1
00:00:00,160 --> 00:00:08,000
so everyone knows that for a CPU to last longer and perform optimally it needs to

2
00:00:05,600 --> 00:00:13,440
stay cool which makes it all the more confusing that Intel over the last few

3
00:00:11,040 --> 00:00:18,000
generations at least according to comments from the

4
00:00:15,040 --> 00:00:23,439
tech community has been using seemingly the worst possible thermal compound

5
00:00:20,720 --> 00:00:28,640
between the dyes of their processors and the integrated heat spreaders that go on

6
00:00:26,080 --> 00:00:33,200
top of them but the problem is that the only way to evaluate that thermal

7
00:00:30,560 --> 00:00:36,800
compound that we have now is to take off the spreader replace it with something

8
00:00:34,719 --> 00:00:40,559
else and check the results but we're actually changing a lot of variables

9
00:00:39,200 --> 00:00:48,160
when we do that so we devised a test to find out once and

10
00:00:44,320 --> 00:00:50,480
for all just how bad is Intel's thermal

11
00:00:48,160 --> 00:00:56,640
paste alex here has got an entire bag of Intel cpus with

12
00:00:54,239 --> 00:01:00,640
their junk thermal compound and we're going to be harvesting all of it and

13
00:00:58,960 --> 00:01:06,720
then comparing it with the well-respected nth1 from Noctua a

14
00:01:03,840 --> 00:01:13,760
premium thermal paste on a d-lidded liquid metaled extreme edition CPU

15
00:01:09,920 --> 00:01:15,920
so come along for the ride

16
00:01:13,760 --> 00:01:19,840
tunnelbear makes really simple privacy apps so people can enjoy a more open

17
00:01:18,080 --> 00:01:31,759
internet to try it out for free visit tunnelbear.com LTT

18
00:01:31,759 --> 00:01:38,720
to get a baseline we first needed to apply some nth1 thermal paste from

19
00:01:35,840 --> 00:01:44,479
Noctua and mount the nhd15 cooler that we'd be using for our tests it performs

20
00:01:41,280 --> 00:01:47,280
as well if not better than large aios

21
00:01:44,479 --> 00:01:50,799
but is way easier to mount all of our testing was done in our conference room

22
00:01:49,040 --> 00:01:55,200
to keep ambient temperatures locked at 24 degrees at stock speeds with

23
00:01:53,280 --> 00:02:00,560
multi-core enhancement turned off and the CPU fan at full blast

24
00:01:58,000 --> 00:02:05,200
30 minutes starting now while our control thermal test was running we had

25
00:02:02,880 --> 00:02:10,399
some time to begin what we're calling the grand tim harvest

26
00:02:08,479 --> 00:02:14,640
we aren't totally sure if these processors work but

27
00:02:12,480 --> 00:02:18,640
we don't want to just destroy them like so many things d-litting is pretty

28
00:02:16,720 --> 00:02:23,120
simple when you have the right tools place the CPU in the little vise tighten

29
00:02:21,040 --> 00:02:27,680
until the top moves a bit and then remove the lid which normally you would

30
00:02:25,280 --> 00:02:32,640
do by hand but these ones seemed pretty reluctant to let go and we wanted to get

31
00:02:30,319 --> 00:02:36,800
this over with once all the cpus were opened we scraped off all the thermal

32
00:02:34,959 --> 00:02:41,200
interface material we could get at with a guitar pick and then carefully placed

33
00:02:39,280 --> 00:02:46,239
it on an esd bag now you might be thinking guys

34
00:02:43,760 --> 00:02:51,599
those are third gen core series chips that came out six years ago

35
00:02:48,720 --> 00:02:56,560
how can this be a fair comparison well the longevity of Intel's paste is one of

36
00:02:54,319 --> 00:03:01,519
the main justifications that we have heard for them using it also it's a heck

37
00:02:59,599 --> 00:03:06,800
of a lot easier to get a bag of third-gen chips than uh 8th gen

38
00:03:05,200 --> 00:03:11,519
chips oh the savagery

39
00:03:09,440 --> 00:03:14,080
look at this this is like they came in and

40
00:03:12,480 --> 00:03:17,599
wrecked up the place well you don't have to do that to them

41
00:03:15,680 --> 00:03:22,800
well that's a fair point okay so oh so this is it yeah that's

42
00:03:20,720 --> 00:03:29,280
all of it right there don't blow on it i'm not gonna blow on it so this is our

43
00:03:26,080 --> 00:03:31,760
harvested thermal interface material

44
00:03:29,280 --> 00:03:39,040
from four cpus this looks like it should be more than enough for our 7980xe

45
00:03:39,040 --> 00:03:45,360
after inspection we found that our first application of thermal paste was

46
00:03:42,959 --> 00:03:49,120
excellent so these results should be perfect for our comparison do you think

47
00:03:47,440 --> 00:03:53,680
there's like a good way to get this on there i think we're in uncharted waters here

48
00:03:52,080 --> 00:03:58,640
right now i hadn't really given it a lot of thought to be perfectly honest with you

49
00:03:56,799 --> 00:04:03,599
several methods of applying the old thermal goop were considered including

50
00:04:00,720 --> 00:04:07,280
just placing it onto the CPU but without a way to spread it around we

51
00:04:05,680 --> 00:04:11,519
were concerned that we wouldn't get adequate coverage

52
00:04:08,799 --> 00:04:14,799
mashing it into a ball seemed promising but it just kind of

53
00:04:13,040 --> 00:04:19,919
flaked out and turned into a weird powder finally we turned to what seemed like

54
00:04:17,919 --> 00:04:24,639
the soberist idea soaking it in alcohol

55
00:04:23,040 --> 00:04:30,080
is that doing anything i'm not sure yet alex although this

56
00:04:27,199 --> 00:04:35,280
doesn't seem like the best of ideas it's actually not the worst ever either the

57
00:04:32,720 --> 00:04:41,440
alcohol should allow what is currently thermal dust to turn back into more of a

58
00:04:38,880 --> 00:04:45,360
paste for application and then quickly evaporate away

59
00:04:43,120 --> 00:04:50,320
our spudger ended up being too fine of a tool so we swapped it out using the

60
00:04:47,759 --> 00:04:54,720
allen key from the d-litting tool as a makeshift pestle for our bottle cap

61
00:04:52,720 --> 00:04:58,800
mortar making a reconstituted thermal paste

62
00:04:57,600 --> 00:05:06,000
here oh i don't miss it what's going on i

63
00:05:01,360 --> 00:05:06,000
spilled the alcohol so ridiculous

64
00:05:06,320 --> 00:05:11,360
this alcohol spill might have been a blessing in disguise

65
00:05:09,360 --> 00:05:14,320
this may be one of the dumbest things we've ever done yeah i was trying to

66
00:05:12,560 --> 00:05:18,240
think of dumber things but this is really up there

67
00:05:16,240 --> 00:05:22,160
should we try to spread it out more or i mean we got the particle size pretty

68
00:05:20,240 --> 00:05:26,479
small ish with the tim spread out to the best of

69
00:05:23,919 --> 00:05:33,120
our abilities all we could do was wait for the alcohol to dry

70
00:05:33,120 --> 00:05:39,360
like i guess we should just go for it now

71
00:05:36,800 --> 00:05:43,520
this we have no choice but we've committed this much time and

72
00:05:41,039 --> 00:05:48,880
energy to this now initial temperatures at idle had us at just over six degrees

73
00:05:46,400 --> 00:05:53,039
over ambient meaning that our application worked at least to some

74
00:05:51,039 --> 00:05:57,759
degree well it might just be more oddly now it was like 30 watts before outfit

75
00:05:55,360 --> 00:06:02,160
like eight okay so you don't think it's the

76
00:05:59,919 --> 00:06:06,400
material heating up and spreading out i think it's just Windows as all the way

77
00:06:03,919 --> 00:06:10,639
started but we wouldn't know how well it worked until we hit it with an intense

78
00:06:08,639 --> 00:06:15,199
load uh it didn't i mean it didn't instantly

79
00:06:12,800 --> 00:06:19,319
overheat but this is probably a fair bit hotter

80
00:06:16,319 --> 00:06:19,319
88.

81
00:06:20,880 --> 00:06:29,120
now our earlier test gave us a maximum temperature of 60 degrees and an average

82
00:06:26,479 --> 00:06:36,000
temperature across the course of 50 degrees while maintaining a CPU power of

83
00:06:31,919 --> 00:06:38,880
about 150 watts with our reconstituted

84
00:06:36,000 --> 00:06:45,120
Intel tim we almost immediately blew past that reaching a maximum of up to 89

85
00:06:42,240 --> 00:06:52,240
degrees ouch before settling into an average of 75 degrees to be clear this

86
00:06:49,680 --> 00:06:56,160
is a totally acceptable number that shouldn't harm the processor within its

87
00:06:54,479 --> 00:07:03,599
expected lifespan but 25 degrees more than the enthusiast

88
00:07:00,400 --> 00:07:06,560
solution still isn't great but

89
00:07:03,599 --> 00:07:10,080
is the problem the paste or the application this is the real moment of

90
00:07:08,880 --> 00:07:18,000
truth because we weren't expecting it to perform as well anyway right

91
00:07:15,039 --> 00:07:22,880
did it spread actually it did i think if we described that as

92
00:07:20,160 --> 00:07:27,680
sub-optimal but adequate it doesn't seem to have um

93
00:07:25,919 --> 00:07:32,880
liquided up at all so with our decent enough application we

94
00:07:31,039 --> 00:07:38,000
feel at this point that it's safe to say that the 25 degree difference was at

95
00:07:35,440 --> 00:07:44,960
least largely due to the poor thermal conductivity of the paste that is inside

96
00:07:41,599 --> 00:07:47,120
the CPU and supporting our theory is the

97
00:07:44,960 --> 00:07:50,800
fact that this difference is actually pretty close to the temperature

98
00:07:48,639 --> 00:07:56,560
difference that you can achieve after deleting a CPU and replacing that stock

99
00:07:54,000 --> 00:08:01,599
thermal interface material so to answer our original question there are a number

100
00:07:59,280 --> 00:08:07,039
of variables that you end up changing when swapping out thermal paste in your

101
00:08:04,000 --> 00:08:08,720
delidded processor including mounting

102
00:08:07,039 --> 00:08:13,199
pressure the thickness of the application the silicone seal the

103
00:08:11,199 --> 00:08:20,639
flatness of the heat spreader but as everyone expected it's Intel's choice

104
00:08:17,039 --> 00:08:23,039
of a longevity optimized tim rather than

105
00:08:20,639 --> 00:08:29,680
a performance optimized one and their decision to eschew solder that

106
00:08:26,000 --> 00:08:31,039
makes up the bulk of the difference

107
00:08:29,680 --> 00:08:35,200
speaking of difference maybe you're looking for a

108
00:08:33,120 --> 00:08:39,680
way to make a difference to your small business or you do freelance work well

109
00:08:37,519 --> 00:08:44,240
freshbooks is the online cloud-based accounting solution that's built for how

110
00:08:42,240 --> 00:08:48,800
you work it allows you to create and send professional looking invoices in

111
00:08:46,480 --> 00:08:53,120
just a couple of clicks set up online payments easily so you can get paid up

112
00:08:50,640 --> 00:08:56,959
to four days faster see when your client has seen your invoice putting an end to

113
00:08:54,720 --> 00:09:00,959
the guessing games and take the full functionality of the freshbooks platform

114
00:08:59,200 --> 00:09:06,720
out with you on the go so you can track expenses and track hours and all of that

115
00:09:03,440 --> 00:09:07,920
good stuff on Android and iOS so check

116
00:09:06,720 --> 00:09:12,640
them out at the link in the video description that's freshbooks.com

117
00:09:10,240 --> 00:09:16,640
tech tips and just enter tech tips in the how did you hear about us section

118
00:09:14,160 --> 00:09:21,040
you can get a free 30-day trial so thanks for watching guys a huge shout

119
00:09:18,800 --> 00:09:25,760
out to free geek vancouver for providing us with all these cpus to harvest uh

120
00:09:23,760 --> 00:09:29,600
thermal interface material from if this video sucked you guys know what to do

121
00:09:27,279 --> 00:09:32,480
but it was awesome get subscribed hit that like button or check out the link

122
00:09:31,040 --> 00:09:35,519
to where to buy the stuff we featured in the video description also link down

123
00:09:34,240 --> 00:09:41,440
there is our merch store which has cool shirts like this one and our community

124
00:09:37,519 --> 00:09:41,440
forum which you should totally join