NVIDIA GeForce GTX 580 & SLI MSI & EVGA Graphics Card Video Review Linus Tech Tips
Linus Tech Tips
·Linus Tech Tips
·2011-05-08
·
1,834 words · ~9 min read
0:00
this will be a quick overview of the testing setup that I'll be using for my
0:05
GTX 580 review so as you can see here I
0:09
do have two GeForce GTX 580s I have one
0:13
from MSI and then I also have one from
0:17
EVGA so that means that I will be including SLI testing for the GTX 580 as
0:24
well as crossfire testing for the radon
0:27
5870 which as of right now is the
0:30
closest competitor for the GTX 580 although I don't expect that to be the
0:34
case for uh for very much longer in the future now one of the other comparisons
0:39
I also want to make just to include it in my charts is the GTX 480 which the
0:45
580 is replacing and just because it
0:48
represents such a good value I also have
0:51
two MSI GeForce GTX 460 Hawk edition
0:56
cards because two 460s when they came out on the market with the exceptional
1:01
scaling that you get in crossfire as well as with the outstanding
1:04
overclocking of the hawk Edition card became a great value for a performance
1:09
Enthusiast because two 460s was actually
1:13
cheaper than a 480 but provided better performance so I really want to see
1:18
where the GTX 580 Falls in relation to
1:22
two overclocked 460s which are out there on the market for a very good price
1:26
these days now I want to introduce you to the test bench really quick I have my
1:30
3.8 GHz core i7
1:33
875k I have that running with 4 gigs of
1:36
Kingston hyperx RAM I've got an ASUS
1:40
Crossfire and SLI ready motherboard down there somewhere so I've got a new CPU
1:45
Cooler now actually it keeps my uh chip
1:48
quite a bit cooler than the A50 I was using before this is a silent X aphysio
1:52
extreme now down here I have a cougar SX
1:56
series 850 W power supply which is more than capable of power in all the cards
2:01
down here below mind you not all at the same time that would be quite
2:04
unreasonable and then for my boot drive I'm using an Intel x25 M 80 gig drive
2:10
now I have changed my monitor setup slightly as you may or may not be able
2:15
to tell that is a 30in monitor so I'm going to be doing all of my benchmarks
2:20
other than 3D Mark at 2560 x600 3D Mark
2:24
is going to be run in the extreme profile because the reality of it is the
2:28
gpus I'm testing here today are all
2:31
premium high-end products so we're going to use them as they're intended to be
2:35
used and that is at extremely high resolution and find out how they all
2:39
compare to each other stay tuned for
2:48
more now for Starcraft 2 it's pretty obvious that with every setup from the
2:54
GTX 480 up through the 460s and SLI and
2:59
the 580 580s in SLI we got almost exactly the same frame rate so what that
3:03
comes down to is a CPU bottleneck it means that this game cannot take
3:07
advantage of the extra GPU power that we're throwing at it now I was not
3:11
running Starcraft 2 with anti-aliasing and if we turned on anti-aliasing we'd
3:15
probably see some improvement however with that unit count bear in mind we
3:19
were using a very action intensive part of the game to do our benchmarking we
3:24
were not able to find any difference between well one very high-end solution
3:28
and another even more Uber high-end
3:35
solution Mafia 2 is being tested with physx so we have not included any
3:39
results for Radeon cards they didn't fail the test they just can't run it
3:43
with
3:47
fizx for Metro 2033 with fizx you
3:50
probably noticed the GTX 460 SLI config is not in there and that's because it
3:55
failed the reason is that the GTX 460
3:58
only has 1 gig of video memory and what
4:02
that means is because the two cards in SLI can't share that video memory you
4:06
have effectively one gig of video memory and at that resolution with those
4:10
settings with that particular game you
4:13
need more RAM so that's what the GTX 480
4:16
and the GTX 580 had as an advantage over
4:20
those lower-end cards so that's why I wasn't able to include any 460 SLI
4:25
results
4:29
I did another run through of Metro 2033 with no physx so that I could include
4:33
the radon cards and as you can see the GTX 460 SLI config still didn't make it
4:40
I got a benchmark result and it kind of fell in line with everything else but I
4:44
can't present that to you guys because when I was playing through the game it
4:48
did not feel right like I had times where it dipped down and it felt like 2
4:52
FPS even though at the end The Benchmark reported something like very very
4:56
playable so uh that's why I could not include that result
5:04
there Battlefield Bad Company 2 is actually one of the games that uh scaled
5:09
extremely well with all of the different cards that we ran and we were able to
5:13
generate very consistent results I did a little playthrough of the uh in between
5:19
going through the town here and then where at the end you find a tank uh
5:24
that's the part of this game that was used for the uh benchmarking
5:28
purposes and uh try and take out the out at the rocket launcher now you can't get
5:33
a consistent run through every time so I did have to do multiple run throughs
5:36
with each GPU and average them amount but I think overall we get a result
5:39
that's fairly representative of what each graphics card is capable of in this
5:44
particular shooter
5:50
game going to be completely honest with you guys I didn't have a whole lot of
5:54
time to spend with the GTX 580s my radon
5:58
5870 Crossfire setup was doing some
6:01
weird stuff in crisis where I was getting worse performance than a single
6:05
card at 2560 x600 I didn't even have the
6:08
time to really look up and find out if that's a common issue cuz there aren't
6:11
too many people who are trying to run crisis with anti-aliasing at 2560 x600
6:16
with radon 5870 Crossfire so it's uh
6:20
probably not all that well documented but basically sorry I couldn't include
6:23
those results you can check out everything else though
6:33
for power consumption bear in mind all of my measurements are taken from the
6:36
wall and they are an absolute Peak value so you can expect that usually while
6:42
gaming it'll be quite a bit lower but this is the maximum power consumption
6:46
that I was able to pull from the wall with this system during the entire
6:50
benchmarking process for each GPU so that means that if I was doing something
6:54
CPU intensive at the same time as gaming it could probably be higher so make sure
6:58
that when you're budgeting a PS you for your build if you're going to use one of
7:01
these new cards like say for example the GTX 580 that you plan
7:07
accordingly so here's the settings that uh we ran Metro 2033 on although
7:11
actually as you saw from my introduction when I was at home and I was not at work
7:16
I had a 30-in monitor at my disposal so the benchmarks were done remember at two
7:21
different resolutions for Metro 23 2560
7:24
x600 as well as 1920 x 1200 although
7:27
these are the other settings that are being used everything is enabled so
7:31
we're going to do some quick gaml here while I just talk a little bit about the
7:35
GTX 580 and what this particular game
7:38
means to the market so this is the run through that I did in order to test
7:43
Metro 2033 it's on the level the bridge
7:46
and I follow this guy around kill some dudes basically I've got the GTX 580
7:52
setup running right now and I'm just wanting to show
7:55
you how beefy this setup is it really
7:59
does blow away everything else especially in the 2560 x600 resolution
8:06
because so many of the other setups that I tried to run this game at those
8:11
settings with just outright failed they
8:14
just couldn't do it and here you can see it's butter smooth even at 1920 x 1200
8:20
everything maxed out I'm seeing you know 50 frames per second hold on I better
8:24
put on my gas mask in just a second you guys will have some combat to check out
8:28
and so uh that's one of the things I really
8:31
noticed about the Lesser setups is that especially when the action gets a little
8:35
bit more intense the frame rate dips and it can even sort of skip a couple and
8:40
that's where it becomes unplayable very very quickly so I got a monster rushing
8:44
at me can kill the guy stays very smooth
8:47
no matter what I'm doing shotgun 40 FPS still anyway so all of
8:55
that aside GTX 580 what does it mean well it's really really expensive so not
9:00
everyone can afford one it's going to be up over $500 Canadian at launch however
9:06
unlike its predecessor the 480 NVIDIA has really gotten a few
9:11
things under control they've completely fixed the problem with the excess heat
9:16
output for the power compared to the competition that's on the market today
9:20
they fixed the problem with the noise both by using their Innovative Vapor
9:24
chamber cooler as well as their adaptive fan profile so make sure you do check
9:28
out my videos that are dedicated to those two aspects of this card I show
9:33
how the fan ramps up and down very slowly so it sounds very smooth and I
9:36
also disassemble the MSI card so you can see the vapor chamber cooler up close
9:41
and personal and they've added more power and it comes in at the same price
9:46
as a GTX 580 so it's really hard to uh
9:49
to find anything wrong with it I guess it's a super premium product like I said
9:54
so it's quite expensive but it's pretty
9:57
much a gamech Cher and it is without a doubt whether it's a loan or an SLI the
10:04
fastest graphics card solution that you can buy today so thank you for checking
10:08
out my performance review of the GTX 580 don't forget to check out a lot of the
10:12
other videos that I've done around this important launch for NVIDIA and thanks
10:16
for checking out lineus Tech tips don't forget to subscribe