WEBVTT 00:00:00.960 --> 00:00:09.280 Today I will be reviewing my NVIDIA pen. It declares that you should insist 00:00:06.799 --> 00:00:13.120 on NVIDIA and it also has some other pretty good stuff on it. First of all, 00:00:11.040 --> 00:00:16.960 NVIDIA is the number one choice in PCs and consumer electronics. Also, all 00:00:15.120 --> 00:00:20.800 GeForce GPUs are Vista certified. Shows you how old this pen is. And it is the 00:00:18.960 --> 00:00:26.119 number one choice for extreme gamers. And NVIDIA defined the GPU category, or 00:00:23.519 --> 00:00:30.000 at least they acquired whoever did. Neat. Actually, that's not what 00:00:28.800 --> 00:00:34.880 we're doing today. Today, we're going to be having a look at the GeForce GTX 590. 00:00:32.880 --> 00:00:39.360 This is going to be only one of my performance review videos of this 00:00:37.120 --> 00:00:44.879 particular card. I'm going to do a few. One of them is going to be all about 3D 00:00:42.719 --> 00:00:49.600 3D Vision. So, I'll only be comparing against other GeForce cards in that one. 00:00:47.200 --> 00:00:56.079 I am going to be doing one heavyweight showdown between the GTX 590 and the 00:00:52.719 --> 00:00:59.120 Radeon HD 6T990. And that will actually 00:00:56.079 --> 00:01:01.320 be at a different resolution. So, that 00:00:59.120 --> 00:01:07.680 one's going to be at 2560 x600 with this beast of a monitor right 00:01:04.920 --> 00:01:11.920 here. Okay. So, that's my my 30-in monitor. But this one right here that 00:01:10.000 --> 00:01:18.159 you're watching is going to be all about 1080p performance. Now, the GTX 590 and 00:01:15.280 --> 00:01:21.840 the 6T90 for that matter are both a little bit overpowered for 1080p. So, 00:01:20.240 --> 00:01:26.159 what you're going to see is that they will dominate the performance charts in 00:01:24.080 --> 00:01:29.840 all likelihood, but we're not going to see them really stretch their wings 00:01:27.600 --> 00:01:37.560 until you start to get into things like 3D, for example, as well as uh as well 00:01:33.439 --> 00:01:41.360 as um HD plus res resolutions like 2560 00:01:37.560 --> 00:01:42.960 x600, what that 30-in monitor uses. So, 00:01:41.360 --> 00:01:47.600 uh I'm going to talk a little bit about my testing methodology. For most of 00:01:44.960 --> 00:01:51.799 these games, I'm using 4x anti-aliasing at the max. I am running all of the 00:01:49.600 --> 00:01:56.320 games at extreme maximum settings except for Crisis. I turn motion blur to medium 00:01:54.799 --> 00:02:01.960 because it bothers me when it's any higher than that and I don't like playing the game that way. So, I just 00:01:59.680 --> 00:02:05.280 turn it off. Um, yeah, I think that's all there is to 00:02:04.159 --> 00:02:11.280 say. So, we're going to be looking at Hawk 2, Lost Planet 2, Metro 2033. So, 00:02:08.879 --> 00:02:15.440 that's uh Civilization 5. So, those are some DirectX11 titles for us. We've got 00:02:13.360 --> 00:02:19.920 Battlefield Bad Company 2, Far Cry 2, and Crisis to have a look at some DX10 00:02:18.000 --> 00:02:24.080 gaming. Any one of these GPUs is going to run any DirectX9 title. Great. So, I 00:02:22.959 --> 00:02:28.080 personally don't find it all that relevant. We're going to be looking at 00:02:26.160 --> 00:02:32.640 nextG features like tessellation more than anything else. All of those 00:02:29.840 --> 00:02:37.120 DirectX11 titles do support tessillation in some way, shape, or form. So, that's 00:02:34.640 --> 00:02:40.319 going to be a an emphasis of what we are doing today. So, in terms of the 00:02:38.640 --> 00:02:44.000 configurations we're looking at, this is going to be extreme performance stuff 00:02:42.319 --> 00:02:49.120 only. We're not going to be looking at any dinky stuff. So, we've got the 590, 00:02:46.640 --> 00:02:55.440 the Radeon 6990. These are the two flagship uh $700 MSRP cards from NVIDIA 00:02:53.440 --> 00:03:00.640 and AMD, respectively. We're also going to be looking at a GTX 580 SLI, which is 00:02:58.800 --> 00:03:06.000 another solution that gives you extreme performance on dual cards rather than on 00:03:02.959 --> 00:03:08.040 a single card because the two GTX 580s 00:03:06.000 --> 00:03:14.159 are clocked higher than the GPUs on the 590, but 00:03:11.840 --> 00:03:19.360 they have the drawback of consuming more power, kicking off more heat, and taking 00:03:17.040 --> 00:03:26.519 up four slots in your computer rather than only two. Okay, so there's that. 00:03:21.760 --> 00:03:30.239 Uh, and then we've also got uh a single 00:03:26.519 --> 00:03:34.000 6T970, a single GTX 580, and then I've 00:03:30.239 --> 00:03:35.680 also thrown in GTX 560Ti SLI. So, these 00:03:34.000 --> 00:03:40.239 two cards don't appear to match, but actually they do. And that is it's just 00:03:38.239 --> 00:03:45.280 an interesting configuration because it's a pretty good value for a high-end 00:03:42.720 --> 00:03:50.360 gaming system. and I just wanted to see how it stacks up against what is 00:03:47.680 --> 00:03:57.840 essentially a downclocked GF10 dual GPU solution versus a fully 00:03:54.000 --> 00:04:00.560 clocked uh GTX 560 Ti dual GPU solution. 00:03:57.840 --> 00:04:02.720 So uh stay tuned guys for charts and graphs. 00:04:37.040 --> 00:04:43.360 So, one of the things I did not mention in my introduction was my test bench. I 00:04:41.120 --> 00:04:50.160 should probably cover that. So, I've got a Core i7 2600 K Sandy Bridge processor 00:04:46.560 --> 00:04:53.600 at 4.7 GHz. I've got 8 gigs of Mushkin 00:04:50.160 --> 00:04:56.880 Redline DDR3600 RAM. I've got an ASRock 00:04:53.600 --> 00:05:00.960 Fatality motherboard. I've got an Intel 00:04:56.880 --> 00:05:02.400 um 510 series 120 gig SSD and that's 00:05:00.960 --> 00:05:07.600 pretty much it for the relevant performance related stuff. So, let's 00:05:05.680 --> 00:05:13.039 talk the overall results. So, I'll go through the games individually. Hawk 2 00:05:09.759 --> 00:05:15.600 scales great. Um the CPU is able to 00:05:13.039 --> 00:05:19.520 allow pretty much any of these cards to reach their full potential and shows us 00:05:17.759 --> 00:05:24.320 how they all perform relative to each other to an extent. The Radeon cards 00:05:21.919 --> 00:05:29.120 don't perform nearly as well in Hawks 2 as the GeForce cards. And that's an 00:05:27.120 --> 00:05:34.919 architectural thing. Could be driver related. Either way, they get left in 00:05:32.000 --> 00:05:43.039 the dust. They get creamed. Even the 6990 loses to GTX 560 SLI and a single 00:05:40.000 --> 00:05:44.400 GTX 580. So, that's uh that's kind of an 00:05:43.039 --> 00:05:48.400 interesting thing about that. But, it does allow us to show the relative 00:05:46.560 --> 00:05:52.479 performance of the GeForce solutions against each other at the very least. 00:05:50.240 --> 00:05:56.400 Lost Planet 2 is another one. It's a very demanding game. We are not CPU 00:05:54.560 --> 00:06:01.280 limited in any way in this particular title. So, each of the GPUs is able to 00:05:59.280 --> 00:06:06.560 really stretch their wings and show us what it is capable of. So, you see the 00:06:03.199 --> 00:06:09.600 590 come out ahead of the 6990 in this 00:06:06.560 --> 00:06:12.039 particular game as well. Metro 2033 is 00:06:09.600 --> 00:06:18.720 another example of a game that all of the GPUs are able to perform 00:06:15.120 --> 00:06:20.880 quite well in. So, the 580 SLI um 00:06:18.720 --> 00:06:26.880 actually performs about the same as the 6T990, which means that the GTX 590, 00:06:23.520 --> 00:06:29.520 which doesn't perform as well as 580 SLI 00:06:26.880 --> 00:06:34.880 due to the lower clock speed, actually falls behind a little bit in that title. 00:06:33.199 --> 00:06:39.280 That's the first one, so I can't say as well. Battlefield Bad Company 2 is 00:06:37.520 --> 00:06:44.560 another one where we actually see pretty good scaling. And we also see fairly 00:06:42.560 --> 00:06:51.039 close. I would say this is pretty much a tie. fairly close performance between 00:06:46.880 --> 00:06:53.120 the 590 and the 6T990. So, Civ 5, this 00:06:51.039 --> 00:06:57.199 one seems to be fairly CPU limited, although we do get a little bit of a 00:06:55.120 --> 00:07:02.800 performance increase with the 580. The anomaly here is 560 SLI, which performs 00:06:59.919 --> 00:07:06.560 just as well as 580 SLI. I got no idea why that is, but I ran it again and 00:07:04.639 --> 00:07:10.479 again and again. So, my for my Civ 5 benchmark, basically, I take a very very 00:07:08.560 --> 00:07:15.120 late game uh save game that I have uh from 00:07:12.800 --> 00:07:20.000 actually playing the game. And then I've got about 30 some odd cities on the 00:07:18.080 --> 00:07:24.639 screen. And then I zoom out all the way and I take a reading over a period of 00:07:22.319 --> 00:07:27.440 time and then average it out. So, pretty much we're just looking at the same 00:07:25.919 --> 00:07:31.599 static image. We're centered over one of the cities. So, each of the cards is 00:07:29.199 --> 00:07:35.759 rendering exactly the same thing, but that particular solution performs really 00:07:33.440 --> 00:07:39.599 well. could be a driver thing um that's preventing the single card solutions 00:07:37.919 --> 00:07:44.720 from performing as well as the dual card solutions. Not really sure what we're 00:07:41.840 --> 00:07:49.759 looking at there. For Far Cry 2, yeah, we're pretty much CPU limited at that 00:07:46.639 --> 00:07:51.360 point. We are not in any way GPU bound. 00:07:49.759 --> 00:07:57.280 Any one of these graphics cards is going to be able to give you a very good experience in Far Cry 2 if you actually 00:07:54.720 --> 00:08:01.599 still play that game. Crisis is another one. It looks like we are not going to 00:07:59.120 --> 00:08:05.919 be able to observe much in the way of scaling. The one that really falls 00:08:02.960 --> 00:08:11.039 behind in Far Cry 2 is the 6T970, but the 6T90 actually performs uh out it 00:08:09.199 --> 00:08:17.039 performs on top. It actually outperforms even the GTX 580s. So, at 1080p, it 00:08:14.479 --> 00:08:21.919 looks like for the most part, the 590 and the 6T90 are going to be trading 00:08:19.039 --> 00:08:28.240 blows. So, it's going to take our testing at 2560 00:08:24.840 --> 00:08:31.520 x600 to really separate the men from the 00:08:28.240 --> 00:08:34.479 boys in this case. Remember, 2560 x600 00:08:31.520 --> 00:08:38.959 is going to be an advantage for anyone who's got better multiGPU scaling, 00:08:36.560 --> 00:08:42.880 whether it's NVIDIA or AMD. It's also going to be an advantage for whoever has 00:08:40.800 --> 00:08:46.279 the most GPU power because you're going to be far less likely to run into CPU 00:08:44.880 --> 00:08:51.040 limitations at that kind of a resolution. But it should give AMD a 00:08:49.120 --> 00:08:55.519 distinct advantage because more resolution means you need more video 00:08:52.640 --> 00:09:01.920 RAM. And the 6T90 actually has 4 gigs of RAM, two per GPU, whereas the 590 has 3 00:08:58.640 --> 00:09:03.920 gigs of RAM, 1.5 gigs per GPU. So, that 00:09:01.920 --> 00:09:07.600 may be a bit of an advantage. The only way to know is to try out some titles at 00:09:05.920 --> 00:09:13.519 that res. So, thank you for checking out my video review of the GTX 590 at 1920x 00:09:11.279 --> 00:09:17.440 1080 1080p resolution. Don't forget to subscribe to Linus Tech Tips for more 00:09:14.959 --> 00:09:19.920 unboxings, reviews, and other computer videos.