WEBVTT

00:00:07.120 --> 00:00:14.839
so this is going to be a quick follow-up video to my GTX 660ti content now you

00:00:12.080 --> 00:00:21.800
can see here I've got two 660 Ti running I've got the MSI 660ti power Edition as

00:00:18.160 --> 00:00:24.599
well as Galaxy's custom cooled GC card

00:00:21.800 --> 00:00:30.080
now for those of you who don't know yes you can mix vendors in SLI yes you can

00:00:27.760 --> 00:00:34.200
mix different clock speeds in SLI different pcbs in SLI I mean this one's

00:00:32.399 --> 00:00:38.120
even running two six pin powers and this one's running a six and an eight

00:00:35.960 --> 00:00:42.039
different coolers in SLI the thing that's important is that the GPU is the

00:00:40.719 --> 00:00:49.520
same so that means you have to have a 660 Ti and a 660 Ti no 660 Ti and a you

00:00:46.960 --> 00:00:54.359
know 670 or anything like that no flexibility on the NVIDIA side unlike

00:00:51.640 --> 00:01:01.160
the AMD side breaker in a 6990 with the 6970 and trifire this doesn't work that

00:00:57.239 --> 00:01:04.559
way so just cuz a 690 has the same GPU

00:01:01.160 --> 00:01:05.479
as not that I think there's a 680 around

00:01:04.559 --> 00:01:11.000
here somewhere but I'm not going to worry too much about it ah there it is right was hiding um as a 680 that

00:01:09.360 --> 00:01:14.840
doesn't mean you can run them in SLI got to be matching gpus also it's been

00:01:13.479 --> 00:01:18.200
suggested to me in the past that mismatched RAM configurations are okay

00:01:16.880 --> 00:01:23.240
but I haven't tried it in a long time and the last time I tried it it didn't

00:01:20.000 --> 00:01:25.520
work so there you go all right so we are

00:01:23.240 --> 00:01:29.920
going to be doing like a value analysis because people ask me about this all the

00:01:27.400 --> 00:01:33.799
time should I buy one powerful card or should I buy two less powerful cards and

00:01:31.720 --> 00:01:37.880
run them in SLI so there are some trade-offs right off the bat two cards

00:01:36.000 --> 00:01:41.240
is going to use more power period it's going to be noisier period because you

00:01:39.560 --> 00:01:45.719
got more fans in this case you got four fans versus one but uh which isn't to

00:01:44.159 --> 00:01:52.360
say necessarily that two of these fans is louder than one of these fans but

00:01:48.520 --> 00:01:55.079
well whatever moving on um so yes power

00:01:52.360 --> 00:02:02.920
noise both worse heat is worse for having multiple cards and in 90% of

00:01:58.600 --> 00:02:04.920
cases multiple cards is bad never run

00:02:02.920 --> 00:02:08.119
two of these instead of one more powerful one that is a low-end card well

00:02:06.920 --> 00:02:14.000
well it's a gaming card but it's a low-end gaming card and you're definitely better off getting something

00:02:11.959 --> 00:02:18.560
in the sort of more bang for the buck $250 to $350 range that's where you

00:02:16.800 --> 00:02:24.360
usually find the best value so we're going to take two 660 Ti versus a single

00:02:21.920 --> 00:02:28.560
680 and find out what delivers the best performance and the best value so here's

00:02:26.599 --> 00:02:32.120
my first graph this is Crisis 2 1080p I'm using all my standard settings so

00:02:30.120 --> 00:02:36.840
this is like ubered out with high reses textures and all that good stuff so you

00:02:34.000 --> 00:02:42.040
can see the 660 Ti configuration actually well destroys the 680 however

00:02:39.840 --> 00:02:46.959
it is more expensive it's about 20% more expensive and when we actually go ahead

00:02:44.640 --> 00:02:52.280
and look at the performance differential it looks like the 660 Ti comes ahead so

00:02:49.080 --> 00:02:54.480
it's about 26% faster and 20% more

00:02:52.280 --> 00:02:58.959
expensive and at minimum FP if you measure minimum FPS it's about 20%

00:02:56.519 --> 00:03:02.319
faster there as well so let's go ahead and you know what I'll do the graphs off

00:03:00.319 --> 00:03:08.239
camera Battlefield 3 again was running at like uber settings 1080P and you can

00:03:05.239 --> 00:03:11.920
see the 660ti again just destroys it in

00:03:08.239 --> 00:03:13.879
terms of average FPS but the minimum FPS

00:03:11.920 --> 00:03:18.440
was a little bit closer so it doesn't look like as compelling a value as as it

00:03:16.879 --> 00:03:23.519
at first appears when you look at the averages however that still does show

00:03:20.760 --> 00:03:28.840
that the 660ti does present a better or SLI does present a better value than a

00:03:26.040 --> 00:03:32.879
single 680 in this particular title as long as you're willing to make those

00:03:30.000 --> 00:03:37.920
trade-offs in terms of power consumption as well as

00:03:34.959 --> 00:03:45.239
noise again running a game at 1080p with ultra high details and again we see at

00:03:41.720 --> 00:03:49.519
that minimum FPS level a 20% improvement

00:03:45.239 --> 00:03:51.599
from the 660 Ti SLI config over the 680

00:03:49.519 --> 00:03:56.439
we see about a 47% Improvement in average FPS but it almost looks like

00:03:54.280 --> 00:04:02.200
NVIDIA knew what they were doing when they were pricing out a single GTX 680

00:03:59.280 --> 00:04:08.319
versus 660 Ti SLI because they're delivering pretty much the performance

00:04:04.799 --> 00:04:10.799
that you'd expect for the Dual uh the

00:04:08.319 --> 00:04:17.880
Dual GPU configuration versus the single one um you know what here I'll just do

00:04:14.680 --> 00:04:19.400
this last one uh on okay no I won't do

00:04:17.880 --> 00:04:23.680
this on camera I need two hands and there's Metro 2033 where we actually see

00:04:21.440 --> 00:04:27.960
about a 30% Improvement in both minimum and average FPS so this is the clear

00:04:26.600 --> 00:04:32.680
winner so what's neat about this cross-section of games is that these are all well they're all Shooters they're

00:04:30.759 --> 00:04:37.680
all demanding uh which or two isn't a shooter actually they're all demanding

00:04:34.800 --> 00:04:42.639
and they all demonstrate that the 660 Ti not only scales consistently in SLI but

00:04:40.919 --> 00:04:47.600
presents a compelling value versus buying a single higher in GPU that is to

00:04:45.440 --> 00:04:51.400
say that's not to say though that it's necessarily the better option because

00:04:49.080 --> 00:04:55.880
two of these is 600 bucks one of these is 500 bucks and for that 20%

00:04:54.039 --> 00:05:00.479
Improvement you are trading off extra heat and extra power consumption so it's

00:04:58.280 --> 00:05:04.520
up to you personally I think SLI cards look more bling bling

00:05:04.639 --> 00:05:11.919
and yeah I think that's probably all else that it's got going for it but uh

00:05:10.520 --> 00:05:17.520
remember guys when you're trying to figure out if a dual GPU configuration

00:05:15.240 --> 00:05:20.840
makes sense you just want to look at performance per dollar that's pretty

00:05:18.960 --> 00:05:24.560
much the only Factor here because they've now got the same features they

00:05:22.560 --> 00:05:28.639
both support 3+ 1 videos out even when you add another card if you're running

00:05:26.280 --> 00:05:34.639
an SLI you can only plug the uh you can only plug in 3 plus one in SLI and this

00:05:32.080 --> 00:05:38.759
one also supports 3+ one off of a single card um you also want to remember that

00:05:36.919 --> 00:05:43.319
things like memory configurations are important so when you're running two

00:05:40.120 --> 00:05:47.160
cards in SLI two 2 gig cards does not

00:05:43.319 --> 00:05:48.840
equal 4 gigs of video memory okay so

00:05:47.160 --> 00:05:52.000
they both have to share the frame buffer which means that you still only get

00:05:50.160 --> 00:05:56.000
effectively two kicks so bear that in mind as well so I know this really

00:05:54.319 --> 00:05:59.440
didn't have a clear-cut conclusion because basically performance per dollar

00:05:58.199 --> 00:06:04.440
they're very very similar it just comes down down to do you want a little bit more performance or do you not want a

00:06:02.600 --> 00:06:07.840
little bit more performance thanks for checking out this episode of L's Tech

00:06:05.880 --> 00:06:13.680
tips and don't forget to subscribe for more unboxings reviews and

00:06:10.160 --> 00:06:13.680
other computer videos