WEBVTT

00:00:00.240 --> 00:00:09.440
now this video is going to be a bit of a look at sli scaling on the gtx 550 ti

00:00:06.879 --> 00:00:15.519
now sli has been a big story for NVIDIA since they first released it with the

00:00:11.360 --> 00:00:17.760
6800 series however not every

00:00:15.519 --> 00:00:22.240
not every card necessarily makes sense to run an sli so

00:00:19.840 --> 00:00:27.920
i've always been of the mind that if there is

00:00:24.560 --> 00:00:29.039
a better card than the one that you have

00:00:27.920 --> 00:00:33.680
that is as good as two of the one that you have

00:00:31.119 --> 00:00:39.040
running in sli you're probably usually better off to go with the single more

00:00:36.000 --> 00:00:42.320
powerful card versus the two slower

00:00:39.040 --> 00:00:44.480
weaker cards running in sli so i wanted

00:00:42.320 --> 00:00:47.840
to take a quick snapshot of some directx 11 titles so the ones that i've gone

00:00:46.719 --> 00:00:53.520
with are hawks 2 lost planet 2 and metro 2033 and

00:00:51.360 --> 00:01:01.120
i compared 4 different scenarios so i compared a single gtx 550 ti cyclone 2

00:00:58.320 --> 00:01:07.600
so that is a slightly overclocked card i compared that running an sli with a

00:01:04.320 --> 00:01:11.200
gtx 550 ti reference card

00:01:07.600 --> 00:01:12.400
i compared a gtx 560 twin frozen two so

00:01:11.200 --> 00:01:16.320
once again this is a slightly overclocked card but since neither of

00:01:14.640 --> 00:01:20.560
these carry a price premium over the reference cards i see no reason to

00:01:18.400 --> 00:01:25.360
necessarily bench with reference cards and the last one that i included was the

00:01:22.320 --> 00:01:26.560
gtx 570. so the reason i've gone this

00:01:25.360 --> 00:01:31.920
route is that the cost of a single 550 ti

00:01:30.720 --> 00:01:36.720
is well significantly lower than anything

00:01:33.759 --> 00:01:41.119
else on the table however the cost of two of them together

00:01:38.320 --> 00:01:45.680
actually lands somewhere in between a 560

00:01:42.479 --> 00:01:48.640
and a 570 so that's why i wanted to test

00:01:45.680 --> 00:01:53.840
the single gpus that are available on either side of the price bracket of

00:01:50.960 --> 00:01:57.360
this dual GPU solution now the upgrading guys will say well yeah but what if i

00:01:55.200 --> 00:02:00.240
wanted to upgrade to one later then my response would be well then what you

00:01:58.719 --> 00:02:04.560
really should have done was saved a couple extra bucks to buy something else

00:02:02.799 --> 00:02:07.840
uh in the first place if you need more performance then buy something with more

00:02:06.240 --> 00:02:13.680
performance so here we go why don't we have a look at the benchmarks that i was able to run

00:02:11.280 --> 00:02:19.360
here and we'll start with hawks too so in hawks 2 you can see that the gtx 550

00:02:16.879 --> 00:02:24.480
ti at about 85 frames per second actually scales incredibly well in this

00:02:22.080 --> 00:02:30.800
particular game so that's like uh 90 percent scaling or so and

00:02:28.560 --> 00:02:37.599
actually performs right exactly where it's supposed to be in between the 560

00:02:33.440 --> 00:02:40.319
ti and the 570. however even this is an

00:02:37.599 --> 00:02:46.959
ideal scenario an ideal scenario has the 550 ti performing about where it should

00:02:43.680 --> 00:02:48.319
in terms of the pricing of this solution

00:02:46.959 --> 00:02:53.280
so you're going to see that the problem with sli is that it doesn't always scale

00:02:50.800 --> 00:02:58.640
in an ideal fashion so right here our 550 performs about where it should with

00:02:55.440 --> 00:03:00.400
respect to a 560 and even a 570 but the

00:02:58.640 --> 00:03:05.440
sli solution while costing more than a 560

00:03:03.360 --> 00:03:11.519
and consuming more power than a single 560 actually doesn't even perform as

00:03:08.159 --> 00:03:13.760
well and remember the 550 ti sli

00:03:11.519 --> 00:03:17.599
solution is about as expensive as in between these two so really it should be

00:03:16.400 --> 00:03:23.599
right in between them it should be around 50 FPS but it's uh it's about 10

00:03:21.200 --> 00:03:28.000
to 12 percent slower than that in lost planet 2. and now let's have a look at

00:03:25.760 --> 00:03:32.480
the last one that i ran here which is a metro 2033 benchmark so here i've got it

00:03:30.879 --> 00:03:36.640
running at normal quality with tessellation on and zero times

00:03:34.879 --> 00:03:43.040
anti-aliasing all of these benchmarks are at 1920x1200 by the way so you can

00:03:39.120 --> 00:03:44.959
see the 550 ti is right here 550 ti sli

00:03:43.040 --> 00:03:49.920
really doesn't scale that well compared to where it should be which is around 90

00:03:47.040 --> 00:03:54.879
FPS so once again we're looking at like a 20 performance delta

00:03:53.040 --> 00:03:59.519
difference between where it should be to justify the price versus where it

00:03:58.000 --> 00:04:04.640
is when it doesn't scale quite as well as

00:04:01.519 --> 00:04:08.480
it could so i basically the conclusion

00:04:04.640 --> 00:04:10.000
of this is that the 550 ti if you do buy

00:04:08.480 --> 00:04:13.920
one and you find yourself needing some more performance yes you can get

00:04:12.159 --> 00:04:18.160
additional performance and you can go from what i would consider basic

00:04:15.920 --> 00:04:22.960
playable to very very playable in a demanding game like metro 2033 at full

00:04:20.560 --> 00:04:25.919
hd resolution however if you know you need the performance

00:04:24.720 --> 00:04:32.160
you're going to be playing demanding games at hd resolution then just get a

00:04:29.520 --> 00:04:35.759
560 ti do yourself a favor and buy the right card in the first place because

00:04:34.479 --> 00:04:42.080
even though once in a while and this is sort of the

00:04:38.320 --> 00:04:44.800
exception not the rule the 550 ti sli

00:04:42.080 --> 00:04:49.280
setup will outperform a 560 ti you're gonna get a more consistent overall

00:04:46.720 --> 00:04:54.880
experience with a single more powerful card versus two

00:04:51.840 --> 00:04:56.639
more entry level cards running in tandem

00:04:54.880 --> 00:05:00.400
so thanks for checking out this video online as tech tips don't forget to

00:04:58.080 --> 00:05:07.000
subscribe and for more right for more unboxings

00:05:02.479 --> 00:05:07.000
reviews and other computer videos