New OWC SSD is Fast!

[mangrove]

Just did xBench tests on my 2009 Mini-3,1.

Old drive=stock 5200 320GB, 4 GB RAM, 2.53gHz.
xBench Disk Test=38.28

New drive=40GB OWC clean OSX custom install (about 4GB) with Applications (about 4GB) Library (about 11GB), Users (about 9GB), 4GB RAM.
xBench Disk Test=144.32-via FW800, not SATA

 

[Baby Mac]

I was wondering how much improvement there was between drives. I put in my SSD before doing any tests (I never tested a computer's speed before). The stock drive specs look the same as the 2010 mini and our OWC drives should have the same performance.

I'd like to hear about differences you notice in day-to-day use.

 

[dolphin842]

Any chance you could post a screenshot or list the results for the individual disk tests (sequential/random reads/writes, etc.)?

 

[mangrove]

Any chance you could post a screenshot or list the results for the individual disk tests (sequential/random reads/writes, etc.)?

No problem. Ok disk tests rounded off for simplicity. Left is stock 320GB HDD. Right side is OWC 40GB SSD via FW800.

overall disk test=38-------------------------------------144
Sequential=------77-------------------------------------94
--uncached write=103----------------------------------122--4k blocks
--uncached write=99------------------------------------117--256k blocks
--uncached read=43------------------------------------51--4k blocks
--uncached read=115----------------------------------158-256k blocks
Random=---------25------------------------------------315
--uncached write=8------------------------------------193--4k blocks
--uncached write=72-----------------------------------216-256k blocks
--uncached read=68------------------------------------1811--4 k blocks
--uncached read=112-----------------------------------424-256k blocks

Trust this helps. Next time I do it, I'll download to xBench site and give details. Let me know how it compares and to which name/type of drive.

 

[mangrove]

I was wondering how much improvement there was between drives. I put in my SSD before doing any tests (I never tested a computer's speed before). The stock drive specs look the same as the 2010 mini and our OWC drives should have the same performance.

I'd like to hear about differences you notice in day-to-day use.

No problem. Just give me a few more days to follow up completing upgrade, etc. and I'll give you a run down. I agree that real world experience is more important than tests, but they are a start and an initial indication.

Incidentally, I also have a 2010 Mini 2.66mHz with 8GB RAM and will soon install same SSD in it. Just did a clean/hybrid install of the SSD today and its size was less than the 2009 at 27.2 GB with everything except iTunes, iPhoto and iMovie data, but left all Pages and Numbers files on SSD. Come to think about it the 2010's downloads came to about 3GB which is slightly larger than the 2009's downloads. Weird!

Which OWC SSD did you end up getting?

 

[dolphin842]

Thanks mangrove for posting your full stats.

It's kind of sad (though not surprising given the specs) that the stock drive for the Mini is slower than the 3.5" drive in my 4.5-year-old iMac:

Overall: 70

Sequential: 116
4K-write: 114 (70 MB/s)
256k-write: 117 (66 MB/s)
4k-read: 103 (30 MB/s)
256k-read: 133 (68 MB/s)

Random: 50
4K-write: 19 (1.98 MB/s)
256k-write: 121 (38.8 MB/s)
4k-read: 91 (0.65 MB/s)
256k-read: 128 (23.8 MB/s)

I think something's up with your OWC drive though. My 4k sequential reads are about twice as fast (!), and most of your stats are roughly half the speed of the benches I'm seeing for the OCZ Agility 2. Here's one of the better benches I've seen for the Agility:

Overall: 311

Sequential: 189
4K-write: 205 (126 MB/s)
256k-write: 232 (131 MB/s)
4k-read: 105 (30 MB/s)
256k-read: 402 (202 MB/s)

Random:
4K-write: 1148 (122 MB/s)
256k-write: 422 (135 MB/s)
4k-read: 3293 (23.34 MB/s)
256k-read: 1045 (194 MB/s)

In this (probably optimistic) case the 4k reads are on par with my old iMac's mechanical drive. Even for the Agility though, the reads don't look to be anywhere near what is advertised. Is this a problem with OS X or is Xbench just not testing the SSDs in a way that reveals their speed?

 

[mangrove]

Thanks mangrove for posting your full stats.



I think something's up with your OWC drive though. My 4k sequential reads are about twice as fast (!), and most of your stats are roughly half the speed of the benches I'm seeing for the OCZ Agility 2. Here's one of the better benches I've seen for the Agility:

I haven't had much time of late to look at comparative tests. Thanks for that heads up.

There could good reason since the SSD is not installed yet. I'm running via FW800 for now. You know, hold down Option key on startup then switch to SSD from stock drive.

Got my iTunes content copied over to what will be my #2 internal, but getting the SSD Boot to see it over there is giving me fits. Until I get that worked out I 'm not going to open it up to put in 2 drives and more RAM because if I have to take back out the #1 drive (the SSD) it'll be like starting over.

So maybe once I do that the SATA interface will speed things up. What do you think?

 

[dolphin842]

There could good reason since the SSD is not installed yet. I'm running via FW800 for now.

Ah, that would explain it. FW800 would give you 100MB/s at most. SATA II tops out at 300MB/s. The drive should bench much better once you get it inside.

Incidentally, just of out curiosity, how does the system feel (responsiveness/app loading times) with the SSD hooked up externally? Any faster than a generic internal desktop drive?

 

[mangrove]

Ah, that would explain it. FW800 would give you 100MB/s at most. SATA II tops out at 300MB/s. The drive should bench much better once you get it inside.

Incidentally, just of out curiosity, how does the system feel (responsiveness/app loading times) with the SSD hooked up externally? Any faster than a generic internal desktop drive?

Much better, yes. On boot up it is like 12-15 secs from gray screen-spinning gear start to dock appearing and on the stock it's like 35 seconds up to 45 seconds sometimes. But the longer one is after I have been screwing around changing boot up drives so it has to see where I'm pointing it to, I guess.

For apps with the SSD I click and it's there versus the stock it's 3-4 bounces.

I'll let you know how it does once I'm using SATA. I guess that 311 score above is what I should shoot for. But you know with FW that does bot seem too bad to me. What you say considering?

 

[dolphin842]

That's actually quite good... I mean, faster boots and instantaneous app launches are the reason for going SSD, and you've achieved that. My assumption was that an SSD hooked up via firewire was going to be severely hampered, but it seems you're getting most of the responsiveness of an SSD via FW800.

Maybe someone more knowledgable about file structure can shed some light on this: when OS X is booting and when launching apps, what kind of reads are prevalent (sequential vs. random, 4k vs. 256k)? I'm just trying to figure out which of the Xbench specs we should be paying most attention to. If it's 4k random reads, then the 10-20MB/s we're seeing from these SSDs could even be handled by FW400.

 

[mangrove]

That's actually quite good... I mean, faster boots and instantaneous app launches are the reason for going SSD, and you've achieved that. My assumption was that an SSD hooked up via firewire was going to be severely hampered, but it seems you're getting most of the responsiveness of an SSD via FW800.

Maybe someone more knowledgable about file structure can shed some light on this: when OS X is booting and when launching apps, what kind of reads are prevalent (sequential vs. random, 4k vs. 256k)? I'm just trying to figure out which of the Xbench specs we should be paying most attention to. If it's 4k random reads, then the 10-20MB/s we're seeing from these SSDs could even be handled by FW400.

Very good point. You know on some threads many ask for a certain test other than xbench, but to me all tests are just that. I just don't remember what it was called. Your real world question is the most important to me.

 

[dolphin842]

Indeed, I've been doing some digging this afternoon to try and track down some answers.

Perhaps the most concrete information I found was this video benchmark done by OWC. It compares their SSD in a variety of MacBooks, but the most instructive comparison for our purposes is the Core 2 Duo MacBook Pro with SATA II vs. the old Core Duo MBP with SATA I, which takes over 10 seconds longer to run through the boot/app-launch sequence.

Unfortunately since the rest of the specs (RAM/CPU) are different, a direct comparison would be unfounded, but it certainly seems as if the SSD is taking advantage of the faster 300MB/s interface in the Core 2 Duo machine. If this suggestion is accurate, it would follow that one would also notice a real, visual improvement in a SATA II vs. FW800 scenario. Of course, once you get the OWC in your Mini, you'll have some first-hand experience with what the difference is actually like in normal, day-to-day use.

Elsewhere, OWC actually sells a FW800 external solution that appears to offer a noticeable improvement over both desktop and laptop hard drives. Aside from OWC's claims (comment #18), I've found two reports that attest to the performance boost, so you're not alone. Along those lines, one of my thoughts earlier today was that, hey, if you can get most of the benefits through a FW800 bus, why not just get a cheaper SSD that's not rated as fast (anything above ~80MB/s would do), and use that? Well, one article actually simulated several months' worth of everyday usage on several brands of SSD, and it turns out all except the OWC suffered moderate to severe degradation in speed over time. So it looks like I'll have to keep saving my pennies after all....

 

[iDave]

That's actually quite good... I mean, faster boots and instantaneous app launches are the reason for going SSD, and you've achieved that.

I sometimes wonder if SSD proponents sit around all day just launching applications and rebooting their machines. What am I missing? I reboot three or four times a week and launch each application maybe twice a day. So, let's say I wait 60 seconds a day doing such things. The rest of the time is spent working with my data. If the data is on an external drive vs. internal, seems 99.99% of what I do would actually be slower if I replaced my large internal drive with a tiny SSD.

 

[dolphin842]

Well for RAM-limited systems, it's not just launching apps, but also bringing them back out of virtual memory. With a lot going on, the swap-related slowdown can be a big performance drag. In this case, an SSD can make a system that's limited to 4-8GB of RAM run much better and last a lot longer than it otherwise could. At the same time, if you look over in the Mac Pro forum, a lot of people see improvements on higher-end as well.

Also, for geeks who have grown up waiting ages for systems to boot and apps to load, having the system be so responsive is a little unreal and perhaps a bit cathartic .

 

[Baby Mac]

Incidentally, I also have a 2010 Mini 2.66mHz with 8GB RAM and will soon install same SSD in it.

Which OWC SSD did you end up getting?

I am thinking about getting another 2010 mini for home use

I got the DIY Upgrade Bundle: 120GB Mercury Extreme Pro Solid State Drive, OWC Express USB 2.0 Enclosure Kit + Includes OWC 5 Piece Tool Kit at the current $329 price. The additional $10 for a nice enclosure for the stock drive and all the necessary tools made the bundle a no-brainer. It also came with a 1-foot USB cable, which I hook up for Win7 backups. The nice thing is when I plug the drive in, Win7 asks if I want to start my backup using Win7's built-in tool.

 

[Giuly]

Is this a problem with OS X or is Xbench just not testing the SSDs in a way that reveals their speed?

Doesn't the drive double-check every data some time after first use? I guess that's what it does and becomes faster as it verifies that the drive is OK.

Let me know how it compares and to which name/type of drive.

Probably every other SandForce-1200 drive out there, some of them only have the firmware modified to do more IOPS.

 

[mangrove]

I am thinking about getting another 2010 mini for home use

I got the DIY Upgrade Bundle: 120GB Mercury Extreme Pro Solid State Drive, OWC Express USB 2.0 Enclosure Kit + Includes OWC 5 Piece Tool Kit at the current $329 price. The additional $10 for a nice enclosure for the stock drive and all the necessary tools made the bundle a no-brainer. It also came with a 1-foot USB cable, which I hook up for Win7 backups. The nice thing is when I plug the drive in, Win7 asks if I want to start my backup using Win7's built-in tool.

It's amazing to me that Apple has refurbed new 2010 models so soon after introduction. That might be the way to go right now if you're in the market for a 2010 Mini..

 

[mangrove]

So I got tired of seeing what would happen if I installed the SSD-actually 2 SSD's and upped the RAM to 8GB. Did it today.

Well just did an xbench tested new setup and it appears I might have to open it up again to check my install. The CPU test actually drops in half sometimes. Could it be due to the 2nd internal SSD using the cable where the SD used to sit?

Can't figure it out. Most disk test data about doubles except random read 4k block is just shy of 3000. Overall disk is around 255. Best CPU at 181 and Memory with 192 being the best. Best overall results only at 219.

Something still does not sit right with me. Biggest problem is that on restarts I am now loosing Pictures and ability to point to iTunes content. bBut this is my doing.

Now boot times are down to around 5 seconds from gear wheel until I see the dock. Dock apps load in no more than 1 bounce, but mostly a half a bounce. I work with large Numbers files a lot and they load much faster than in the past. Heck, everything loads faster and saves faster.

 

[evilcat]

I think most people forget that the tests posted tend to be for uncommon uses. If you are regularly transferring 200Gb of data from drive to drive, then SATA or eSATA is going to kick FW800's arse.

Go to OWC's SSD page (http://eshop.macsales.com/shop/internal_storage/Mercury_Extreme_SSD_Sandforce/Solid_State_Pro) and look at their every day usage stats (4k random read/writes) and you'll see that an internal hard disk gets 0.5MB/s read and 1Mb/s writes, where the SSD gets 26MB/s read and 39MB/s write speeds.

That's for real world acts like saving files, playing video or ripping a CD... and it doesn't saturate the FW800 bus by any means.

So according to that, until you come to copy files, your SSD should work just as well over FW as it would installed internally.

I'd be interested to see if the FW bus causes any other issues regarding latency, but if not it seems an SSD in a FW800 enclosure is a good bet for the mini.