1
00:00:00,080 --> 00:00:06,319
it feels like every driver update we get

2
00:00:03,120 --> 00:00:09,200
from NVIDIA prompts a huge spike in

3
00:00:06,319 --> 00:00:17,320
forum posts and angry tweets saying NVIDIA ruined my FPS in

4
00:00:12,160 --> 00:00:17,320
game x with driver yyyy.

5
00:00:28,960 --> 00:00:34,239
it and we're gonna tell you after these messages

6
00:00:41,920 --> 00:00:48,800
browse privately and securely with tunnelbear the simple vpn app start your

7
00:00:46,879 --> 00:00:52,399
seven-day free trial at the link in the video description

8
00:00:50,640 --> 00:00:57,760
now before we really get into this i want to clarify that i've been a long

9
00:00:54,800 --> 00:01:04,640
time skeptic when it comes to NVIDIA and AMD conspiracy theories

10
00:01:00,399 --> 00:01:07,040
they are almost always half baked fanboy

11
00:01:04,640 --> 00:01:12,479
purchase decision reinforcing rants and frankly they make my brain hurt

12
00:01:09,920 --> 00:01:16,799
seriously Jake tried a lot of different search terms all the way from NVIDIA

13
00:01:14,720 --> 00:01:22,880
driver hertz performance to NVIDIA broken how please

14
00:01:19,439 --> 00:01:25,119
and not even one post contained concrete

15
00:01:22,880 --> 00:01:29,840
evidence of a legitimate decrease in performance over time i mean don't get

16
00:01:27,680 --> 00:01:33,840
me wrong there were a lot of posts about the 375.86

17
00:01:31,680 --> 00:01:38,720
drivers memory clock issue but no one has gathered sufficient or

18
00:01:36,159 --> 00:01:44,159
really any evidence showing that NVIDIA has purposely harmed performance of

19
00:01:41,600 --> 00:01:49,720
older cards to force their users to upgrade through a driver update so

20
00:01:48,000 --> 00:01:54,479
i decided to painstakingly

21
00:01:51,520 --> 00:01:58,640
assign Jake the task of testing it ourselves

22
00:01:55,920 --> 00:02:03,759
first we needed to pick some games we wanted a mix of ones that use only

23
00:02:00,880 --> 00:02:08,879
standard features of the directx API and at least one example that uses NVIDIA's

24
00:02:06,560 --> 00:02:12,959
proprietary gameworks then to ensure we weren't missing out on

25
00:02:11,200 --> 00:02:17,200
a particular architecture that got kneecapped because it was like too good

26
00:02:15,599 --> 00:02:24,239
compared to the new thing they needed to sell we grabbed a flagship from every

27
00:02:20,400 --> 00:02:27,040
generation from gtx 680 all the way to

28
00:02:24,239 --> 00:02:32,239
1080 ti with games and cards selected the next challenge was drivers we can't

29
00:02:30,000 --> 00:02:39,040
test all of them so we eventually settled on the drivers one release after

30
00:02:35,440 --> 00:02:41,440
the launch of each card that we used and

31
00:02:39,040 --> 00:02:47,040
the most recent one available at this time so the idea then was that this

32
00:02:44,000 --> 00:02:50,959
would show us a clear progression over

33
00:02:47,040 --> 00:02:53,200
time is NVIDIA improving ignoring or

34
00:02:50,959 --> 00:02:57,519
outright harming the performance of older cards once the new generation

35
00:02:55,840 --> 00:03:03,280
comes out so driver by driver

36
00:02:59,200 --> 00:03:05,920
card by card Jake ran almost 200 rounds

37
00:03:03,280 --> 00:03:11,760
of benchmarks collecting three metric ass tons of data then he sat down to

38
00:03:08,800 --> 00:03:15,680
organize it into some charts and the results are astonishingly boring the

39
00:03:14,000 --> 00:03:20,239
most recent release seems to be a bit of a turret across the board so there's

40
00:03:17,599 --> 00:03:25,200
that but in general not only did performance not decrease there were

41
00:03:22,480 --> 00:03:30,000
small but noticeable increases over time in almost every case

42
00:03:28,000 --> 00:03:33,200
due to the minuscule nature of these improvements

43
00:03:31,120 --> 00:03:38,319
it's hard to see where they actually start and end so we're not much closer

44
00:03:35,840 --> 00:03:43,280
to a clear picture of NVIDIA's internal policy with respect to product software

45
00:03:41,040 --> 00:03:48,159
support but it does at least tell us that NVIDIA has never rolled out a

46
00:03:45,920 --> 00:03:53,560
sandbag edition driver which doesn't mean they've never rolled

47
00:03:50,959 --> 00:03:59,760
out a bad driver 375.86 hurt performance and sli issues

48
00:03:57,760 --> 00:04:05,120
seem to come and go like shadow from final fantasy 6.

49
00:04:01,760 --> 00:04:07,519
so what is going on then where are

50
00:04:05,120 --> 00:04:10,959
people getting these ideas we think the perception mostly stems

51
00:04:09,760 --> 00:04:16,880
from this when a new game launches

52
00:04:13,519 --> 00:04:19,440
older gpus don't get the full

53
00:04:16,880 --> 00:04:22,800
optimization treatment the way that the current ones would

54
00:04:20,959 --> 00:04:29,440
that means that you might run into situations where older cards with more

55
00:04:26,320 --> 00:04:32,400
raw power might be beat by newer ones

56
00:04:29,440 --> 00:04:36,320
with less tech spot showcased this in their articles on project cars and

57
00:04:34,720 --> 00:04:41,600
battlefield 1 performance which are linked down in the video description

58
00:04:38,160 --> 00:04:44,800
where you can see a gtx 970

59
00:04:41,600 --> 00:04:47,600
outperforming a 780 ti

60
00:04:44,800 --> 00:04:51,520
and there's another major factor the tech

61
00:04:48,720 --> 00:04:56,720
not all graphic solutions are the same the goal with every new generation is to

62
00:04:54,320 --> 00:05:00,880
tailor it for the features of the game engines that are in wide use or are

63
00:04:59,120 --> 00:05:05,919
right on the horizon examples would be NVIDIA specific

64
00:05:03,360 --> 00:05:11,120
features like hair works or general rendering techniques like hbao

65
00:05:08,400 --> 00:05:17,120
plus ambient occlusion older silicon doesn't handle either of these things as

66
00:05:13,280 --> 00:05:19,840
well because they either didn't exist

67
00:05:17,120 --> 00:05:25,280
when it was being designed or NVIDIA guessed it wouldn't be widely

68
00:05:21,680 --> 00:05:28,000
implemented the work of designing a GPU

69
00:05:25,280 --> 00:05:31,199
starts years before you'll be able to buy one

70
00:05:29,199 --> 00:05:35,759
i mean if you were in the pc scene a few years back you might remember this

71
00:05:32,960 --> 00:05:42,240
happening with tessellation NVIDIA cards were and still seem to be much faster at

72
00:05:39,199 --> 00:05:45,680
rendering tessellated meshes to such a

73
00:05:42,240 --> 00:05:47,840
degree that AMD had to add AMD optimized

74
00:05:45,680 --> 00:05:53,440
tessellation to their drivers basically removing a bunch of it and when crytek

75
00:05:50,840 --> 00:05:58,639
controversially put an overkill amount of tessellation in crisis too

76
00:05:55,680 --> 00:06:04,400
it completely tanked the ability of similarly performing AMD cards compared

77
00:06:01,680 --> 00:06:08,960
to their NVIDIA counterparts so in this case your older hardware

78
00:06:06,479 --> 00:06:13,680
literally wasn't made for the questionable decisions of game devs

79
00:06:11,919 --> 00:06:18,400
in newer games so in conclusion then from our testing

80
00:06:16,160 --> 00:06:23,840
NVIDIA does not appear to be slowing down older cards to encourage upgrades

81
00:06:21,840 --> 00:06:28,080
the better performance from newer cards probably has more to do with their

82
00:06:25,600 --> 00:06:32,000
engineers not just sitting on ass all day i mean look at it from their

83
00:06:29,280 --> 00:06:36,000
perspective if they don't optimize this from generation to generation jensen

84
00:06:34,400 --> 00:06:41,039
will yell at them what does seem like it's happening is

85
00:06:38,160 --> 00:06:46,000
that newer game features aren't getting as much optimization for older cards

86
00:06:44,400 --> 00:06:51,360
but i also think it's a little unreasonable to get too mad at NVIDIA

87
00:06:48,479 --> 00:06:56,400
for this they support cards as far back as the 8000 series on Windows 10 and

88
00:06:54,400 --> 00:06:59,840
while they might not perform clock for clock the same as they used to they

89
00:06:58,080 --> 00:07:04,319
simply weren't built for modern rendering technologies so if you want

90
00:07:02,160 --> 00:07:07,199
bleeding edge performance you need to be on the bleeding edge in terms of

91
00:07:05,360 --> 00:07:11,120
hardware whether or not that's worth bleeding your wallet dry

92
00:07:09,120 --> 00:07:14,800
that one's up to you you know what the funny thing about

93
00:07:12,800 --> 00:07:19,759
squarespace is all of their gorgeous website templates

94
00:07:17,360 --> 00:07:26,160
feature responsive design so your website looks great whether it's

95
00:07:22,400 --> 00:07:29,039
on a square screen or a rectangular one

96
00:07:26,160 --> 00:07:33,280
or even if you turn it this way and the best thing about squarespace is

97
00:07:30,639 --> 00:07:38,400
they make it so anyone can build a website it starts at just 12 bucks a

98
00:07:35,520 --> 00:07:43,759
month they've got 24 7 support via live chat and email and their cloud-based

99
00:07:40,800 --> 00:07:48,880
website building tool has tons of great features like their logo designer

100
00:07:46,400 --> 00:07:53,520
commerce every website comes with a free online store their cover pages feature

101
00:07:51,199 --> 00:07:59,440
that lets you build a simple one-page online presence in minutes and the fact

102
00:07:56,800 --> 00:08:03,280
that you can try it with no strings attached start a trial with no credit

103
00:08:01,759 --> 00:08:08,400
card required and start building your website today then when you sign up for

104
00:08:05,919 --> 00:08:14,560
squarespace go to squarespace.com LTT and use offer code LTT to get 10 off

105
00:08:15,039 --> 00:08:20,800
so thanks for watching guys let us know if there are any other brands or

106
00:08:18,960 --> 00:08:24,319
conspiracy theories like this that you'd like us to investigate if this video

107
00:08:23,039 --> 00:08:28,720
sucked you know what to do but if it was awesome get subscribed hit that like button or check out the link to where to

108
00:08:26,879 --> 00:08:32,240
buy the stuff we featured in the video description also link down there is our

109
00:08:30,720 --> 00:08:35,770
merch store which has cool shirts like this one and our community forum which

110
00:08:34,320 --> 00:08:42,080
you should totally join 384.76 wait they make my brain hurt

111
00:08:45,760 --> 00:08:50,000
there were small but no

112
00:08:50,000 --> 00:08:57,040
i always hate that after these messages these messages

113
00:08:54,880 --> 00:09:00,040
and now i am these messages damn