Page 1 of 1

Samsung F4 Raid 0 vs Caviar Black

Posted: Wed Jun 06, 2012 9:58 pm
by dsrtegle
I am interested in Raid 0 with two F4 2TB drives since it is quiet and fast, but it is 5400RPM. Will the Raid 0 be slower than Caviar black 2TB with even higher noise level? How can they be compared in terms of speed and noise, regardless of capacity?

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 10:28 am
by mkk
What do you need to go faster? Only copying large files in strict sequence will benefit from striping 5400 RPM drives like the HD204UI's together. Running programs like games off them can even do worse than a single drive, since their weakness is seek times and seeks get at least somewhat worse when striping. I keep my HD204UI's as single drives and wouldn't consider any 7200 RPM 3,5" drives for their noise.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 10:55 am
by dsrtegle
I have 128G SSD for Mac OS and APPs. All other stuff is stored on HDD, including documents, pictures, video, music, VMware images and tons of source codes. The apps operating on these data are of a wide variety, like Adobe CS, MS Office, iTunes, programming tools, video player and editor, etc. In this case, I am worrying about the performanc of a single 5400RPM, even if with high density drive like F4. Any experience is more than welcome.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 11:58 am
by HFat
Drive performance is complicated.

rule #1: most people have no clue and generalizations are wrong
rule #2: generic benchmarks are useless
rule #3: don't use RAID if you don't know what you're doing

I haven't used a 7200rpm in years, except as backup drives when they were cheaper than the slower versions.

What you want to avoid above all with hard drives is concurrent access. A faster hard drive makes concurrent access marginally less likely but it's software that causes concurrent access. That means you can usually solve performance issues on single-user systems in software and/or by moving data between drives.

If you have a dumb OS, you will experience slowness regardless of the drive you use. But RAM is cheap. If you have loads of RAM and allow a reasonably smart OS to make use of that RAM to speed up your data drives, it will speed up many operations. So change your OS and upgrade your RAM if necessary. Also make sure you have not tweaked your OS with silly settings (lots of tweaking tips you find on the web actually slow you down).
There are additional software tricks you can use to speed up particular operations if you actually have performance issues with the default settings. So post about your issues if you have any.
Moving select parts of your data drives to your SSD could also speed stuff up obviously (I bet your SSD is filled with stuff which has no buisness being there). Upgrading your SSD might also help.
Partitioning is also free and could also help with performance if you know what you're doing.
Using more than one drive could also help with performance, depending on what your issues are. RAID is not the only way to do this but sometimes a RAID-type setup is necessary (the only alternative being a very expensive SSD).
You could conceivably set up a Flash drive as cache but that would probably be unnecessarily complicated and risky.

Since you use a Mac, this is probably a topic for a Mac forum.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 1:00 pm
by washu
Will a RAID 0 of Samsung F4s beat the raw transfer rate of a single WD Black? Probably. Will it beat (or even match) the access time? Not a chance. In many situations access time is far more important. Compare a "slow" SSD like an X25-V. It will have less transfer rate than a WD black, but will absolutely destroy it in access time. For at least two of your tasks (VMs and source code) you want access time over transfer.

Really the only thing that might benefit is if you are doing video editing in Adobe CS with big files. That also assumes nothing else is causing disk contention.

As HFAT said, you want to avoid concurrent accesses, ie disk contention. If have multiple data sets that you are going to access concurrently you'll see a much greater speed up by putting them on separate stand alone drives then together on a single RAID 0.

A simple example you can do yourself. Put a bunch of files that are much bigger than your RAM (to avoid caching) on the RAID 0. Now copy them to itself and take note of the speed. Repeat the test but with the drives setup stand alone and copy from one drive to the other. You'll see the second copy go much faster as the drives are not facing concurrent accesses.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 1:22 pm
by HFat
washu wrote:For at least two of your tasks (VMs and source code) you want access time over transfer.
Except you don't want the access time found on spec sheets and generic benchmark results but effective access time, the kind that's subject to caching, contention and the location of the data on the drive.

What is source code doing on the hard drive to begin with?

Setting up the VMs so that it they can be put on a relatively small SSD may be trickier but it'd also be a much better solution than getting a marginally faster hard drive.
If you really need them on a hard drive, putting (some of) the VMs on a separate hard drive would be a better solution in many cases, especially with clever partitioning.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 1:42 pm
by washu
HFat wrote: Except you don't want the access time found on spec sheets and generic benchmark results but effective access time, the kind that's subject to caching, contention and the location of the data on the drive.
This is true, but in general a drive with a faster access time will be faster at these kinds of tasks. Lots of variables at play of course.
What is source code doing on the hard drive to begin with?
If he's just developing at home or working on local checkout then this is reasonable. Setting up a source control server may be a bit overkill for some home uses.

Someone at work (not me) ordered a bunch of PCs for our developers without checking the specs first and they came with WD green drives. Source control checkin/checkouts and compiling are downright painful on those drives. While SSDs are awesome for this task, even standard 7200 RPM drives made a huge difference.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 2:28 pm
by HFat
washu wrote:While SSDs are awesome for this task, even standard 7200 RPM drives made a huge difference.
Lacking an SSD or effective caching, RAM drives would be fine as well...
But maybe you have way bigger projects than I've ever seen.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 3:53 pm
by dsrtegle
Thank you so much guys. The computer is mainly for websites development, so that is why I have all kinds of data. Based on your comments, I am now thinking to use a separate Raid 0 for video only, and a single drive for other files. Now the question is, which HDD to go, F4 or Black, for the single drive to store documents, codes, pictures and audio? Vmware images may be too different. Maybe I can buy a secondary SSD for it, but let's ignore them for now. For other file types, should I go with F4 for lower noise, or Caviar Black for shorter access time? Is it possible to use some noise reducing methods, like elastic suspension or grommets, to lower Caviar Black noise level close to F4?
BTW, the computer configuratio is: i7 950, 12GB RAM, GT240. I bought this very low-end video card for its fan-less feature, so I do want a quiet working environment.

Re: Samsung F4 Raid 0 vs Caviar Black

Posted: Thu Jun 07, 2012 4:06 pm
by dsrtegle
Concurrent access is probably not a problem. I may open many different programs and a lot of windows at the same time, but actually only work on one, except for waiting for video rendering to finish.