Final Word on the Samsung HD204UI

[cdika17]

I just recently bought 5 of these, what should i do?

[Talos]

Fair enough, but Lime-Tech has to make it work for everyone.  There has not been a lot of communication from Lime-Tech lately.

This change will hopefully make it into 5.0 (as we are seeing more and more drives with the 4K sector setup) but I am not sure if it will make it into a 4.5.x release.  My gut says it probably will but will probably be the last change to the 4.5.x series.  Every change after that will only be done to the 5.x releases.

It will need to be there sooner rather than later if the 1 January deadline comes to fruition. Its been decreed that all drive manufactured post 1 January should use the 4k cluster size. Its already hard enough as it is to get drives with the 512b cluster size.

[PeterB]

As I understand it, the 'extra space' is invisible to the user.  Basically, advanced format drives are able to cram more data into the same physical platter surface area (4x as much, if I understand it correctly).

More or less correct.  I don't think the gain is 4x ... maybe closer to 4%.  Perhaps the 'wasted' space is reduced by 4x.

The gains come about because the overheads (sector gaps, sector headers, parity bits etc) are reduced. Changing sector size from 512B to 4kB means that there are 8x less sectors.  However, the improvement is not 8x because the sector headers have to be bigger.

[johnieutah]

Yep I've just bought one of these in complete ignorance thinking it would work fine as long as it wasn't one of the 4k sector drives.  So what's the score exactly?  I understand the alignment issues, but is it a definite no-go or will it work?  I need to use this as my parity drive since it's larger than the other drives in the array.  Will there be some sort of patch to make these drives operate optimally? - updates can be kinda slow around here so I wondering if I should just send it bake for a refund now that I have the chance.  On the otherhand I do need the space now...

Thoughts and suggestions?

Matt.

[t2ffn]

For UK users, there's still Samsung 2TB HD203WI instock here: http://www.scan.co.uk/products/2tb-samsung-hd203wi-spinpoint-f3eg-ecogreen-sata-3gb-s-32mb-cache-89-ms-ncq-oem

These are on the approved hardware wiki.

[BRiT]

So what's the score exactly?  I understand the alignment issues, but is it a definite no-go or will it work?  I need to use this as my parity drive since it's larger than the other drives in the array.  Will there be some sort of patch to make these drives operate optimally? - updates can be kinda slow around here so I wondering if I should just send it bake for a refund now that I have the chance.  On the otherhand I do need the space now...

Thoughts and suggestions?

From what others have reported it will be very slow as a parity drive and degrade the performance of your entire array. On top of the slow performance issues, it also reports temps that are impossible to achieve (below room temp). I would not and will not trust this Samsung drive. It would be best if you return it now.

There is no known or estimated arrival date of changes required in unRAID to make good use of the drive.

[ohlwiler]

Although I would not buy an F4 drive for use in an unraid array right now, if I had one, I may consider using it. How much would it cost to return it? What kind of a deal did you get on it? Can you wait for a deal on something else? If you decide to keep it, I would not use it as a parity drive. I would only use it as a media drive, install it in your array, fill it up, and leave it alone. In that scenario it would be fine. I have three F3 drives and the temperatures they report are in line with the others in the array. I don't see that as reason alone to reject a drive. Reliability, low power and ease of warranty replacement are what I seek.

[LaurentHU]

FYI I am now using 2 of those in my array since 3 weeks, not as parity though. Good performance, low temp and stability as far as I can tell, but of course we'll have to see if they stand the test of time.
Really not sure though where the issues were coming from for others, as those drives seem to offer perfect 512 sector compatibility.

[Blofeld]

FYI I am now using 2 of those in my array since 3 weeks, not as parity though. Good performance, low temp and stability as far as I can tell, but of course we'll have to see if they stand the test of time.
Really not sure though where the issues were coming from for others, as those drives seem to offer perfect 512 sector compatibility.

interesting - can you tell us anything about the performance? is is on par with your other disks?

[LaurentHU]

Well that the thing - I didn't notice any difference with other recent disks, and they are much better than the old slow 300GB and 500GB they replaced - but that's not performance per se!
What's the best way to really test? Parity check speed for example didn't change vs when I had the old disks, which I think is good because it should mean they perform as good as the 2*203WI that were the only 2TB I had so far.

[prostuff1]

Well that the thing - I didn't notice any difference with other recent disks, and they are much better than the old slow 300GB and 500GB they replaced - but that's not performance per se!
What's the best way to really test? Parity check speed for example didn't change vs when I had the old disks, which I think is good because it should mean they perform as good as the 2*203WI that were the only 2TB I had so far.

If you added 2 new disks that should by all rights perform better than the disks they replaced, but you are getting the SAME parity check speeds then something may be not quite right.

If you have unmenu installed try the Disk Management and Disk Performance links in the upper left corner.

[BRiT]

low temp and stability as far as I can tell, but of course we'll have to see if they stand the test of time.

You can not trust the temps that they report. At best you need to add an adjustment offset to what they do report, likely 5C - 10C. It has been confirmed that they report temps that are impossible to achieve (below room temps).

Also, if you replaced antiquated disks with brand new disks, your performance should have increased. If you're seeing the same performance, then something is wrong with the drives or the system or your observations. As prostuff1 said, you should try unMenu's Disk Management and Disk Performance links.

[Joe L.]

As prostuff1 said, you should try unMenu's Disk Management and Disk Performance links.

They too might report inconsistent results... (Disk Performance link in unMENU is based on what bwm-ng reports.  bwm-ng NEVER reports the same rate as unRAID's management console when doing a parity check.)

Remember, parity check speed is determined on the SLOWEST disk involved through the SLOWEST interface involved.  A brand new "fast" disk will do nothing to change the effective parity speed if it was not replacing the disk that was the slowest.

[madpoet]

I have 4 in my array right now.  I guess I don't know enough to know if they are impacting parity!  Ultimately it hasn't mattered to me yet.

[LaurentHU]

As prostuff1 said, you should try unMenu's Disk Management and Disk Performance links.

They too might report inconsistent results... (Disk Performance link in unMENU is based on what bwm-ng reports.  bwm-ng NEVER reports the same rate as unRAID's management console when doing a parity check.)

Remember, parity check speed is determined on the SLOWEST disk involved through the SLOWEST interface involved.  A brand new "fast" disk will do nothing to change the effective parity speed if it was not replacing the disk that was the slowest.

Exactly - I had only 1 disk going until 2TB, now I have 3 - 1 203, 2 204 - and speed is the same. My conclusion is that 204 is as fast as 203 on parity check...

As for the Disk Performance, well, it's an instant measure and depends on a lot of parameters, I'd rather use a Linux level test... Is there any?