1
00:00:00,240 --> 00:00:08,480
Intel improved the thermal performance of the core i9 10900k by actually

2
00:00:05,440 --> 00:00:10,800
reducing the thickness of the CPU die

3
00:00:08,480 --> 00:00:15,759
itself improving thermal transfer but as crazy as it sounds enthusiasts had

4
00:00:13,519 --> 00:00:20,320
already tried this technique on the previous generation 9900k with a degree

5
00:00:18,960 --> 00:00:25,039
of success pun intended but

6
00:00:22,720 --> 00:00:30,560
this kind of tomfoolery doesn't come without risk i mean i'm talking deleting

7
00:00:27,680 --> 00:00:36,719
the CPU lapping down the integrated heat spreader lapping the silicon dye itself

8
00:00:34,559 --> 00:00:41,280
to improve its flatness and thermal conductivity and putting it all back

9
00:00:38,800 --> 00:00:47,520
together sam's warranty of course for our attempt we're going to use haha this

10
00:00:44,320 --> 00:00:49,920
two tools from nude cnc that will

11
00:00:47,520 --> 00:00:56,879
hopefully give us the precision that we need to get the performance we crave

12
00:00:52,719 --> 00:00:58,480
without ruining our chip in the process

13
00:00:56,879 --> 00:01:02,800
you know what i never ruined though my segways to sponsors like ridge wallet

14
00:01:00,879 --> 00:01:07,040
rage wallet wants to redefine the wallet with its compact frame and rfid blocking

15
00:01:05,119 --> 00:01:11,520
plates check out how they can help keep your wallet bulge down and use offer

16
00:01:09,520 --> 00:01:14,400
code lioness to save 10 and get free worldwide shipping

17
00:01:21,360 --> 00:01:29,280
we get into it in more detail here but in a nutshell keeping heat under control

18
00:01:26,240 --> 00:01:31,119
makes a processor last longer and or run

19
00:01:29,280 --> 00:01:37,520
faster by allowing you to increase the voltage and overclock it now as for why

20
00:01:34,960 --> 00:01:41,600
a very small reduction in thickness can help thermal transfer a lot the reason

21
00:01:40,159 --> 00:01:47,759
is that silicon is much much much less thermally

22
00:01:44,400 --> 00:01:49,759
conductive than copper so the more of

23
00:01:47,759 --> 00:01:54,479
our distance that we need to cover to get to our large heat dissipating

24
00:01:51,840 --> 00:01:57,600
heatsink can be done through copper the better off we are

25
00:01:56,479 --> 00:02:02,000
now let's have a look at the tools that

26
00:01:59,520 --> 00:02:05,600
we're going to be using today yeah we got a lot of them

27
00:02:03,360 --> 00:02:09,440
it's going to take this whole gamut of tools we have

28
00:02:07,600 --> 00:02:13,360
calipers we need to precisely measure our ihs uh allen keys that's actually

29
00:02:12,080 --> 00:02:17,599
the right one you did good we got thermal grizzly conductor knot paper

30
00:02:15,599 --> 00:02:22,239
this is just for figuring out how close the ihs is to the die

31
00:02:19,280 --> 00:02:26,640
silicon for resealing the the ihs to the CPU uh scraper we've got to scrape off

32
00:02:24,400 --> 00:02:30,400
the old uh stuff lots of scrapers lots of scrapers all the scrapers cotton

33
00:02:28,239 --> 00:02:33,280
swabs cleaning tools all this all this knock to a stuff there's

34
00:02:31,840 --> 00:02:37,200
cool debra is a cool guy go check out his channel dude but first before we can

35
00:02:35,519 --> 00:02:41,519
get to any of that we need to get a baseline measurement to tell us if what

36
00:02:39,680 --> 00:02:45,599
we've done at the end all of it was actually effective what we're looking at

37
00:02:43,360 --> 00:02:50,239
here is our temperatures as recorded by the on CPU sensors what we're going to

38
00:02:48,080 --> 00:02:53,920
do to get a baseline reading is we're going to allow it to heat up until it

39
00:02:52,080 --> 00:02:58,720
reaches a steady state which it already did about 10 minutes ago then we're

40
00:02:55,920 --> 00:03:03,360
going to take five minutes of data and average it so the ones that we are most

41
00:03:00,720 --> 00:03:08,959
interested in are both our hottest core as well as our CPU package temp so both

42
00:03:06,720 --> 00:03:14,720
of those are sitting in the neighborhood of around 81 82 degrees so have you

43
00:03:12,480 --> 00:03:18,800
deluded a CPU before yeah because i haven't

44
00:03:15,840 --> 00:03:23,680
well thing is i haven't done one of these soldered ones so Intel has been

45
00:03:22,239 --> 00:03:28,480
resorting to more and more elaborate schemes to bring

46
00:03:26,560 --> 00:03:33,360
the temperatures of their cpus down over the last couple of generations and with

47
00:03:31,040 --> 00:03:38,480
the 10 series of course it was thinning the die but with the 9 series they

48
00:03:35,599 --> 00:03:42,720
finally went to a soldered ihs which means that instead of just using a

49
00:03:40,400 --> 00:03:48,560
delitting tool to kind of slide the ihs off with a little bit of

50
00:03:44,959 --> 00:03:50,879
force we actually need to soften the

51
00:03:48,560 --> 00:03:54,560
solder between the dye and the integrated heat spreader in order to

52
00:03:53,120 --> 00:03:59,040
perform that removal that i haven't done because the

53
00:03:56,799 --> 00:04:02,720
solder performed so much better than the previous thermal compound that i didn't

54
00:04:01,120 --> 00:04:07,040
honestly really see the point uh it's better to clean that off once you've got it out of the socket because that way

55
00:04:05,439 --> 00:04:10,239
you won't accidentally lift it up and mangle any pins oh okay fair enough

56
00:04:09,040 --> 00:04:15,200
there's a free tech tip for you everybody gets oh look at that i got free ones

57
00:04:13,040 --> 00:04:18,639
i even pay you to give them to you just if you think about it i see there's

58
00:04:17,199 --> 00:04:24,320
a heat gun that should be there is a heat gun we're gonna give her give her some beans

59
00:04:21,600 --> 00:04:28,160
and then uh hopefully it'll help are we just basically following roman's guide

60
00:04:25,919 --> 00:04:32,639
on this pretty much good yeah now we gotta heat this puppy up i guess hey i'm

61
00:04:30,160 --> 00:04:36,639
gonna set that to 1 50ish yeah try 150 to start

62
00:04:37,120 --> 00:04:43,520
if i apply too much force without it being melted enough we will break it and

63
00:04:41,840 --> 00:04:48,479
this video will be over i saw a guy do it cold but he had to

64
00:04:45,440 --> 00:04:48,479
really reef on it

65
00:04:48,560 --> 00:04:54,240
what temp are we at now 190 that's a

66
00:04:51,680 --> 00:04:59,919
full half turn that we've done there now so it's an m6 screw so m6 by one so one

67
00:04:58,160 --> 00:05:06,320
one turn is one millimeter i think we can just give her okay

68
00:05:03,759 --> 00:05:11,039
there we go hey look at that that actually looks like we got it off

69
00:05:08,000 --> 00:05:12,479
clean wow that's coming off real easy

70
00:05:11,039 --> 00:05:16,880
yeah well it's still warm it's definitely like not

71
00:05:14,560 --> 00:05:20,639
perfect but it's something and actually most of the stuff on the underside of

72
00:05:18,240 --> 00:05:24,800
the ihs came off pretty well too is it normal that my anxiety is really

73
00:05:22,160 --> 00:05:27,919
high watching you guys do this yes yeah okay

74
00:05:25,759 --> 00:05:32,880
well that's probably off enough now i just got to remove the silicone

75
00:05:30,639 --> 00:05:37,680
not to be confused with silicon we need the silicon the CPU is super convenient

76
00:05:35,600 --> 00:05:41,520
to remove the silicone from because there's no surface mount components up

77
00:05:39,360 --> 00:05:45,360
here so you can just go ham with a

78
00:05:43,360 --> 00:05:48,960
plastic razor blade here this looks sketchy as heck but this is actually a

79
00:05:47,280 --> 00:05:52,160
very safe way to do this we are pretty darn close to this being stripped i

80
00:05:50,320 --> 00:05:56,960
think we're good to go so we need the end lamp this is the ihs one okay so we

81
00:05:55,280 --> 00:06:02,720
need the die and we're going to take this little piece of blue tack so we want the

82
00:06:00,880 --> 00:06:06,560
blue tac to sit completely in this pocket we don't want any blue tack

83
00:06:04,319 --> 00:06:10,639
around the actual contacting surface of the tool remember what you just like

84
00:06:08,639 --> 00:06:14,160
yeah pretty much so i brought this from uh

85
00:06:12,560 --> 00:06:18,960
from my home stash this is a granite surface plate it's not

86
00:06:16,479 --> 00:06:23,600
super high quality but it's flat and consistent so we're going to use

87
00:06:20,720 --> 00:06:30,720
this as our base reference surface and our first step is we're going to take

88
00:06:25,680 --> 00:06:32,880
some 220. did you just say 220 grit yeah

89
00:06:30,720 --> 00:06:38,400
so our first step is we need to create our zero reference so as we start it

90
00:06:36,000 --> 00:06:46,479
won't actually be touching the die oh i see so we actually start by rubbing

91
00:06:42,720 --> 00:06:48,639
away the outside plastic here

92
00:06:46,479 --> 00:06:54,319
yeah and then that tells us through these steps how far we've gone exactly

93
00:06:51,680 --> 00:06:54,319
genius

94
00:06:55,440 --> 00:07:03,360
we're getting into the first bar here which means we're approaching our kind of zero plane because this bar is larger

95
00:07:01,120 --> 00:07:07,520
than this lower bar that means that as i'm lapping i'm applying more force to

96
00:07:05,120 --> 00:07:11,360
the lower portion it's wearing away here first the beauty of these bars is that i

97
00:07:09,759 --> 00:07:16,080
can now adjust and watch from the top these bars and

98
00:07:14,080 --> 00:07:21,520
make sure that the lapping is going on very flat across the entire plane i'm not

99
00:07:19,360 --> 00:07:25,759
tilting it onto one corner or the other in that interesting

100
00:07:23,360 --> 00:07:31,280
we're approaching 0.2 millimeters now we can tell from the red marks but uh

101
00:07:28,800 --> 00:07:36,240
from our initial go with the 220 grit we've actually still got a little bit uh

102
00:07:34,160 --> 00:07:40,639
that we haven't touched again should we go

103
00:07:37,680 --> 00:07:45,120
farther go to point three more power is that a crack die

104
00:07:42,560 --> 00:07:48,639
wait what let's not go any further

105
00:07:47,199 --> 00:07:52,960
okay now we have a problem to solve because

106
00:07:50,720 --> 00:07:58,000
we wore away part of the dye we need to wear away part of the outer

107
00:07:55,360 --> 00:08:01,680
boundary of the ihs to make sure that they're sitting

108
00:08:00,000 --> 00:08:09,440
right against each other this is real

109
00:08:04,560 --> 00:08:12,319
too real man train wreck can't look away

110
00:08:09,440 --> 00:08:15,520
that ain't a good sign sure giant is it

111
00:08:13,759 --> 00:08:20,000
possible that because the ihs isn't contacting

112
00:08:18,400 --> 00:08:24,080
the core it's not holding it down oh it's got

113
00:08:22,080 --> 00:08:28,160
yeah no no pressure well there will there's some there's some but it's not

114
00:08:25,680 --> 00:08:31,199
even okay this is among the crazier things i've

115
00:08:29,759 --> 00:08:33,599
ever done but i'm gonna hold it in the socket i'll let you know if it's getting

116
00:08:32,800 --> 00:08:38,959
hot ow quick off off off off please off

117
00:08:39,599 --> 00:08:43,440
okay it's hot

118
00:08:43,760 --> 00:08:50,160
did you just do that what did i just do what are you burned

119
00:08:47,519 --> 00:08:55,279
what did that really hurt okay so let's sand the ihs but like okay

120
00:08:53,600 --> 00:08:59,680
right because well there was no heat when we tried it before

121
00:08:56,959 --> 00:09:03,839
that means no contact okay we're gonna reduce it by the same amount so we went

122
00:09:01,680 --> 00:09:07,040
to two bars on that guy we're gonna go two bars on this guy okay

123
00:09:07,519 --> 00:09:12,720
it's a much more even wear pattern on this guy but we haven't even touched the

124
00:09:10,720 --> 00:09:16,160
ihs wait yes we have have we yeah it's totally oh

125
00:09:14,240 --> 00:09:20,160
i thought there's a gap right here i can feel it

126
00:09:17,600 --> 00:09:24,160
oh it looks like it and there's no silicone left on it do

127
00:09:22,320 --> 00:09:28,160
you want some water on that um not really we're not at the point okay um

128
00:09:26,720 --> 00:09:32,080
we're not getting any clogging of the abrasives okay because we have water we

129
00:09:30,080 --> 00:09:35,440
keep it in ltteststore.com water bottles really yeah you gotta go check it out

130
00:09:33,680 --> 00:09:39,120
lttstore.com i'd say that's a pretty even four bars

131
00:09:37,200 --> 00:09:43,600
that actually looks pretty good thank you all right wow i think we're

132
00:09:41,120 --> 00:09:46,600
good yeah i think we're ready to go okay

133
00:09:54,080 --> 00:09:57,120
that's not a good noise

134
00:09:58,080 --> 00:10:04,959
what a day this one confirmed dead toast

135
00:10:02,080 --> 00:10:08,800
the second one oh it's over here we had a little bit of uh i had a little bit of

136
00:10:07,200 --> 00:10:14,480
it get a moment we had a moment let's call it a moment colin removed the ihs without testing

137
00:10:12,560 --> 00:10:18,160
the before temperatures it's fine it's not dead we can just put

138
00:10:16,640 --> 00:10:23,120
it back together we haven't sanded anything we just don't have a baseline we just

139
00:10:20,720 --> 00:10:28,000
don't have a baseline reading and as we discussed you unfortunately can't

140
00:10:25,680 --> 00:10:32,079
take your baseline reading from one CPU and apply it to another one because part

141
00:10:29,519 --> 00:10:37,680
of the whole point of this exercise is that due to the manufacturing tolerance

142
00:10:34,640 --> 00:10:40,720
issues around things as as you would

143
00:10:37,680 --> 00:10:42,399
think precise as the die height you get

144
00:10:40,720 --> 00:10:48,720
a lot of variance from one chip to another so we went to mx

145
00:10:45,200 --> 00:10:50,640
and bought yet another 99 3k we're doing

146
00:10:48,720 --> 00:10:55,440
this for you this one has now been lapped two

147
00:10:53,120 --> 00:11:01,839
millimeters on the die 0.2 millimeter excuse me 0.2 yup big difference

148
00:10:59,120 --> 00:11:07,040
the ihs has been lapped 0.4 millimeters we've got a bit of a bigger gap but

149
00:11:04,959 --> 00:11:11,920
i told colin i think it's probably okay by gap we mean between the ihs and the

150
00:11:08,880 --> 00:11:14,480
pcb yeah so what we're gonna do now

151
00:11:11,920 --> 00:11:20,079
is we're gonna liquid metal this puppy we're not going to try and do like a a a

152
00:11:17,519 --> 00:11:24,800
rando test without it properly glued on so colin's going to glue it on with

153
00:11:22,160 --> 00:11:29,040
silicone and we are finally ready to find out if this works fortunately

154
00:11:27,680 --> 00:11:33,839
no cracks don't want to use too much less is more

155
00:11:31,600 --> 00:11:36,959
according to Linus this stuff reeks

156
00:11:39,760 --> 00:11:47,279
so now we should have i did my job right perfect contact between liquid metal

157
00:11:45,200 --> 00:11:50,079
now we wait we'll do that magic wipe thing to

158
00:11:48,240 --> 00:11:52,800
three hours from now moment of truth

159
00:11:52,959 --> 00:11:58,480
what do you think well i think my silicon job was uh her

160
00:11:56,320 --> 00:12:02,320
silicone job was a little bit uh messy we might have to clean up a little

161
00:12:00,079 --> 00:12:08,160
bit so we had a little bit of silicone ooze out but uh all in all i would

162
00:12:04,880 --> 00:12:10,079
actually say yes colin did a good job of

163
00:12:08,160 --> 00:12:14,160
gluing this back together you can see we've got a little bit of a gap there

164
00:12:12,320 --> 00:12:17,839
which is exactly what we would expect and we've got what looks like an

165
00:12:16,000 --> 00:12:22,639
identically sized gap on the other side as long as shaving down the dye didn't

166
00:12:20,800 --> 00:12:26,720
kill the CPU i have a pretty high degree of

167
00:12:24,399 --> 00:12:30,320
confidence right now big if think of it this way colin

168
00:12:28,800 --> 00:12:36,320
you can't make an omelette without breaking some eggs

169
00:12:32,880 --> 00:12:36,320
the goal wasn't to make an omelet

170
00:12:36,839 --> 00:12:43,519
but contact okay

171
00:12:40,639 --> 00:12:48,880
there's postcodes that's a good sign it's a lot better than zero zero

172
00:12:45,200 --> 00:12:50,560
postcode which means zero CPU detected

173
00:12:48,880 --> 00:12:53,560
whoa hey well i saw yes

174
00:12:54,240 --> 00:12:59,360
you have to say now if it's going to be better if it's going

175
00:12:57,680 --> 00:13:04,399
to be better i think it's going to be within margin of error i didn't see how you applied

176
00:13:02,399 --> 00:13:08,560
the liquid metal but i'm going to say you were sure to consult with alex you

177
00:13:06,639 --> 00:13:13,200
did a great job and it's gonna be better i did it exactly how you did it okay so

178
00:13:10,560 --> 00:13:17,360
as like as little as possible just like you know wet it down okay

179
00:13:15,440 --> 00:13:21,440
so i ran the last bench for 12 minutes so that gives it two minutes to kind of

180
00:13:18,959 --> 00:13:27,680
acclimatize and what was the average last time i think it was 92 91 okay so

181
00:13:25,040 --> 00:13:31,120
if we're under 90 that basically tells us

182
00:13:28,720 --> 00:13:34,240
we did it the system worked yeah we did it read it

183
00:13:32,560 --> 00:13:38,560
in case it wasn't obvious by this point in the video even if this works it's not

184
00:13:36,480 --> 00:13:43,279
something that we would recommend not even a little okay so our per core

185
00:13:41,200 --> 00:13:49,120
temps all look 34 very normal yeah but they're actually

186
00:13:45,519 --> 00:13:51,200
really even they're within three degrees

187
00:13:49,120 --> 00:13:54,800
well that looks very promising very promising

188
00:13:52,399 --> 00:13:57,600
i'm not gonna celebrate yet yet

189
00:13:56,399 --> 00:13:59,920
well i'm celebrating because it's working

190
00:14:00,000 --> 00:14:06,399
it's not going up anymore i think it might have worked

191
00:14:04,399 --> 00:14:09,920
it doesn't take that long for an air cooler like this to

192
00:14:08,560 --> 00:14:13,600
um what is it heat reach heat soak or

193
00:14:12,399 --> 00:14:17,920
whatever like what would you call it yeah equilibrium oh we should check our

194
00:14:15,839 --> 00:14:20,240
clocks oh 4.8 okay

195
00:14:19,040 --> 00:14:24,800
so we're we're turboing boys we're turboing yep i

196
00:14:22,880 --> 00:14:27,680
mean 88.87 that's a four degree drop with liquid metal on a lap so i wonder

197
00:14:26,800 --> 00:14:34,639
what that would equate to an overclocking very little don't be a downer

198
00:14:32,320 --> 00:14:38,079
i'm not a downer i'm a realist well data wow

199
00:14:37,040 --> 00:14:43,680
so red is good bread is good this is great

200
00:14:41,680 --> 00:14:49,120
so this is this is a five degree gap here so as expected

201
00:14:46,000 --> 00:14:50,399
we are three to four degrees better over

202
00:14:49,120 --> 00:14:53,120
the time of our run yep

203
00:14:52,320 --> 00:14:58,000
huh yeah and i mean it did soak up a little bit i mean that's like a degree so up

204
00:14:56,560 --> 00:15:03,760
here we're probably you know closer to three degrees but

205
00:15:00,639 --> 00:15:05,120
it worked it did work and theoretically

206
00:15:03,760 --> 00:15:09,920
the uh sorry i'm just

207
00:15:07,279 --> 00:15:13,920
and theoretically the flatter you lap it and the better your contact the more it

208
00:15:12,240 --> 00:15:18,720
will work and if you win another 0.1 millimeter you might even get another

209
00:15:16,000 --> 00:15:22,959
degree fraction of a degree or a degree again

210
00:15:19,760 --> 00:15:25,279
not recommending it but uh

211
00:15:22,959 --> 00:15:29,440
darn it roman you were right and i guess you've already got your

212
00:15:27,440 --> 00:15:36,160
validation because Intel went in did exactly this on their next generation

213
00:15:31,680 --> 00:15:38,480
CPU but this was fun anyway

214
00:15:36,160 --> 00:15:38,480
fun

215
00:15:39,360 --> 00:15:45,279
speaking of fun you're going to have fun if you check

216
00:15:42,639 --> 00:15:50,399
out our sponsor micro center micro center is open to supply your work or

217
00:15:48,160 --> 00:15:55,759
learn from home technology needs the main gear element 3 laptop is available

218
00:15:52,320 --> 00:15:58,160
now for 19.99 on amazon or 18.99 on

219
00:15:55,759 --> 00:16:04,399
micro center it's got an Intel core i7 10 875 h processor it's got an NVIDIA

220
00:16:00,880 --> 00:16:06,959
GeForce rtx 2070 super 32 gigs of ddr4

221
00:16:04,399 --> 00:16:12,959
memory a terabyte solid state drive and a 17.3 inch narrow bezel

222
00:16:09,800 --> 00:16:14,959
240 hertz display check out this and

223
00:16:12,959 --> 00:16:19,440
other micro center offers over at the link in the video description if you

224
00:16:17,360 --> 00:16:24,079
guys enjoyed this video maybe you would enjoy uh oh how about checking out our

225
00:16:22,240 --> 00:16:28,959
liquid metal adventure where we put liquid metal in a phone that phone

226
00:16:26,160 --> 00:16:28,959
eventually died

227
00:16:30,240 --> 00:16:32,720
i'm fine

228
00:16:33,279 --> 00:16:38,079
i'm not burned yet

229
00:16:36,399 --> 00:16:42,639
actually this one's a little red we do need to fill out a report later why

230
00:16:40,880 --> 00:16:46,959
i'm not burned look look see look even if you weren't injured it would be a near miss

231
00:16:45,279 --> 00:16:51,560
i mean i could have taken my hand off it at any time we're gonna fill it up