intentionally have a demanding
2:03
application running in the background so the test i set up is based on a worst
2:09
case scenario for a prosumer like a game
2:12
streamer someone who's heavily into gaming and uses their gaming rig for
2:17
some content creation as well so here's my hypothetical scenario our pro
2:22
streamer just finished editing together a highlight reel in sony vegas pro 12
2:27
and is rendering the file then he or she wants to fire up a game we'll use far
2:32
cry 3 at 1080p ultra details with 8x msaa wants to stream aforementioned game
2:38
on twitch using the fast preset with an hd webcam feed in the corner using
2:42
xsplit wants to use screen capture to have a buddy skype in to join the stream
2:47
and then while all that's happening an antivirus scan starts running in the
2:51
background so for the test i used an ASUS rampage 5
2:55
extreme motherboard with a core i7 5960x then i used the handy dandy
3:00
multiplier adjustments that all k and x series chips have along with ASUS's
3:05
manual turbo boost controls and hyper threading toggle to simulate the
3:09
performance of all the other cpus in the
3:12
test so i'm using this the same platform same os
3:15
everything so there are two different ways to
3:19
evaluate the performance of our system here first up we can look strictly at
3:23
average FPS in our game over a two minute span at the beginning of my
3:27
benchmark the overclocked eight core runs away with this but not everyone
3:32
overclocks and at stock speed our quad core 4790k with hyper threading does
3:38
just as well as our stock
3:42
5960x which costs about three times as much presumably thanks to its high turbo
3:47
clock speed of 4.2 to 4.4 gigahertz
3:50
adding minimum frame rates to the mix gives us a little bit more insight
3:54
though now we can see that our six core
3:57
seems to be struggling a bit thanks to its lower core count or lower clock
4:01
speed compared to the other competitors depending on which one we're looking at
4:04
it against and even though on average our quad cores hold up against our eight
4:09
core the dips in frame rate are 20 lower
4:13
which would normally be okay but what the numbers don't even tell us here is
4:17
how noticeable the hitching and stuttering is in each of those configs
4:22
the a-core configs on the other hand both ran the game with smoother
4:26
animations and the six core stuttered occasionally but better than the quad
4:30
cores the quad core with hyper threading was noticeable but usable and the quad
4:36
core without hyper threading was basically unplayable and you can see a
4:41
full explanation for why these FPS numbers might look okay even though the
4:45
animations are terrible in the video linked here
4:49
but there's another completely different angle that we can use to examine this
4:53
torture test scenario to give us a clearer idea of how the chips are
4:57
holding up to the load how well was the stuff in the background running well
5:02
first let's have a look at dropped frames in the twitch stream only our
5:06
non-hyper-threaded quad-core dropped frames here so we can see that our
5:10
simulated 4690k was really struggling to multitask like
5:15
this in a way that the others weren't
5:18
now finally we can look at render times
5:21
it's a one minute chunk of my Corsair gaming k70 RGB keyboard video that i
5:27
exported at 1080p with CPU encoding while everything else was running and
5:32
this is where we see the men separated from the boys our hyper threaded six
5:37
core stutters less in-game and renders
5:40
the video a full two minutes faster than the hyper-threaded quad core even though
5:45
its average frame rates are not as good then our a-core improves frame rates
5:50
smooths out gameplay completely and knocks another 40 seconds off the render
5:54
time with our overclocked 4.4 gigahertz 8 core delivering smooth gameplay the
5:59
best frame rates out of all the configs and also it manages to render the video
6:04
in a third the time of our similarly clocked quad core with hyper threading
6:10
so in conclusion am i saying any of this is
6:13
the kind of thing a normal user will encounter every day or am i saying that
6:16
this is the only way to achieve butter smooth gaming while working on projects
6:20
in the background absolutely not
6:24
using a device with an onboard h.264 encoder setting up a separate machine
6:29
for capture or utilizing a network render farm are all strategies that
6:33
might actually be more elegant depending on the exact workload now the point of
6:37
this video was just to explore the effects of an overwhelming workload on a
6:42
variety of different configs and have a look at what those lucky dogs with eight
6:47
core extreme editions can throw at their machines before they even really start
6:51
to feel it and the answer i guess is a whole heck of a lot
6:57
guys thanks for watching like this video if you liked it dislike it if you
7:00
disliked it leave a comment letting me know was this about what you were
7:04
expecting did you think it would be able to handle much more did you think even
7:07
the a core would suffer love to hear from you guys also check
7:11
out the link in the video description you can give us a monthly contribution to help support us to make videos you
7:15
can buy a t-shirt cool one like this which does the same thing or you can
7:19
change your amazon bookmarks one with our affiliate code so we get a small kickback whenever you buy cpus or
7:25
tape or stands or whatever else it is you buy on amazon i actually got these on amazon
7:29
great great product very inexpensive anyway thanks again for watching and as
7:33
always don't forget to subscribe
7:53
you