Gadgetory

okay do you remember that project I was working on where for the better part of six months I built up this badass 36 core dual Xeon server machine to handle our video encoding and transcoding tasks over the network here well fast-forward almost a year and many many hours spent on diagnosis not to mention a kick in the right direction from this post over on Puget systems I think I finally figured out why we never got quite the performance that I expected is it possible then that a four thousand dollar twenty two core CPU could be outperformed by one that costs only a few hundred bucks for video encoding is it possible that I made a mistake nothing to hold on to fails all I read in the sign I'm definitely getting their attention so does one of the recurring themes of these laptop or bus videos become - failure montages and I mean aside from those ones let's find out freshbooks is the super simple invoicing solution that lets you get organized save time and get paid faster click now at the link in the video description to try it for free ok so to open this video up we need to take a closer than usual look at my test bench i wanted to eliminate bottlenecks wherever possible so that the cpu is the only factor in my performance evaluation so for that reason most of the performance testing was done on an Intel 750 series 1.2 terabyte nvme SSD a gtx titan X 128 gigs of DDR 4 quad channel memory on an X 99 deluxe 2 motherboard and the CPUs tested are as follows intel's top of the server line $26.99 v4 22 cores e on their top of the high-end desktop line 10 core Core i7 extreme 69 50 X the 8 core and 6 core 6900 K and 6800 K and finally I decided to throw in their flagship mainstream 6700 K quad core to give us the most complete picture possible at the end of the day as for the video tests I apologize in advance if the codec or encoder application that you personally prefer wasn't covered but this was done as much to optimize the linus media group workflow as it was for the purposes of creating a video so i'm looking at four different scenarios that we encounter pretty much daily 1 transcoding a 4k mxf off of our sony FS 5 to 1080p sinha form our mezzanine codec of choice for editing to exporting a finished project in this case a green-screen episode of fast as possible directly to h.264 for publication to youtube 3 a quick export in cinah form how we normally export so that a network media encoder machine with a watch folder can transcode it to h.264 and automatically upload it to the channel and for finally the performance of that Sinha form to h.264 conversion with the 1080p to 4k up sampling that we perform for the reasons we covered more thoroughly in this video here so I ran every test with and without CUDA acceleration enabled in Adobe Media encoder and used a second machine to capture the screen output with CPU and GPU usage displayed so I could review it later let's begin then with scenario 1 this is what most people probably expect from a multi-core CPU in a video encoding benchmark traditionally this is one of the easiest workloads to scale across more course and our CPU usage graph indicates that all is working beautifully throwing a GPU into the mix levels the playing field somewhat but this won't surprise anyone who knows how deep you dependent of video codecs Sena form is and how that bastard law of diminishing returns works moving on to exporting a project directly from our Sena form timeline in CPU only mode we see nice scaling with more cores but maybe not quite the dominance we'd expect from a chip with and yes I know it doesn't quite work this way like 60 gigahertz of theoretical total performance this is a hint of things to come and been throwing a GPU into the mix paints a much more extreme picture here the CUDA accelerated code path not only reaps very little benefit from more than 6 course it punishes CPUs with lower clock speeds in a way that I really didn't expect observed GPU usage is much lower than any other processor in this test for our four thousand dollar chip and the CPU usage we see of about 25% tells us this is not a heavily threaded workload oops alright so let's break that down then into the individual steps and find out where our heavy multi-thousand dollar investment in an uber xeon falls apart exporting the project from a center form 1080p timeline to a Sena form 1080p file theoretically elsewhere on the network but I'm using my nvme drive as a target for these benchmarks for consistency sake is pretty flat across the board and curiously this is true with or without CUDA acceleration enabled in media encoder GPU usage is 85% regardless of which drop-down so this is clearly nearly 100% GPU dependent which leads us then to the second step in the process converting from Sena form 1080 to h.264 4k in CPU mode only we see a similar trend to our initial ingest test more horses is better but only to a point then in GPU assisted mode there it is we are almost entirely bound by per core performance with a lowly quad-core costing one tenth as much handily beating our xeon beast so then did i horribly miss configure our video encoding ingest stations and output server are Zeon's basically pointless in video work well if you're looking simply at the graphs i just showed you along with these charts of approximate CPU and GPU usage in all the different scenarios i tested then it's pretty clear that these lower clocked many-core chips are being underutilized and the money though i fortunately didn't pay for them would be better invested almost anywhere else but as always the real world isn't really that simple and it's going to come down to the needs and workflow of each individual or organization virtualization can be used to get damn near 100% scaling out of as many cores as you please encoding software like Sorenson squeeze can process many files at a time and on the subject of different software testing any given codec in any given software could yield very different results from what you're looking at here so there's no way around testing just make sure that when you do so for yourself you go in without any assumptions about what the right tool for the job will end up being so you can avoid pulling a Linus speaking of tools for the job it's summer apparently something-something boarding planes trains driving a car leave your worries behind okay I don't know what any of that stuff in my notes is but today's sponsor is tunnel bear and if today's lack of online privacy brings out your inner grizzly bear rawr rawr then you can try Tunnel bear it's simple and it is free to try at the link in the video description it's the easy to use VPN that makes it so you can browse privately and enjoy a more open Internet without all that hassle associated with more complex VPN solutions any you know port forwarding your DNS or any nonsense like that you just click the button and boom you can tunnel in to up to 20 different countries and it will appear to the websites and services that you are using as though you are coming from that country and Tunnel bear has a top rated privacy policy and does not log your activity so try it free with 500 megabytes and no credit card required and if you decide you like it and you want to get a year of unlimited data you can save 10% by going to tunnel bear comm slash LTT linked in the video description so thanks for watching guys if this video sucked you know what to do but if it was awesome get subscribed hit that like button or even check out the link to where to buy the stuff we featured and Amazon in the video description also linked in the description is our merch store which has cool t-shirts just like this one and our community forum which you should totally join now that you're done doing all that stuff you're probably wondering what to watch next so check out that little button on the top right to check out our latest video over on channel super fun