1 00:00:00,000 --> 00:00:05,600 Super news everyone, NVIDIA just announced their RTX 40 Super series of GPUs and we are getting a 2 00:00:05,600 --> 00:00:11,120 super early look at the 4080 Super. I've got everything I need. This super screwdriver to 3 00:00:11,120 --> 00:00:15,680 unbox it and install it, this super computer to put it in, and this super laptop with all my 4 00:00:15,680 --> 00:00:21,600 notes about it, except I can't really read it from here. I have a super idea. Ah, much better. 5 00:00:21,600 --> 00:00:27,520 Let's take a look at this thing, shall we? At the top end, the main difference is not in the spec. 6 00:00:27,520 --> 00:00:33,200 For your CUDA cores, tensor cores, shader cores, and ray tracing cores, you are getting increases 7 00:00:33,200 --> 00:00:44,240 of 5.26, 7.17, 6.12, and 7.07% respectively. You're getting a 1.59% increase in boost clock 8 00:00:44,240 --> 00:00:51,840 speed and the same 16 gigs of GDDR6X memory. The big difference here is the price tag. It's no 9 00:00:51,840 --> 00:00:59,360 longer as super overpriced, dropping from $11.99 for the 4080 down to $9.99, meaning that if you 10 00:00:59,360 --> 00:01:06,000 want to run your games at 4K, high refresh rate on the latest monitors, this is basically going to 11 00:01:06,000 --> 00:01:10,160 be the card to power that. To demonstrate, we're going to get this card installed in the system, 12 00:01:10,160 --> 00:01:13,840 then we're going to play some games. But to do that, I have to get back to the other side of this 13 00:01:13,840 --> 00:01:21,520 table, which is a much more onerous process than you would probably imagine. While we walk, 14 00:01:21,520 --> 00:01:27,680 why don't I tell you about the 4070 Ti Super and the 4070 Super? For these cards, we're not really 15 00:01:27,680 --> 00:01:34,880 getting any pricing decreases. So they're still going to be at $7.99 and $5.99 respectively, 16 00:01:34,880 --> 00:01:41,920 but the difference is that we get much bigger bumps to the spec. So the 4070 Ti Super gets to be 17 00:01:41,920 --> 00:01:50,880 about 10% faster across the board and gets a bump from 12 gigs to 16 gigs of GDDR6x memory. 18 00:01:50,880 --> 00:01:58,320 As for the 4070, well, it doesn't get the memory bump, unfortunately, but it gets by far the biggest 19 00:01:58,320 --> 00:02:05,280 spec bump, increasing upwards of 20% across CUDA cores, Tensor Cores, Shader Cores, and Ray Tracing 20 00:02:05,280 --> 00:02:11,360 cores. One moment, please. Now all I got to do is put together my, I don't want to call it TSA 21 00:02:11,440 --> 00:02:15,520 safe because I think it's at their discretion, but I got it through Customs and Security. 22 00:02:15,520 --> 00:02:23,040 Screwdriver, get this bad boy installed. This is a fun fact. While the idle power consumption 23 00:02:23,040 --> 00:02:28,960 rated for this card goes up by two watts, the load power consumption actually goes down a little 24 00:02:28,960 --> 00:02:34,240 bit. And this is in spite of it being rated to run at a higher boost clock. So the only conclusion 25 00:02:34,240 --> 00:02:38,480 I can draw is that NVIDIA has managed to refine their manufacturing on these a little over the 26 00:02:38,560 --> 00:02:43,920 last year. Speaking of things that have needed some refinement, let's go ahead and get our power 27 00:02:43,920 --> 00:02:48,880 connector installed. This has been a bit of a thorn in their side, but my understanding is if 28 00:02:48,880 --> 00:02:53,440 you're using the first party NVIDIA adapters, there haven't been any reports of any kind of melting 29 00:02:53,440 --> 00:03:03,280 lately. Let's get this bad boy installed. And I don't know, bend it a lot with a heat gun. 30 00:03:03,280 --> 00:03:08,640 Don't bend it at all. Depends who you ask. Let's just plug it in. I have not personally 31 00:03:08,640 --> 00:03:13,520 witnessed any of these melting, but I mean, we've all seen the Reddit threads giving myself props 32 00:03:13,520 --> 00:03:23,760 for installing connectors properly. Yeah, cool. Let's fire it up. And yeah, we're in a different 33 00:03:23,760 --> 00:03:29,280 room. That system was running some kind of conversational AI and PC demo thing. This 34 00:03:29,280 --> 00:03:34,400 is Call of Duty Modern Warfare 3. No, not that Modern Warfare 3. The other Modern Warfare 3, 35 00:03:34,400 --> 00:03:42,320 the one that is actually difficult to drive at 4k, 240 frames per second. And yet, 36 00:03:42,320 --> 00:03:50,640 my friends, here we are. Is DLSS on? Why do you have DLSS on at this kind of performance level? 37 00:03:50,640 --> 00:03:55,200 I get it. NVIDIA wants to show off the performance benefits of DLSS and to anyone who doesn't cover 38 00:03:55,200 --> 00:03:59,600 the stuff for a living, the difference probably wouldn't have been immediately noticeable. In 39 00:03:59,600 --> 00:04:04,960 fact, it looks really, really good these days. But here's the thing about the 4080 Super. You 40 00:04:04,960 --> 00:04:12,560 don't really need it. I am still getting over 100 frames per second. And this, to me, is one of 41 00:04:12,560 --> 00:04:17,200 the most beautiful things about PC gaming. You can run the game however you want, rather than how 42 00:04:17,200 --> 00:04:23,440 the console manufacturer or the game developer decides you should run it. For me, oh man, 43 00:04:23,440 --> 00:04:28,640 I'm torn on this one because it's already really smooth at 100 frames per second. But on this 44 00:04:28,640 --> 00:04:38,240 monitor, this is the new Alienware 4k, 240Hz OLED. I'm really torn. I'm going to look at it again. 45 00:04:40,320 --> 00:04:47,360 Good. Lord, is that ever smooth. And unlike when you're using DLSS, you know, to go from a native 46 00:04:47,360 --> 00:04:55,360 35 to 55 or 60, you are much less likely to notice any kind of motion artifacts because 47 00:04:55,360 --> 00:05:00,080 the natural blur of your eyes makes them pretty hard to pick out. I don't even know who I'm fighting, 48 00:05:00,080 --> 00:05:05,040 but it's still cool. Probably that guy, I hope. What? How does he not did? I saw something. 49 00:05:05,040 --> 00:05:08,400 What? How many times do I have to shoot this guy? Oh, he's probably on my team then. No, 50 00:05:08,400 --> 00:05:11,920 wait, he's totally not on my team. Where is he? No, this guy's not on my team. How many times do 51 00:05:11,920 --> 00:05:16,160 you have to shoot this guy? Oh, brutal. What a violent game. I'm not going to let my kids play 52 00:05:16,160 --> 00:05:20,800 this game. It is definitely the mark of a good hardware demo that I just kind of lose myself 53 00:05:20,800 --> 00:05:24,320 and start playing video games. But I can't do that right now because I have something else 54 00:05:24,320 --> 00:05:28,480 that's really important to show you guys. This right here is the next evolution in the work that 55 00:05:28,480 --> 00:05:33,600 NVIDIA has been doing over the last 10 years on variable refresh rate gaming and, more recently, 56 00:05:33,600 --> 00:05:39,600 low persistence displays. So what we're looking at is a prototype display from a Seuss that runs at 57 00:05:39,600 --> 00:05:46,000 1440p 360 Hertz. But unlike the ones that we checked out recently with those OLED panels from 58 00:05:46,000 --> 00:05:54,800 Samsung display, this is an IPS, which raises the question of how can it possibly have motion 59 00:05:54,800 --> 00:06:00,160 clarity this good? Realistically, there's no way to show this to you guys without a full pursuit 60 00:06:00,160 --> 00:06:05,040 camera setup, and we didn't bring one to the show. So you'll have to trust me when I say that 61 00:06:05,120 --> 00:06:12,240 Turning the switch on and off goes from a blurry, smeary mess to clarity that I've 62 00:06:13,040 --> 00:06:18,640 rarely, if ever, seen on an LCD. And what's really wild is they've got the demo only running at 120 63 00:06:18,640 --> 00:06:23,760 Hertz with the GPU underclocked. I asked them if they can turn it up, and the results were really 64 00:06:23,760 --> 00:06:30,000 shocking. Oh, there we go. Now, compared to an OLED, you're still going to get a little bit of pixel 65 00:06:30,000 --> 00:06:35,920 overdrive artifacting, which manifests as kind of like a faint shadow that leads an object as it 66 00:06:35,920 --> 00:06:42,480 moves across the screen. But the main image has clarity, the likes of which I have never seen on 67 00:06:42,480 --> 00:06:50,880 an LCD. And to NVIDIA's credit, maybe darn close to those 360 Hertz OLEDs, which shouldn't even be 68 00:06:50,880 --> 00:06:56,320 possible with the difference in pixel response time. But here's the thing, some of the blurriness 69 00:06:56,320 --> 00:07:02,240 that you perceive in moving objects is actually because of your eye, and strobing the backlight 70 00:07:02,240 --> 00:07:08,480 helps to alleviate that. So until we can strobe the backlight of an OLED, which to my knowledge 71 00:07:08,480 --> 00:07:15,920 has not been combined with variable refresh rate at this time. No, yeah, you would have to drive an 72 00:07:15,920 --> 00:07:21,600 OLED much, much more faster than you would an LCD where you have a backlight that you can strobe. 73 00:07:21,600 --> 00:07:24,960 And the special sauce of getting this all to work together, variable refresh rate, 74 00:07:24,960 --> 00:07:29,280 with all the different, oh man, all the different overdrives that you would have to have for all 75 00:07:29,280 --> 00:07:35,360 the different color transitions. It's not ready yet, but it's going to be a contender. 76 00:07:35,360 --> 00:07:39,680 Lots of other cool stuff in here, but nothing as cool as getting subscribed to ShortCircuit. 77 00:07:39,680 --> 00:07:44,160 If you guys enjoyed this video, don't miss any of our other show coverage. I could be that guy 78 00:07:44,160 --> 00:07:47,840 and be like, oh, we did a really cool video in AMD's booth. I mean, they didn't have anything 79 00:07:47,840 --> 00:07:52,000 that really competes with you guys anyway.