WEBVTT 00:00:00.040 --> 00:00:07.600 so this is exciting we are finally going to be validating if our idea for the 00:00:05.640 --> 00:00:12.880 rendering machine over there is going to work but first we need a long Ethernet 00:00:10.120 --> 00:00:12.880 cable 00:00:21.840 --> 00:00:26.760 so freshbooks is the online invoicing 00:00:24.920 --> 00:00:32.760 solution that lets you get organized save time and get paid faster click on 00:00:29.480 --> 00:00:37.040 this bought right here for a free trial 00:00:32.760 --> 00:00:38.800 now um all right so one of the few 00:00:37.040 --> 00:00:41.960 pieces of the puzzle still well there's a few pieces of the puzzle still missing 00:00:40.399 --> 00:00:47.000 okay one of the pieces of the puzzle still missing is getting a 10 GB 00:00:44.399 --> 00:00:50.760 Ethernet connection to that machine so we've swapped out to the super micro 00:00:48.520 --> 00:00:56.840 board that doesn't have it on board but the good news is oh I just dropped one 00:00:53.680 --> 00:01:01.079 okay well the good news is my x540s and 00:00:56.840 --> 00:01:02.920 other 10 gbit cards are here now so 00:01:01.079 --> 00:01:06.920 we're going to go ahead and get that puppy linked up then we have to actually 00:01:05.400 --> 00:01:10.240 build the machine I've just had it in a test bench up till now 00:01:12.320 --> 00:01:20.600 so and this this was the thinking behind 00:01:17.439 --> 00:01:20.600 the carabiners in the 00:01:25.200 --> 00:01:29.200 ceiling okay 00:01:31.600 --> 00:01:34.320 so now in 00:01:35.520 --> 00:01:39.119 theory we are 10 00:01:40.759 --> 00:01:48.159 gbit sick all right we are good 00:01:45.200 --> 00:01:54.079 so let's let's actually let's actually build this machine 00:01:50.040 --> 00:01:57.280 finally finally okay so I thought that 00:01:54.079 --> 00:02:00.200 this RPC 00:01:57.280 --> 00:02:06.039 4224 was actually going to be one for to grow into but it looks like we need it 00:02:03.439 --> 00:02:12.280 now so this is the last of my four U cases from Norco that they graciously 00:02:09.479 --> 00:02:17.959 provided for our new server room and we are going to be using this one for the 00:02:15.200 --> 00:02:22.480 rendering machine so there's all of our mounting 00:02:19.319 --> 00:02:24.319 Hardwares there's our nice soft foam 00:02:22.480 --> 00:02:30.120 that my cat would eat if he could get at it and he would try to swallow it and he 00:02:28.120 --> 00:02:34.680 would then get it stuck in his stomach which he did again costing me another 00:02:32.599 --> 00:02:39.560 $1,200 if he does it again I'm going to kill him that's twice in 4 months he's 00:02:36.640 --> 00:02:39.560 eating foam and needed 00:02:41.640 --> 00:02:46.159 surgery so there point is it cheaper to 00:02:44.760 --> 00:02:51.280 just get a new cat it would have been cheaper to get two new cats the first 00:02:48.480 --> 00:02:55.080 time now I could have had three new cats and like a pretty cool 00:02:55.159 --> 00:02:59.360 lizard all right 00:03:00.159 --> 00:03:06.280 but he's my baby and I love him that's 00:03:03.599 --> 00:03:10.480 what I tell myself it's like gee why doesn't lonus have a Lambo with all that 00:03:08.159 --> 00:03:17.000 YouTube money cuz he has the world's most expensive cat that's why 00:03:14.920 --> 00:03:20.239 okay um I'm going to need a minute to clear some space here I 00:03:21.879 --> 00:03:27.040 think um band machine 00:03:27.080 --> 00:03:30.280 gun um 00:03:30.280 --> 00:03:40.439 power supply now I could leave the 860i that I 00:03:36.799 --> 00:03:42.159 have in there but then if I ever want to 00:03:40.439 --> 00:03:45.640 actually put graphics cards in cuz I'm going to take those graphics cards out 00:03:43.640 --> 00:03:52.079 if I ever wanted to put proper graphics cards in then I would have to replace it 00:03:49.959 --> 00:03:58.439 with something else which is kind of a pain in the 00:03:53.360 --> 00:04:01.480 butt so H you know what no I'm going to 00:03:58.439 --> 00:04:03.120 stick with the 860 ey the odds of 00:04:01.480 --> 00:04:05.439 throwing graphics cards in here are pretty 00:04:08.000 --> 00:04:14.920 slim do you know how long it's been since I've seen a motherboard where I 00:04:12.599 --> 00:04:22.959 actually have to pop the covers on the io plate off 00:04:19.320 --> 00:04:22.959 myself like are you for 00:04:25.400 --> 00:04:29.880 real YOLO man 00:04:30.280 --> 00:04:34.840 oh that's really tight okay I need 00:04:36.160 --> 00:04:41.600 pliers okay just doing a quick test fit here finding out where the devil all the 00:04:40.759 --> 00:04:46.639 holes are so that one over there and this one 00:04:44.639 --> 00:04:54.199 right here what okay so there's that 00:04:50.960 --> 00:04:56.759 one there's that one there's that one 00:04:54.199 --> 00:04:56.759 where's this 00:04:57.320 --> 00:05:02.960 one do you need all of them well it's 00:05:00.520 --> 00:05:05.360 optimal get away with two I think that's what I 00:05:05.440 --> 00:05:14.120 do um this show is all about the uh right 00:05:10.919 --> 00:05:14.120 way to do things 00:05:14.360 --> 00:05:18.280 so so at least 00:05:19.400 --> 00:05:24.919 three I guess we'll need a power supply at some 00:05:23.360 --> 00:05:30.080 point okay then oh look at that there's fits 00:05:28.160 --> 00:05:38.360 over at the front here it's perfect just like Christmas kind of a week Christmas a fan 00:05:34.639 --> 00:05:40.280 header yeah your expectations are so 00:05:38.360 --> 00:05:43.960 high how did you know that I wasn't going to do fan headers for Christmas 00:05:41.639 --> 00:05:47.400 bonuses this year now my feelings are hurt they should be now that my feelings 00:05:46.240 --> 00:05:53.319 are hurt maybe there will be no Christmas bonus did you ever think of that bet you'd rather have a fan header 00:05:51.080 --> 00:05:59.520 than nothing no I think I'd rather nothing than fan headers just because 00:05:56.039 --> 00:05:59.520 that's kind of insulting 00:06:01.199 --> 00:06:09.199 duly noted Merry Christmas everyone here 00:06:06.680 --> 00:06:11.720 poop oh you don't like poop oh you want poop 00:06:12.840 --> 00:06:18.960 nothing I think everyone in the comments will agree with me that fan headers are 00:06:16.880 --> 00:06:25.840 better than poop that doesn't even make sense at 00:06:22.520 --> 00:06:27.800 least you could use poop for what 00:06:25.840 --> 00:06:32.400 leaving on your boss's desk for giving youan bosses give you 00:06:32.759 --> 00:06:41.560 poop I actually ordered a game with uh little turds and velcro helmets that you 00:06:38.400 --> 00:06:43.240 wear that they stick to on Amazon that I 00:06:41.560 --> 00:06:49.520 was thinking we'd play on Channel Super Fun so it occurs to me now I need a 00:06:46.120 --> 00:06:50.880 graphics card I didn't really want to 00:06:49.520 --> 00:06:56.319 put one in here but this motherboard has no onboard 00:06:57.240 --> 00:07:05.039 graphics and you're sure you can't use a Titan so I'm thinking as I install the 00:07:02.080 --> 00:07:09.759 power supply this is the one piece of Hardware in this machine that I would 00:07:06.720 --> 00:07:12.360 really like to change um not because the 00:07:09.759 --> 00:07:16.280 860 watt isn't enough for for what we're going to do and that may be the case in 00:07:14.360 --> 00:07:20.199 the future if we find some kind of application that can take advantage of 00:07:17.840 --> 00:07:24.160 GPU which is why I wanted a motherboard in the first place that would support 00:07:21.840 --> 00:07:28.400 multiple gpus but the reason I would like to have a different power supply in 00:07:26.160 --> 00:07:35.000 here is actually more to do with redundancy um but finding a a mini 00:07:33.639 --> 00:07:39.520 redundant so that's what they're called a mini redundant power supply one in 00:07:36.919 --> 00:07:42.560 this form factor of sufficient wattage for this system with a couple of high 00:07:41.560 --> 00:07:48.800 power gpus actually not not easy um there was 00:07:46.039 --> 00:07:52.400 a 600 watt there may be something higher than that feel free to leave a message 00:07:50.240 --> 00:07:58.400 in the comments if you guys can can find something better than that but uh yeah 00:07:55.319 --> 00:08:00.120 I'm pretty much stuck with a single 00:07:58.400 --> 00:08:03.960 non-redundant power supply the good news is this is a very reliable power supply 00:08:02.240 --> 00:08:08.960 and this machine won't actually be storing anything Mission critical data 00:08:06.599 --> 00:08:12.840 goes onto another server gets ripped off of it and then transcoded over to 00:08:11.159 --> 00:08:19.240 another server by this one it never actually actually touches the data so um 00:08:17.240 --> 00:08:23.720 it should be fine it's just I would rather I would rather it was actually 00:08:21.919 --> 00:08:27.800 redundant and it'll be on a UPS as well so the power supply would have to fail 00:08:25.280 --> 00:08:33.120 to take it down even temporarily I guess it's all right all 00:08:30.000 --> 00:08:38.479 right in theory this boots now any time 00:08:33.120 --> 00:08:39.680 hey all right okay so here we are here's 00:08:38.479 --> 00:08:44.279 our 72 CPU threads only half of which are 00:08:43.360 --> 00:08:52.160 actual cores weak sauce I know and this my 00:08:48.480 --> 00:08:54.640 friends is the reason for some of the 00:08:52.160 --> 00:09:00.640 confusion around this particular system had a lot of people asking things like 00:08:57.720 --> 00:09:05.480 why do you need so many CPU cores to which I replied nothing because 00:09:03.920 --> 00:09:09.279 I wasn't really ready to talk about what the solution was yet the reason we need 00:09:07.519 --> 00:09:14.880 so many CPU corses we're going to be using it for transcoding and then doing 00:09:11.839 --> 00:09:17.040 the final outputs of our videos to make 00:09:14.880 --> 00:09:21.399 our videos easier to work on on our workstations without leg as we're 00:09:18.959 --> 00:09:26.720 scrubbing through the timeline and to make our exports more seamless of the 00:09:23.800 --> 00:09:32.880 finished videos and hopefully faster question number two 00:09:28.200 --> 00:09:36.440 lonus what aren't you running Linux on a 00:09:32.880 --> 00:09:39.720 server it's a server to which my reply 00:09:36.440 --> 00:09:41.440 is well gee I guess if you'd maybe asked 00:09:39.720 --> 00:09:46.560 what software we were using I might have been able to reply that the software 00:09:43.079 --> 00:09:49.920 we're using doesn't run on Linux and in 00:09:46.560 --> 00:09:52.279 fact there isn't a Linux version so 00:09:49.920 --> 00:09:57.720 we're going to be using soron squeeze server and how this works is pretty much 00:09:55.560 --> 00:10:03.240 it's that's not server this isn't server well where's server we are going going 00:09:59.560 --> 00:10:04.959 to be using Sorenson squeeze server and 00:10:03.240 --> 00:10:11.720 the way that works is we pretty much have a watch folder that we drop files 00:10:07.760 --> 00:10:13.000 into and it transcodes them for us so we 00:10:11.720 --> 00:10:17.600 uh had to do a fair bit of experimentation by we I mean Ed who's 00:10:15.839 --> 00:10:24.079 behind the camera had to do a fair bit of experimentation to find out exactly 00:10:20.959 --> 00:10:26.800 how much we could get out of our 36 CPU 00:10:24.079 --> 00:10:33.360 cores and if we did would it even do anything positive for our workflow so 00:10:29.360 --> 00:10:35.320 the exact workflow is not 100% finalized 00:10:33.360 --> 00:10:40.040 yet but what we do know is that if we use ENC code and watch folders and we 00:10:37.600 --> 00:10:44.720 set ourselves up a watch folder and a destination when we ingest footage off 00:10:42.959 --> 00:10:48.200 of our cards which will be over Thunderbolt so we'll actually be able to 00:10:46.279 --> 00:10:53.680 ingest very quickly even off multiple cards automatically our watch folder is 00:10:51.760 --> 00:10:57.760 going to take those and apply a particular preset to them so we can take 00:10:55.959 --> 00:11:05.519 a big whack of footage which we've actually got on the local disc here and 00:11:01.680 --> 00:11:08.600 then we are going to copy here and this 00:11:05.519 --> 00:11:10.600 will do two things so number one it will 00:11:08.600 --> 00:11:14.959 take the footage off of our SD cards which we would cut paste instead of copy 00:11:12.600 --> 00:11:20.680 pasting and then number two now that it's safely on our storage array it will 00:11:18.600 --> 00:11:24.800 take these and transcode them to a different easier to work with format and 00:11:23.279 --> 00:11:29.360 then boot them out somewhere else so in this case we are using another local 00:11:27.160 --> 00:11:32.800 folder which oops I just paused that which may end up being a bottleneck but 00:11:31.200 --> 00:11:36.839 we'll we'll do a quick experiment before this video is over to find out if our 10 00:11:34.880 --> 00:11:41.240 GB network is going to benefit us here what we're waiting for now as the files 00:11:38.920 --> 00:11:47.519 copy is we've set up our watch folder to monitor itself every 1 minute and then 00:11:45.279 --> 00:11:52.959 start processing so we can have a look at our current jobs nothing has happened 00:11:50.839 --> 00:11:58.639 yet but we'll know soon enough because our CPU is going to start going bananas 00:11:56.000 --> 00:12:02.519 here oh here we go here we go I think something's happening 00:12:00.120 --> 00:12:06.639 5% 7% CPU let's see if the uh the job is 00:12:04.959 --> 00:12:11.760 showing up yep there we go so you can see it's taking all these clips and 00:12:09.040 --> 00:12:14.760 importing them then transcoding them then it does something else and there's 00:12:13.079 --> 00:12:21.680 a lot of there's a lot of waiting involved but uh our CPU usage should 00:12:17.800 --> 00:12:23.760 start to slowly ramp up as more of them 00:12:21.680 --> 00:12:28.680 get past the importing stage and make it to the encoding stage excuse me 00:12:26.560 --> 00:12:31.160 transcoding which I guess is also encoding 00:12:31.279 --> 00:12:38.720 76 oh it's going up to 85% oh now it's 00:12:35.160 --> 00:12:41.320 working well what the Sam heck I guess 00:12:38.720 --> 00:12:46.320 it just takes a while to get all fired up or something look at it go we're 00:12:44.040 --> 00:12:52.040 actually getting a fairly reasonable turbo frequency out of this whole 00:12:48.079 --> 00:12:53.360 process like 2.7 plus GHz right now so 00:12:52.040 --> 00:12:59.639 there's actually a couple of different views we can get of this it looks like 00:12:56.720 --> 00:13:04.120 cpu0 is getting most of the love but we can actually change it to look at the 00:13:01.639 --> 00:13:11.480 two different num noes and you can see that CPU Z's utilization is higher but 00:13:07.800 --> 00:13:14.199 it's not that much uh well it is quite a 00:13:11.480 --> 00:13:18.639 bit higher so it is significantly higher than uh than node one and then we can 00:13:17.000 --> 00:13:24.040 also change the graph to overall utilization so you can see as it 00:13:21.199 --> 00:13:31.040 climbs really high CPU one kind of catches up to zero 00:13:28.000 --> 00:13:33.519 but it is it is sign significantly 00:13:31.040 --> 00:13:38.199 lower none of these jobs are finished they're all at the importing or 00:13:35.360 --> 00:13:44.720 transcoding stage still he's very interesting okay so 00:13:42.079 --> 00:13:49.639 we're only seeing like Max highest peak 70% CPU utilization and we realized that 00:13:47.600 --> 00:13:53.120 the last time we validated any of this it was actually in this server with one 00:13:52.160 --> 00:13:59.160 of the processors plugged into the motherboard 00:13:56.000 --> 00:14:01.720 so maybe the advantage that this one has 00:13:59.160 --> 00:14:06.560 has over the one we're using now is much much faster storage so if we can connect 00:14:04.360 --> 00:14:11.880 to this one over the network then we should be able to leverage that storage 00:14:09.040 --> 00:14:16.240 in theory so let's fire up the 24s SD server and see if uh using that as a 00:14:14.639 --> 00:14:23.079 Target helps 1 Gigabyte per second over the 00:14:21.040 --> 00:14:28.279 network and what's interesting about being able to see network utilization 00:14:25.440 --> 00:14:33.000 here is that unlike local dis utilization 00:14:29.720 --> 00:14:34.880 uh we we know where the hard cap is and 00:14:33.000 --> 00:14:39.160 it gives us a really clear idea of how much CPU utilization corresponds to how 00:14:36.959 --> 00:14:43.160 much Network utilization you can see network is actually not even changing in 00:14:41.639 --> 00:14:47.839 spite of the fact that CPU's gone from 13 to 00:14:44.880 --> 00:14:53.120 45% but what remains to be seen is whether our CPU usage will go higher 00:14:50.560 --> 00:14:57.680 than it was able to before with less of a storage bottleneck so like so many of 00:14:55.320 --> 00:15:02.519 our videos this one ends up actually raising more questions then it answers 00:15:00.160 --> 00:15:07.320 but hopefully we were at least able to answer the question of why we wanted 00:15:04.839 --> 00:15:11.600 such a high-powered server it's to enable a smoother workflow when multiple 00:15:09.880 --> 00:15:17.279 people are hitting the files at the same time and uh in order to have an easy 00:15:15.279 --> 00:15:22.120 quick way to replicate our data as soon as we ingest it onto our server so we 00:15:19.079 --> 00:15:24.120 can avoid the current workflow which is 00:15:22.120 --> 00:15:28.279 just leave it on the SD card just in case cuz there's only one copy on the 00:15:25.920 --> 00:15:33.680 server this way we'll actually have a copy on on one server and a copy on 00:15:31.440 --> 00:15:37.880 another one as soon as we've ingested the footage and it has gone through this 00:15:35.000 --> 00:15:42.720 transcoding process okay so we're back at 00:15:39.880 --> 00:15:45.880 the the current office for now but that doesn't mean I'm not going to take an 00:15:44.079 --> 00:15:51.560 opportunity to tell you guys about our sponsor today here and that is h e re 00:15:49.440 --> 00:15:55.639 not h e a r in spite of the fact that these things go in your ears is a pretty 00:15:53.920 --> 00:16:02.360 unique little campaign that's up on Kickstarter right now now to be clear 00:15:59.000 --> 00:16:04.600 these are not earbuds they're not uh a 00:16:02.360 --> 00:16:10.560 Bluetooth handset they're not any of those things what they are is they're a 00:16:06.720 --> 00:16:13.560 real world real time uh sort of audio 00:16:10.560 --> 00:16:15.759 effects applicator so you can adjust the 00:16:13.560 --> 00:16:20.440 volume of the real world you can change EQ settings of the real world you can go 00:16:18.800 --> 00:16:24.639 ahead and you know like say for example you're at a concert and you're like you 00:16:22.120 --> 00:16:28.680 know what this could use some Reverb you whip up the app on your phone and if all 00:16:26.600 --> 00:16:33.959 goes according to plan you'll be able to add add like a concert hall sound to 00:16:31.519 --> 00:16:37.040 that dude in the cubicle next to you if you so desire or you could do more 00:16:35.279 --> 00:16:41.839 useful things with it for example like tuning out a baby on an airplane or even 00:16:39.720 --> 00:16:45.399 the engine noise on an airplane simply by making adjustments through the app 00:16:43.639 --> 00:16:48.680 they're quoting a 6-hour battery life and it comes with a charging case that 00:16:46.759 --> 00:16:52.839 holds an additional two charges so over the course of your day I'm sure you'll 00:16:49.920 --> 00:16:57.240 come up with a lot of interesting effects application things to do to it 00:16:55.920 --> 00:17:03.480 now you might be asking yourself okay well hold on a second what about the latency they're quoting 30 micros 00:17:01.600 --> 00:17:07.799 between the microphone on the outside and the speaker on the inside taking 00:17:05.240 --> 00:17:10.360 noises changing them and then blasting them into your it doesn't have to be 00:17:09.000 --> 00:17:14.240 blasting because there's a volume control and then blasting them into your 00:17:12.319 --> 00:17:17.360 ears for you to hear so if you're curious to learn more or you want to 00:17:15.720 --> 00:17:22.199 support them check out the kickstarter page which is linked in the video 00:17:20.160 --> 00:17:27.480 description thanks for coming along for the ride you guys um if you disliked the 00:17:25.079 --> 00:17:30.720 video I think you know what to do but if you liked it then go ahead and press 00:17:29.080 --> 00:17:34.400 that like button helps us out a lot and get subscribed and all that good stuff 00:17:33.039 --> 00:17:38.200 if you really liked it then you can go ahead and check out the links in the video description where you can buy a 00:17:36.440 --> 00:17:41.360 t-shirt give us a monthly contribution or change your Amazon bookmark to one 00:17:39.799 --> 00:17:45.520 with our affiliate code that helps us out a lot and I guess that's pretty much 00:17:43.760 --> 00:17:48.760 it thanks again for watching and if you are looking for something else to watch 00:17:47.120 --> 00:17:53.480 after having done all that stuff then go ahead and click the little I in the top 00:17:51.160 --> 00:18:00.360 right corner and check out the video on Channel Super Fun that we did where uh a 00:17:57.919 --> 00:18:05.240 certain Teran van hemert and an electrical vehicle and some Thug Life 00:18:08.159 --> 00:18:13.880 involved I don't think that's how you hold it and you should probably have 00:18:16.880 --> 00:18:23.840 both yeah there you 00:18:19.960 --> 00:18:26.970 go I don't know what I'm doing 00:18:30.480 --> 00:18:33.480 he