WEBVTT

00:00:02.720 --> 00:00:11.920
this is the 8 gig stick i ordered the 16 Gigabyte kit

00:00:07.680 --> 00:00:13.519
the chef insists sir your chef is a fool

00:00:11.920 --> 00:00:19.840
because some game developers are even recommending 32 gigabytes told you this

00:00:16.560 --> 00:00:19.840
was the kind of place you leave hungry

00:00:20.080 --> 00:00:26.960
unless eight gigabytes is enough

00:00:25.039 --> 00:00:31.199
to find out crucial sponsored this video where we're

00:00:28.640 --> 00:00:36.800
going to be testing 8 and 16 Gigabyte kits in both ddr4 and ddr5 flavors to

00:00:34.559 --> 00:00:39.800
ensure that you are not wasting your money

00:00:47.039 --> 00:00:55.120
for gamers system memory or RAM plays a crucial role see what i did there

00:00:52.320 --> 00:01:00.079
its main job is to ensure that your CPU and for gamers using onboard graphics

00:00:57.039 --> 00:01:02.160
your GPU get the high speed access to

00:01:00.079 --> 00:01:07.040
assets that are required to run your game smoothly because the second you

00:01:04.320 --> 00:01:11.920
have to swap files out to an SSD or even worse a hard drive you're likely to

00:01:09.439 --> 00:01:16.720
encounter problems ranging from FPS drops to slower loading times to random

00:01:14.640 --> 00:01:22.640
hitching or stuttering but once you have enough RAM to eliminate

00:01:20.080 --> 00:01:27.200
these problems there's often no performance benefit to adding more

00:01:24.560 --> 00:01:33.520
meaning it is really easy to blow away money that could have been spent on a

00:01:29.520 --> 00:01:36.479
better CPU better GPU or even a faster

00:01:33.520 --> 00:01:40.159
rather than higher capacity RAM kit to isolate memory as our performance

00:01:38.240 --> 00:01:45.759
variable then we chose a top tier test bench with a core i9 CPU and an rtx

00:01:42.720 --> 00:01:48.720
3080ti we also tested each game at both

00:01:45.759 --> 00:01:52.640
1080p and 4k to see if our RAM would affect performance more or less

00:01:50.640 --> 00:01:57.680
depending on whether we're CPU or GPU bottlenecked as for our memory we tested

00:01:55.439 --> 00:02:02.640
eight gigs of ddr4 in both single and dual channel configurations and 16 gigs

00:02:00.399 --> 00:02:07.360
in dual channel and crucial also asked us to throw in their new ddr5 sticks

00:02:04.719 --> 00:02:10.879
with the black pcb that's right no unicorn vomit lighting effects on these

00:02:09.119 --> 00:02:15.040
boys but they're pretty sharp looking in an understated kind of way and according

00:02:13.200 --> 00:02:19.440
to crucial's internal testing they don't need heat spreaders so it turns out some

00:02:17.040 --> 00:02:24.879
of the concerns about ddr5's on dim power management requiring extra cooling

00:02:21.599 --> 00:02:27.200
don't apply to at least jdx spec modules

00:02:24.879 --> 00:02:30.560
anyway whether you like RGB or blacked out builds at the end of the day it's

00:02:28.800 --> 00:02:34.959
performance that counts so let's kick things off with csgo at 1080p it's our

00:02:33.200 --> 00:02:39.120
tried and true CPU limited game and you'll be happy to know that yes 10

00:02:37.200 --> 00:02:45.280
years later it will still run perfectly fine with only 8 gigs of RAM however the

00:02:42.319 --> 00:02:50.080
advantage of going dual channel is huge at this resolution with our single 8 gig

00:02:47.760 --> 00:02:52.959
stick of ddr4 performing terribly by comparison

00:02:51.280 --> 00:02:57.120
this advantage mostly disappeared however once we scaled up to 4k and

00:02:55.440 --> 00:03:01.280
adding more memory hardly made a difference at either resolution which

00:02:59.599 --> 00:03:06.879
makes sense given the developer calls for only two gigabytes of system memory

00:03:03.920 --> 00:03:12.159
curiously while our ddr5 numbers are a little lower than ddr4 we don't see the

00:03:09.680 --> 00:03:15.920
same massive drop in performance for dropping from dual to single channel

00:03:14.720 --> 00:03:20.080
what ah that's because ddr5 is already dual

00:03:19.120 --> 00:03:28.000
channel or sort of instead of using a single 72-bit bus on

00:03:24.720 --> 00:03:30.720
a dim like ddr4 with 64 bits for data

00:03:28.000 --> 00:03:37.519
and 8 for ecc or error correction code each ddr5 module gets two 40-bit buses

00:03:35.599 --> 00:03:41.519
again with eight bits on each going towards ecc

00:03:39.040 --> 00:03:46.879
so you still get the same total 64-bit data width but with each half of it

00:03:44.640 --> 00:03:51.920
operating semi-independently now some workloads like running complex

00:03:48.959 --> 00:03:56.879
simulations may still need at least two sticks for peak performance but for

00:03:54.080 --> 00:03:59.760
bandwidth light applications like games well the results kind of speak for

00:03:58.640 --> 00:04:04.239
themselves csgo is old though let's turn to far cry

00:04:02.159 --> 00:04:08.159
6 which launched late last year with a minimum requirement of 8 gigs of RAM

00:04:06.480 --> 00:04:12.720
specifically dual channel and a recommendation of 16 gigs

00:04:10.959 --> 00:04:17.600
unsurprisingly eight gigs in dual channel did perform markedly better than

00:04:15.120 --> 00:04:22.560
a single channel stick at 1080p but curiously this uptick applied only to

00:04:20.239 --> 00:04:26.960
average frame rates the far more important one percent lows were nearly

00:04:24.720 --> 00:04:33.199
identical between the two at either resolution doubling our memory to a 16

00:04:29.919 --> 00:04:36.639
gig kit however saw our lows skyrocket

00:04:33.199 --> 00:04:38.720
improving over 40 percent at full hd

00:04:36.639 --> 00:04:45.759
we saw improvements at 4k resolution as well with a 16 gig kit bringing our lows

00:04:41.199 --> 00:04:47.199
from 47 FPS to 61 on ddr4 these are the

00:04:45.759 --> 00:04:51.120
kinds of performance differences that you will feel while gaming

00:04:49.440 --> 00:04:54.960
Floatplane.com by the way has a great behind the scenes of alex feeling pretty

00:04:53.199 --> 00:04:58.160
upset about spilling water on a system and trying to fix it you can check it

00:04:56.479 --> 00:05:02.720
out at the link down below moving up a generation our single stick

00:05:00.240 --> 00:05:07.759
of ddr5 marginally outperforms our single stick of ddr4 at 1080p but then

00:05:05.520 --> 00:05:12.800
does slightly worse at 4k however when we add a second eight gig

00:05:09.680 --> 00:05:14.560
ddr5 module we see this same trend of

00:05:12.800 --> 00:05:19.520
increased capacity leading to better performance at both resolutions lifting

00:05:17.199 --> 00:05:26.400
our one percent lows by 25 percent in 1080p and 33 at 4k getting us really

00:05:23.520 --> 00:05:29.919
close to that magic 60fps number where most people consider animation to be

00:05:28.479 --> 00:05:33.919
very smooth remember though guys just because a game

00:05:31.759 --> 00:05:36.800
runs at an average of 60 frames per second that doesn't mean you're not

00:05:35.280 --> 00:05:42.320
going to notice hitching when you're hitting those one percent lows of 30 or

00:05:39.520 --> 00:05:46.560
40 FPS so for the smoothest experience possible you want your lows at 60 or

00:05:45.280 --> 00:05:52.720
even higher tiny tina's wonderland saw similar if

00:05:49.440 --> 00:05:54.960
lower results at 1080p with our poor sad

00:05:52.720 --> 00:05:57.680
single stick of ddr4 underperforming by about 18

00:05:56.560 --> 00:06:02.479
compared to our dual channel configuration and here again increasing

00:06:00.800 --> 00:06:08.800
capacity doesn't really net us improvements to our average FPS but it

00:06:05.440 --> 00:06:10.960
does impact our one percent lows though

00:06:08.800 --> 00:06:14.720
only by about five percent this time it's up to you to decide if a five

00:06:12.720 --> 00:06:20.000
percent gain is worth doubling up your RAM budget when we scale up to 4k though

00:06:17.600 --> 00:06:25.600
all three configurations of ddr4 saw nearly identical results since our CPU

00:06:22.960 --> 00:06:28.639
usage dropped and our GPU became the limiting factor

00:06:27.120 --> 00:06:33.039
what's interesting about wonderlands however is that we actually saw the best

00:06:30.800 --> 00:06:38.319
results with ddr5 in 1080p our single stick of ddr5 beat

00:06:35.680 --> 00:06:43.360
our single stick of ddr4 by 11 and doubling up improved it even more

00:06:41.280 --> 00:06:48.000
at 4k the story is pretty similar with our single channel ddr4 being the worst

00:06:45.919 --> 00:06:52.479
adding capacity beyond 8gig not really seeming to help and ddr5 being the

00:06:50.479 --> 00:06:58.080
absolute king for performance in this game but why though

00:06:55.440 --> 00:07:01.199
well wonderlands is the newest game we tested so

00:06:59.360 --> 00:07:05.280
it could be that newer games coming out over the next few years are going to be

00:07:03.120 --> 00:07:08.720
designed targeting current gen consoles which have

00:07:06.479 --> 00:07:13.360
much faster memory than the playstation 4 and xbox one x generation

00:07:11.440 --> 00:07:17.680
these newer games may be able to better take advantage of what ddr5 has to offer

00:07:16.000 --> 00:07:22.560
and crucial would like us to point out that with up to 87 more bandwidth than

00:07:20.080 --> 00:07:27.039
ddr4 at launch performance is enhanced not just during testing but in real

00:07:24.800 --> 00:07:31.440
world conditions and with support for xmp 3.0 we can expect to see even more

00:07:29.440 --> 00:07:37.440
performance out of crucials ddr5 on future platforms like AMD's upcoming am5

00:07:34.639 --> 00:07:41.840
and Intel's 13th gen core thanks again crucial for sponsoring this video

00:07:39.360 --> 00:07:46.800
and for understanding that our audience needs the whole picture which might

00:07:44.240 --> 00:07:51.680
include situations where your product won't have much of an impact

00:07:48.639 --> 00:07:54.639
total warhammer 3 saw almost identical

00:07:51.680 --> 00:07:59.919
performance across the board at 4k regardless of single dual channel

00:07:57.520 --> 00:08:04.639
capacity and even the generation of memory then at 1080p improvements to our

00:08:02.560 --> 00:08:09.520
hardware yielded only minimal performance uplifts meaning that if you

00:08:06.800 --> 00:08:13.120
only play tw3 you could just pick up a wan hoodie from ltdstore.com instead

00:08:11.759 --> 00:08:17.280
check them out at the link below because we've got lots in stock finally

00:08:15.440 --> 00:08:21.840
back to our results assassin's creed valhalla saw clear improvements going

00:08:19.280 --> 00:08:27.360
from single to dual channel as expected but increasing capacity only seemed to

00:08:24.000 --> 00:08:28.560
help ddr5 and not by much regardless of

00:08:27.360 --> 00:08:33.120
resolution now unfortunately because every game is

00:08:30.800 --> 00:08:36.959
developed so differently even if you're looking at two titles that use the same

00:08:35.120 --> 00:08:40.880
engine it's hard to say whether or not upgrading your memory capacity will

00:08:38.640 --> 00:08:45.360
improve your gaming experience we've known for years that dual channel at the

00:08:42.800 --> 00:08:49.200
bare minimum is the way to go and that's still crystal clear here but unless

00:08:47.519 --> 00:08:53.519
you're playing newer games that actually take advantage of more memory you're

00:08:51.279 --> 00:08:57.920
probably better off buying a faster dual channel configuration and dumping any

00:08:55.680 --> 00:09:03.519
extra you have left over into your CPU GPU combo now to be clear there are

00:09:01.040 --> 00:09:08.320
games out there now that will utilize more than 8 gigs of RAM but if you're

00:09:05.920 --> 00:09:11.680
not playing the newest aaa titles it may not be worth the upgrade for you

00:09:10.720 --> 00:09:17.040
yet if you want to be ready though we're gonna have all the products we use today

00:09:14.640 --> 00:09:20.080
linked in the video description below thanks for watching guys if you enjoyed

00:09:18.480 --> 00:09:24.880
this video make sure to check out some of our other RAM videos like that time

00:09:21.920 --> 00:09:29.159
we tested mixed memory modules with mixed results