16th July 2008 - OCZ Core SSD review

The first (to my knowledge) interactive review of the new OCZ Core SSD drive, in this case a 64GB model. Please feel free to suggest tests or data you'd like to see to blueficouk@googlemail.com and I'll do as much as I can.

UPDATES: For those of you who are checking back for updates, I'll put a note here when something changes on the page below, to save you scanning through.

10am BST - Phil from the Netherlands has suggested running PC Mark 05, which I'm currently downloading. I was also asked if I had any background processes running when the write test was being carried out as that may have caused the strange performance. Well I run the minimum of utlities and nothing should have been running in the background. I would run another write test but it would require wiping the drive which I've slowly been setting up an OS on, maybe I'll make an image of it later and do more testing that way.

17th July - 9.40am BST - I used my Thinkpad with the Core last night and battery life does seem to have increased a fair bit, I wish I could do a more scientific power test but I'm just going on impressions from real-world usage. The overall system temperature has decreased massively and I was only using a 5400rpm drive before. The Thinkpad is silent in use, apart from when the CPU fan comes on sometimes, which it doesn't seem to need to, as the underside casing feels cool to touch. The killer application for these first-generation low cost SSD's has to be in notebooks where they'll probably make the biggest difference.

7.49pm BST - Don't think there'll be any more updates today sorry. My last notes for the day are: That going back to systems with traditional disks in them is annoying both accoustically and in terms of performance. If my Thinkpad / SSD combo had a three monitor display ability on it like my Desktop, I'd quite happily ditch the Raptor + Quad core desktop for it. Oh and the overall performance per pound / dollar of the Core is IMHO excellent and I don't regret buying it at all.

6.26pm BST - HD Tach test can be viewed here (Run on Thinkpad, earlier HD Tune were done on Desktop).

5.50pm BST - HD Tune write stats added further down page, these were done on the Vista desktop, not the Thinkpad

5.30pm BST - Running XDD tests shortly and have just run h2benchw.

If you want to run h2benchw on your own drive to compare, I ran it with: h2benchw.exe 0 -s -c 0 -p -english

Here are the results (will have them for older Hitachi soon):

UDMA mode 5 active.
acoustic management not supported.
Capacity: 118117440 sectors=57675 MByte, CHS=(7812/240/63)
Checking timer for 10 seconds (Win32) ............. Ok.
timer resolution: 0.279 µs, 3.580 MHz
timer statistics: 1748819 calls, min 1.40 µs, average 3.01 µs, max 501.74 µs
Reading some sectors to warm up... done.
interface speed test with block size 128 sectors (64.0 KByte):
sequential read rate medium (w/out delay): 70.6 MByte/s
sequential transfer rate w/ read-ahead (delay: 0.97 ms): 63.7 MByte/s
Repetitive sequential read ("core test"): 71.7 MByte/s
Measuring random access time (whole disk):
reading... 0.43 ms (min. 0.37 ms, max. 5.67 ms)
random access time in lower 504 MByte
reading... 0.43 ms (min. 0.36 ms, max. 0.53 ms)
Running application profile `swapping' ...20562.6 KByte/s
Running application profile `installing' ...79275.9 KByte/s
Running application profile `Word' ...78362.1 KByte/s
Running application profile `Photoshop' ...67571.0 KByte/s
Running application profile `copying' ...92068.0 KByte/s
Running application profile `F-Prot' ...33102.5 KByte/s
Result: application index = 53.3
!!! WARNING: application profiles inaccurate since measured read-only

4.40pm BST - If you know of any (preferably free) good disk benchmark tools, please let me know via email.

I'm getting an initial impression that write performance is relatively not going to be as great as read performance. Which is to be expected for MLC memory at this price point. Just how not as great isn't clear yet. I'm also having a problem with my Thinkpad that I'm running SATA in compatability mode as I get BSOD when in UHCI, I'm not sure what difference this will make to performance but I'm testing the Hitachi drive in the same compatability mode. I do have SATA drivers installed and don't know why it is BSOD'ing with the Core but not the Hitachi. Might just be a Thinkpad BIOS issue.

DISCLAIMER: As you can tell I don't review drives as my day job so apologies for this not being as professional as it could be, but hopefully some of the info will be useful. Also, any performance figures seen here are not necessarily representive of the performance you will get with a Core SSD on your system.

8.30am (BST) The unboxing

Up bright and early at 6am this morning, the delivery man dropped off my OCZ Core SSD.

As soon as I first read of this new lower-cost SSD drive, I pre-ordered straight away. I plan on using it in my Thinkpad X61s, with the aim of extending battery life, running the laptop cooler and not worrying about shocks losing data.

So to start with, here is an unboxing video and I couldn't resist, I gave it a little shock test too, by dropping it a couple of inches on the floor. Well that is supposed to be one of the advantages of SSD isn't it...

Note: After doing the little drop, I noticed on the drive label: "Warning Delicate product sensitive parts inside. Damage may occur if shocked." So this isn't something I'd recommend doing and I've yet to see if the drive still works...

First impressions

Putting the Core into the Thinkpad that it will live in and starting it up was a slightly surreal experience. You get so used to the drive spin up and clunking when turning on PC's that its abscence is very noticeable when it's not present.

A fresh install of XP didn't seem much quicker than normal and passed uneventfully. With Windows installed I set about installing drivers and tweaking. Now the performance improvement over the old 2.5" 5400rpm drive the Thinkpad had previously becomes obvious. Boot-up time seems to have halved (will do timings later). Installing drivers and Windows update all seem faster than normal. Just going to finish getting XP set up and will report back with some numbers.

Booting, installing, shutting down

My first test on the Thinkpad is to compare booting speed, installing of an application and shutting down times between the Core and the Hitachi 5400rpm the Thinkpad came with.

	OCZ Core SSD	Hitachi 5400rpm
Booting to XP	26sec	58sec
Installing paint.net	44sec	52sec
Shutting down XP	34sec	33sec

So my first impressions were correct and the Core boots in half the speed of the old drive. Installing of Paint, the gap narrows, presumably due to write speed not being so great on the Core.

Power usage

Another basic test, this time the difference on power usage between the Core and the Hitachi.

Looking at the above image, you can pretty much ignore the remaining battery figure, as this changes every couple of seconds as the Thinkpad battery software does its estimating.

The key figures are in the Battery details section and you can see the OCZ Core is using slightly less power. Both screenshots were taken when the system was at idle, so this could explain why the difference isn't great and I'm hoping the Core will offer better performance relatively under load.

HD Tune results

Here's some HD Tune stats I ran before putting the drive in my Thinkpad, the benchmark system is a Quad-core Q6600 with 4GB RAM on Vista.

The first drive to compare the Core to is the oldish WD1500ADFD Raptor, which I use on my desktop:

Raptor read stats

OCZ Core SSD read stats

OCZ Core SSD write stats

So...the write stats are all over the place. OCZ mention write speeds of up to 80MB/sec and the Max figures do approach that. The 0.1MB min write speed isn't as worrying as it might appear, at least not to me anyway. Using the drive on my Thinkpad is so much better than the drive it replaces.

I'll hold my hands up and admit I know little about how drive controllers work, but I do wonder if the Core controller has a write buffer and this is being used to good effect until 25% when it fills. Then there is the all over the place performance as the buffer is used and written from repeatedly over and over. Whereas with an SLC SSD, because the controller doesn't need to do as much write-checking and can write faster, they're more consistent with their results.

Please check back soon and any digg's are appreciated. :)