Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
SATA 3 Anyone!
- Thread starter Vapor matt
- Start date
- Sort by reaction score
I don't think the performance difference will be noticeable. Adding that extra 100MB/s or so to sequential reads/writes will help with copying or moving large files, but unless you're moving them to/from other SSD's those kinds of operations are often constrained by the speed of the other media anyway (often much less than SATA2).
Most desktop usage is random I/O at speeds well within the constraints of SATA2. Even the fastest SSD's can only process random reads at about 25MB/s at QD=1 (although this is 10-100 times faster than HD's its still well within SATA2).
Most desktop usage is random I/O at speeds well within the constraints of SATA2. Even the fastest SSD's can only process random reads at about 25MB/s at QD=1 (although this is 10-100 times faster than HD's its still well within SATA2).
having the extra read and write speed will speed things up greatly when editing video.
It will speed things up... I'm not sure I'd expect great improvements there either unless you're starved for RAM and then you've got bigger issues
Honestly, if you guys are convinced that sequential reads/writes are critical, then you should be running two drives in RAID0 on the SATA2 ports instead. That way you can get close to 600MB/s throughput (2xSATA2 ports) instead of the 400MB/s this card offers.
Bottom line is that this card is a nice DIY proof of concept, but the added value is negligible.
Or you add TWO of these cards and then RAID across both controllers.
That's how I will do it. Not because it's really necessary, but simply because I can :-D These cards cost nothing.
Keep in mind, if you're using a 2009/10 model, slots 3 and 4 share the same 4x PCIe lanes via a switch. So there's a performance penalty if this implementation is used.Or you add TWO of these cards and then RAID across both controllers.
That's how I will do it. Not because it's really necessary, but simply because I can :-D These cards cost nothing.
Well there's a few aspects to this...
First, there's your workflow. As I mentioned above, SATA3 is only an improvement over SATA2 in sequential I/O where large files are being transferred. The cases where this can actually impact a person's workflow is fairly limited. If you work on manipulating a lot of large files, then it can help, but even then, there's often a lot of other system resources that come into play.
Next, there's the nature of the files you're working on and the SSD technology used to achieve those high-speed sequential reads and writes. Most people here wanting to enable their SSD's with SATA3, are using SandForce based drives. As you may or may not know, these drives rely on compression to achieve these high sequential speeds. When working with files like images and videos that are already compressed, their performance is somewhat muted... often well within the constraints of SATA2. So even if you work with large files, you have to ask yourself, what will the benefits be if my files are highly compressed already and my drive uses compression to achieve SATA3 like I/O transfer rates?
Also, SATA3 is not the only or best way to achieve high sustained transfer rates. RAID0 arrays have long been the answer to improving STR and that hasn't changed with SSD's. Two SATA2 drives in RAID0 will outperform a single SATA3 drive. But then again, Two SATA3 drives in RAID0 will probably give four SATA2 drives a run for their money. But again, as per the above, is your workflow one that can actually benefit?
Finally, this SATA3 card is not the best SATA3 solution. It's bottlenecked by the x1 PCIe lane and perhaps even the controller itself. It's certainly not opening up your SSD to it's max potential. I wouldn't waste my time on SATA2 card unless it has two ports and 4x lanes of PCIe and a controller that can handle two drives in RAID0 for about 1GB/s throughput.
The bottom line is, I wouldn't invest much effort, money, or time, in this SATA3 card. The real benefit of SSD's is in their random I/O performance and this is well within the limits of SATA2.
yes. as was mentioned by me earlier in this thread.
the point was to find a cheap and easy solution to boot from sata3. this card does just that. it's neither a lot of effort nor 'wasting money'. it will never come close to a full blown 4-lane PCIe controller that's for sure. but that's not what anyone was asking for. put two of these in your mac, raid0 them and you'll see roughly 800 MB/sec. - for a tenth of the price of a non-bootable 4-lane controller. a pretty damn good option, if you ask me.
Yes it will, as the PCIe lane is slower than the SATA 3 controller (gets into the specifications between PCIe 2.0 and SATA 3).
Real world sequential throughputs on a SATA 3 controller would be ~540 - 550MB/s, but since it's a single PCIe 2.0 lane, you're bottlenecked to 500MB/s for a single drive with the PCIe lane all to its self.
And in your specific intended configuration, that single lane is used for both cards simultaneously (stripe set where both cards share the same PCIe lane via a switch). So yes, there will be a net performance degradation.
Now whether or not you'd notice is another matter, as your workflow will have a significant effect here. For example, if you're using the disks for random access performance, you'd never notice it, as the random access performance of an SSD is no where near capable of saturating the PCIe bandwidth.
But for large files on drives that can exceed SATA 2 (sequential throughputs), you may be able to notice it, particularly if you've experienced this with such SSD's that didn't need to share bandwidth (PC), unlike a 2009/10 MacPro when running cards in slots 3 & 4.
You can think of it as splitting the bandwidth in half, but in reality, it's a bit under that due to switching latency (bit under 250MB/s per drive for sustained throughputs <= large sequential files>).
Hope this helps.
Nano... as I understand it, slot 3 and 4 are both 4-lane. if you put single lane cards in both of them, shouldn't the switching be intelligent enough to give each card it's full lane? there are 4 of them available.
I seriously doubt it, as that increases the complexity = increased cost for the part (intelligent switches aren't nearly as common, so I'm figuring only for a MUX to handle the switching function).
But if you're willing to read off the part numbers (it's on the backplane board and located up where slots 3 & 4 are), I'll look for the datasheet.
Wow! Ordered on the 4th and received today (7th). That's fast service!
Installed card, connected SSD, and it boots Lion! Woohoo!
There's a noticeable speed difference vs. SATA II setup. However, I cannot do a proper disk speed test as I've moved all Users volume from the SSD to my RAIDed HD drives. Only OS, system files and apps are on my SSD. I have the Black Magic Disk Speed Test and it is analyzing my RAID drives.
Any ideas/suggestions?
Installed card, connected SSD, and it boots Lion! Woohoo!
There's a noticeable speed difference vs. SATA II setup. However, I cannot do a proper disk speed test as I've moved all Users volume from the SSD to my RAIDed HD drives. Only OS, system files and apps are on my SSD. I have the Black Magic Disk Speed Test and it is analyzing my RAID drives.
Any ideas/suggestions?
Thanks for this thread and OP! I've been waiting for a cheap "bootable" sata III card. My Vertex 3 MIOPS SSD has been running from the MP mobo's sata II port.
Just ordered! Can't wait to see the speed difference vs now!
As I've glanced over this thread, one main question pops into my head with regard to practical concerns. Forgive my naivete, BTW, as I am one of those "it just works" Mac users who seldom does any hardware hacking to my systems other than rudimentary RAM/HDDs/ExpressCard mods:
Question: Uber-Fast Boot Drives? From strictly a productivity standpoint, does a super-fast boot drive really provide continual real-world productivity gains/speed for actual work? If one is using any one of the RAID solutions cited in this thread for apps and document files with enough RAM installed on the system to keep up with the demands of the work being done, how often and to what degree would a "slow" SSD boot drive on a legacy SATA II bus affect performance? Or is this just a matter of a) embracing the challenge of locating "bootable" cards, and/or b) loving the "instant-on" feel of a MacBook Air and wanting that gee-whiz experience for one's main desktop?
I would like to put together a new system for music production in the near future. As I'll likely be depending upon fast i/o for use with huge orchestral and percussion sample libraries, finding a robust way to RAID SSDs inexpensively is a primary concern. Seeing that I know I'll need to spend substantial coin on RAM alone, I am just curious as to what i/o plan would make the most sense when balancing cost effectiveness vs. future-proofing from potential OSX-upgrade-based conflicts vs. speed.
So set me straight on this?
Question: Uber-Fast Boot Drives? From strictly a productivity standpoint, does a super-fast boot drive really provide continual real-world productivity gains/speed for actual work? If one is using any one of the RAID solutions cited in this thread for apps and document files with enough RAM installed on the system to keep up with the demands of the work being done, how often and to what degree would a "slow" SSD boot drive on a legacy SATA II bus affect performance? Or is this just a matter of a) embracing the challenge of locating "bootable" cards, and/or b) loving the "instant-on" feel of a MacBook Air and wanting that gee-whiz experience for one's main desktop?
I would like to put together a new system for music production in the near future. As I'll likely be depending upon fast i/o for use with huge orchestral and percussion sample libraries, finding a robust way to RAID SSDs inexpensively is a primary concern. Seeing that I know I'll need to spend substantial coin on RAM alone, I am just curious as to what i/o plan would make the most sense when balancing cost effectiveness vs. future-proofing from potential OSX-upgrade-based conflicts vs. speed.
So set me straight on this?
I am also using my Mac Pro for music production with large sample libraries. While I have a SATA II SSD boot drive, it offers absolutely no performance advantage once you have booted the computer and started your apps. What does make a huge difference is using SSD's to stream your samples from. As I posted in another recent thread, I am using a Transintl DX4 bracket to hold the drives (up to 4) and an OWC mini-SAS host bus adapter to connect up to 4 internal drives at SATA III speeds as well as 4 external eSATA devices. The card also has a RAID controller on board, but for our application, RAID 0 is no more desirable than having multiple independent drives.
I went this route since I needed both internal and external ports and only had a single PCIe slot available (UAD cards). THe SAS HBA is an 8 lane PCIe card, so plenty of bandwidth is available for running multiple SSD's at SATA III speed.
There was a somewhat noticeable difference in load times with larger libs in Kontakt changing my Crucial M4 SSD's from internal SATA II ports to the SATA III ports on the SAS HBA. In my case I needed the extra drive capacity as well as wanted the SATA III link speed.
Mini-SAS to SATA breakout cables can be had cheap from Monoprice. THe external SAS to eSATA cable is not widely available and I had to pay the $55 price tag from OWC for that.
Be sure to install this card in slot 2 (the other 16X PCIe slot right above the graphics card if you plan on using multiple drives with it.
Last edited:
It all completely depends on the work you do and with what kind of data you work with.
SSDs won't speed up your computer per se (as in compute). They will speed up data transfers. If your work includes moving around or loading a lot of data all the time (e.g. scrubbing video files) you will notice a huge improvement. if you only do office work (email, browsing, whatever) you won't see any improvement in actual application performance.
what you will notice in any case - and it is very significant - is overall system response. apps basically start instantanous. all the background stuff that your OS is doing (logging, paging, maintenance etc.) won't be noticable anymore, giving you the impression that your system feels faster all around. putting your system to sleep and waking it up is instantanous (macbook air'ish). these things alone make SSDs a worthy upgrade, even if your actual workflow doesn't benefit from it. you will definitely "feel" an improvement basically anywhere.
SSDs won't speed up your computer per se (as in compute). They will speed up data transfers. If your work includes moving around or loading a lot of data all the time (e.g. scrubbing video files) you will notice a huge improvement. if you only do office work (email, browsing, whatever) you won't see any improvement in actual application performance.
what you will notice in any case - and it is very significant - is overall system response. apps basically start instantanous. all the background stuff that your OS is doing (logging, paging, maintenance etc.) won't be noticable anymore, giving you the impression that your system feels faster all around. putting your system to sleep and waking it up is instantanous (macbook air'ish). these things alone make SSDs a worthy upgrade, even if your actual workflow doesn't benefit from it. you will definitely "feel" an improvement basically anywhere.
This drive is no longer made? I didn't see it on the site.
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Ah, I see the issue... Series 1 is EOL, but they have released a Colossus Series 2 (2x models; 460GB and 960GB).
At least the newer series is cheaper (here).
Technically, it's not 1TB anymore (960GB), but for ~$1200 less ($1245 by Amazon's prices), I don't think they'll get a lot of complaints over it.
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Do you really need to RAID the SSDs ? I'm pretty sure you'll see ample speed benefits to having an SSD for your boot drive, and another for your libraries. You could also get a third smaller one for scratch if you DAW uses that.
Not to hijack the thread, but based on what you're saying, do you think that I could get away with a set-up like this?:
- 128 GB SSD on stock SATA II bus for Boot/Apps only
- 256 GB SSD for each major sample library (example: EWQL's Platinum Symphony Orchestra alone is 194 GB), all of which would be hosted on one of the PCI card solutions mentioned on this thread--whatever is the most reliable and cost effective.
- 128 GB SSD for WIP mixes and final project files
- 128 GB SSD for application scratch disk and for general non-music production files
- 2 or more larger, slower HDDs for redundant back-ups of the entire system
It would be nice to have something robust; a system that has 99.99% uptime, no system overloads, and with redundant data back-up so that replacing a toasted SSD would be a quick and straightforward affair.
For huge sample libraries, my understanding is that RAM is the single most important asset to running things without hiccups, but being able to load and swap out samples quickly from disk also comes into play. Any further advice welcome.
This card has been around here a couple of weeks only. I don't know if I'd rig a 24/7 machine on it yet. And who knows what will happen at the next OS update, and the next, and the one after that,...
What you ask can almost be completely taken care of by the Mac Pro without extra cards. You can rig 4 SSDs (and more) on the SATA2 controller and already have a significant performance to load your library, and it's not even sure SATA3 would best that. RAM is what you need most of all. Backup can go through firewire or eSATA.