WEBVTT

00:00:07.319 --> 00:00:16.720
welcome to the power consumption heat and Acoustics oriented

00:00:13.120 --> 00:00:21.320
segment of my sort of series of video

00:00:16.720 --> 00:00:23.080
reviews on the GeForce GTX 680 uh this

00:00:21.320 --> 00:00:27.160
one's from Galaxy although this is a reference board and I'll get into a

00:00:25.160 --> 00:00:31.840
little bit more the advantages and disadvantages of our reference board

00:00:29.439 --> 00:00:38.280
very very shortly so out of the cards that you guys see here so I've got the

00:00:34.280 --> 00:00:40.640
GTX 680 I've also got a Radeon hd7970

00:00:38.280 --> 00:00:48.399
which price-wise as well as Market positioning wise is its closest

00:00:42.920 --> 00:00:48.399
competitor and what on Earth did he just

00:00:48.760 --> 00:00:57.559
drop my cat is like dropping

00:00:53.680 --> 00:01:00.000
things all right moving on closest

00:00:57.559 --> 00:01:03.440
competitor in terms of positioning and and uh so we're going to see how these

00:01:01.680 --> 00:01:07.960
two stack up head to-head both of these are reference cards and then there's one

00:01:05.479 --> 00:01:14.280
card that I haven't tested yet so is not part of my graphs but I want to include

00:01:10.400 --> 00:01:18.720
it uh less for the power consumption and

00:01:14.280 --> 00:01:22.119
the here we go and the idle temperature

00:01:18.720 --> 00:01:24.720
purposes and more for the load

00:01:22.119 --> 00:01:28.159
temperature and load noise purpose because I mean the thing about any

00:01:26.520 --> 00:01:33.799
launch card is it's always a reference card which means it's an NVIDIA designed

00:01:31.119 --> 00:01:38.720
or an AMD designed card using a reference or AMD or NVIDIA designed

00:01:35.880 --> 00:01:43.159
cooler with a reference AMD or NVIDIA designed PCB so the advantage of that of

00:01:41.240 --> 00:01:47.719
that is that it's all validated by AMD or NVIDIA so that's good usually these

00:01:45.960 --> 00:01:51.200
reference coolers tend towards the blower types of coolers which exhaust

00:01:49.680 --> 00:01:55.719
air from the case which some people consider better whereas others prefer

00:01:53.640 --> 00:02:00.280
the more open style of coolers which deliver significantly better cooling

00:01:57.920 --> 00:02:04.680
performance for the GPU itself but do not necessarily keep the rest of your

00:02:02.079 --> 00:02:10.319
system components as cool

00:02:07.399 --> 00:02:15.160
um reference boards are pretty standard so you're not going to see a lot of

00:02:11.599 --> 00:02:17.400
extra features like for example on um

00:02:15.160 --> 00:02:21.879
you know highend SKS you sometimes see non-reference pcbs with more power

00:02:19.519 --> 00:02:25.519
phases things like that better coolers so there's a non-reference card so at

00:02:23.879 --> 00:02:30.360
launch we are pretty much stuck with reference boards I'm looking forward to

00:02:27.160 --> 00:02:32.440
seeing GeForce GTX 680s with reference

00:02:30.360 --> 00:02:36.720
designs but for now this is what we got and what the reason I want to compare

00:02:34.560 --> 00:02:41.040
this non-reference 7970 is cu we're starting to see non-reference 797s now

00:02:39.680 --> 00:02:45.480
and I want to look at what kind of an advantage it has in terms of especially

00:02:43.239 --> 00:02:49.800
Acoustics as well as load temperatures compared to a reference card so let's do

00:02:47.680 --> 00:02:53.800
charts and graphs first since that's everybody's favorite part of these

00:02:51.040 --> 00:03:00.560
videos right so remember not including the sapphire uh dual x fan 7970 here

00:02:57.840 --> 00:03:07.159
let's talk system idle temp to start with and my cat really needs to get out

00:03:03.000 --> 00:03:07.959
of the way go on go away go away please

00:03:07.159 --> 00:03:13.360
thank you all right here we go system idle

00:03:10.879 --> 00:03:17.640
temperatures in degrees Cel so here I've got a wide variety of different cards

00:03:15.200 --> 00:03:21.959
you can see that out of all of them the 6990 uh compared to its closest

00:03:20.360 --> 00:03:28.200
competition in terms of performance which is the 680 and the 7970 as well as

00:03:25.080 --> 00:03:30.280
the GTX 590 is by far the worst idle

00:03:28.200 --> 00:03:34.920
temperatures but that kind of to be expected the 6990 is a bit of a beast

00:03:32.480 --> 00:03:39.480
when it comes to uh noise as well as power consumption so I mean honestly so

00:03:37.120 --> 00:03:44.799
is the 590 not as loud but definitely in terms of power consumption and heat

00:03:40.959 --> 00:03:46.799
output the uh so the 680 comes in at 33°

00:03:44.799 --> 00:03:50.720
honestly all of these are close enough and they're all quiet enough at idle

00:03:48.599 --> 00:03:54.439
that none of it's really a concern now instead of using furmark for load

00:03:53.120 --> 00:03:59.920
temperature testing which I have traditionally done in the past I am

00:03:57.000 --> 00:04:03.000
using games now so I am recording temperatures during games and then I'm

00:04:01.640 --> 00:04:07.720
tracking whatever was the highest temperature that was reached and I'm

00:04:04.760 --> 00:04:12.439
going with that in fact you know what oh I guess I actually don't have to I don't

00:04:09.959 --> 00:04:17.759
have to do the dual x 7970 because we can kind of extrapolate the results from

00:04:14.280 --> 00:04:19.799
the dual x 7950 that I did use for my

00:04:17.759 --> 00:04:23.400
testing I sort of forgot about that okay well either way I'll use this one for

00:04:21.199 --> 00:04:28.320
the acoustic testing since I don't know if I have that 7950 on me right now so

00:04:26.639 --> 00:04:36.960
reference boards all of these are reference cards the 7970 OC the 680 the

00:04:33.080 --> 00:04:40.440
7970 non overclock 6990 570 580 590 and

00:04:36.960 --> 00:04:43.560
GTX 680 non overclocked they all hover

00:04:40.440 --> 00:04:45.320
around that 75 to 80° range although

00:04:43.560 --> 00:04:49.000
they vary wildly in terms of Noise with the 6990 being by far the loudest check

00:04:47.720 --> 00:04:54.360
out what you get by going with an aftermarket cooler you can drop 20 to

00:04:51.800 --> 00:04:58.039
25° off of your load temperatures on your card we're talking real world

00:04:56.400 --> 00:05:04.039
gaming load by the time you put it in a closed case you're probably going to gain another 3 to 5° but we're still

00:05:01.759 --> 00:05:10.160
talking a very significant Improvement in performance of the cooler on your

00:05:06.759 --> 00:05:12.720
card so that's something to bear in mind

00:05:10.160 --> 00:05:16.880
um so that was with the 7950 OC from Sapphire which is using their dual x

00:05:14.520 --> 00:05:21.360
cooler next we've got system Idol power draw so this is where this latest

00:05:18.319 --> 00:05:22.840
generation of cards just obliterates the

00:05:21.360 --> 00:05:30.080
previous generation cards check out these numbers 680 and

00:05:25.479 --> 00:05:32.479
7970 consume up to a 100 or up to 50 or

00:05:30.080 --> 00:05:37.560
60 watts less power than previous generation similarly performing cards

00:05:34.400 --> 00:05:40.240
like the 6990 and the GTX 590 what that

00:05:37.560 --> 00:05:45.440
means is basically the equivalent of not having a light bulb on at all times when

00:05:43.600 --> 00:05:48.440
you have your PC on this is like having like just sort of like a random light

00:05:47.000 --> 00:05:52.720
bulb on somewhere in your house at all times when your PC is idling in terms of

00:05:50.560 --> 00:05:56.800
the extra power consumption and then at load the story gets even more dramatic I

00:05:55.199 --> 00:06:03.240
mean check this out guys so the 6990 and the GTX X 590 are just

00:06:00.680 --> 00:06:07.280
absolute power Hogs whereas these latest generation cards consume a trivial

00:06:05.720 --> 00:06:12.280
amount I mean we're talking about 3% more power than a last generation 580

00:06:09.639 --> 00:06:18.160
for the 680 but in spite of that the performance of this card is just worlds

00:06:15.759 --> 00:06:23.120
better and I uhoh I just clicked something bad okay it's back uh is

00:06:20.880 --> 00:06:27.599
world's better so the 680 in many scenarios is outperforming the GTX here

00:06:25.479 --> 00:06:32.280
let's take Battlefield 3 for an example is outperforming the GTX 580 a factor of

00:06:30.120 --> 00:06:38.800
more than double so in terms of the performance per watt it just just yeah

00:06:36.440 --> 00:06:43.840
it just rocks I mean the 7970 is no slouch either although it does consume

00:06:40.479 --> 00:06:46.080
more power under load versus the GTX 680

00:06:43.840 --> 00:06:49.039
the 7950 consumes significantly less power and hopefully we'll see this

00:06:47.479 --> 00:06:53.759
something similar from something like a GTX 670 or whatever the derivative is

00:06:51.800 --> 00:06:59.800
going to be that comes out of it and I'm very very excited about

00:06:56.400 --> 00:07:00.800
that system idle power draw in so yeah

00:06:59.800 --> 00:07:07.080
we went through that so that's pretty much it so now let's just do some acoustic testing I'll give you guys some

00:07:03.879 --> 00:07:09.919
idle and then gaming load uh reading so

00:07:07.080 --> 00:07:13.440
my testing methodology as unscientific as this may be basically involves taking

00:07:11.840 --> 00:07:17.039
my camera holding it a fixed distance away from the cards and then letting you

00:07:15.479 --> 00:07:19.280
guys listen to it for a period of about 5

00:07:28.319 --> 00:07:31.319
Seconds

00:07:53.520 --> 00:08:00.639
honestly it's too bad that this particular 7970 has uh a significant

00:07:58.800 --> 00:08:04.919
amount of coil wine it's based on a reference PCB while it does have a

00:08:02.960 --> 00:08:10.280
reference cooler so uh you can't really tell how quiet the cooler is um I did

00:08:08.319 --> 00:08:16.840
not have the same issue with the 7950 dual x uh that one was very very

00:08:14.840 --> 00:08:22.400
quiet cuz it was using a sapphire designed PCB as opposed

00:08:19.159 --> 00:08:26.639
to a reference one anyway don't forget

00:08:22.400 --> 00:08:26.639
to subscribe yay