Jump to content


Photo

Intel 530 vs Samsung 840 pro


  • You cannot start a new topic
  • Please log in to reply
5 replies to this topic

#1 aXurde

aXurde

    Member

  • Member
  • 3 posts

Posted 27 December 2013 - 08:02 AM

Hi all,
I'm new to the forum and I'm Italian, so I apologize now for my bad English =)
however, I'm going to change SSD because my OCZ vector is giving problems for the second time!
However, in terms of quality, which SSD is better? Intel 530 or Samsung 840 PRO?
Thanks for the help =)

#2 Bytebite

Bytebite

    Member

  • Member
  • 60 posts

Posted 27 December 2013 - 09:42 AM

Intel and Samsung build both great SSDs, I personally don't like Sandforce, so I would go with the 840 Pro.


#3 aXurde

aXurde

    Member

  • Member
  • 3 posts

Posted 27 December 2013 - 09:55 AM

Thanks very much for the answer,

A lot of people do not like the SandForce controller, why? Another question, Samsung and Intel SSDs are the best or is there any other brand that produces quality SSD? I have read that Mushkin  SSDs are very good, what do you think?


#4 MRFS

MRFS

    Member

  • Member
  • 190 posts

Posted 27 December 2013 - 11:55 PM

>  A lot of people do not like the SandForce controller, why?

 

Evidently, when SandForce Nand Flash controllers were initially designed,

the designers were acutely aware that Nand Flash memory supports

a limited number of Program / Erase ("P/E") cycles before it fails.

 

At ~34nm, the number of P/E cycles with MLC was about 5,000.

At ~24nm, the number of P/E cycles with MLC was about 3,000.

At ~19nm, the number of P/E cycles with TLC is now about 1,000.

 

(We'll leave SLC (single-level cell) Nand Flash out of this discussion, because

that particular breed of Nand Flash is much more expensive

than MLC and TLC varieties.)

 

 

Therefore, SandForce controllers were designed to do real-time compression

of raw data output by a host controller, as a way of minimizing

the actual amount of raw data that is stored in Nand Flash chips.

 

The net effect of this design decision is that SandForce controllers

have historically performed noticeably better with raw data that is compressible,

and not as well with raw data that is incompressible.

 

 

The Nand Flash controllers in Samsung's SSDs do not perform real-time compression;

and, moreover, they perform equally well with compressible and incompressible data.

 

 

Another very annoying feature of certain firmware installed with early

SandForce controllers was a very abrupt and serious decline in performance

if the Media Wear Index ("MWI") reached zero before the end of the factory warranty period.

 

Presumably, that particular firmware logic was enabled to keep such SandForce SSDs

still running during their factory warranty periods, even though they should

have started to exhibit "wear-out" symptoms because of an excessive number of P/E cycles

that reduced MWI to zero.

 

 

For reasons like these, I have shifted away from SandForce SSDs in favor of Samsung's SSDs,

and I'm very interested in trying 2x or 4x Plextor M5 Pro 256GB SSDs in a RAID-0 array:  consequently,

I don't know for sure whether or not current SandForce controllers still exhibit the latter problem.

 

 

In general, it is fair to say that Nand Flash SSDs benefit a lot whenever

superior "wear leveling" logic is implemented in their firmware, and

this is particularly true of MLC and acutely true of TLC Nand Flash memory.

 

 

Plextor's best M5 Pro SSDs also do very well "After 30 Min. Idle"

as reported by the excellent reviewers at xbitlabs.com :

 

http://www.xbitlabs....nh_5.html#sect0

 

Notice how widely those SSDs vary on that one metric.

 

Plextor's excellent scores can be very important on systems that do not support

the TRIM command, and particularly on RAID arrays that do not support TRIM.

 

 

In my own opinion, the industry needs to adopt a standard protocol

which executes the TRIM command on any and all RAID arrays of SSDs, e.g.

by using the "boot mode" that is supported in Windows and in third-party software

like the excellent freeware Partition Wizard:  whenever the C: system partition

is checked by requesting the OS to "automatically fix file system errors",

Windows will schedule that task during the next STARTUP and

launch a "boot mode" task that takes exclusive control of that partition

before reaching the Windows Desktop.

 

Such a "boot mode" is an excellent place for TRIM commands to be issued,

perhaps on a User-defined schedule.


Edited by MRFS, 27 December 2013 - 11:58 PM.

#5 aXurde

aXurde

    Member

  • Member
  • 3 posts

Posted 28 December 2013 - 05:04 AM

Thank you very much for your answer, finally someone explained to me very well why SandForce are so problematic . So , after reading your response , I begin to think that my two OCZ Vector 128 GB , even if they don't use SandForce controller but Indilinx controller, have died after two months due to a limitation in the rewriting of NANDs . Now , as before, this vector is no longer detected by the system, after a sharp drop in performance and a lot of blue screens. Now , in fact, I don't want the best SSDs in terms of performance, I want an SSD that will last a long time. Or at least the time specified in the warranty. I read some reviews of the Samsung SSD ( 840 Pro) of users who have seen their SSD fail after only two months or even days. I wouldn't let the same thing happen to me again , even with the Samsung SSD and although they have a 5 year warranty . I also read that if a samsung SSD fails, there is no way to get it replaced because Samsung employees are incompetent , or find excuses to make sure you don't replace it.
I have read , however, that Intel SSDs use SandForce controller but are heavily modified . What do you think? Intel SSD could last five years as specified in the warranty?

#6 MRFS

MRFS

    Member

  • Member
  • 190 posts

Posted 28 December 2013 - 10:12 AM

I'd recommend that you consider Samsung 840 Pro, Plextor M5P Extreme and Corsair Neutron GTX.

 

http://www.newegg.co...N82E16820147193

 

http://www.newegg.co...N82E16820249033

 

http://www.newegg.co...N82E16820233398

 

 

 

 


Edited by MRFS, 28 December 2013 - 10:13 AM.




3 user(s) are reading this topic

0 members, 3 guests, 0 anonymous users