1 00:00:00,240 --> 00:00:05,600 No Doom Vulcan benchmarks, eh? No DX12 or Vulcan. Oh, I see you guys followed 2 00:00:03,760 --> 00:00:09,519 NVIDIA's review guide to a te. Sad to see Linus Tech Tips benchmarking APIs, 3 00:00:07,759 --> 00:00:14,559 which is close to its end of life. What about RX480 Vulcan performance? Fine, 4 00:00:12,000 --> 00:00:18,480 let's talk about the new APIs, DirectX12 and Vulcan. How they work, how they 5 00:00:16,480 --> 00:00:29,439 perform, what they mean for the future, everything. 6 00:00:29,439 --> 00:00:35,520 FreshBooks is the super simple invoicing solution that lets you get organized, 7 00:00:33,200 --> 00:00:40,160 save time, and get paid faster. Click now to try it for free. Now, what is an 8 00:00:38,399 --> 00:00:44,239 API or application programming interface? In layman's terms, it's a 9 00:00:42,480 --> 00:00:48,559 facilitator. It handles requests similarly to a waiter at a restaurant 10 00:00:46,480 --> 00:00:54,160 taking orders. You can think of the customer as system A. System A wants to 11 00:00:51,520 --> 00:00:59,520 access some functionality or information that system B, which we'll call the 12 00:00:56,399 --> 00:01:01,440 kitchen, controls. Now, system B doesn't 13 00:00:59,520 --> 00:01:07,200 want to just let system A directly access all of its information. There are 14 00:01:04,159 --> 00:01:09,520 security and liability issues here. But 15 00:01:07,200 --> 00:01:14,520 not only that, having all these different customers trying to talk to 16 00:01:11,520 --> 00:01:16,960 the cook directly unless timed 17 00:01:14,520 --> 00:01:22,080 nearperfectly has the potential to be more confusing than helpful. Enter the 18 00:01:19,520 --> 00:01:26,479 waiter or in this analogy, the API. The waiter can serve the customer with a 19 00:01:23,840 --> 00:01:31,040 menu of subruine definitions, protocols, and tools that the customer can use to 20 00:01:28,560 --> 00:01:35,920 achieve his task. The customer then sends that request through the waiter to 21 00:01:33,119 --> 00:01:40,880 the kitchen where system B will act out those instructions and then send a 22 00:01:38,240 --> 00:01:46,960 response back to system A through the API. Clean, simple, and delicious. 23 00:01:44,880 --> 00:01:50,960 Theoretically, this is how it works, but they can be more bloated and therefore 24 00:01:49,200 --> 00:01:54,799 slow. Have you ever been to a restaurant that has like four different menus, one 25 00:01:53,200 --> 00:01:59,040 for just all the different kinds of freaking water with many pages 26 00:01:56,880 --> 00:02:03,520 containing a multitude of items which are essentially the same damn thing? 27 00:02:01,680 --> 00:02:08,160 Then you can relate to the frustration of not being able to find what you're 28 00:02:05,759 --> 00:02:11,520 looking for at all. Do you ever just want to throw the damn menus all in the 29 00:02:10,080 --> 00:02:16,239 air, stomp over to the kitchen, and scream, "Can you just render me a 30 00:02:13,680 --> 00:02:20,480 chicken burger and some yam fries? Is that too much to ask?" 31 00:02:18,640 --> 00:02:24,640 Okay, let's come back to that later. First, what are they used for? Lots of 32 00:02:22,640 --> 00:02:29,440 things. APIs can be used for web-based systems, operating systems, databases, 33 00:02:26,959 --> 00:02:33,599 software libraries, or in this case, computer hardware. But as important and 34 00:02:31,840 --> 00:02:38,319 ubiquitous as they are, they're not something the average user will ever 35 00:02:35,920 --> 00:02:43,360 interface with. So, why do we suddenly care so much about DirectX12 and Vulcan 36 00:02:41,280 --> 00:02:48,000 as selling points for games and video cards? Well, hype machine aside, 37 00:02:45,440 --> 00:02:52,879 DirectX11 and OpenGL, the predecessors to the shiny new graphics APIs, are 38 00:02:50,640 --> 00:02:57,400 rather bureaucratic. Like the complicated menu analogy, but worse 39 00:02:55,200 --> 00:03:02,480 because there are even more degrees of abstraction. There are many systems and 40 00:02:59,840 --> 00:03:08,480 devices talking to each other at a time, passing instructions around with many of 41 00:03:04,800 --> 00:03:10,720 them being redundant or outdated. It's 42 00:03:08,480 --> 00:03:15,680 just not as efficient. And there are other problems. For instance, one of 43 00:03:13,519 --> 00:03:21,440 your processor cores carries the burden of managing the vast majority of all of 44 00:03:19,040 --> 00:03:25,440 the critical time-sensitive tasks. One of the main ways that the new graphics 45 00:03:23,200 --> 00:03:30,319 APIs are more efficient is that they can use previously untapped hardware 46 00:03:27,680 --> 00:03:35,200 resources. Multi-core CPUs are a great example here. Look at this relatively 47 00:03:32,560 --> 00:03:41,040 realistic hypothetical scenario given to us by AMD. Thanks, guys. They show that 48 00:03:38,400 --> 00:03:46,879 OpenGL stacks all of the work of presenting to the user and processing 49 00:03:43,360 --> 00:03:49,680 the OpenGL driver onto core 1 along with 50 00:03:46,879 --> 00:03:54,799 a large portion of processing the OpenGL runtime and the largest portion of game 51 00:03:52,640 --> 00:03:59,040 code processing. This leaves little for cores 2 to four to work on, almost 52 00:03:56,720 --> 00:04:03,680 nothing for five and six to work on, and literally nothing for seven and 8. And 53 00:04:01,439 --> 00:04:06,959 for your reference, DirectX11 works in a very similar manner. Now, let's look at 54 00:04:05,519 --> 00:04:11,519 Vulcan. Though this is generally applicable to both of the new generation 55 00:04:09,040 --> 00:04:15,519 APIs, one of your processing cores is still responsible for a bigger part of 56 00:04:13,519 --> 00:04:20,400 the work than any of the others, but overall utilization is noticeably 57 00:04:18,079 --> 00:04:24,320 higher, even all the way down to core 8. Now, this is a hypothetical example, so 58 00:04:22,560 --> 00:04:28,800 I'm not going to spend time analyzing the exact improvements to the 59 00:04:26,400 --> 00:04:33,759 millisecond, but in a nutshell, Vulcan and Direct X12 do a much better job 60 00:04:31,440 --> 00:04:38,720 overall of sharing the load across all cores. Another key objective is to shed 61 00:04:36,240 --> 00:04:42,160 old layers of unneeded abstraction. They've done this by simplifying 62 00:04:40,240 --> 00:04:47,840 protocol routes, minimizing graphical driver overhead, focusing heavily on 63 00:04:44,880 --> 00:04:52,080 preventing timeheavy draw calls sent linearly to a single core of the CPU 64 00:04:50,240 --> 00:04:57,919 instead of parallelizing draw call packages in order to ease up on CPU load 65 00:04:55,120 --> 00:05:02,800 and let the GPU function as it should. Kind of like multiple waiters attending 66 00:05:00,160 --> 00:05:07,919 to the needs of each diner at a table instead of having one waiter with many 67 00:05:05,440 --> 00:05:12,479 plates on his ARM, dropping them off one at a time. Another key is that game 68 00:05:09,919 --> 00:05:17,440 developers can talk more directly to the GPU hardware. This is what is meant by 69 00:05:14,800 --> 00:05:22,240 the term lower level API. Think of it like having a manual transmission in a 70 00:05:19,680 --> 00:05:25,360 car instead of an automatic. But to all the gear heads out there watching this, 71 00:05:23,840 --> 00:05:30,320 don't just assume that all your DirectX11 and OpenGL games are outdated 72 00:05:28,000 --> 00:05:36,479 junk at this point. Bear in mind that poor operation of a manual transmission 73 00:05:33,120 --> 00:05:39,680 can be worse than an automatic or simply 74 00:05:36,479 --> 00:05:42,160 not better. Just because a game has 75 00:05:39,680 --> 00:05:47,680 Vulcan or DirectX12 support doesn't mean that efficiency and performance 76 00:05:44,759 --> 00:05:52,639 automatically improve. And this is true especially for games that were released 77 00:05:49,680 --> 00:05:58,160 with it early on. Though of course there are a few exceptions. For the best 78 00:05:55,120 --> 00:06:00,560 results, the new APIs need to be part of 79 00:05:58,160 --> 00:06:05,759 the core design of not only the game itself, but the underlying engine as 80 00:06:03,360 --> 00:06:12,080 well. And you do the math. It can take easily two, three, or even more years to 81 00:06:08,479 --> 00:06:13,680 make a good AAA title. And these APIs 82 00:06:12,080 --> 00:06:18,639 haven't even been available for that long. Meaning the implementation of 83 00:06:15,759 --> 00:06:24,000 these APIs was likely added to tick a marketing checkbox, appease a longtime 84 00:06:20,880 --> 00:06:25,919 partner, or hopefully, and very likely, 85 00:06:24,000 --> 00:06:30,880 just to gain the valuable experience that you can only get by working 86 00:06:27,919 --> 00:06:35,199 directly on them. But when more true DirectX12 and Vulcan titles drop, this 87 00:06:33,039 --> 00:06:39,520 level of control for developers has the potential to be awesome, but a little 88 00:06:37,520 --> 00:06:44,400 scary, too, because we're putting a lot of responsibility on the game and more 89 00:06:42,000 --> 00:06:49,120 specifically the game engine developers. And this is a major shift. So, while I 90 00:06:46,960 --> 00:06:54,479 just spent like four minutes telling you how interesting and important the new 91 00:06:51,039 --> 00:06:56,240 APIs are, in reality, the truly 92 00:06:54,479 --> 00:07:01,039 interesting component of the equation is going to be the new game engines that 93 00:06:58,319 --> 00:07:07,120 arise because of the freedoms afforded to them by the APIs, not necessarily the 94 00:07:04,720 --> 00:07:11,520 APIs themselves. And from game engine creators who care, the John Carmax and 95 00:07:09,360 --> 00:07:16,479 Tim Sweenies of the world, you'll get amazing new features unlocked by this 96 00:07:13,599 --> 00:07:21,120 additional performance and flexibility. But back to the scary bit from before, 97 00:07:19,199 --> 00:07:26,160 we're also putting this greater degree of control and the responsibility that 98 00:07:23,759 --> 00:07:32,000 comes with it in the hands of gaming companies who tell us things like 30fps 99 00:07:29,039 --> 00:07:36,639 is a good thing or that implement game physics effects that are reliant on the 100 00:07:34,479 --> 00:07:41,120 frame rate. That hasn't been a thing since like Intel 46, and that was 101 00:07:38,479 --> 00:07:45,360 arguable back then. That's inexcusable trash. Back to the APIs, though, since 102 00:07:43,120 --> 00:07:50,080 that's what we're really here for. So far, we've focused on similarities, but 103 00:07:47,440 --> 00:07:55,039 how do DX12 and Vulcan differ from each other? Let's start with DirectX12. A big 104 00:07:52,240 --> 00:07:59,440 focus for Microsoft this time around has been the introduction of their new LDA 105 00:07:57,120 --> 00:08:04,560 and MDA modes for multiple graphics cards. For a rundown on how those work, 106 00:08:02,240 --> 00:08:09,599 check out this video on WTF is going on with SLI. Along with this, Microsoft's 107 00:08:07,120 --> 00:08:14,800 vision to allow developers to support a mixture of graphics card models and even 108 00:08:12,240 --> 00:08:19,120 brands so the user can get as much power as possible out of their available 109 00:08:16,639 --> 00:08:23,680 hardware. Think of this more in terms of using your onboard graphics to get a 110 00:08:21,360 --> 00:08:29,039 little bit more oomph than a compelling reason to build this monstrosity. One of 111 00:08:26,960 --> 00:08:34,159 the ways this could work is split frame rendering or SFR. This is when a portion 112 00:08:31,840 --> 00:08:38,399 of the screen is rendered by one GPU and the rest is rendered by a second. But 113 00:08:36,240 --> 00:08:42,959 we'll have to wait and see how well this performs in the real world. So that's 114 00:08:41,039 --> 00:08:48,640 cool. But being a Microsoft product, DirectX12 will only function on Windows 115 00:08:45,519 --> 00:08:50,880 10 and Xbox. So Linux, OSX, Android 116 00:08:48,640 --> 00:08:55,120 enthusiasts, and actually even people who are still on Windows 7 and Windows 8 117 00:08:52,800 --> 00:08:59,200 are left out in the cold. Which brings us to Vulcan. Vulcan, brought to us by 118 00:08:57,360 --> 00:09:04,399 the Kronos Group, is the primary successor to OpenGL and is proudly 119 00:09:02,200 --> 00:09:09,360 cross-platform, working on everything we just mentioned and more. This is a huge 120 00:09:07,279 --> 00:09:14,320 deal for Steam OS in particular and Linux in general because it should bring 121 00:09:11,920 --> 00:09:20,240 with it a stronger support for a wider range of titles, something Linux has 122 00:09:16,480 --> 00:09:22,320 struggled with ever since, well, ever. 123 00:09:20,240 --> 00:09:27,920 So, Vulcan is a big deal. Google has been using it as Android's low-level API 124 00:09:24,480 --> 00:09:30,240 since 2015. And Dan Ginsburgg from Valve 125 00:09:27,920 --> 00:09:34,720 has talked about it on stage and said that Vulcan is the future. Although 126 00:09:32,959 --> 00:09:40,959 that's not surprising considering DirectX12 doesn't work on Steam OS and 127 00:09:37,920 --> 00:09:43,279 Valve Microsoft relationshipness has 128 00:09:40,959 --> 00:09:47,600 been getting a little tense lately. Moving on, there was one last thing that 129 00:09:45,600 --> 00:09:51,560 they had in common. Asynchronous compute. Now, this has been a highly 130 00:09:49,440 --> 00:09:56,720 controversial subject, making it unfortunately outside the scope of this 131 00:09:53,760 --> 00:10:00,320 video. But if you guys want to see a similar video dedicated to it, let me 132 00:09:58,800 --> 00:10:04,240 know in the comments down below. In a nutshell, though, it allows additional 133 00:10:02,080 --> 00:10:08,240 lightweight work to run in parallel alongside the main graphics thread. So, 134 00:10:06,320 --> 00:10:13,600 a specific lighting technique or post-process anti-aliasing method like 135 00:10:11,000 --> 00:10:18,160 TSSA. Now, this dynamic scheduling introduces some challenges for NVIDIA 136 00:10:15,680 --> 00:10:23,440 and AMD that did not exist in a more static ecosystem, but that's for them to 137 00:10:20,880 --> 00:10:28,320 worry about and for me to maybe cover in another video. That being said, it does 138 00:10:25,839 --> 00:10:32,720 seem that AMD has been pushing harder than NVIDIA, and it shows 139 00:10:30,279 --> 00:10:38,160 performance-wise as AMD is currently seeing more of a benefit than NVIDIA in 140 00:10:35,440 --> 00:10:43,200 the 3D Mark Time Spy benchmark, which allows you to switch asynchronous on or 141 00:10:40,800 --> 00:10:47,200 off. When reviewing the numbers for these APIs, you'll notice that they're 142 00:10:45,040 --> 00:10:51,760 kind of all over the place. Looking at Rise of the Tomb Raider, a DirectX game, 143 00:10:49,839 --> 00:10:57,760 we can see that while it does make a significant improvement at 1080p, once 144 00:10:54,480 --> 00:11:00,320 you step up a bit to 1440p or even 4K, 145 00:10:57,760 --> 00:11:04,079 things seem to fall apart a little bit, and 146 00:11:01,160 --> 00:11:07,920 DirectX11 actually pulls ahead. Hitman, on the other hand, did not share this 147 00:11:05,760 --> 00:11:12,000 funky scaling pattern and seemed to improve when using DirectX12, or at the 148 00:11:09,680 --> 00:11:16,160 very least stay the same across all the graphics cards I tested it with, which 149 00:11:13,760 --> 00:11:22,560 should represent GP 104 with the GTX 1080, GP 106 with the GTX 1060 6 GB 150 00:11:19,760 --> 00:11:27,920 edition for the NVIDIA side, and the power of AMD's new Polaris architecture 151 00:11:24,880 --> 00:11:30,000 with the RX 480. Moving on to Ashes of 152 00:11:27,920 --> 00:11:34,880 the Singularity. This game is like half benchmark, half game. So, it shows the 153 00:11:32,720 --> 00:11:39,760 biggest improvement by far when using DirectX12 instead of DirectX11. And for 154 00:11:37,839 --> 00:11:43,680 the Vulcan fans out there, I've heard there will be a patch coming for you as 155 00:11:41,519 --> 00:11:48,640 well. One great additional piece of insight that Ash's displays at the end 156 00:11:45,760 --> 00:11:53,360 of their benchmark is CPU frames versus GPU frames. What this essentially means 157 00:11:51,279 --> 00:11:58,720 is how many frames your CPU could potentially push compared to what your 158 00:11:55,800 --> 00:12:03,360 GPU actually pushed. So, if the CPU number is higher, you could get more 159 00:12:00,959 --> 00:12:06,800 overall FPS with a graphics card upgrade. Notice that when we run the 160 00:12:05,040 --> 00:12:11,279 more CPU focused version of the benchmark, the numbers are at par 161 00:12:08,959 --> 00:12:17,760 because now you're intentionally CPU bottlenecked. Moving on to Vulcan. Doom 162 00:12:14,720 --> 00:12:20,320 numbers for the NVIDIA side are just a 163 00:12:17,760 --> 00:12:26,240 mess. For the GTX 1080, Vulcan was similar to Tomb Raider worse at 4K than 164 00:12:23,120 --> 00:12:27,680 OpenGL 4.5. dropping down to 1080p, 165 00:12:26,240 --> 00:12:32,240 however, and it's a whole different story with massive improvements shown. 166 00:12:30,000 --> 00:12:37,440 The RX 480 does show the performance trend that we would expect, improving 167 00:12:34,880 --> 00:12:41,839 considerably when running the Vulcan API and utilizing asynchronous compute 168 00:12:39,440 --> 00:12:46,639 shaders. Good. So, the features sound pretty great. The performance, depending 169 00:12:44,160 --> 00:12:52,240 on your configuration, is probably good, but might not be great. What about 170 00:12:49,440 --> 00:12:56,480 supported games? On the DirectX12 side of things, the list seems rather 171 00:12:54,079 --> 00:13:00,800 populated and there is a fair number of titles on the way that are claiming they 172 00:12:58,639 --> 00:13:06,800 will support it. But this doesn't mean that DirectX12 integration hasn't been 173 00:13:03,519 --> 00:13:09,680 riddled with issues. Quantum Break was a 174 00:13:06,800 --> 00:13:13,839 mess. Tomb Raider's backend barely even functioned for a while to the point 175 00:13:11,519 --> 00:13:19,440 where they added a warning if you enable it. and the upcoming DSX Mankind 176 00:13:17,040 --> 00:13:23,279 Divided, which I think is out now. It was supposed to launch with support for 177 00:13:21,279 --> 00:13:26,959 it, and that's been pulled for a while, so they can fix it up. And Vulcan side 178 00:13:25,279 --> 00:13:31,360 of things is rough, too. Their support list has four items on it. One of which 179 00:13:29,519 --> 00:13:37,320 is upcoming. One of which is Dota Freaking 2. Not sure about you, but I 180 00:13:34,399 --> 00:13:42,480 sure needed more FPS in Dota on my DirectX12 capable hardware. Another is 181 00:13:40,560 --> 00:13:47,680 the rather wonderful single player game from 2014. Possibly not that many 182 00:13:45,360 --> 00:13:52,160 current players, unfortunately. And then lastly, we have Doom. A beautiful 183 00:13:50,079 --> 00:13:57,360 looking game with proper built-in support and asynchronous compute 184 00:13:54,160 --> 00:13:59,839 capabilities. Cool, but that's singular. 185 00:13:57,360 --> 00:14:03,839 So there you go. In conclusion, the future is bright with physics and AI 186 00:14:02,000 --> 00:14:08,079 heavy games being among the most exciting things we have to look forward 187 00:14:05,279 --> 00:14:12,160 to, which is awesome. But the present is more a light at the end of the tunnel 188 00:14:09,839 --> 00:14:16,720 situation. Get excited for new game engines that support these APIs and be 189 00:14:14,240 --> 00:14:23,440 on the lookout for awesome new game engine technology that will likely arise 190 00:14:20,079 --> 00:14:26,079 from the improvements made here. This 191 00:14:23,440 --> 00:14:31,120 may finally signal the return of our old cores for gaming episodes. Maybe we'll 192 00:14:28,720 --> 00:14:37,600 finally have an episode that doesn't end in four cores is the best. Because by 193 00:14:34,480 --> 00:14:41,320 that time, your $1,700 10 core Extreme 194 00:14:37,600 --> 00:14:43,639 Edition will finally matter. 195 00:14:41,320 --> 00:14:48,959 Maybe. Today, we're highlighting the K7XX limited edition Ruby Red 196 00:14:46,240 --> 00:14:53,279 headphones, of course, from Massdrop. They also have a bunch of other cool 197 00:14:50,959 --> 00:14:57,199 things that you can buy like with other people so that the price goes down and 198 00:14:55,519 --> 00:15:00,959 if more people buy it, the price goes down further till a logical minimum. And 199 00:14:59,519 --> 00:15:04,959 it it's it's really cool. You should check out Massrop regardless. The 200 00:15:02,800 --> 00:15:09,600 products we're showcasing today is the same specwise as the K7XX headphones 201 00:15:08,079 --> 00:15:14,000 that Lionus reviewed and you guys all liked sometime last year. You can check 202 00:15:11,360 --> 00:15:16,800 that video out here or all over my I don't know where it's going to be. The 203 00:15:15,440 --> 00:15:21,839 real difference, however, is that this run uses red accents on the ear cups and 204 00:15:19,880 --> 00:15:25,760 headband. Remember that this is a limited drop, so if you want a pair, 205 00:15:23,920 --> 00:15:31,199 you're going to have to act relatively fast. These headphones were configured 206 00:15:28,240 --> 00:15:35,440 by Massdrop and manufactured by AKG. Just a note for international orders, if 207 00:15:32,959 --> 00:15:39,360 you're outside of the US, there's a $25 shipping and handling fee put on top. 208 00:15:37,199 --> 00:15:43,920 And that's it. Check out Massdrop. Thanks for watching, guys. If this video 209 00:15:41,959 --> 00:15:47,360 sucked, but if you liked it, you know what to do. So, hit the like button, get 210 00:15:45,360 --> 00:15:49,680 subscribed, do all that kind of stuff. Uh, check out the link in the video 211 00:15:48,560 --> 00:15:53,759 description down below where you can buy like GPUs or something. I guess we talked about those in this video. Also, 212 00:15:52,800 --> 00:16:00,720 check out the link in the video description down below to buy a shirt and go to the forum to talk about 213 00:15:58,000 --> 00:16:04,639 everything I said in this video. There's might be something wrong. File the crap 214 00:16:03,120 --> 00:16:09,920 out of me for that and then I'll learn and that'll be good. 215 00:16:06,839 --> 00:16:12,160 But, man, anyways, watch this video 216 00:16:09,920 --> 00:16:15,360 which is about WTF is going with SLI because we're making a series of these 217 00:16:13,360 --> 00:16:19,680 now. Although continuing to do these might kill me, so I don't know.