1
00:00:02,560 --> 00:00:10,160
It seems this RAM crisis just won't stop. Earlier today, a customer's wallet

2
00:00:07,680 --> 00:00:15,200
was brutalized at the local computers RS. Live at the scene is special

3
00:00:12,559 --> 00:00:19,520
correspondent Riley [music] Murdoch. Thank you, Lionus. The scene here is

4
00:00:17,440 --> 00:00:24,240
chaos. There is affected customers everywhere. Excuse me. Excuse me, sir. I

5
00:00:21,439 --> 00:00:27,840
I see you're holding a high-speed DDR5 kit there. Can you explain why you paid

6
00:00:25,840 --> 00:00:31,760
extra for that instead of a lower speed kit? >> I don't know, man. Uh, everybody tells

7
00:00:30,320 --> 00:00:37,680
me I need the more expensive stuff, but I It's just so expensive. I'm just happy

8
00:00:34,079 --> 00:00:39,040
they didn't take my other kidney.

9
00:00:37,680 --> 00:00:43,040
>> Should we tell them? >> Are you a gamer, sir?

10
00:00:41,440 --> 00:00:47,840
>> Yes. [laughter] >> Then we've got some fantastic news for

11
00:00:45,280 --> 00:00:52,559
you because we tested four CPUs at eight different speeds of DDR5 running as slow

12
00:00:50,079 --> 00:00:57,920
as 4,800 mega transfers per second. And it turns out you might not need to spend

13
00:00:54,719 --> 00:01:01,280
as much on memory as you thought.

14
00:00:57,920 --> 00:01:02,800
Ooh, that does not look great. Good

15
00:01:01,280 --> 00:01:08,880
thing he was born with two of them. But hey, it's not all bad. Let me cheer you

16
00:01:05,199 --> 00:01:10,240
up with this segue to our sponsor.

17
00:01:08,880 --> 00:01:14,240
Can we get this guy some help? >> Synergy. Synergy lets you use one

18
00:01:12,320 --> 00:01:18,400
keyboard and mouse to control multiple computers on your local network. You can

19
00:01:16,400 --> 00:01:22,240
even do things like copy and paste in between computers. Check them out with

20
00:01:20,400 --> 00:01:28,560
our link in the video description. The TLDDR of our extensive testing is that

21
00:01:25,439 --> 00:01:30,799
while some CPUs do obviously benefit

22
00:01:28,560 --> 00:01:36,720
from faster memory, AMD's recent statement that their popular X3D gaming

23
00:01:33,280 --> 00:01:38,799
CPUs are super down to shack up with

24
00:01:36,720 --> 00:01:43,119
even the slowest memory seems to be pretty much accurate. We found that at

25
00:01:40,960 --> 00:01:47,840
typical gaming resolutions and settings, there was almost no difference between

26
00:01:45,360 --> 00:01:53,200
kits that run at industry standard or JDEX speeds and faster overclocked

27
00:01:50,799 --> 00:01:57,040
gaming RAM. Now, this is not going to completely solve the problem. Even

28
00:01:54,960 --> 00:02:02,240
slower memory has been impacted by recent price increases, but hey, at

29
00:01:59,759 --> 00:02:06,640
least it gives us some options. Let's find out then how much performance we

30
00:02:04,159 --> 00:02:13,120
might be giving up. Starting with Intel, where unfortunately the story is not

31
00:02:09,840 --> 00:02:15,760
quite as rosy as AMD's X3D chips. We

32
00:02:13,120 --> 00:02:19,840
chose both a lastg and a current gen representative, and we used our standard

33
00:02:17,599 --> 00:02:23,599
Labs test benches. Testing took forever, by the way, because we had to use the

34
00:02:21,760 --> 00:02:28,319
same memory kit across each platform, which meant swapping out a lot of RAM.

35
00:02:25,840 --> 00:02:32,400
Let's start with gaming on Raptor Lake. Looking strictly at our Geo Mean across

36
00:02:30,319 --> 00:02:39,200
all games at 1080p, it's pretty clear that our 14900K loves fast DDR5. So, if

37
00:02:37,200 --> 00:02:42,879
your grandma made a strategic DDR5 investment last year and you can get

38
00:02:40,720 --> 00:02:47,840
some 8,400 megat transfer per second CL40, to the moon it goes. At least in

39
00:02:46,160 --> 00:02:52,160
certain games like Ashes of the Benchmark, Cyberpunk, and Red Dead

40
00:02:49,840 --> 00:02:56,160
Redemption 2. The good news is though, as long as you don't pair it with the

41
00:02:53,760 --> 00:03:01,440
slowest kit you can find, it won't suffer too much. As for Intel's newer

42
00:02:58,319 --> 00:03:03,599
Core Ultra 285K, good news, it couldn't

43
00:03:01,440 --> 00:03:08,080
even run our fastest kit. And as for the rest, as long as you're using something

44
00:03:05,200 --> 00:03:12,959
that's faster than JDK, basically don't worry about it. When we look at the

45
00:03:09,760 --> 00:03:15,200
allimportant 1% lows, there is at most a

46
00:03:12,959 --> 00:03:19,680
handful of FPS difference across every game in resolution that we tested to the

47
00:03:17,360 --> 00:03:23,519
point where I would say that unless you have an FPS counter in the corner, you

48
00:03:21,760 --> 00:03:27,200
would never know the difference. And this is especially true as your

49
00:03:25,200 --> 00:03:31,280
resolution and your in-game details scale up. Now, you're probably wondering

50
00:03:29,120 --> 00:03:36,080
at this point, if these differences are so small, why were all of you reviewers

51
00:03:33,519 --> 00:03:41,599
acting like RAM speed mattered before? It's all about the context, guys. When

52
00:03:38,400 --> 00:03:43,519
we were talking about 20 or 30 bucks to

53
00:03:41,599 --> 00:03:49,360
get a few% performance improvement on your $1,500 or $2,000 gaming PC, that's

54
00:03:47,440 --> 00:03:55,200
a no-brainer. Who doesn't want bang for the buck? But we're not talking 20 or 30

55
00:03:51,519 --> 00:03:57,280
bucks. We're talking $200 or $300 for a

56
00:03:55,200 --> 00:04:00,560
higher speed kit. That changes the math completely. Or does it? Maybe

57
00:03:59,120 --> 00:04:06,480
productivity will tell a different story. Okay, not in Blender. But our

58
00:04:03,599 --> 00:04:10,879
GDAU compile test does see a decent improvement with higher speed memory on

59
00:04:08,080 --> 00:04:15,120
the 14900 K. And that's not a completely isolated case. While many of our

60
00:04:12,959 --> 00:04:20,000
productivity tests do mirror our gaming tests across both Intel's current and

61
00:04:17,519 --> 00:04:24,720
last gen chips, there are a significant number of results here where spending

62
00:04:22,000 --> 00:04:28,160
far more on your memory can yield a meaningful difference in performance to

63
00:04:26,240 --> 00:04:32,160
the point where if you're making money with your machine, you might want to

64
00:04:30,080 --> 00:04:37,280
consider ponying up for a high-speed, low latency kit, even at today's

65
00:04:34,560 --> 00:04:42,000
extortionate pricing, which is exactly the problem that consumers are facing

66
00:04:39,199 --> 00:04:46,479
right now. You want to play video games and you got to budget accordingly. Well,

67
00:04:44,479 --> 00:04:51,120
Joe Workstation user, he's not playing games. He wants to compile his code

68
00:04:48,560 --> 00:04:53,759
faster. So, he submits a requisition to the IT department who goes,

69
00:04:52,479 --> 00:04:57,919
>> "Yeah, if it'll make you more productive, go for it." >> We're going to have a companion

70
00:04:56,000 --> 00:05:02,160
lttabs.com article, by the way, that dives a little deeper into which

71
00:04:59,919 --> 00:05:05,759
business use cases might benefit and which ones won't. On the subject of

72
00:05:03,759 --> 00:05:09,280
business, check out the new sand colored tech pants at ltstore.com. They offer

73
00:05:07,759 --> 00:05:13,280
loads of utility with hidden pockets galore while keeping you stylish on the

74
00:05:11,360 --> 00:05:17,520
job. Now, let's change gears and talk a little bit about AMD. Don't you guys

75
00:05:15,199 --> 00:05:22,160
love it when twice a day the corporate puffery clock gets it right? Just look

76
00:05:20,000 --> 00:05:27,520
at those beautifully aligned graphs on our 9800 X3D. Okay, there is a little

77
00:05:25,199 --> 00:05:33,759
bit of variance. Our 4,800 megat transfer per second kit does sit in last

78
00:05:29,919 --> 00:05:35,919
place and 6,000 mega transfer CL30 is

79
00:05:33,759 --> 00:05:41,199
still the sweet spot. But turn on ray tracing or up the resolution and things

80
00:05:38,639 --> 00:05:46,000
become even more seemier than before. Like on Intel, Cyberpunk and Ashes show

81
00:05:43,600 --> 00:05:50,000
the biggest spread in our 1080p results, but it's a much smaller gap. And

82
00:05:48,320 --> 00:05:55,360
realistically, there are cases here where using 16 gig versus 24 gig modules

83
00:05:53,199 --> 00:06:01,039
is going to make as much of a difference or more. 3dB cache for the win. As for

84
00:05:58,080 --> 00:06:06,240
the regular cache, it might be for the win, too. It turns out that while our

85
00:06:03,039 --> 00:06:08,880
9950X did stumble a little bit with our

86
00:06:06,240 --> 00:06:14,000
slowest JDK kit, overall, as long as you can get your hands on something,

87
00:06:11,120 --> 00:06:18,080
anything with a tuned expo profile, you probably won't notice a difference.

88
00:06:16,319 --> 00:06:22,000
Especially at risk of sounding like a broken record, at risk of sounding like

89
00:06:19,600 --> 00:06:25,919
a broken record. Especially if you turn up your in-game details or kick your

90
00:06:24,000 --> 00:06:29,919
resolution up to 4K. Moving on to productivity, we've got a similar story

91
00:06:27,440 --> 00:06:33,680
to what we saw in Intel, just through a different colored lens. The JDK kit is

92
00:06:31,919 --> 00:06:38,244
definitely the slowest, demonstrating that overclocked memory does have a

93
00:06:35,600 --> 00:06:41,199
reason to exist, just not [clears throat] at the current prices.

94
00:06:39,440 --> 00:06:46,240
Beyond that, there were occasional cases where faster memory, or especially sweet

95
00:06:44,000 --> 00:06:51,199
spot memory that runs in sync with AMD's Infinity Fabric at nice low latencies,

96
00:06:48,880 --> 00:06:55,600
did result in measurable gains. But outside of those specific use cases,

97
00:06:53,360 --> 00:07:00,639
like GDAU, the differences are not going to be enough to ruin your day. So, in

98
00:06:57,440 --> 00:07:02,639
conclusion, don't be this guy. If you're

99
00:07:00,639 --> 00:07:06,960
a gamer and high-speed memory is available for a modest price delta, it

100
00:07:05,120 --> 00:07:12,240
can make a lot of sense, especially if you're on an older Intel platform. But

101
00:07:09,199 --> 00:07:14,160
right now, I just can't recommend it,

102
00:07:12,240 --> 00:07:18,000
especially to the tens of thousands of you that are buying X3D processors for

103
00:07:16,319 --> 00:07:21,360
gaming. By the way, we're going to have X3D processors and all the stuff that we

104
00:07:19,759 --> 00:07:26,240
used on our benches linked in the video description if you guys wanted to check something out for your next rig. Just uh

105
00:07:24,560 --> 00:07:30,080
don't expect any great deals on the RAM down there. As for the non-gaming folks,

106
00:07:28,240 --> 00:07:35,120
man, I really feel for the cottage industry professionals right now. The

107
00:07:31,840 --> 00:07:38,479
folks who do work at home that requires

108
00:07:35,120 --> 00:07:40,800
faster RAM or especially a lot of RAM.

109
00:07:38,479 --> 00:07:45,039
Like, it's hard to care too much about corporates's workstation budget, but if

110
00:07:42,639 --> 00:07:50,479
you're an atome developer or you're a contract creative professional, it is a

111
00:07:47,599 --> 00:07:54,080
super sucky time to need an upgrade. The only good news I have for you is that

112
00:07:52,160 --> 00:07:57,840
you can shave a few bucks off of your budget for now and then with the money

113
00:07:55,759 --> 00:08:02,039
you saved, hopefully get something a little faster once the bubble pops.

114
00:08:03,680 --> 00:08:08,039
Anytime now would be nice.

115
00:08:08,560 --> 00:08:13,919
All right. Well, I guess it's not happening yet, but um while we continue

116
00:08:11,840 --> 00:08:17,680
to wait, here's a word from our sponsor, >> Synergy. If you have multiple computers

117
00:08:16,000 --> 00:08:21,680
on your local network and you have to run from system to system to get any

118
00:08:19,520 --> 00:08:26,479
work done, you should check out Synergy. They let you use a single keyboard and

119
00:08:23,759 --> 00:08:31,840
mouse to control multiple computers. And you can move your cursor in between

120
00:08:28,800 --> 00:08:34,000
devices as well as copy and paste. And

121
00:08:31,840 --> 00:08:38,719
unlike KVM switching, there isn't those few seconds of waiting. It supports

122
00:08:36,080 --> 00:08:42,320
hotkeys, TLS encryption, and clipboard sharing. Synergy isn't a subscription

123
00:08:40,719 --> 00:08:46,560
service. There are no reoccurring payments. Just choose which one of their

124
00:08:44,320 --> 00:08:50,640
three tiers fits your needs most and you go from there. In the end, it's just

125
00:08:48,800 --> 00:08:54,959
less clutter and confusion. With Synergy, one keyboard and mouse is all

126
00:08:53,040 --> 00:08:59,839
you need to control every computer in your setup. Grab 50% off using code LTT

127
00:08:57,839 --> 00:09:03,360
at checkout with our link down below. If you guys enjoyed this video, make sure

128
00:09:01,519 --> 00:09:09,040
to check out the last time we did this, but with DDR4. So, if you want to pick

129
00:09:06,080 --> 00:09:13,600
up like an older Intel 12th or 13th gen chip and pair that with some nice fancy

130
00:09:11,279 --> 00:09:16,720
DDR4, you can decide for yourself exactly how much to spend.