1
00:00:00,260 --> 00:00:01,540
RAM

2
00:00:01,540 --> 00:00:07,250
Whether you bat for the red team or the blue team

3
00:00:08,210 --> 00:00:09,930
You need it

4
00:00:09,930 --> 00:00:15,250
but there are so many options from 2133 to

5
00:00:16,290 --> 00:00:18,490
4266 megahertz

6
00:00:19,190 --> 00:00:25,150
But could it possibly be worth emptying your wallet for the fastest memory on the market?

7
00:00:26,420 --> 00:00:28,420
Let's find out

8
00:00:28,420 --> 00:00:44,940
We're opening up a new merch store. You can check out some of our new gear today at the link in the video description

9
00:00:46,490 --> 00:00:48,550
First a primer on RAM

10
00:00:49,250 --> 00:00:56,470
DDR4 is the latest version of the form of DRAM that our computers almost universally use

11
00:00:57,230 --> 00:01:02,310
programs get loaded into it from our SSDs and hard drives because it's

12
00:01:02,630 --> 00:01:08,850
Blazing fast performance is needed to feed our CPUs data to munch on

13
00:01:08,850 --> 00:01:11,990
Turn in those ones and zeros into games

14
00:01:12,670 --> 00:01:17,210
Boring PowerPoint presentations or this video that you're watching right now

15
00:01:17,910 --> 00:01:22,690
Cool, so the faster the RAM the better the performance, right?

16
00:01:23,720 --> 00:01:28,360
Not quite there comes a point where even if you could

17
00:01:28,940 --> 00:01:36,520
Force feed your CPU more data. It couldn't possibly chew through it fast enough to make a difference

18
00:01:36,940 --> 00:01:38,940
so the sweet spot as

19
00:01:38,940 --> 00:01:44,420
Always probably lies somewhere in the middle. We just need to find it

20
00:01:45,100 --> 00:01:51,160
Let's establish a baseline here the joint electron device engineering council or

21
00:01:51,840 --> 00:01:53,700
Jdec for short

22
00:01:53,700 --> 00:01:55,220
specifies

23
00:01:55,220 --> 00:02:02,360
2133 megahertz for DDR4 while on both Intel's Z270 and X99 platforms

24
00:02:02,640 --> 00:02:08,440
Running anything over DDR4 2400 megahertz is outside the official spec

25
00:02:08,940 --> 00:02:16,760
AMD for their part has specified a range of speeds with the most frequently used configuration also sitting at 2400 megahertz

26
00:02:16,760 --> 00:02:18,760
But wait a minute what?

27
00:02:18,760 --> 00:02:21,800
Well, why the heck are companies like G skill?

28
00:02:22,560 --> 00:02:24,560
advertising speeds that are up to

29
00:02:25,180 --> 00:02:26,980
almost double that

30
00:02:26,980 --> 00:02:28,920
Well, the answer is

31
00:02:28,920 --> 00:02:30,920
sanctioned overclocking

32
00:02:30,960 --> 00:02:38,860
Intel's extreme memory profile or XMP for short has become something of an industry standard for running high-end

33
00:02:39,000 --> 00:02:41,120
memory to its rated spec on

34
00:02:41,740 --> 00:02:43,740
supported motherboards

35
00:02:43,840 --> 00:02:52,300
Every RAM module even the highest-end XMP ones then have the Jdec standard speeds baked in then along with that a

36
00:02:52,860 --> 00:02:57,980
faster XMP profile that's stored on a chip that identifies each

37
00:02:58,440 --> 00:03:02,960
Individual module in a standard that's called serial presence detect or SPD

38
00:03:03,420 --> 00:03:08,320
for compatibility reasons Jdec comes first and it's only when you specifically

39
00:03:09,000 --> 00:03:15,800
Enable XMP in your computer's BIOS that it will run at those speeds or at least in theory it will

40
00:03:16,280 --> 00:03:22,700
We've had tons of issues over the years most recently during the Ryzen 7 launch where we couldn't get past

41
00:03:23,460 --> 00:03:27,380
2666 megahertz no matter what ASUS AMD or G skill said

42
00:03:27,960 --> 00:03:34,000
our board with our memory has since been fixed which is great with a firmware update, but

43
00:03:34,480 --> 00:03:38,260
sanctioned overclocking is still overclocking and it can cause

44
00:03:38,260 --> 00:03:45,480
These kinds of issues from chip to chip or board to board or RAM stick to RAM stick depending on your luck

45
00:03:45,480 --> 00:03:48,540
It can also cause less severe

46
00:03:48,540 --> 00:03:53,360
But equally annoying issues to crop up at random at some point in the future

47
00:03:53,360 --> 00:04:00,680
But whatever Linus enough of this warning label crap. I'm prepared to spend a weekend bashing my head

48
00:04:01,260 --> 00:04:03,260
against the wall

49
00:04:03,340 --> 00:04:07,600
Troubleshooting my new gaming rig so tell me what do I stand to gain?

50
00:04:08,580 --> 00:04:10,540
for my pain

51
00:04:10,540 --> 00:04:17,600
Great question. We tested a variety of configurations on AMD the maximum we could achieve was ddr4

52
00:04:18,120 --> 00:04:20,240
3200 and on Intel we hit

53
00:04:20,820 --> 00:04:24,320
3600 megahertz before we just stopped turning the dial because

54
00:04:25,080 --> 00:04:26,320
spoiler alert

55
00:04:26,320 --> 00:04:31,100
Well, no wait actually no you'll see using 2133 megahertz as a baseline

56
00:04:31,100 --> 00:04:37,540
We ran Deus Ex mankind divided and rise of the tomb raider in both DirectX 11 and DirectX 12

57
00:04:37,640 --> 00:04:38,740
modes

58
00:04:38,740 --> 00:04:43,040
Some games clearly won't care much about RAM speeds

59
00:04:43,200 --> 00:04:50,060
But others can end up being more memory sensitive than we'd initially expected in for honor

60
00:04:50,580 --> 00:04:58,220
average frame rates looked to be within margin of error regardless of RAM speed, but looking closely er the

61
00:04:58,600 --> 00:05:07,480
97th percentile minimum frame rate does seem to have been affected as far as synthetics go we tested seven zip

62
00:05:07,640 --> 00:05:15,460
The RAM hungry why cruncher Cinebench r15 and ASUS real bench where we had much more

63
00:05:15,880 --> 00:05:21,700
consistent results if by consistent we mean eerily similar across the board

64
00:05:22,550 --> 00:05:29,050
Which leads us then into the price and performance summary in almost every case?

65
00:05:29,510 --> 00:05:34,170
the largest improvement in performance over 2133 is

66
00:05:35,470 --> 00:05:37,470
achieved by 2666

67
00:05:37,930 --> 00:05:44,650
Megahertz then there's another modest improvement when you jump to 3200 and from there

68
00:05:45,590 --> 00:05:46,970
marketing and

69
00:05:46,970 --> 00:05:53,070
purely synthetic benchmarks aside things go downhill pretty quickly on both Intel and

70
00:05:53,670 --> 00:05:59,350
AMD leading us to conclude that there's no point going higher even if you could

71
00:06:00,170 --> 00:06:05,170
So then by 3200 megahertz RAM since that's the sweet spot, right?

72
00:06:05,980 --> 00:06:07,900
well, but shhhhhh

73
00:06:07,960 --> 00:06:09,400
Sure

74
00:06:09,400 --> 00:06:13,960
But it actually depends to what kind of a budget you're working with for your system

75
00:06:13,960 --> 00:06:18,760
If you've already got the best of everything then sure yeah

76
00:06:18,800 --> 00:06:23,620
You might want to go higher on your RAM to eke out that slim improvement

77
00:06:23,680 --> 00:06:32,740
But if you're on any kind of fixed budget well for 64 bucks the difference between 2133 and 3600 RAM

78
00:06:32,860 --> 00:06:36,640
you could go from a core i5 7500 to a

79
00:06:36,640 --> 00:06:44,140
7600 K netting you higher CPU clock speeds and overclocking support or from a Ryzen 5

80
00:06:44,640 --> 00:06:50,800
1500 X to a 1600 X giving you two whole additional processing cores and

81
00:06:52,340 --> 00:06:55,300
That only accounts for the financial investment

82
00:06:55,300 --> 00:07:02,740
What we haven't even talked about is the time that this can end up consuming and going back again to look at the total

83
00:07:03,680 --> 00:07:06,440
possible improvement from a bare bones

84
00:07:06,440 --> 00:07:14,200
J deck 2133 kit to the fastest we tested the full conclusion becomes crystal clear

85
00:07:15,510 --> 00:07:23,890
4.7% for $64 more without any guarantee that it'll actually work means that we wouldn't recommend it

86
00:07:24,090 --> 00:07:30,590
Unless you're out of stuff to spend money on that will give you more meaningful performance or usability

87
00:07:31,230 --> 00:07:36,190
Improvements and you have not only the time and know-how to tinker with it

88
00:07:36,470 --> 00:07:42,230
But also the patience to drop everything and deal with it when something isn't working

89
00:07:42,230 --> 00:07:48,590
And this is especially true if you're using the latest and greatest platform with that

90
00:07:49,170 --> 00:07:51,170
fastest and shiniest RAM

91
00:07:52,520 --> 00:07:58,260
Tunnel bear is the simple VPN app that makes it easy to browse the internet more

92
00:07:58,740 --> 00:08:05,760
Privately and securely you just turn it on. It's like really really easy you press like one button and boom

93
00:08:06,880 --> 00:08:11,300
your connection gets encrypted with AES 256 bit encryption and

94
00:08:11,920 --> 00:08:20,060
You appear to the services and websites that you use as though you are coming from a completely different country

95
00:08:20,060 --> 00:08:25,340
Which in many cases allows you to use services that would otherwise be geo blocked

96
00:08:25,660 --> 00:08:30,720
Another cool thing is that your ISP can't track exactly what it is that you're doing online

97
00:08:31,540 --> 00:08:36,780
Tunnel bear has a top rated privacy policy and you can try it for free with 500 megabytes and no

98
00:08:36,880 --> 00:08:42,240
Credit card required and you can get 10% off by going to tunnel bear dot com slash LTG

99
00:08:42,940 --> 00:08:46,040
So thanks for watching guys if this video sucked

100
00:08:46,060 --> 00:08:47,160
You know what to do

101
00:08:47,160 --> 00:08:52,220
But if it was awesome get subscribed hit that like button or check out the link to where to buy the stuff we featured in

102
00:08:52,300 --> 00:08:57,140
The video description also down there is our merch store which I own hurt

103
00:08:57,660 --> 00:09:03,160
Which has cool shirts like this one and our community forum, which you should totally join