WEBVTT 00:00:00.480 --> 00:00:09.280 At long last, here is my performance review for the GTX 570, which I never 00:00:05.680 --> 00:00:11.519 did one for, as well as AMD's new Radeon 00:00:09.280 --> 00:00:16.080 6T900 series graphics card. So, that's the 6T70, which I have on my test bench 00:00:14.320 --> 00:00:22.400 right now. I'm running Battlefield Bad Company 2 for my uh my sort of pre 00:00:20.080 --> 00:00:26.640 premeditated run through that I do on all of the cards to get you guys some uh 00:00:24.640 --> 00:00:32.880 Fraps recorded results. So, I have the 6T970. I have the 6T950 here. I've also 00:00:30.320 --> 00:00:40.320 got on the GeForce side, the green side, the GTX 570 as well as the GTX 580. And 00:00:36.640 --> 00:00:44.559 then I've included in my graphs the 6870 00:00:40.320 --> 00:00:47.280 as well as a Hawk Edition GTX 460. So, 00:00:44.559 --> 00:00:54.960 that should give you guys some uh some sort of a a measuring stick against a 00:00:52.000 --> 00:01:00.399 performance series card. So, like a 460 or a 6870 versus one of these enthusiast 00:00:57.920 --> 00:01:06.080 grade cards, including the sixtyn00 series as well as the GTX 500 series. 00:01:04.239 --> 00:01:11.680 So, the games that I'm going to be showing today are Metro 2033, 00:01:08.720 --> 00:01:17.439 Battlefield Bad Company 2, uh 3D Mark 11, which is not a game 00:01:13.840 --> 00:01:20.799 technically, as well as, uh the Heaven 00:01:17.439 --> 00:01:23.840 benchmark, and I will be running all of 00:01:20.799 --> 00:01:26.159 them at 1080p. Oh, yeah. Crisis 2. and 00:01:23.840 --> 00:01:32.240 running at fairly high detail settings. That means I have enabled anti-aliasing 00:01:28.080 --> 00:01:33.840 in all tests. And the idea was to really 00:01:32.240 --> 00:01:40.400 stress these cards. So, at the settings I was running in Metro 2033, actually 00:01:36.240 --> 00:01:42.880 both the 6870 and the GTX 460 uh are 00:01:40.400 --> 00:01:46.880 were what I considered a fail. Um like it was 14 frames per second. I could 00:01:45.280 --> 00:01:52.320 measure frames per second, but it was just too choppy, completely unplayable. 00:01:49.280 --> 00:01:55.200 So anyway, stay tuned for boring charts 00:01:52.320 --> 00:01:59.520 and graphs and uh other things coming soon. So other than straight 00:01:57.759 --> 00:02:03.600 performance, let's talk a bit of a feature comparison between the green 00:02:01.840 --> 00:02:07.520 team and the red team for this particular generation of products. So 00:02:05.759 --> 00:02:12.000 there's a few different things that are similar. First of all, these are both 00:02:09.440 --> 00:02:16.959 direct X11 graphics cards with beefy tessellation performance, support for 00:02:14.160 --> 00:02:21.440 all the latest standards and all of that noise and hoo-ha. Okay, so they are both 00:02:19.520 --> 00:02:26.640 feature complete. Now, we have had times in the past when one company actually 00:02:23.840 --> 00:02:31.599 has a more advanced product than the other. For example, AMD was ahead of 00:02:28.879 --> 00:02:35.840 NVIDIA for DirectX11 and NVIDIA was way ahead of AMD for DirectX10, but we're 00:02:34.000 --> 00:02:40.080 not in a situation like that right now. So, the feature differences that we 00:02:37.840 --> 00:02:44.800 actually run into are some different things. So, first of all, AMD has their 00:02:42.319 --> 00:02:48.879 iinity technology. You can see they have a good number of outputs on the back of 00:02:46.800 --> 00:02:55.920 their card. two DVI, two mini DisplayPort, one mini rather, one HDMI, and 00:02:53.440 --> 00:03:00.959 that is what enables AMD as well as some special configuration of the card to run 00:02:57.599 --> 00:03:02.720 up to four displays off of a single card 00:03:00.959 --> 00:03:06.480 with affinity. So that means you could do three displays with your crosshair in 00:03:04.959 --> 00:03:10.319 the middle middle peripheral vision and then you could have like your uh your 00:03:08.560 --> 00:03:14.239 chat on another screen above it or whatever you want to do or you can even 00:03:12.159 --> 00:03:20.000 run just three displays off one card for triple HD resolution off of one card. 00:03:17.280 --> 00:03:26.159 Now to counter that NVIDIA has their NVIDIA surround which isn't quite there 00:03:24.319 --> 00:03:30.879 compared to iffinity because you do need two cards. So, if you run SLI on any 00:03:29.120 --> 00:03:35.680 supported GPU, which includes anything back to, as far as I know, the GTX 200 00:03:33.280 --> 00:03:39.599 series, you can run two monitors off the top card, one monitor off the bottom 00:03:37.200 --> 00:03:43.840 card, and you can get surround gaming, but it doesn't run off one card. Now, 00:03:41.760 --> 00:03:48.159 NVIDIA also has a couple other features, including their 3D vision. Okay, so I 00:03:46.799 --> 00:03:52.000 have the glasses up here, and this is actually a 3D vision ready monitor, so 00:03:49.840 --> 00:03:59.519 you can play in stereoscopic 3D. Mind you, AMD also has their competing HD 3D 00:03:55.040 --> 00:04:00.879 now. So, while you have to 00:03:59.519 --> 00:04:06.720 bear in mind that both of these approaches are slightly different, with NVIDIA, you're bound to certain 00:04:04.799 --> 00:04:10.959 standardized components, including the glasses and the displays. The AMD 00:04:09.439 --> 00:04:14.080 approach is a little bit different in that you are not bound to a standard. 00:04:12.879 --> 00:04:18.799 There are a bunch of different ways to go about it, but not all of them are 00:04:16.239 --> 00:04:23.120 created equal. Finally, last but not least, we have CUDA and PhysX on the 00:04:20.560 --> 00:04:27.680 NVIDIA side, as well as full support for direct compute on the AMD side. So, 00:04:26.000 --> 00:04:31.440 these are both a couple of competing standards, and it really remains to be 00:04:29.759 --> 00:04:36.160 seen which one is going to emerge as victorious. So, you kind of have to uh 00:04:33.840 --> 00:04:40.080 pick a road and walk down it at this point and hope you made the right 00:04:37.520 --> 00:04:44.560 choice. So, that is my feature summary. They both Oh, yeah, of course. They both 00:04:41.840 --> 00:04:48.560 support uh SLI, Crossfire, multiGPU configurations. Well, they don't support 00:04:46.080 --> 00:04:51.759 both support SLI. This one supports SLI. And they don't both support Crossfire. 00:04:50.160 --> 00:04:56.720 Only this one does, but they both support uh two-way and 3-way GPU 00:04:54.160 --> 00:05:00.320 configurations except on their performance grade cards. So, that is on 00:04:58.720 --> 00:05:04.240 the enthusiast grade cards, support up to 3-way. But you can see both of the 00:05:02.560 --> 00:05:11.199 performance grade cards I have here, that is the GTX 460 as well as the 00:05:07.280 --> 00:05:12.960 Radeon 6870 only have a single multiGPU 00:05:11.199 --> 00:05:18.240 connector each. So, those ones only support two-way multi-GPU 00:05:15.759 --> 00:05:22.639 configurations. Graphs to come. Crazy Russian also 00:05:20.800 --> 00:05:27.600 brought to my attention that the Radeon 6000 series also supports adaptive 00:05:24.880 --> 00:05:33.440 anti-aliasing. And I missed one feature that they both have, and that is support 00:05:29.680 --> 00:05:36.240 for HDMI 1.4A, which enables you to play 00:05:33.440 --> 00:05:39.639 back 3D Blu-ray with supported software and display. 00:05:54.080 --> 00:06:00.000 So, in conclusion, I'm not really going to declare there to be a winner because 00:05:58.479 --> 00:06:05.360 I think it's a bit of a mistake for graphics card reviews. Um because what 00:06:03.280 --> 00:06:09.919 happens is these guys are both so competitive, AMD and NVIDIA, that as 00:06:07.600 --> 00:06:13.120 soon as the market condition changes, they'll both be adjusting their 00:06:11.440 --> 00:06:17.759 strategies and their pricing structures in order to compete better with each 00:06:15.199 --> 00:06:21.199 other. So all you really need to do is look at the performance. And I do 00:06:19.520 --> 00:06:25.120 recommend checking out written reviews in addition to any video reviews you 00:06:22.960 --> 00:06:28.800 might see because quite frankly a lot of the written reviews have a lot more time 00:06:27.280 --> 00:06:33.520 spent on crunching the numbers and getting into the nitty-gritty of the 00:06:30.960 --> 00:06:37.520 technology. So check out for example the review on hardware that is 00:06:35.039 --> 00:06:43.280 www.hardkconnects.com. The GPU review guy over there is a total 00:06:40.000 --> 00:06:44.720 guru. So please do check them out. And 00:06:43.280 --> 00:06:50.560 so, yeah, I'm not going to declare a winner because what you really have to do is whether you're watching this video 00:06:48.000 --> 00:06:54.639 right now or watching it 6 months down the road, you can look at these cards 00:06:52.800 --> 00:06:58.319 and compare them in terms of their performance, but you also have to bear 00:06:56.319 --> 00:07:01.840 in mind the price. So, you'll have to just see what are the prices at that 00:07:00.000 --> 00:07:06.720 time and how do they stack up against each other when you factor in the bang 00:07:04.000 --> 00:07:10.960 for the buck aspect of the equation. So, thanks for checking out my Linus Tech 00:07:08.160 --> 00:07:15.680 Tips review of the GTX 570 as well as the Radeon 6900 00:07:13.919 --> 00:07:20.479 series. And don't forget to subscribe to Linus Tech Tips for more unboxings, 00:07:17.199 --> 00:07:20.479 reviews, and other videos.