1
00:00:00,240 --> 00:00:07,440
these two computers have the exact same specifications which means their

2
00:00:04,560 --> 00:00:13,920
performance is exactly the same right wrong when we tested the Corsair one

3
00:00:10,139 --> 00:00:16,859
A200 we noticed some fishy results so we

4
00:00:13,920 --> 00:00:21,600
built a near exact replica except for the case motherboard radiator and memory

5
00:00:19,220 --> 00:00:26,400
and don't you worry we'll explain why the memory is different later then we

6
00:00:23,640 --> 00:00:31,380
ran the same benchmarks and I'm pretty sure the results will surprise you

7
00:00:28,800 --> 00:00:36,180
No actually that's a lie I am completely sure the results will surprise you but

8
00:00:34,440 --> 00:00:39,660
won't surprise you is this message from our sponsor glassware are you worried

9
00:00:38,340 --> 00:00:43,920
that someone else is using your Wi-Fi glasswire can alert you anytime a new

10
00:00:42,059 --> 00:00:48,860
device joins so use offer code Linus to get 25 off glassware at the link in the

11
00:00:46,260 --> 00:00:48,860
video description

12
00:00:55,879 --> 00:01:02,879
our Corsair 1a200 boasts impressive

13
00:00:59,399 --> 00:01:06,600
specs it's got an AMD 5900x 12 core CPU

14
00:01:02,879 --> 00:01:08,700
a full fat RTX 3080 GPU one terabyte of

15
00:01:06,600 --> 00:01:14,460
NVMe storage with two terabytes of hard drive-based storage and 32 gigs of 3200

16
00:01:12,060 --> 00:01:19,439
megahertz RAM all in this Nifty little case with separate 170 and 240

17
00:01:17,760 --> 00:01:25,920
millimeter all-in-one liquid coolers for each the CPU and the GPU wait 170

18
00:01:23,759 --> 00:01:29,939
millimeter well should be better than 120 at least so then

19
00:01:28,560 --> 00:01:35,880
what the heck is up with this performance I mean to be clear 60 FPS in

20
00:01:33,360 --> 00:01:42,479
Shadow of the Tomb Raider at 4K with RTX enabled is a lot of performance but the

21
00:01:38,939 --> 00:01:45,180
Corsair 1a200 also costs a lot of money

22
00:01:42,479 --> 00:01:50,340
so I expect it to leave absolutely nothing on the table and something

23
00:01:47,340 --> 00:01:53,119
smells a little funny here

24
00:01:50,340 --> 00:01:53,119
you smell that

25
00:01:53,460 --> 00:02:01,079
it smells like heat it's going up my nose

26
00:01:57,860 --> 00:02:04,079
ah there's your problem despite this

27
00:02:01,079 --> 00:02:07,560
machine being liquid cooled it's CPU

28
00:02:04,079 --> 00:02:09,840
idles at around 50 degrees and check

29
00:02:07,560 --> 00:02:14,520
this out just about the second you put it under any kind of load with a print

30
00:02:11,940 --> 00:02:21,239
screen and this paint and we're going to paste this you're gonna see it's

31
00:02:17,160 --> 00:02:24,239
actually jumped a full 10 degrees

32
00:02:21,239 --> 00:02:28,500
just since I took the screenshot

33
00:02:24,239 --> 00:02:31,200
what even is that and get this rcpu

34
00:02:28,500 --> 00:02:36,000
never saw the Max rated boost speed of 4.8 gigahertz throughout our testing I

35
00:02:33,840 --> 00:02:41,040
mean it is important to note that AMD's fine print specifies that that is a

36
00:02:38,340 --> 00:02:47,700
single core boost speed but still seeing 4.8 gigahertz on the box and only

37
00:02:43,860 --> 00:02:50,340
actually seeing 4.5 in games and 4.0 in

38
00:02:47,700 --> 00:02:56,220
cinebench can be very confusing to the average user especially when I'm pretty

39
00:02:53,340 --> 00:02:59,580
sure that this chip can do better and we're not the only ones experiencing

40
00:02:57,540 --> 00:03:05,400
issues we've found a few posts on Corsair's Forum discussing poor thermals

41
00:03:01,739 --> 00:03:08,160
on the 5900x and 5950x equipped models

42
00:03:05,400 --> 00:03:14,220
it's not surprising considering how small form factor builds do tend to run

43
00:03:10,800 --> 00:03:16,019
harder but it's certainly not ideal the

44
00:03:14,220 --> 00:03:20,640
thing is though many modern computers are actually designed to run at or near

45
00:03:19,080 --> 00:03:24,239
their thermal limits without it significantly impacting their

46
00:03:22,140 --> 00:03:28,980
performance so the only way for us to know just how much this is hindering us

47
00:03:26,760 --> 00:03:33,319
is to build up another machine to compare and thanks to movie Magic

48
00:03:40,620 --> 00:03:45,620
foreign

49
00:03:45,620 --> 00:03:54,000
this one is already done she's a chunky boy but

50
00:03:50,940 --> 00:03:56,879
spec for spec this sucker is identical

51
00:03:54,000 --> 00:04:01,319
to our Corsair 1a200 except for a couple things as I mentioned before we used a

52
00:03:59,700 --> 00:04:07,620
different motherboard to match our case size and our 3200 megahertz memory

53
00:04:04,739 --> 00:04:13,080
actually had to be manually adjusted to be slower to match the loose timings on

54
00:04:10,319 --> 00:04:18,900
the RAM included with the one and since we don't have a 170 millimeter red we

55
00:04:15,900 --> 00:04:23,460
went with a 240 since 120 is definitely not enough during a 10 minute stress

56
00:04:21,060 --> 00:04:28,979
test running cinebench R23 the Corsair one reached a sweltering peak of 90

57
00:04:26,520 --> 00:04:35,280
degrees Celsius running at just over 4 gigahertz meanwhile our clone here with

58
00:04:32,040 --> 00:04:37,440
a similar Corsair mind you 240

59
00:04:35,280 --> 00:04:44,460
millimeter radiator ended up in the mid 70s at just over 4.1 gigahertz that's a

60
00:04:40,740 --> 00:04:48,060
difference of just over 15 degrees

61
00:04:44,460 --> 00:04:50,400
but the thing is that clock speed Delta

62
00:04:48,060 --> 00:04:56,400
doesn't fully explain the performance that I feel like is missing here what

63
00:04:53,639 --> 00:05:02,400
else is going on in both of our systems the RTX 3080 gpus easily reached their

64
00:04:59,820 --> 00:05:07,620
maximum boost clock of 1950 megahertz and maintained nearly identical average

65
00:05:05,400 --> 00:05:12,960
clock speeds throughout our testing but while our air cooled car barely

66
00:05:10,560 --> 00:05:18,960
pushed past 60 degrees running shadow of the Tomb Raider maxed out the liquid

67
00:05:15,180 --> 00:05:21,780
cooled Corsair one peaked at over 70 in

68
00:05:18,960 --> 00:05:26,580
the same test now 70 degrees under load is fine for a GPU and it's clear from

69
00:05:25,020 --> 00:05:32,820
the narrow two and a half percent difference in results that 80 degrees on

70
00:05:29,039 --> 00:05:34,320
the CPU is also fine that is if you

71
00:05:32,820 --> 00:05:38,400
don't mind trading two and a half percent of your performance for a super

72
00:05:36,419 --> 00:05:42,720
gorgeous and compact design like this one but hold on a second

73
00:05:40,320 --> 00:05:47,759
not all games behave the same take a look at CS go yeah yeah sure both

74
00:05:45,479 --> 00:05:53,039
machines manage around 600 frames per second so you're going to be having a

75
00:05:49,620 --> 00:05:56,479
pretty good time but that is a six

76
00:05:53,039 --> 00:05:59,280
percent difference in raw performance so

77
00:05:56,479 --> 00:06:03,960
it's clear that the more CPU bottlenecked our application is the more

78
00:06:01,560 --> 00:06:09,479
the Corsair ones borderline cooling system is likely to affect it

79
00:06:06,240 --> 00:06:11,580
and this Rabbit Hole goes even deeper

80
00:06:09,479 --> 00:06:17,960
yeah man I thought this was just going to be a simple system review

81
00:06:14,039 --> 00:06:19,740
that makes sad Linus even sadder Linus

82
00:06:17,960 --> 00:06:25,560
lttstore.com we actually built our clone system here

83
00:06:22,139 --> 00:06:27,960
with a fatal flaw watch as we transform

84
00:06:25,560 --> 00:06:30,380
it into a system that performs even better

85
00:06:49,979 --> 00:06:52,979
dude

86
00:06:54,680 --> 00:07:00,180
holy hot monkey bananas going back to CS

87
00:06:58,560 --> 00:07:05,880
go here there's something I didn't mention last time average FPS not really

88
00:07:04,139 --> 00:07:11,880
a great indication of the gaming experience because you don't notice

89
00:07:09,120 --> 00:07:17,759
animation smoothness when it's good you notice it when it sucks that's why a CPU

90
00:07:15,000 --> 00:07:22,979
that's thermal throttling at all is such a problem as it speed fluctuates you'll

91
00:07:20,460 --> 00:07:27,900
experience extremely short but sometimes drastic dips in performance and we can

92
00:07:26,819 --> 00:07:33,180
measure those so looking at the worst one percent of

93
00:07:30,479 --> 00:07:38,940
our frame times our clone here beat the Corsair one by almost 15 right out of

94
00:07:36,060 --> 00:07:46,319
the gate and our upgraded clone takes that lead to almost 20 percent just for

95
00:07:43,440 --> 00:07:54,060
adding a bigger radiator I mean the Corsair one is still over 300 FPS so all

96
00:07:50,699 --> 00:07:56,039
right no problem in this game but these

97
00:07:54,060 --> 00:07:59,639
effects can be measured and more importantly felt in other situations as

98
00:07:58,620 --> 00:08:05,160
well let's take a look at cinebench our clone

99
00:08:02,880 --> 00:08:10,139
with the Dual radiator performed nearly identically to our course R1 but all of

100
00:08:07,979 --> 00:08:14,099
a sudden we throw a triple Radiator in and it manages a three percent advantage

101
00:08:12,360 --> 00:08:20,280
over both of them this is probably because our 12 core

102
00:08:16,440 --> 00:08:23,220
5900x is right at the limit of what a

103
00:08:20,280 --> 00:08:28,020
dual 120 millimeter AIO can handle so bringing that coolant temperature down

104
00:08:25,259 --> 00:08:33,360
lets the CPU opportunistically boost up higher and more often

105
00:08:30,360 --> 00:08:36,240
of course this raises the question why

106
00:08:33,360 --> 00:08:41,279
is it so hard for a 170 millimeter radiator to cool a CPU That's rated at

107
00:08:38,219 --> 00:08:43,320
just 105 Watts TDP when a slightly

108
00:08:41,279 --> 00:08:48,480
larger Radiator on the other side of the same computer is cooling a GPU That's

109
00:08:45,480 --> 00:08:50,160
rated at 320 Watts that's over double

110
00:08:48,480 --> 00:08:56,100
well there's a couple of reasons for this first is that the die of the GPU

111
00:08:52,920 --> 00:08:58,920
like this one is much larger and more

112
00:08:56,100 --> 00:09:03,540
surface area means more thermal transfer second GPU coolers are pre-installed at

113
00:09:01,980 --> 00:09:09,180
the factory where they can be placed directly that's a pun on the GPU itself

114
00:09:07,200 --> 00:09:13,680
rather than having a heat spreader in between our die lapping Adventure video

115
00:09:11,339 --> 00:09:18,779
by the way is a great look at just how many degrees you can literally shave off

116
00:09:16,320 --> 00:09:22,200
with a steady hand and some risky work now some of you have probably noticed

117
00:09:20,459 --> 00:09:26,820
I've been very focused on thermal differences between our systems here it

118
00:09:24,779 --> 00:09:33,180
is possible that the power delivery on the ASRock b550 Phantom gaming ITX AC is

119
00:09:30,180 --> 00:09:35,399
holding us back but given that our

120
00:09:33,180 --> 00:09:40,440
performance continues to improve by going to a bigger radiator it looks like

121
00:09:37,740 --> 00:09:45,120
thermals are the culprit and in any case it changes nothing about this machine

122
00:09:42,300 --> 00:09:50,760
costing 400 dollars more than Corsair's own similarly equipped Vengeance a 7200

123
00:09:47,700 --> 00:09:53,040
while underperforming compared to what

124
00:09:50,760 --> 00:09:59,220
the hardware is capable of I mean we poked at some ideas to fix the

125
00:09:55,920 --> 00:10:01,800
one A200 we updated the BIOS we tried

126
00:09:59,220 --> 00:10:06,120
undervolting with one osmus's tool which resulted in hard crashes we tried tuning

127
00:10:04,260 --> 00:10:11,279
the fan curve although that gave us some really funky behavior that doesn't seem

128
00:10:08,100 --> 00:10:13,500
to be isolated to our unit and I mean

129
00:10:11,279 --> 00:10:16,920
I don't know we could reapply a high quality thermal paste or Put liquid

130
00:10:15,180 --> 00:10:24,240
metal on it or something but the thing is this is a pre-built For Better or For

131
00:10:21,300 --> 00:10:28,560
Worse most people are not going to do any of that stuff and it should just

132
00:10:26,940 --> 00:10:33,420
work out of the box and besides some people might not even

133
00:10:31,620 --> 00:10:38,580
care about the trade-off they're making maybe for you form means more than

134
00:10:36,060 --> 00:10:43,440
function and the courser rate 200 might be worthwhile from your perspective I

135
00:10:41,040 --> 00:10:47,160
mean it still is a downright sexy machine that uses standardized Parts

136
00:10:45,060 --> 00:10:51,660
which allows you to upgrade it down the line if you're so inclined it's just

137
00:10:49,320 --> 00:10:55,560
that I do think that Corsair should revise the case in cooling design if

138
00:10:54,180 --> 00:10:59,040
they're going to keep packing it with the latest and greatest Hardware going

139
00:10:57,420 --> 00:11:02,880
forward just like I'm going to keep packing this video with sponsors thanks

140
00:11:01,200 --> 00:11:06,959
to Ting mobile for sponsoring this video Ting mobile has new rates that make it

141
00:11:04,920 --> 00:11:10,980
easier to see how much you can save by switching they've got unlimited talk and

142
00:11:08,880 --> 00:11:17,820
text for ten dollars data plan starting at 15 with their new set 12 plan with 12

143
00:11:14,339 --> 00:11:20,100
gigs of data for 35 bucks and unlimited

144
00:11:17,820 --> 00:11:23,519
data for 45 dollars a month if you're looking for a great deal Ting mobile is

145
00:11:21,779 --> 00:11:27,240
offering their unlimited plan for just 25 a month for the first three months

146
00:11:25,500 --> 00:11:31,440
right now and if you don't prefer those package plans and you just want pay for

147
00:11:29,339 --> 00:11:34,920
what you use guess what those old plans they're still here they're called Ting

148
00:11:32,940 --> 00:11:38,700
Mobile's Flex plans and they're just five dollars a gig data can even be

149
00:11:36,959 --> 00:11:42,720
shared if you have a family plan so you can connect more phones to save more you

150
00:11:40,920 --> 00:11:46,740
get the same Nationwide coverage in the US and award-winning customer service

151
00:11:44,700 --> 00:11:50,940
and pretty much any phone will work with Ting mobile so check them out at

152
00:11:48,800 --> 00:11:55,860
Linus.ting.com and you'll get a 25 credit

153
00:11:52,440 --> 00:11:57,240
so did you guys buy an A200 I mean even

154
00:11:55,860 --> 00:12:04,140
though it's expensive it might just be the easiest way to get your hands on a 5900x and a 3080 so we understand

155
00:12:01,800 --> 00:12:07,200
entirely if you did just hey make sure you keep it well ventilated and let us

156
00:12:06,240 --> 00:12:12,899
know in the comments if you're running into similar issues if you guys like this video make sure to check out my

157
00:12:11,040 --> 00:12:18,120
easiest PC build in years the machine looks a lot like the Corsair

158
00:12:15,000 --> 00:12:20,180
one but it's not it's like super easy to

159
00:12:18,120 --> 00:12:20,180
build