1
00:00:00,280 --> 00:00:02,080
When NVIDIA released SLI,

2
00:00:02,080 --> 00:00:04,780
it enabled gamers to enjoy next generation levels

3
00:00:04,780 --> 00:00:06,760
of quality today.

4
00:00:06,760 --> 00:00:07,640
Assuming of course,

5
00:00:07,640 --> 00:00:10,000
that they could afford literally twice

6
00:00:10,000 --> 00:00:11,720
as many graphics cards.

7
00:00:11,720 --> 00:00:15,000
But as monitor resolutions have grown,

8
00:00:15,000 --> 00:00:18,080
the scalable part of the scalable link interface,

9
00:00:18,080 --> 00:00:20,280
which has been with us for over a decade,

10
00:00:20,280 --> 00:00:23,600
hasn't been able to keep up in spite of fancy,

11
00:00:23,600 --> 00:00:26,580
high bandwidth bridges like this one.

12
00:00:26,580 --> 00:00:29,900
Meanwhile, over on the professional side of things,

13
00:00:29,900 --> 00:00:30,980
NVIDIA has been pushing

14
00:00:30,980 --> 00:00:35,980
a newer inter-GPU communication protocol called NVLink.

15
00:00:38,350 --> 00:00:41,650
This essentially turns SLI up to 11.

16
00:00:41,650 --> 00:00:46,940
But why would you, the general consumer, care about that?

17
00:00:46,980 --> 00:00:51,560
Well, because NVLink is coming to consumers

18
00:00:51,560 --> 00:00:54,340
with the GeForce RTX series.

19
00:00:54,340 --> 00:00:58,600
So it is time then to ask the big question.

20
00:00:58,600 --> 00:01:00,960
Does it make gaming

21
00:01:00,960 --> 00:01:01,800
better?

22
00:01:01,800 --> 00:01:06,090
Whew, that's a lot of hardware.

23
00:01:06,090 --> 00:01:07,530
Speaking of big questions,

24
00:01:07,530 --> 00:01:09,050
have you tried GlassWire?

25
00:01:09,050 --> 00:01:12,150
Detect malware and block badly behaving apps

26
00:01:12,150 --> 00:01:14,010
on your PC or Android device.

27
00:01:14,010 --> 00:01:17,870
Use offer code Linus to get 25% off GlassWire 2.0

28
00:01:17,870 --> 00:01:28,240
at the link below.

29
00:01:28,240 --> 00:01:29,780
So one of the first things you'll notice

30
00:01:29,780 --> 00:01:31,940
about a card equipped with NVLink

31
00:01:31,940 --> 00:01:35,500
is just how big the connector fingers are

32
00:01:35,500 --> 00:01:37,240
compared to traditional SLI.

33
00:01:37,240 --> 00:01:40,340
They are more than three times as wide with way,

34
00:01:40,340 --> 00:01:41,900
way more pins.

35
00:01:41,900 --> 00:01:45,440
Like seriously, a single NVLink finger

36
00:01:45,440 --> 00:01:49,460
is wider than the entire SLI connector setup.

37
00:01:49,460 --> 00:01:52,900
It almost even looks like they're little

38
00:01:52,900 --> 00:01:55,580
PCI Express connectors, which,

39
00:01:55,580 --> 00:01:58,860
as we're about to see, isn't by accident.

40
00:01:58,860 --> 00:02:03,620
So the way that SLI works is actually a lot like,

41
00:02:03,620 --> 00:02:06,060
oh, here, I have a good prep for this.

42
00:02:06,060 --> 00:02:09,460
It's actually a lot like the older SCSI and IDE.

43
00:02:09,460 --> 00:02:14,280
One card functions as the master in the relationship

44
00:02:14,280 --> 00:02:16,580
and the other one as a slave,

45
00:02:16,580 --> 00:02:18,740
or in the case of multiple other cards,

46
00:02:18,740 --> 00:02:20,860
they would all then be slaves.

47
00:02:20,860 --> 00:02:24,420
So that means that because the master alone

48
00:02:24,420 --> 00:02:27,140
is directing the workload for those slave cards

49
00:02:27,140 --> 00:02:31,520
with at best two gigabytes per second of bandwidth

50
00:02:31,520 --> 00:02:34,300
using one of NVIDIA's high bandwidth bridges,

51
00:02:34,300 --> 00:02:36,940
you've got enough for the render results

52
00:02:36,940 --> 00:02:38,820
to be returned to the master

53
00:02:38,820 --> 00:02:43,080
and honestly, not a whole lot more.

54
00:02:43,080 --> 00:02:47,040
This is the reason why you can't simply add together

55
00:02:47,040 --> 00:02:49,540
the memory of your SLI graphics cards,

56
00:02:49,540 --> 00:02:51,520
taking two 11 gig cards and saying,

57
00:02:51,520 --> 00:02:54,620
well, I've got 22 gigs of RAM now.

58
00:02:54,620 --> 00:02:57,840
And the same is true for Team Red's Crossfire.

59
00:02:57,840 --> 00:03:02,780
By contrast, NVLink is bi-directional

60
00:03:02,780 --> 00:03:05,780
and it's configured as a mesh,

61
00:03:05,780 --> 00:03:08,900
which means that no one card is the master,

62
00:03:08,900 --> 00:03:11,420
and there are no slaves.

63
00:03:11,420 --> 00:03:14,980
Think of it more like if you were plugging computers

64
00:03:14,980 --> 00:03:18,240
into a router or a switch.

65
00:03:18,240 --> 00:03:21,540
So this, along with the extra pins

66
00:03:21,540 --> 00:03:23,820
and newer signaling protocol,

67
00:03:23,820 --> 00:03:27,380
gives these cards a lot more bandwidth,

68
00:03:27,380 --> 00:03:29,440
more than even PCI Express,

69
00:03:29,440 --> 00:03:34,440
at a total of up to 160 to 300 gigabytes per second.

70
00:03:35,620 --> 00:03:37,360
That kind of speed lets them pool resources,

71
00:03:37,360 --> 00:03:38,360
that kind of speed lets them pool resources,

72
00:03:38,360 --> 00:03:39,060
that kind of speed lets them pool resources,

73
00:03:39,060 --> 00:03:43,080
in a way that allows access to each card's memory

74
00:03:43,080 --> 00:03:46,740
and CUDA cores as though they were a single card.

75
00:03:46,740 --> 00:03:49,220
And that's perfect for the scientific

76
00:03:49,220 --> 00:03:51,240
and high-end render stations

77
00:03:51,240 --> 00:03:55,480
that NVIDIA has traditionally targeted with NVLink.

78
00:03:55,480 --> 00:03:58,040
Now, you might be thinking to yourself,

79
00:03:58,040 --> 00:04:02,160
awesome, NVLink is coming to GeForce RTX cards.

80
00:04:02,160 --> 00:04:03,640
We're gonna get those benefits.

81
00:04:03,640 --> 00:04:06,190
I'm doubling my pre-order.

82
00:04:06,190 --> 00:04:08,490
Hold your horses there, Tom.

83
00:04:08,490 --> 00:04:09,170
Yeah, it's awesome.

84
00:04:09,170 --> 00:04:12,030
But the number of links provided on RTX

85
00:04:12,030 --> 00:04:13,450
is relatively minimal,

86
00:04:13,450 --> 00:04:18,450
and the RTX cards only support SLI over the NVLink bus.

87
00:04:19,610 --> 00:04:24,010
So there will be no fancy resource pooling going on here.

88
00:04:24,010 --> 00:04:28,750
So our plan today then is to take our Quadro GP100s

89
00:04:28,750 --> 00:04:31,150
and run them both in compute mode,

90
00:04:31,150 --> 00:04:33,450
which actually disables the graphics engine,

91
00:04:33,450 --> 00:04:34,810
like we couldn't plug a display

92
00:04:34,810 --> 00:04:36,730
into these things right now if we tried,

93
00:04:36,730 --> 00:04:38,470
and in what's called SLI mode,

94
00:04:38,470 --> 00:04:41,350
to look at their gaming performance.

95
00:04:41,350 --> 00:04:43,070
Yes, yes, I know.

96
00:04:43,070 --> 00:04:45,750
This card isn't intended for gaming,

97
00:04:45,750 --> 00:04:49,730
but if you look closely at the spec of it,

98
00:04:49,730 --> 00:04:53,610
it's got HBM2 memory, yes, and more of it,

99
00:04:53,610 --> 00:04:58,610
but it's otherwise actually very similar to the GTX 1080 Ti.

100
00:04:59,750 --> 00:05:04,050
So this is probably as close as we will ever get

101
00:05:04,050 --> 00:05:08,050
to an apples-to-apples comparison between SLI and NVLink.

102
00:05:08,470 --> 00:05:13,270
Since Pascal is likely to be the only generation of products

103
00:05:13,270 --> 00:05:16,530
where both of these technologies are present.

104
00:05:16,530 --> 00:05:18,010
First up, some pre-flight tweaks

105
00:05:18,010 --> 00:05:19,550
to get everything working though.

106
00:05:19,550 --> 00:05:22,370
We needed a Quadro SLI-certified motherboard,

107
00:05:22,370 --> 00:05:26,170
so our ASUS X299 Deluxe with a Core i9-7900X

108
00:05:26,170 --> 00:05:27,630
worked nicely for this.

109
00:05:27,630 --> 00:05:31,450
And to look at NVLink's non-gaming performance,

110
00:05:31,450 --> 00:05:33,310
we needed to configure both cards

111
00:05:33,310 --> 00:05:35,750
in Tesla compute cluster mode,

112
00:05:35,750 --> 00:05:38,390
which we can check by going ahead and running...

113
00:05:38,470 --> 00:05:41,370
this command in the Windows PowerShell.

114
00:05:41,370 --> 00:05:42,790
So you can see right here,

115
00:05:42,790 --> 00:05:46,230
links one to three, or zero to three, excuse me,

116
00:05:46,230 --> 00:05:48,090
or one to, whatever the point is,

117
00:05:48,090 --> 00:05:52,610
they're all running, and that's good.

118
00:05:52,610 --> 00:05:55,930
Unfortunately, many of our benchmarks

119
00:05:55,930 --> 00:05:59,650
actually didn't cooperate very well

120
00:05:59,650 --> 00:06:03,740
with this particular setup,

121
00:06:03,740 --> 00:06:07,580
though the latest experimental Blender build managed it,

122
00:06:07,580 --> 00:06:08,420
and, whew!

123
00:06:09,360 --> 00:06:14,110
The results pretty much speak for themselves.

124
00:06:14,110 --> 00:06:17,200
Three and a half minutes for Gooseberry?

125
00:06:17,200 --> 00:06:20,700
20 seconds for BMW?

126
00:06:20,700 --> 00:06:22,300
In spite of these tests

127
00:06:22,300 --> 00:06:24,960
not being particularly memory intensive,

128
00:06:24,960 --> 00:06:28,180
we are seeing a clear advantage here.

129
00:06:28,180 --> 00:06:31,660
As for Luxmark's lower OpenCL performance scaling,

130
00:06:31,660 --> 00:06:35,320
that suggests that CUDA is a necessary ingredient

131
00:06:35,320 --> 00:06:38,260
if we wanna take full advantage of NVLink.

132
00:06:38,260 --> 00:06:39,800
Big surprise, of course.

133
00:06:40,100 --> 00:06:41,480
Not all there is to it, though.

134
00:06:41,480 --> 00:06:43,900
Remember how NVLink allows us to utilize

135
00:06:43,900 --> 00:06:45,960
all of the available memory on our cards

136
00:06:45,960 --> 00:06:48,470
as though they were one big card?

137
00:06:48,470 --> 00:06:50,310
Well, because of that,

138
00:06:50,310 --> 00:06:52,870
we can now work with much larger data sets

139
00:06:52,870 --> 00:06:55,930
than would have been possible on smaller configurations.

140
00:06:55,930 --> 00:06:58,610
And trust us, we tried on smaller configurations.

141
00:06:58,610 --> 00:07:02,610
You can see here, even our twinned GP100s

142
00:07:02,610 --> 00:07:05,130
couldn't handle this particular workload.

143
00:07:05,130 --> 00:07:09,870
So, it's time to bring out the big guns.

144
00:07:09,870 --> 00:07:14,410
Our GV100s with their new NVLink bridges

145
00:07:14,410 --> 00:07:19,410
will give us a total of 64 gigs of HBM2 memory.

146
00:07:19,710 --> 00:07:21,750
That's more than the system memory

147
00:07:21,750 --> 00:07:24,170
of even many workstations.

148
00:07:24,170 --> 00:07:25,810
And there it is.

149
00:07:25,810 --> 00:07:29,630
Our GV100s handle this just fine.

150
00:07:29,630 --> 00:07:33,910
So, that's super impressive and extremely useful

151
00:07:33,910 --> 00:07:37,630
for people with huge data sets.

152
00:07:37,630 --> 00:07:39,410
But the real thing we were after here was

153
00:07:39,410 --> 00:07:44,410
evaluating the SLI mode that is coming with the RTX series.

154
00:07:44,570 --> 00:07:48,170
So, and here we go.

155
00:07:48,170 --> 00:07:50,990
So, in a massive surprise to no one,

156
00:07:50,990 --> 00:07:55,990
the GV100s are the fastest solution on the block, for now.

157
00:07:56,410 --> 00:07:59,590
In SLI, even at 4K Ultra,

158
00:07:59,590 --> 00:08:04,030
the average frame rate never dipped below 60, which is huge.

159
00:08:04,030 --> 00:08:07,510
Nothing else can even come close to claiming that.

160
00:08:07,510 --> 00:08:09,250
What's more interesting, though,

161
00:08:09,250 --> 00:08:12,950
is when we look at the scaling figures side by side.

162
00:08:12,950 --> 00:08:15,910
So, our GP100s here,

163
00:08:15,910 --> 00:08:20,750
these guys seem to scale better than the GV100s

164
00:08:20,750 --> 00:08:22,290
in gaming and productivity,

165
00:08:22,290 --> 00:08:24,590
giving them the best scaling overall,

166
00:08:24,590 --> 00:08:28,210
which may suggest some kind of CPU bottleneck.

167
00:08:28,210 --> 00:08:30,510
As for the GTX 1080 Ti,

168
00:08:30,510 --> 00:08:33,530
well, there's huge gains to be made in gaming,

169
00:08:33,530 --> 00:08:35,850
but not as much in productivity.

170
00:08:35,850 --> 00:08:38,570
So, as you might expect with anything new,

171
00:08:39,250 --> 00:08:42,870
I think SLI doesn't scale the same way for everything,

172
00:08:42,870 --> 00:08:46,490
but it does look to be a pretty decent improvement

173
00:08:46,490 --> 00:08:48,210
over traditional SLI,

174
00:08:48,210 --> 00:08:51,750
about 10 to 23% better by our measure,

175
00:08:51,750 --> 00:08:53,930
with the potential to dramatically improve

176
00:08:53,930 --> 00:08:56,970
undesirable behavior like micro-stuttering as well,

177
00:08:56,970 --> 00:08:59,970
or even enable more than two-way SLI

178
00:08:59,970 --> 00:09:02,490
with decent scaling in the future.

179
00:09:02,490 --> 00:09:05,970
That is, depending on how much NVIDIA decides to neuter it

180
00:09:05,970 --> 00:09:08,650
compared to its professional-grade cousin.

181
00:09:08,650 --> 00:09:09,010
You, too, never know.

182
00:09:09,010 --> 00:09:10,070
You never know with those guys.

183
00:09:10,070 --> 00:09:11,990
I mean, one thing we discovered

184
00:09:11,990 --> 00:09:13,650
in the course of our testing for this video

185
00:09:13,650 --> 00:09:16,410
is that the new NVLink bridges here

186
00:09:16,410 --> 00:09:20,770
don't work with the old NVLink cards, even the pro ones.

187
00:09:20,770 --> 00:09:23,310
So, NVIDIA told us something about

188
00:09:23,310 --> 00:09:26,330
consumer NVLink bridges having fewer pins,

189
00:09:26,330 --> 00:09:29,050
or more importantly, a slightly different pin-out,

190
00:09:29,050 --> 00:09:30,590
but could they have made it work?

191
00:09:30,590 --> 00:09:31,430
I don't know.

192
00:09:31,430 --> 00:09:32,330
I don't know with those guys.

193
00:09:32,330 --> 00:09:36,150
Either way, NVLink has lots of potential

194
00:09:36,150 --> 00:09:38,310
and looks like a significant hardware upgrade

195
00:09:38,310 --> 00:09:39,550
that should only improve

196
00:09:39,550 --> 00:09:42,330
as the drivers themselves continue to improve.

197
00:09:42,330 --> 00:09:46,650
So maybe, just maybe, SLI isn't dead.

198
00:09:46,650 --> 00:09:48,450
Yet.

199
00:09:48,450 --> 00:09:50,250
Maybe.

200
00:09:50,250 --> 00:09:51,310
But you know what's not maybe?

201
00:09:51,310 --> 00:09:54,330
FreshBooks, being the small business accounting software

202
00:09:54,330 --> 00:09:56,510
custom-built for how you wanna work.

203
00:09:56,510 --> 00:09:58,350
If you're a freelancer or a small business owner,

204
00:09:58,350 --> 00:10:00,110
you need to check out FreshBooks.

205
00:10:00,110 --> 00:10:01,850
It's the simple way to be more productive,

206
00:10:01,850 --> 00:10:04,310
more organized, and to get paid faster.

207
00:10:04,310 --> 00:10:06,610
You can create and send professional-looking invoices

208
00:10:06,610 --> 00:10:08,070
in less than 30 seconds.

209
00:10:08,070 --> 00:10:09,410
You can set up online payments,

210
00:10:09,410 --> 00:10:10,570
with just a couple of clicks,

211
00:10:10,570 --> 00:10:12,470
to get paid up to four days faster,

212
00:10:12,470 --> 00:10:14,910
and you don't have to take my word for it.

213
00:10:14,910 --> 00:10:17,290
Go try FreshBooks for 30 days for free

214
00:10:17,290 --> 00:10:19,450
at freshbooks.com slash techtips.

215
00:10:19,450 --> 00:10:20,770
Just enter Linus Tech Tips

216
00:10:20,770 --> 00:10:22,270
in the How You Heard About Us section,

217
00:10:22,270 --> 00:10:24,450
so they'll know who sent you.

218
00:10:24,450 --> 00:10:25,430
So thanks for watching, guys.

219
00:10:25,430 --> 00:10:27,230
If this video sucked, you know what to do.

220
00:10:27,230 --> 00:10:28,910
But if it was awesome, get subscribed,

221
00:10:28,910 --> 00:10:32,330
hit that like button, or check out the NVLink.

222
00:10:32,330 --> 00:10:33,990
Oh, Lordy, that's awful.

223
00:10:33,990 --> 00:10:35,910
Anthony, come on.

224
00:10:35,910 --> 00:10:37,250
To where to buy the stuff we featured

225
00:10:37,250 --> 00:10:38,330
in the video description.

226
00:10:38,330 --> 00:10:39,250
Also, NVLink.

227
00:10:39,410 --> 00:10:40,250
NVLinked.

228
00:10:40,250 --> 00:10:41,450
In the description is our merch store,

229
00:10:41,450 --> 00:10:43,870
which has cool shirts, and our community forum,

230
00:10:43,870 --> 00:10:45,370
which you should totally join.

231
00:10:48,540 --> 00:10:50,300
She made me do it twice.

232
00:10:50,300 --> 00:10:52,420
I mean, it's really, it's my fault.

233
00:10:52,420 --> 00:10:54,420
I'll read anything on the teleprompter.