WEBVTT

00:00:00.439 --> 00:00:09.960
3d crosspoint or as Intel prefers to call it probably

00:00:05.819 --> 00:00:12.389
for sex appeal reasons octane now we had

00:00:09.960 --> 00:00:18.330
a hands-on look at this technology back in april with these tiny low capacity

00:00:15.719 --> 00:00:23.550
models that were designed to make a hard drive that they accompanied performed

00:00:20.760 --> 00:00:28.260
more like a traditional SSD and it worked pretty well so since then we've

00:00:25.890 --> 00:00:34.739
been itching to get our hands on one of the full fat obtained drives that Intel

00:00:31.470 --> 00:00:37.579
has been promising for years and get our

00:00:34.739 --> 00:00:44.280
hands on one we did but what does the real-world performance of a storage

00:00:40.530 --> 00:00:48.420
medium whose raw performance is claimed

00:00:44.280 --> 00:00:52.399
to be up to a thousand times faster than

00:00:48.420 --> 00:00:54.600
man flash look like well let's find out

00:00:52.399 --> 00:00:59.670
after I tell you today's video is brought to you by be quiet! the dark bass

00:00:56.969 --> 00:01:04.890
700 is extremely versatile and features RGB LED lighting plus a tempered glass

00:01:02.579 --> 00:01:11.959
side panel window and more check it out at the link below

00:01:14.390 --> 00:01:20.100
the octane 900p is a four lane PCI

00:01:17.909 --> 00:01:24.509
Express 3.0 card with an all-metal heatsink and an easily removable

00:01:22.380 --> 00:01:29.789
backplate that's full of helpful information about the drive on a sticker

00:01:27.030 --> 00:01:34.530
but what's more interesting is actually what's underneath so here we see our

00:01:32.490 --> 00:01:39.810
first indication that there is more to this card than just a PCI Express 2 m

00:01:37.710 --> 00:01:45.390
dot 2 converter board like you might expect then when we flip it over and

00:01:42.810 --> 00:01:51.689
gently spudge off the heatsink proper it reveals more of the same this is

00:01:47.819 --> 00:01:55.289
absolutely a purpose-built card and the

00:01:51.689 --> 00:01:57.840
comparably immense cooler and beefy

00:01:55.289 --> 00:02:03.119
onboard processor should tell you this SSD means business which is kind of

00:02:01.200 --> 00:02:10.170
strange considering the star citizen promo on the box now you can learn more

00:02:06.990 --> 00:02:13.739
about obtained here but for now the TL

00:02:10.170 --> 00:02:16.799
DR is this the exact details of obtains

00:02:13.739 --> 00:02:20.069
arrangement are a trade secret but the

00:02:16.799 --> 00:02:22.680
end result is a storage latency of a

00:02:20.069 --> 00:02:28.799
theoretical thousand times faster than traditional mend making it only 10 times

00:02:25.290 --> 00:02:31.590
slower than in that sense then D RAM so

00:02:28.799 --> 00:02:38.700
while Intel's rated sequential read and write speeds don't even get us fast as a

00:02:34.620 --> 00:02:41.639
fast SSD like a samsung 960 pro it

00:02:38.700 --> 00:02:47.639
handles significantly more operations per second with seven times the write

00:02:44.790 --> 00:02:52.319
endurance on this consumer product and a mind splitting

00:02:49.069 --> 00:02:55.910
34 times the write endurance for the

00:02:52.319 --> 00:02:59.459
data center oriented obtain P 4800 X

00:02:55.910 --> 00:03:01.859
making the finite lifespans of NAND

00:02:59.459 --> 00:03:06.989
based products a concern that we can bury in the past so the high-end video

00:03:04.560 --> 00:03:12.840
editors and data scientists out there are salivating right now because this

00:03:10.170 --> 00:03:19.739
kind of speed and endurance makes these larger drives amazing for use as scratch

00:03:16.410 --> 00:03:23.009
disks as for what that means for gamers

00:03:19.739 --> 00:03:26.349
well Intel says that load times will be

00:03:23.009 --> 00:03:28.840
significantly reduced and

00:03:26.349 --> 00:03:35.430
suddenly that star citizen promo makes more sense because this game recommends

00:03:31.510 --> 00:03:38.799
a 200 gig SSD

00:03:35.430 --> 00:03:41.019
so to test this thing we need a CPU and

00:03:38.799 --> 00:03:47.049
a platform capable of taking full advantage of it we chose our x2 99 bench

00:03:44.290 --> 00:03:52.090
with a mix of synthetic and real-world benchmarks our SSDs by the way were

00:03:49.810 --> 00:03:56.590
secured raced in preparation for testing in Crystal disc mark we see more or less

00:03:54.069 --> 00:04:01.810
as advertised speeds across our drives with octane scoring quite a bit higher

00:03:58.989 --> 00:04:04.989
on the 4k randoms at a queue depth of 1 than the others hinting at its

00:04:03.760 --> 00:04:09.669
capability for much higher responsiveness moving on to performance

00:04:08.019 --> 00:04:16.479
tests we can see that octane managed much more consistent latency that edges

00:04:12.969 --> 00:04:19.509
closer to but not quite RAM disk levels

00:04:16.479 --> 00:04:22.270
of responsiveness this is super

00:04:19.509 --> 00:04:26.410
impressive and it's at this point then that we wondered whether or not obtained

00:04:24.550 --> 00:04:31.330
would make a difference for some of the heavy tasks that we deal with around the

00:04:28.330 --> 00:04:33.099
office like editing 8k red video footage

00:04:31.330 --> 00:04:40.000
and premiere or rendering with SolidWorks but as it turns out no no it

00:04:37.780 --> 00:04:45.159
does not SolidWorks finished its i/o test about 4 seconds faster on obtain so

00:04:42.820 --> 00:04:51.070
loading or saving would be your primary benefit here unless you run out of RAM

00:04:48.849 --> 00:04:57.130
during a simulation that could be a very different story and as for premiere well

00:04:53.800 --> 00:05:00.460
in a gigantic surprise to no one it is

00:04:57.130 --> 00:05:03.280
so poorly optimized for this that when

00:05:00.460 --> 00:05:09.940
we simulated a low memory situation with a 24 gig RAM drive it just choked no

00:05:07.990 --> 00:05:15.280
matter which drive we had our swap file or the footage on so then we secure

00:05:13.360 --> 00:05:19.349
erase the drives again and copied over our Steam library to see if we could

00:05:16.930 --> 00:05:24.820
spot any differences in game load times which unfortunately due to the massive

00:05:22.479 --> 00:05:29.259
install sizes of many modern games means that we say goodbye to our RAM disk in

00:05:26.710 --> 00:05:34.840
our testing from here on out rebooting between runs we ended up with well

00:05:31.690 --> 00:05:37.670
honestly pretty underwhelming results

00:05:34.840 --> 00:05:42.440
compared to the samsung 960 Pro biggest Delta between them was only a

00:05:39.800 --> 00:05:50.150
little over half a second and even our aging Kingston KC 400 and fashioned SATA

00:05:46.580 --> 00:05:52.880
SSD really doesn't do poorly here though

00:05:50.150 --> 00:05:57.830
this is another non surprise to people who pay closer attention to benchmarks

00:05:54.920 --> 00:06:02.930
than they do to marketing one last secure race and an install of Ubuntu

00:06:00.020 --> 00:06:07.070
later our final test is Houdini a piece of software by side effects that handles

00:06:05.090 --> 00:06:12.440
physics simulations for the high end rendering used by big blockbuster film

00:06:09.650 --> 00:06:18.200
and game companies our test file will be a rendering of a maelstrom that has 1.1

00:06:15.950 --> 00:06:23.300
billion particles interacting with each other so just one frame is gonna eat our

00:06:21.260 --> 00:06:30.440
RAM for breakfast meaning that our RAM disk again is not coming back as for the

00:06:26.360 --> 00:06:34.160
rest well to generate 8 frames our 960

00:06:30.440 --> 00:06:37.760
Pro Kover an hour of i/o thrashing while

00:06:34.160 --> 00:06:40.970
our obtain killed it in just under 29

00:06:37.760 --> 00:06:43.910
minutes as for the KC 400 while we gave

00:06:40.970 --> 00:06:50.800
it a participation trophy so then should you go out and spend $600 on an only 480

00:06:47.900 --> 00:06:56.930
gig octane drive if you're a normal user I'm gonna come right out and say no

00:06:54.160 --> 00:07:02.060
because the chances are pretty slim that you would ever notice a difference

00:06:59.030 --> 00:07:04.880
compared to a traditional NVMe SSD and

00:07:02.060 --> 00:07:09.880
even a SATA one is probably fine for you in most cases as for heavier workloads

00:07:07.640 --> 00:07:14.480
well maybe if you must have the best of the

00:07:12.650 --> 00:07:20.330
best but I'm gonna come out and tell you again that you are not going to feel a

00:07:18.170 --> 00:07:26.780
difference like you did when you went from a hard drive to a conventional SSD

00:07:23.300 --> 00:07:29.090
it's not like that with that said if you

00:07:26.780 --> 00:07:35.600
have a real workload that can benefit from obtains versatility as not just

00:07:31.660 --> 00:07:37.850
hyper fast storage like running many

00:07:35.600 --> 00:07:45.230
virtual machines off of a single Drive but also its low latency and lower cost

00:07:42.950 --> 00:07:51.260
compared to system memory where you can use it as an overflow for your RAM like

00:07:47.900 --> 00:07:53.760
in our Maelstrom demo well then while

00:07:51.260 --> 00:08:00.210
the price of entry is steep the benefits may just outweigh the costs

00:07:57.440 --> 00:08:06.540
speaking of a very high benefit to cost ratio tunnel Bayer VPN is the simple to

00:08:03.930 --> 00:08:11.730
use VPN app that makes it easy to browse privately and enjoy a more open Internet

00:08:09.360 --> 00:08:16.050
with tunnel bear turned on your Wi-Fi connection is secured and your online

00:08:13.620 --> 00:08:19.680
activity is kept private from your internet provider

00:08:17.240 --> 00:08:23.790
advertisers or anyone who's looking to track you or profit from your data

00:08:21.690 --> 00:08:28.050
tunnel bear has a top rated privacy policy and does not log your activity

00:08:25.620 --> 00:08:33.090
and they have servers in over 20 countries with fast connections with

00:08:30.330 --> 00:08:37.080
their own DNS servers so once you're in your tunnel everything you do is

00:08:34.920 --> 00:08:41.160
encrypted social media pages won't be able to track you your region blocks

00:08:39.720 --> 00:08:46.740
that don't want you to watch certain videos won't be active and even your isp

00:08:44.400 --> 00:08:50.520
can't see what you're browsing so for more information at a free trial go to

00:08:48.510 --> 00:08:54.840
Tunnel bear comm slash Linus we're gonna have that linked below so thanks for

00:08:53.160 --> 00:08:58.830
watching guys if this video sucks you know what to do but if it was awesome

00:08:56.640 --> 00:09:01.710
get subscribed hit that like button or check out the link to where to buy the

00:08:59.910 --> 00:09:07.290
stuff we featured in the video description though only if your workload

00:09:04.550 --> 00:09:10.530
merits it also in the description is a link to our merch store which has cool

00:09:08.550 --> 00:09:14.000
shirts like this one and our community forum which you should totally join