Gadgetory

one of the newest benchmarks we've been doing lately pertains to live-streaming while gaming it's a new type of CPU test where we eliminate any type of GPU support in the form of encoding and rely entirely on h.264 encoding through OBS leaning on the CPU we haven't yet conducted this for a low or mid range CPU we've only looked at the i7 the i9 and then the r7 today we're looking at the Intel i5 8400 coffee-like cpu matched against the r5 at 1500 X CPU this is a six core sixth red chip versus a four core eight thread CPU with a lower frequency so should make for a good pairing before getting to that this video is brought to you by synergy the software that lets you share a keyboard and mouse between multiple systems if you have limited desk space and multiple computers to command synergy removes the need for separate peripherals or a KVM and works as over the network software use our link below to get 50% off the home or provision with SSL both conducting and understanding stream benchmarks is more difficult than the usual process of conducting and understanding just standalone game benchmarks with streaming we have a few different ends of the spectrum to look at on 1/2 you have the streamer experience that would be the person playing the game what do they see from their game in terms of frame rate and frame latency or frame pacing and then you have the viewer experience on YouTube or twitch or whatever it may be for our testing we're presently using YouTube that we have used twitch in the past but right now we're testing with 1080p 60 for our streamed output and we switch around between different h.264 encoding presets depending on how the test is going faster has been our go to preset thus far because when we conducted a survey of you guys earlier this year we found that most of those in the audience who do stream use either fast or very fast and so we picked the middle and went with faster today we're going to be doing a bit of faster but also we'll be leaning on a super fast in the h.264 preset and for those of you not familiar with this basically it goes down terms of encoding quality versus speed to encode so when you push a frame from the cpu out to the stream it has to be encoded and the speed of that encoding depends on the h.264 preset which you choose as a user based on how fast your processor or your GPU acceleration may be at the very lowest end of quality but the fastest end in terms of getting the frames out is ultra fast then super fast and then I believe very fast faster fast and medium and then some slower ones which we don't really get into because once you get down to medium and below it enters into placebo quality territory where something like an i-9 or a threader trip or CPU might be able to handle it but it kind of gets questionable whether you can see that quality as a user through a stream though perhaps you could notice it if you're doing something like hand brake encoding or something like that where you're ripping transcoding or encoding videos not just putting them out to YouTube where the quality will eventually be compressed at some level anyway so that's what we're going through today how there's a lot more to this you can check the article linked in the description below if you want to read more about the testing approach it does get complicated but we've really only listed the critical parts that you need to know to keep it simple as far as it can be anyway and ultimately with streaming you can use it as a synthetic test but in terms of the received output to the stream there's only so good that it can get because you can't push more than 100% of the encoding target which is 60 FPS for 1080p 60 and of course bear in mind that all of these CPU numbers will look better if you step down to 720p 60 if for example you are bottlenecked on your upload data rate through your isp rather than bottlenecked by the cpu but once you get into that territory basically everything encodes at 100% for the most part with at least super fast if not faster so that kind of gets to be a pointless test because you can there's there's no measurement beyond 100% encoded frames it just stops because that's effectively a perfect output at that point let's get started with dirt rally we'll move Doda in a bit we recently added dota and have done some work with csgo recently but dirt rally remains our go-to for stream tests it's not that a lot of people are streaming it it's just that the game is easy to work with and provides a trustworthy baseline that's easy to replicate which is what you want for your baseline at 1080p 60 Altru settings and advanced blending and a 1080p output to youtube streamed with a 10 megabit per second bit rate and again faster for the preset we get the results shown here from OBS the i5 8400 stock CPU only delivers about 31 percent of its frames showing that this workload is simply too difficult for the 6 core to handle the r5 1500 X priced equivalently at $190 for each CPU only delivers 7.7 percent of its frames we're not overclocking today but it is something to consider for the future we have to drop from the faster h.264 profile to better accommodate these low-end CPUs as our testing thus far has only dealt with i7s r7 x to 99 or X 399 CPUs and even overclocking won't help either of them at this point they're just too slow and to thread limited to be fair that's the bigger aspect here as threads matter when you're dealing with encoding and gaming because this is clearly not usable as a streamed output let's just look at it as a synthetic test for the starter even with these dismal eat low encoded frame counts of 31 percent on the 8400 and 7.7 presenta than 1500 X which means we're dropping about 69 and 92% of frames respectively we can still look at the frame delivery Layton sees the r5 1500 Alex manages to deliver 73 percent of its admittedly very low frame count within a 16 point seven millisecond ideal window the 8400 delivers about 56 percent of its frames in that same sixteen point seven millisecond window we can't make much of this just yet but it is something to keep in mind for the next few charts the 8400 is clearly delivering significantly more frames than the 1500 X though both are completely unusable for this configuration but the 1500 X delivers its lower frame count more consistently out of desperation and before dropping quality we switched to high priority for OBS this previously worked to help save the i7 77 kay from general uselessness in a similar test so it might help here we do end up delivering 100% of frames from the i5 8400 with switching the OBS the high priority but there's a hit to player side or streamer side performance look at that momentarily after this charge the 1500 X manages now to deliver 61% of its frames dropping about thirty eight point seven percent still somewhat unusable but also improved we can also see that again the 1500 X manages to deliver more in the plus or minus 2% window of sixteen point seven milliseconds but it's dropping so many more frames that it's still a poor match or this streaming configuration it's just too much of a workload the player side FPS chart will help better reveal whether the higher priority OBS workload murder is the ability to play and enjoy the game as we talked about in our r7 1700 and I 770 700 K streaming benchmarks previously it's possible to deliver 100% of frames and still have an awful experience if you're only able to render for example 30fps or below on the player side then you obviously have a lot fewer frames available to encode to stream then with a 60 + FPS render output as a refresher these charts consist of two halves we've just gone through some of the OBS numbers for the streaming output the next tab is the streamer view the game performance and that consists of FPS charts and make up the other half of the equation there are 2 FPS numbers per CPU at a minimum there's baseline performance which is performance without streaming and only gaming and then there's streams performance there might be in some cases a third option which is high priority or some affinity performance but that's less of a concern here baseline FPS for again dirt rally still the i5 8400 manages 109 FPS average 86 1% lows and 78 0.1% lows we right up against the limits here the r5 1500 X performs effectively the same with this video card and with these settings this makes the to roughly tied pre stream note that the 80 400 and 1500 X are not tied in gaming benchmarks but this is quite a bit different than those will publish the full review with those game benchmarks shortly hereafter in the meantime though they're equal pre stream during the stream at the i-5 8400 outputs 107 FPS average 81 1% lows and 34 fps 0.1% lows it has lost effectively zero percent of its performance but it's also failing to output any frames to stream if you remember correctly from the previous charge the 8400 was only successfully encoding 31 percent of its frames with the 1,500 acts encoding 8 percent of its frames this is interesting because the 8400 actually gives the streamer a great experience but the viewer is getting unwatchable 1 the r5 1,500 X drops more of its frames falling to 97 FPS average but is still spending more resources than necessary on streamer gameplay and not enough resources on the encoding changing to high priority for each CPU means we lose in streamer side FPS but we improve in coded frame output the i5 8400 is now giving the player 851 FPS average with low somewhat dismal at 11 fps 1% and 7 FPS 0.1% this is effectively unplayable what we do manage to deliver 100% of the frames to be fair there aren't that many frames to deliver anymore because at this point it's just not really usable the r5 1500 X with high priority does even worse at 33 FPS average and with 61% of frames delivered so it doesn't even hit 100% what you're seeing now is some of the streamed outputs from those tests we'll let this play side by side for a moment while I talk through the next section just to reiterate this test has traditionally been a mainstay for us in our limited time with these new stream benchmarks dirt rally has proved adequate for most testing and gives us an easy replicable output with these two CPUs it's just too much work though they can't handle it so we'll have to drop down to an easier setting for mid-range CPUs like encoding with superfast instead of faster let's take a look at superfast vs. faster for video quality but we'll hold the CPU unveil for a moment for this particular game superfast doesn't particularly drop our visual quality in a noticeable way of course ignoring the fact that faster is just stuttering like crazy by the time it gets to YouTube they really kind of look the same at least for this game the actual important part is that superfast allows us to output to an actual stream rather than a choppy slideshow and outweighs any minor quality differences you could perceive here's the chart for dirt rally was super fast with all the same settings except for h.264 encoding preset which is again super fast now we're able to encode 100% of frames at 60fps to the YouTube stream the r5 1500 X on i-5 8400 both job 0% of frames well gonna have frame Layton sees we see that the r5 1500 accent codes 53.85 presenta bits frames within the ideal sixteen point seven millisecond window which is a 60 FPS refresh we were to divide 1000 milliseconds by 60 fps you get that and then it's also got seventeen point seven percent of its frames delivering faster than sixteen point seven milliseconds 28.5% delivering slower being that faster is a valueless here you really just want consistency sort of like vr gaming doesn't matter you're faster slower just be there at the right time as for the I 580 451 point three percent of its frames are encoded within that sixteen point seven millisecond window with 21% faster and twenty seven point five percent slower both CPUs managed to achieve a 100% frame encode marker to their credit but they also achieve that mark more chaotically there is minimal precision between each the 51 percent versus 54 ish percent numbers are so close that they're more or less equal especially since we don't run as many test passes with streaming as we do with these shorter game tests the good news is that we are actually able to stream this game successfully now just with a lower quality output let's take a look at streamer side fps and experience because the two are pretty close so far the baseline performance remains the same as previously of course so we're looking at numbers of roughly 109 FPS average for each CPU against the limits are 5 1500 X drops to 99 FPS average 72 fps 1% low and 29 fps your opponent mows the four core 8 thread r5 CPU is therefore dropping about 10% in streamer side performance versus baseline in order to maintain a fluid output stream while we're still within playable territory well within it I 580 400 meanwhile manages to keep nearly all of its original perform because the superfast encoding and gameplay aren't generating enough work to fully engage and trip up the CPU as the previous test did it's still a heavy workload but it's not nearly as heavy the low frame rates are around where the 1500 X is with a 0.1% frame latency converting to 32 fps this fits the profile we've created for all CPUs when streaming on the host system we lose consistency in low-end frame latencies by initiating a stream which means that ultra-competitive players in games like csgo or dota 2 may be more sensitive to the loss using an external capture system would remove this concern but subjectively speaking we don't really notice the difference too often with this game when dropping just super fast from faster both CPUs are able to keep up with both encoding workloads and gaming workloads the rise in CPU drops about 10% of its baseline fps to accommodate the stream while the Intel CPU manages to hang on and is capped elsewhere we'll explore this in the dota test neither is outright superior to the other in streaming output when we drop to this quality setting before dota 2 here's a look at power consumption during the stream benchmarking each CPU is represented by baseline power consumption when gaming and not streaming then further streams of power consumption after that these power metrics are taken out EPS 12-volt rails not at the wall so they are measuring 12 volt power to the CPU the Intel i5 8400 CPU consumes about 29 to 31 watts when gaming only with the 1,500 X at about 40 to 42 watts ignoring one spike streaming the CPUs land at 52 to 54 watts of the i5 8400 when pushing dirt and a 10 megabit per second stream but the RFI 1500 X pulling 64 to 66 watts for the same workload move on to dota 2 using the same unplayable settings that we tested for dirt rally capturing dota 2 at 10 megabit per second with a 1080p 60 output and faster preset made the stream completely unwatchable we dropped about 98% of frames on both the i-5 and the r5 leaving us with fewer than 3% of frames and code into the stream successfully so achieve the high priority resolved this for each CPU giving us a full 100% encoding rate but we got lower frames in game this is revealed when looking at the FPS chart now the stock 8400 and stock 1500 X both have a significantly higher frame rates when no process is given priority but the stream is useless so they have functionally failed the test 84/100 has a baseline performance ceiling of 163 FPS average with 82 and 40 fps lows the 1500 X baseline is 108 FPS average with lows at 53 and 27 moving to the high prioritization the 8400 manages a 70 FPS average player site throughput but introduced is noticeable frame latency that computes 12 FPS 0.1% lows it's loosely playable in the stream is great but it's just frustrating enough to demand better performance you'd want to drop encoding settings a bit still r5 1500 X meanwhile just can't keep up at 28 FPS average and low sub 10 FPS it's not playable despite the stream encoding all of the limited frames moving the superfast should be more realistic streaming with this version of encoding we're able to output 100% of the frames from both the i-5 8400 and r5 1500 X making them functionally equivalent as far as the stream viewer is concerned that said the r5 1500 X doesn't manage one noteworthy feet of its frames delivered the CPU keeps 94.6% of them within plus or minus 2% of 16.7 milliseconds whereas the i5 8400 is closer to 65% as follows previous trends though they were less noticeable we don't yet have enough data to fully draw conclusions but it may be that when the rise in CPU is under less stress and strain for frequency demand its additional threads are assisting in frame pacing that's something that the i-5 lacks in this case but it does make up for it in speed and gaming instances looking at FPS the i-5 8400 delivers a decisive victory here it manages a 59 percent lead over the r5 streaming test we're at 135 FPS average versus 85 FPS average part of this difference contributes to the i5 more chaotic frame basing and just like with the 7700 K you'd be best off doing some manual tuning on the CPU to get the most out of it we'd recommend some tuning of affinities on the CPU threat Rison still manages reasonably well at 85 FPS average while delivering 100% of the frames to the stream but the i5 is also capable of this task the difference is that the i5 has less precision and you really should be manually tuning it a bit to compete with the rise in CPU more directly the rise CPU seems to be able to handle frame pacing a bit better out-of-the-box as for power consumption the spacing still favors Intel i5 8400 is pushing about 4 to 4.1 amps with the r5 1500 is pushing about 4.8 amps so that's it for our r5 and i-5 stream benchmark there's a lot more to do with this streaming is highly complex there are a lot of aspects we don't or can't test there are a lot of other games to test and this video is dense enough as it is it's hard enough to run these tests as it is so we're working on doing more of these obviously we keep adding them and iterating on them but just keep in mind that this isn't fully definitive so while we can draw conclusions for these settings with these games as to which CPU is better at these particular settings and games it's not enough data to draw conclusions across the board because just looking at dirt and dota alone it's obvious that the two CPUs kind of trade a little bit in performance depending on which game it is and we've seen this in the past with the i7 s the r7 is thread Ripper and I nines and things like that so anyway that's the data for now take it for what it is it's basically still in kind of it's a test it's methodological we've advanced it into the reviews at this point so it's good data that we trust but we're still learning as far as what settings make the most sense for these types of tests superfast appears to me what's basically necessary for the IO fives and our fives faster is just too hard for them to do as far as how visible that quality difference is we're still figuring that out but it looks like both the i5 and r5 can handle with superfast streaming it's just that the i5 in some cases does a bit better the r5 in other cases does a bit better between frame pacing and frame encoding and then the i-5 is just pretty much always ahead in terms of raw FPS particularly when carrying a stream simultaneously sometimes massively sometimes not as much so lots to consider ideally you don't use either of these for your stream but you could make it work if you had to at least a little bit more than some of the previous cpus we've tested it's just you really start tanking the quality as you move down into these lower court couch chips particularly so as always thank you for watching you can go to patreon.com/scishow and exes that's where you can help us out directly or join our discord community and you can go to store that gamers nexus net to pick up a shirt like this one or hopefully soon more hats like that one subscribe for more I'll see you all next time you