WEBVTT

00:00:00.280 --> 00:00:01.680
So you may be asking yourself,

00:00:01.680 --> 00:00:05.040
how on earth did Linus Tech Tips

00:00:05.040 --> 00:00:08.600
get a hold of a Quadro M6000,

00:00:08.600 --> 00:00:13.660
a $5,000 video card?

00:00:13.820 --> 00:00:18.850
And the answer to that is that I'm not going to tell you.

00:00:18.850 --> 00:00:21.510
It's all top secret and .

00:00:21.510 --> 00:00:24.560
No, but what I will tell you is,

00:00:24.560 --> 00:00:36.580
but can it run crisis?

00:00:36.580 --> 00:00:38.300
Intel's Skull Canyon Nook features

00:00:38.300 --> 00:00:40.700
a sixth generation Core i7 processor

00:00:40.700 --> 00:00:42.220
and Thunderbolt 3.

00:00:42.220 --> 00:00:45.500
You can learn more at the link in the video description.

00:00:45.500 --> 00:00:47.620
But before I answer the age old question,

00:00:47.620 --> 00:00:48.940
let's see if we can't figure out

00:00:48.940 --> 00:00:52.080
exactly what makes a $5,000 graphics card

00:00:52.080 --> 00:00:54.440
so freaking specialtastic.

00:00:54.440 --> 00:00:59.440
I mean, it's got a GM 200 Maxwell architecture GPU on it.

00:00:59.920 --> 00:01:01.720
So you can't see it, it's on the other side,

00:01:01.720 --> 00:01:03.360
but you go ahead and try, good for you.

00:01:03.360 --> 00:01:05.500
That's like your effort over there.

00:01:05.500 --> 00:01:07.740
It's got 3,072 CUDA cores.

00:01:07.740 --> 00:01:10.480
It's clocked at somewhere around one gigahertz,

00:01:10.480 --> 00:01:11.560
a little lower on the base,

00:01:11.560 --> 00:01:12.760
a little higher on the boost.

00:01:12.760 --> 00:01:14.120
It's got, ooh, look at this.

00:01:15.570 --> 00:01:17.330
Four DisplayPort outputs.

00:01:17.330 --> 00:01:19.530
So you can have four 4K displays running off it

00:01:19.530 --> 00:01:20.370
at the same time.

00:01:20.370 --> 00:01:22.870
Also there's lonely old DVI up there.

00:01:22.870 --> 00:01:24.930
It can also run VGA if you're into that sort of thing.

00:01:24.930 --> 00:01:26.190
Although I don't think too many people

00:01:26.190 --> 00:01:28.110
are running $5,000 graphics cards.

00:01:28.110 --> 00:01:31.230
Can't afford a monitor that was made in the last 10 years.

00:01:31.230 --> 00:01:34.450
In terms of, okay, you got SLI and sync connectors,

00:01:34.450 --> 00:01:35.810
as well as a stereo connector

00:01:35.810 --> 00:01:37.650
if you work in stereo 3D content.

00:01:37.650 --> 00:01:38.870
So that's pretty cool.

00:01:38.870 --> 00:01:40.570
And this is really exciting.

00:01:40.570 --> 00:01:41.410
Boom!

00:01:41.410 --> 00:01:43.230
The power connector is a single eight pin

00:01:43.230 --> 00:01:45.610
on the back of the card.

00:01:45.610 --> 00:01:46.530
Over there, yes.

00:01:46.530 --> 00:01:48.750
On the back of the card, my friends,

00:01:48.750 --> 00:01:50.930
which I guess is important if you're mounting it

00:01:50.930 --> 00:01:52.170
in a rack mount case,

00:01:52.170 --> 00:01:53.590
where there isn't always clearance

00:01:53.590 --> 00:01:55.690
for power connectors that come out the side,

00:01:55.690 --> 00:01:58.250
like is popular on desktop models.

00:01:58.250 --> 00:02:02.310
The more you know.

00:02:02.310 --> 00:02:05.380
But hold on a second, Linus.

00:02:05.380 --> 00:02:09.580
All of that sounds awfully similar to the Titan X,

00:02:09.580 --> 00:02:13.540
a graphics card that costs $4,000 less.

00:02:13.540 --> 00:02:13.900
I mean, okay.

00:02:13.900 --> 00:02:16.260
There's an HDMI port instead of DisplayPort.

00:02:16.260 --> 00:02:19.820
And like, it's got an extra power connector

00:02:19.820 --> 00:02:20.880
on a different spot,

00:02:20.880 --> 00:02:23.360
but this one has like LEDs and .

00:02:23.360 --> 00:02:25.640
What makes this better?

00:02:25.640 --> 00:02:27.980
Well, we missed one small detail.

00:02:27.980 --> 00:02:32.840
It has 24 gigs of GDDR5 memory.

00:02:32.840 --> 00:02:34.540
And not only that,

00:02:34.540 --> 00:02:37.760
but that memory is capable of running in ECC mode

00:02:37.760 --> 00:02:40.520
for like super, you know,

00:02:40.520 --> 00:02:43.740
less chance of something going bad.

00:02:43.900 --> 00:02:47.260
Let's put it on a bench.

00:02:47.260 --> 00:02:49.060
Angels singing?

00:02:49.060 --> 00:02:52.140
Nothing, all right, fine.

00:02:52.140 --> 00:02:54.700
So maybe we get like lots of amazing options

00:02:54.700 --> 00:02:56.580
inside NVIDIA control panel,

00:02:56.580 --> 00:03:01.320
deciding how to use our GPU, you know.

00:03:01.320 --> 00:03:04.380
We can view system topology.

00:03:04.380 --> 00:03:10.680
We can set up mosaic, which is like surround.

00:03:10.680 --> 00:03:16.000
And we have a dedicated tab for changing the ECC state.

00:03:16.000 --> 00:03:17.480
Hard to make this exciting.

00:03:17.480 --> 00:03:20.460
Let's run a game.

00:03:20.460 --> 00:03:25.880
I'm methodically getting very high right now.

00:03:26.700 --> 00:03:27.860
Whatever, I didn't change anything.

00:03:27.860 --> 00:03:28.880
Screw off.

00:03:28.880 --> 00:03:35.460
Let's do the crisis benchmark.

00:03:35.460 --> 00:03:37.690
So the benchmark's over.

00:03:37.690 --> 00:03:41.440
Prepare to be amazed and awed.

00:03:41.440 --> 00:03:46.110
And I'm talking like, aw, not like aw.

00:03:46.110 --> 00:03:51.400
In crisis three, the Quadro M6000 scores 104 FPS.

00:03:57.340 --> 00:03:58.020
Is that even good?

00:03:58.040 --> 00:03:59.840
Yes, it is.

00:03:59.840 --> 00:04:02.500
So we're back at it with the Titan X.

00:04:02.500 --> 00:04:05.480
And I have to confess,

00:04:05.480 --> 00:04:08.370
this feels like a very similar experience.

00:04:08.370 --> 00:04:12.580
Okay, moment of truth time.

00:04:12.580 --> 00:04:14.860
105.5 FPS.

00:04:14.860 --> 00:04:16.420
Well, okay, then.

00:04:16.420 --> 00:04:17.740
But come on Linus.

00:04:17.740 --> 00:04:18.580
What were you expecting?

00:04:18.580 --> 00:04:21.780
Quadro cards aren't made for gaming.

00:04:21.780 --> 00:04:24.640
Well, truthfully, that's exactly what I was expecting

00:04:24.640 --> 00:04:26.620
because while it is true that Quadro cards

00:04:26.620 --> 00:04:29.040
are not designed for gaming, I mean, the gaming performance,

00:04:29.040 --> 00:04:36.330
of them is right on the spec sheet. The GPU, the clock speed it runs at, I don't know where this

00:04:36.330 --> 00:04:42.830
whole myth comes from, that quadro cards are meant for specific workloads and are not meant to game.

00:04:43.470 --> 00:04:51.470
And like, no, a GPU is a GPU. It just means that quadro cards are also made for other things.

00:04:51.710 --> 00:04:57.290
So let's have a look at some rendering benchmarks or maybe some, you know, high-end video production

00:04:57.290 --> 00:05:05.670
benchmarks and see if we can better justify its high price tag. Okay, so this is a Octane demo,

00:05:05.850 --> 00:05:12.620
a little outside my comfort zone in terms of products that I, oh, control Z. Okay, all good.

00:05:12.620 --> 00:05:24.800
Render target. There we go. So we're using, let's see, 1.6 gigs of memory and we are processing at

00:05:24.800 --> 00:05:28.780
7.2 to 7.3.

00:05:29.080 --> 00:05:39.460
Million pixels per second. Those pixels, I don't know what they are. Okay, well, I'm not going to sit here and watch this go for half an hour.

00:05:39.460 --> 00:05:46.150
So let's try something else before we put the Titan X back on. DaVinci Resolve. Let's watch some frickin' sharks.

00:05:47.540 --> 00:05:54.040
All right, we're getting about 12 frames per second with like denoising on and such.

00:05:54.680 --> 00:06:01.750
Again, presumably that is very good, but I don't know. That is very similar.

00:06:01.930 --> 00:06:02.430
That is...

00:06:02.470 --> 00:06:04.170
faster so far.

00:06:04.170 --> 00:06:05.430
That is...

00:06:05.430 --> 00:06:06.710
the same.

00:06:06.710 --> 00:06:12.510
But maybe DaVinci Resolve then. I want to see big FPS. Oh, now come on.

00:06:13.070 --> 00:06:19.110
That doesn't even make sense. No, that can't be right. Wow, that shark just like ate a fish.

00:06:19.590 --> 00:06:24.470
Okay, so DaVinci Resolve with our 6K footage on a 4K timeline

00:06:25.930 --> 00:06:28.790
performed much better on the Titan X. But of course,

00:06:29.530 --> 00:06:31.530
you could afford much more than one Titan X for the price of the Titan X.

00:06:31.530 --> 00:06:32.230
You could afford much more than one Titan X for the price of the Titan X.

00:06:32.230 --> 00:06:32.730
you could afford much more than one Titan X for the price of the Titan X.

00:06:32.730 --> 00:06:34.090
for the price of an M6000.

00:06:34.090 --> 00:06:37.810
Let's see if the story changes with more Titan Xs.

00:06:38.370 --> 00:06:40.020
Eight pin.

00:06:40.020 --> 00:06:41.460
Two.

00:06:41.460 --> 00:06:42.940
Six pin.

00:06:42.940 --> 00:06:44.710
Two.

00:06:44.710 --> 00:06:46.900
Dual Molex.

00:06:46.900 --> 00:06:49.300
Okay, I have used adapters.

00:06:51.160 --> 00:06:57.100
Wait, what? Why is this not lighting up?

00:06:57.600 --> 00:07:02.390
Is it working? Is like the light just unplugged? Did someone screw up and not plug in the light on this?

00:07:03.030 --> 00:07:05.890
Wait, no, this doesn't make sense, because this was my first one.

00:07:05.890 --> 00:07:06.310
Wait, no, this doesn't make sense, because this was my first one.

00:07:06.310 --> 00:07:09.110
How could that not be working?

00:07:09.230 --> 00:07:14.030
Whatever the indicator LEDs might seem to think all three video cards are showing up

00:07:14.030 --> 00:07:19.790
And we are not running an SLI because a I don't have a three-way SLI bridge handy and B

00:07:20.070 --> 00:07:23.390
Because that's really not the point of what we're doing. We are not looking at gaming performance

00:07:23.390 --> 00:07:29.150
We are seeing how the Titan X compares to the quadro in more quadro like applications

00:07:29.710 --> 00:07:31.710
three Titan X's is

00:07:31.730 --> 00:07:33.270
much faster

00:07:33.270 --> 00:07:36.000
Wow, that's scaling though. Look at that

00:07:36.000 --> 00:07:40.860
It's estimating that it's only gonna take about 11 minutes to finish this scene

00:07:40.860 --> 00:07:44.880
Of course, I still don't have 11 minutes because WAN Show started 28 minutes ago

00:07:45.460 --> 00:07:47.460
Let's fire up resolve and see how that does

00:07:48.100 --> 00:08:06.160
You've got to be kidding me. I think that would have worked. I think we had all three GPUs engaged

00:08:06.780 --> 00:08:10.800
But we just triggered over current protection. I mean

00:08:11.440 --> 00:08:15.260
The truth of the matter is that none of this really affects the conclusion anyway

00:08:16.400 --> 00:08:19.440
Why does the quadro M6000 cost five?

00:08:20.180 --> 00:08:22.280
$5,000 in the olden days

00:08:22.280 --> 00:08:29.360
I would have told you that it's for its fp64 double precision compute performance because prior to the m-series

00:08:29.640 --> 00:08:33.800
That would have actually been true it performed more like a Tesla in that regard

00:08:33.960 --> 00:08:38.660
But these days it actually performs just like a g-force there too with

00:08:39.800 --> 00:08:44.600
1 32nd of the performance of what you can expect in fp 32 computations

00:08:44.600 --> 00:08:49.160
so if you were say for example doing financial analysis or some kind of

00:08:49.560 --> 00:08:52.900
Physical modeling you wouldn't choose one of these either

00:08:53.280 --> 00:08:53.800
No

00:08:53.800 --> 00:08:58.020
my friends the quadro M6000 the reason that NVIDIA and PNY both

00:08:58.200 --> 00:09:03.980
Refused to send one to me even though I explained that I was gonna explain all this that they're not for general consumers

00:09:03.980 --> 00:09:07.100
they're not for gaming in fact, they're not even for this the purpose of these is

00:09:08.080 --> 00:09:16.760
simple if you need to render scenes that require more than the amount of memory that you'll find on a consumer-grade graphics card you

00:09:16.940 --> 00:09:19.320
Will need to buy one of these so let's say for

00:09:19.320 --> 00:09:21.420
example your

00:09:21.420 --> 00:09:26.780
Pixar or you're working on the Angry Birds movie or whatever the case may be if you are anyone else

00:09:27.080 --> 00:09:29.540
then you would be better served by a

00:09:30.260 --> 00:09:33.700
Bunch of these and like a lifetime supply of M&Ms

00:09:33.700 --> 00:09:38.740
And maybe throw in a power supply that doesn't trip the OCP when you try to actually engage all of them

00:09:39.540 --> 00:09:41.700
So that's pretty much it then

00:09:42.220 --> 00:09:44.500
Thanks for watching this episode of holy sh**

00:09:45.290 --> 00:09:48.250
Online is tech tips if you guys just like this video hit the dislike button

00:09:48.250 --> 00:09:49.910
But if you liked it hit that like button and subscribe to my channel and I'll see you in the next video. Bye!

00:09:49.930 --> 00:09:55.370
Get subscribed maybe even consider checking out where to buy one of these or maybe a more sensible graphics card

00:09:55.370 --> 00:09:57.370
We'll have a link in the video description

00:09:57.450 --> 00:09:58.490
While you're down there

00:09:58.490 --> 00:10:02.650
You can also check out our merch store where we have cool shirts like this one as well as our community forum where you can

00:10:02.650 --> 00:10:09.130
Ask all of your technology related questions and get answers better answers than no you can't game on a quadro

00:10:09.330 --> 00:10:13.170
The true answer is yes, you can. No, you probably shouldn't

00:10:13.810 --> 00:10:15.810
Because you just spent

00:10:16.450 --> 00:10:19.650
$5,000 for the gaming performance of it and that's ignoring the

00:10:19.930 --> 00:10:26.770
TDI which performs exactly the same as the Titan X for a thousand dollar graphics card. Thanks again for watching. See you guys next time
