WEBVTT 00:00:00.520 --> 00:00:06.520 video cards they're good for you know 00:00:04.040 --> 00:00:11.040 anything to do with video right I mean it's it's right in the name video card 00:00:09.000 --> 00:00:15.640 they've got features to enhance video playback in real time they've got 00:00:13.440 --> 00:00:20.760 dedicated Hardware to accelerate video games and they're a necessary component 00:00:18.320 --> 00:00:26.960 of any system that needs to Output video to a display at all but what about video 00:00:24.599 --> 00:00:33.120 editing this capability has been featured prominently on the boxes of 00:00:29.439 --> 00:00:35.960 video cards for over a decade but do 00:00:33.120 --> 00:00:40.840 they actually make any bloody difference or should you just get a beefy CPU and 00:00:38.160 --> 00:00:47.609 Tackle your video rendering that way let's find out shall 00:00:50.480 --> 00:00:57.480 we the master case 5 by Cooler Master gives you the freedom to truly make your 00:00:55.120 --> 00:01:01.399 midtower PC case your own with a variety of modular parts and accessories check 00:00:59.519 --> 00:01:06.600 out the l in the video description to learn more let's start with a little bit 00:01:03.920 --> 00:01:12.600 of background on why it is that we are doing this we make videos and one of the 00:01:09.600 --> 00:01:14.520 most tedious and timeconsuming processes 00:01:12.600 --> 00:01:19.600 that slows us down aside from our servers being broken is transcoding 00:01:17.479 --> 00:01:24.479 footage for faster performance in Adobe Premiere a process that you can learn 00:01:21.400 --> 00:01:26.600 more about here and encoding our 00:01:24.479 --> 00:01:30.640 finished videos using the optimal resolutions and video codecs for the 00:01:28.640 --> 00:01:36.200 various platforms that we publish on including YouTube vessel and billybilly 00:01:33.560 --> 00:01:41.520 docomo when we were developing our copor workflow that GPU usage in our 00:01:38.840 --> 00:01:45.880 workstations was much higher than we expected with certain workloads which is 00:01:43.560 --> 00:01:49.240 funny because we originally thought the GTX Titan class cards in our editing 00:01:47.680 --> 00:01:54.799 workstations were basically a because we can thing but something we never really 00:01:51.799 --> 00:01:58.360 had the time to investigate was which 00:01:54.799 --> 00:02:00.360 video cards are best for this process we 00:01:58.360 --> 00:02:05.719 just stuck with what NVIDIA had provided us back when we did our whole room water 00:02:02.399 --> 00:02:07.920 cooling project well now we've got an 00:02:05.719 --> 00:02:13.520 opportunity to answer that question for the Adobe editors out there for an 00:02:10.560 --> 00:02:21.560 unrelated project super micro just sent us one of their super server Sy 4028 gr- 00:02:17.400 --> 00:02:24.000 trt servers this freaking badass of a 00:02:21.560 --> 00:02:30.319 machine is going to be used in a project where we run eight gaming rigs off of a 00:02:26.239 --> 00:02:34.400 single box with gtx980 ti amp card from 00:02:30.319 --> 00:02:37.400 Zotac more to come but in the meantime 00:02:34.400 --> 00:02:40.599 it presented an opportunity for me and 00:02:37.400 --> 00:02:42.280 Ed to quickly and with relative ease 00:02:40.599 --> 00:02:48.440 without any physical access to the machine use on raid as a virtualization 00:02:45.440 --> 00:02:51.000 host and then use this thing as a remote 00:02:48.440 --> 00:02:55.800 test bench so Ed could shut down a Windows Virtual Machine virtually swap 00:02:53.560 --> 00:03:01.080 the video card to any one of the ones in here turn it back on and be ready to 00:02:58.319 --> 00:03:06.440 rock with a new set of bench marks this led us examine how a wide cross-section 00:03:04.000 --> 00:03:11.920 of different video cards on both the AMD and NVIDIA side of the fence behaved in 00:03:08.920 --> 00:03:13.840 our typical video encoding tasks so the 00:03:11.920 --> 00:03:18.440 virtual machines were set up with eight core hyperthreaded virtual processors to 00:03:16.040 --> 00:03:24.680 represent a high-end consumer or mid-range Zeon processor and 32 gigs of 00:03:21.840 --> 00:03:30.040 RAM the tests were done using the 200 gig boot SSD as a source and as a Target 00:03:27.920 --> 00:03:34.760 and we used actual Linus media group video projects for all the scenarios 00:03:32.920 --> 00:03:41.200 let's start by looking at the time it takes to export a timeline containing 00:03:37.280 --> 00:03:43.480 mostly cfor video clips in cfor this is 00:03:41.200 --> 00:03:47.360 how we would normally send the project to the final export server for 00:03:45.280 --> 00:03:52.920 processing into h264 and the only real conclusion that 00:03:50.720 --> 00:03:58.599 we can draw from this is that whether you go with AMD and therefore opencl 00:03:56.040 --> 00:04:02.959 acceleration or NVIDIA and therefore Cuda acceleration for your video 00:04:00.480 --> 00:04:08.840 workstation you're going to be getting a very similar experience just make sure 00:04:05.840 --> 00:04:11.280 that you do have a video card as relying 00:04:08.840 --> 00:04:16.000 on the CPU to do this work on its own is clearly a bad idea for the next scenario 00:04:14.360 --> 00:04:20.519 we're looking at something that would be more applicable to most prosumer and 00:04:18.600 --> 00:04:26.759 professional video editors so we're taking the 1080p cfor export from the 00:04:23.240 --> 00:04:29.800 first test and transcoding it to 4K 00:04:26.759 --> 00:04:33.520 h264 no separate encoding server 00:04:29.800 --> 00:04:36.320 involved our GPU usage hovered around 00:04:33.520 --> 00:04:41.560 30% and the advantage that our video cards enjoyed over the CPU only workflow 00:04:38.960 --> 00:04:46.160 diminished significantly here which might lead one to the conclusion that 00:04:43.919 --> 00:04:50.520 which graphics card you choose doesn't matter or even that a graphics card is 00:04:48.680 --> 00:04:56.280 really optional if you're exporting videos directly from your editing 00:04:52.960 --> 00:04:59.759 workstation except that our third test 00:04:56.280 --> 00:05:03.800 using chroma keed copor 4K footage and 00:04:59.759 --> 00:05:06.680 exporting to 1080p cinor showed us what 00:05:03.800 --> 00:05:12.240 we were finally looking for the scenario in which a powerful video card 00:05:09.120 --> 00:05:15.680 demonstrates not just better performance 00:05:12.240 --> 00:05:18.520 against pure CPU encoding but also 00:05:15.680 --> 00:05:24.400 lesser video cards so the beefiest gaming GPU in this case the 00:05:20.960 --> 00:05:26.960 gtx980ti in the system comes out on top 00:05:24.400 --> 00:05:31.720 with the Lesser cards performing about as much worse as you would expect given 00:05:29.400 --> 00:05:36.800 their overall horsepower and this would be applicable not just to chroma key 00:05:34.479 --> 00:05:42.240 green screen footage but also to any project with other effects like lumry 00:05:39.000 --> 00:05:44.840 color uh or warp stabilizer which leads 00:05:42.240 --> 00:05:51.440 us then to a really interesting final conclusion here AMD versus NVIDIA in 00:05:48.560 --> 00:05:56.840 this case doesn't seem to matter much the 970 and the r9290 which is about 00:05:54.360 --> 00:06:01.840 equivalent to its 300 series replacement are similar enough across the board that 00:05:59.479 --> 00:06:06.759 I I couldn't recommend one brand over the other for editing an Adobe Premiere 00:06:04.400 --> 00:06:12.240 with that said that might not be the case if you use other tools but the 00:06:09.919 --> 00:06:17.720 takeaway here then is that for video editing and Adobe Premiere you should 00:06:14.720 --> 00:06:21.319 get a GPU even if it's just a mainstream 00:06:17.720 --> 00:06:23.160 model and that much like for gaming once 00:06:21.319 --> 00:06:27.800 you get into the performance and Enthusiast product range you are getting 00:06:25.520 --> 00:06:33.280 diminishing returns for the money you spend neat 00:06:30.520 --> 00:06:39.520 speaking of which this is pretty cool this is a brand new sponsor for us Blue 00:06:36.599 --> 00:06:44.520 Apron allows you to create delicious chef-designed recipes at home they 00:06:42.440 --> 00:06:48.199 deliver all the farm fresh ingredients you need right to your doorstep in 00:06:46.400 --> 00:06:52.720 exactly the right proportions so no trips to the grocery store are required 00:06:50.440 --> 00:06:57.800 and there is no waste from unused ingredients they offer two types of 00:06:54.880 --> 00:07:02.000 plans the two-person plan and the family plan the recipes are delivered in a 00:06:59.919 --> 00:07:05.560 refrigerated box so the ingredients stay fresh even if you're not at home when 00:07:03.479 --> 00:07:09.800 your package arrives and they offer a large selection of recipes and are 00:07:07.840 --> 00:07:14.720 always adding new dishes to the menu every week there is no commitment you 00:07:12.240 --> 00:07:20.479 can skip or cancel the service at any time and each menu is between 500 to 700 00:07:18.160 --> 00:07:24.680 calories per person the ingredients are sourced from quality suppliers and 00:07:22.560 --> 00:07:30.360 Artisans and all meals can be prepared in 40 minutes or less so blue apron lets 00:07:28.440 --> 00:07:34.720 you learn to make new new recipes and Cuisines so you can get out of your 00:07:32.199 --> 00:07:38.879 dinner rut of making the same old dishes or relying on takeout and the best part 00:07:37.280 --> 00:07:43.400 is that blue apron is available Nationwide in the US and the first 250 00:07:41.720 --> 00:07:48.199 people to try it using our Link in the video description will get two free 00:07:45.280 --> 00:07:53.120 meals on their first Blue Apron order so go check it out so thanks for watching 00:07:51.280 --> 00:07:57.000 guys if this video sucked you know what to do but if it was awesome get 00:07:54.879 --> 00:08:01.199 subscribed hit the like button or even consider supporting us directly by using 00:07:58.759 --> 00:08:04.879 our affiliate to shop at Amazon it costs you nothing the instructions are up 00:08:02.840 --> 00:08:08.240 there buying a cool shirt like this one which does cost you something but you 00:08:06.039 --> 00:08:11.400 get a cool shirt or by joining our community Forum you can get your 00:08:09.919 --> 00:08:16.039 questions answered you can answer other people's questions you can become a contributor get a cool contributor badge 00:08:14.599 --> 00:08:18.759 all good stuff that's Linked In the video description now that you're done 00:08:17.759 --> 00:08:25.400 doing all that stuff you're probably wondering hm what should I watch next and the answer is this video right here 00:08:23.240 --> 00:08:29.240 I guarantee you it's awesome my personal guarantee of awesome 00:08:27.199 --> 00:08:33.000 quality I hope it's awesome I don't know what going to put there