Sign in to follow this  
Adam_a

Samsung 850 EVO SSD 4TB Review Discussion

Recommended Posts

The latest addition to Samsung’s 850 EVO SSD lineup doubles the previous limit of already-impressive SSD capacity available to consumers. On top of that, this drive is an excellent performer, with extremely few (if any) shortcomings to consider when weighing which drive to buy. With Samsung’s proprietary TurboWrite technology, its ability to sip power, and the standard encryption options available, this is an impressive SSD. Though its initial pricetag is hefty at $1,499.99, an included five-year warranty helps take some of the stress out of shelling out that much money and as all SSDs do, the price will decline significantly over time.

Samsung 850 EVO SSD 4TB Review

Share this post


Link to post
Share on other sites

A while back, I took quite a lot of "heat" from other forum users,

when I proposed a simple metric:

HOW LONG DOES IT TAKE TO READ THE ENTIRE DEVICE JUST ONCE?

(This came up in the context of comparing 1TB HDDs with DDR3 DRAM.)

Using the max theoretical speed of a SATA-III channel, then:

4,000,000,000,000 bytes / 600,000,000 bytes per second = 6,667 seconds

6,667 seconds / 3,600 seconds per hour = 1.85 hours

The same thing that happened with HDDs is now happening with SATA SSDs:

they are expanding rapidly withOUT making any improvements in transmission speed.

Consider now 2 well known changes which are already implemented

in PCIe 3.0 chipsets: 8G data channels + 128b/130b jumbo frames:

8G / 8.125 bits per byte = 984.6 MB/second

Now, compute the same "metric" with that new speed:

4.0 TB / 984.6 MB/second = 4,062 seconds

4,062 seconds / 3,600 seconds per hour = 1.13 hours

If nothing else, storage subsystems should be implemented

with VARIABLE transmission speeds, and an option to

enable jumbo frames too.

These options can be chosen from widely available technologies

like hardware jumpers, auto-detection and Option ROMs.

Let's "leap ahead" as Intel advises and predict the same

METRIC using the 16G clock expected for PCIe 4.0 chipsets,

and the existing 128b/130b jumbo frame:

16G / 8.125 = 1.97 GB/second

4.0 TB / 1.97 GB/second = 2,031 seconds

2,031 seconds / 3,600 seconds per hour = 0.564 hour = 33.8 minutes

The major reason why RAID-0 arrays have become so popular

with top tier SATA SSDs is the effect RAID-0 logic produces

by effectively increasing channel speed in close proportion

to the number of array members, hence the number of

roughly "parallel" data channels.

Fortunately, their improved reliability has helped to reduce

the risk inherent in non-redundant RAID-0 arrays.

Instead of buying one of these 4TB SSDs,

I would configure 2 x 2TB SSDs in a RAID-0 array.

The Samsung 850 EVO 2.5" 2TB SSD is $700 at Newegg:

http://www.newegg.com/Product/Product.aspx?Item=N82E16820147441&Tpk=N82E16820147441

2 x 2TB are cheaper (at current prices) and almost twice as fast.

If my chassis has the available space,

I can save even more money with 4 x 1TB (same model):

http://www.newegg.com/Product/Product.aspx?Item=9SIA12K2H01121&Tpk=9SIA12K2H01121

4 @ $320 = $1,280

and the speed is almost four TIMES as fast as 1 x 4TB SSD.

Yes, if one of those 4 SSDs fails, the entire array fails:

the same thing is true if a single 4TB SSD fails:

the failed drive needs to be replaced, and

the data reloaded.

SAME DIFFERENCE (as we used to say in Grad School).

p.s. Please keep the heat down, ladies and gentlemen:

I'm trying to share with you lessons I've learned

from 45 years enrolled in Hard Knocks University

(2016 - 1971).

Edited by MRFS

Share this post


Link to post
Share on other sites

Interesting point...especially as prices continue to fluctuate. I will say there's a simplicity in a single drive solution though.

Share this post


Link to post
Share on other sites

> there's a simplicity in a single drive solution

No question there, obviously for laptops with only one drive bay.

And, all the more reason to increase the clock speed

and also offer jumbo frames as an option,

as USB 3.1 has now done.

There are plenty of existing ways to enable that option

e.g. hardware jumpers, Option ROMs, ideally a

Plug-and-Play Protocol whereby the controllers

at each end of the data cable negotiate the frame format

and the clock speed without rebooting.

The same could be implemented for SAS devices.

Remember, PCe 4.0 increases the clock to 16G.

If DRAM can store JEDEC settings, so can SSDs.

I honestly believe the SATA standards group

missed the boat when they got preoccupied

with SATA-Express (a/k/a DOA).

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this