OCZ Vertex Drives

[lowfat]

vertex does fine in small random writes, its core/corev2 that were particularly bad at it

It is a significant improvement over the Cores, but still a ways behind.

[F@32]

Where is the raid -0- Vertex benches? I bought one Vertex 30Gb now waiting for some reviews and benches on the raid -0-. But all I see is WW3 against this drive now wow. I see everyone running out buying the Intel E, and I thought this one keeps up with it for a better price am I missing something here? When ever I see a posting of this ssd it is followed by a bashing.

I think I need to do some serious tweaking or forget about canned benchies... My single 30Gb maxes out at 166 in ATTO. Tried 32K and 128K stripe. RAID0 topped at about 300. Will post screen shots tonight after work.

[Rhys]

Your 30GB Vertex' are $4.50/GB
An X25-M is $4.53/GB. I highly call that significantly cheaper.

I bought my X25-M after Vertex' release and don't regret it one bit. The Vertex' are a great budget drive, but remember they still fall significantly short in small random writes and IOPS. So no need to start flame baiting and trolling.

What's performance like compared to 2x vertex raid 0?

[3Z3VH]

It is a significant improvement over the Cores, but still a ways behind.

Proof ?

[lowfat]

What's performance like compared to 2x vertex raid 0?

Well I never found RAID to be all that beneficial to real world tasks. Benchmarks yes, but opening applications/games/etc no. And yes I have used SSD + RAID before. I was using 4 patriot Warp V2's on an ARC-1220 before the X25-M.

Proof ?

I was just going from what I've seen of the Vertex so far. But I'll order a 30GB Vertex now and compare them for myself.

[3Z3VH]

30Gb is the crippled Vertex... less cache. 120Gb and up are their full-speed drives.

As far as RAID goes, in the HDD world I have seen huge real-world improvements, as my seek times and throughput both have improved using RAID. With two 74GB raptors in a RAID0 I saw a large improvement in speed over the single 74GB raptor in that same machine. Windows booted faster, my games loaded quicker, and even opening a browser was peppier. With the low seek times of SSDs, I question RAID's effect on speed, but I do see the point of using SSDs in RAID to speed up the random writes using a RAID controller's cache... I just have no real-world comparison to say the RAID made any difference with an SSD.

[lowfat]

30Gb is the crippled Vertex... less cache. 120Gb and up are their full-speed drives.

I am aware of that but way am I going to spent $550CAD just to compare how it does to an X25-M. And Rhys was calling the Intel SSD guys out on

Too many people upset that they emptied their bank account on X25's!
They don't won't other drives that are cheaper threatening their performance crown! Let's face it no one want's to feel like they wasted a ton of money!

And he was referring to the 30GB drives.

[Griff805]

http://ssdtoolkit.com/2009/03/10/ssd...el-x25-m-80gb/
x25 vs vertex

That's a review of a review. The Firmware they used in that drive is not the retail firmware.

[One_Hertz]

I was just going from what I've seen of the Vertex so far. But I'll order a 30GB Vertex now and compare them for myself.

Nice! I was this >< close to doing that myself, but was waiting for anand review.

[Rhys]

I am aware of that but way am I going to spent $550CAD just to compare how it does to an X25-M. And Rhys was calling the Intel SSD guys out on And he was referring to the 30GB drives.

Hi Lowfat

Apparently a new firmware update should be released next week.
If your testing only a single drive then I'm hoping AHCI will be enabled for small file read writes! Apparently Raid is a superset of AHCI, Could this be why the benchmarks show disproportionately better small file performance when on ICH10R raid 0?

Anyway got to admire your willingness to buy a vertex to compare, are you planning on selling it after or will you have a use for it?

[lowfat]

Hi Lowfat

Apparently a new firmware update should be released next week.
If your testing only a single drive then I'm hoping AHCI will be enabled for small file read writes! Apparently Raid is a superset of AHCI, Could this be why the benchmarks show disproportionately better performance when in ICH10R raid 0?

Anyway got to admire your willingness to buy a vertex to compare, are you planning on selling it after or will you have a use for it?

If they have a firmware that fully supports AHCI by the time the drive arrives I'll test with. I doubt I'll sell it. Either I'll use it as an OS drive and use the X25-M for games (more important to me than OS quickness ) Or throw in in my HTPC as a boot drive.

[koc]

Don't forget to post HDtach, HDtune and CrystalDisk results...

sure

[Halk]

30Gb is the crippled Vertex... less cache. 120Gb and up are their full-speed drives.

As far as RAID goes, in the HDD world I have seen huge real-world improvements, as my seek times and throughput both have improved using RAID. With two 74GB raptors in a RAID0 I saw a large improvement in speed over the single 74GB raptor in that same machine. Windows booted faster, my games loaded quicker, and even opening a browser was peppier. With the low seek times of SSDs, I question RAID's effect on speed, but I do see the point of using SSDs in RAID to speed up the random writes using a RAID controller's cache... I just have no real-world comparison to say the RAID made any difference with an SSD.

OCZs spec for the drives was 32MB cache on the 30 and 60GB drives. It turns out that pre-production ones have arrived with the 32MB cache, but retail come with 64MB cache. Post from Tony here.

I'm not so sure about RAID improving seek times, however when I've looked at the overall performance 2 drives gives a noticable boost, 3 gives a bit of a boost, but 4 or more it really does tail off.

[Halk]

That's a review of a review. The Firmware they used in that drive is not the retail firmware.

Right. Since the original reviewer has had both the high seq firmware (the one that did not ship) and the review on that site still seems to have the old firmware I'd be a bit careful with those figures. They certainly do make the Vertex look like the equal of the Intel, but we're still really waiting for a slew of proper reviews.

[3Z3VH]

OCZs spec for the drives was 32MB cache on the 30 and 60GB drives. It turns out that pre-production ones have arrived with the 32MB cache, but retail come with 64MB cache. Post from Tony here.
...

I stand corrected, thanks for the correction Halk !

[azza21]

OCZs spec for the drives was 32MB cache on the 30 and 60GB drives. It turns out that pre-production ones have arrived with the 32MB cache, but retail come with 64MB cache. Post from Tony here.

I'm not so sure about RAID improving seek times, however when I've looked at the overall performance 2 drives gives a noticable boost, 3 gives a bit of a boost, but 4 or more it really does tail off.

Hey Halk, just want to say you've been really helpful with regards to info on the vertex drives, much appreciated.

You mentioned something quoted above about after more then 3 drives in raid it tails off. I have an adaptec 5405 raid controller and was planning to get 4, 30gb ssd's to use all the ports but do you not think i will get a performance improvement having 4 over 3 drives?

Also is there any reviews of having 4 vertex drives using a hardware controller?

[Halk]

There's sadly only two Vertex reviews I'm aware of. One is quite in depth, on CD Freaks. The reviewer has attempted to make it stutter, but she wasn't able to. There's some real world testing done on it. Sadly though she didn't have an Intel drive available at the time so there was no comparison. The other review was at PC Perspective, however OCZ at the 11th hour changed the firmware due to customer demand for a drive that handles smaller file operations faster, as a result their review doesn't show the correct firmware (as far as I'm aware). They do have the right firmware, so hopefully a review is coming.

There's a general problem with SSD reviews since we're now learning that ATTO and HD Tach etc benchmarks are nigh on useless. What matters is how the drive performs in real life, and we don't appear to have either a plethora of benchmarks available, nor a widely accepted benchmark. Add to that drive alignement has an impact which can make some reviews useless, and finally that performance varies over time. All of these issues are progressing, but it's a bit of a minefield trying to accurately identify the performance of a drive.

The only SSD RAID0 scaling review I've read was not for OCZ drives, it was for (I think?) Samsung drives, it was a very good review and I believe the high level tests were with PC Mark - not sure if it was Vantage or 05. What it showed was that for some tests RAID has little or no impact, for others RAID0 with 2 drives gives a very significant performance boost, less with 3, and actually in some stats RAID0 with 4 drives was worse. I don't think RAID0 with 4 drives will ever be worse in real world performance. However just looking at the numbers...

RAID0 1 drive (really no RAID) - Baseline performance 1.0
RAID0 2 drives - Performance 2.0 (100% increase)
RAID0 3 drives - Performance 3.0 (50% increase)
RAID0 4 drives - Performance 4.0 (33% increase)

As you can see performance increases from the drives give diminishing marginal gains. And that's assuming perfect scaling, which simply isn't going to happen. Those percentage increases will always be worse in real world performance. Simply adding a drive to a RAID array will add some degree of latency to some tasks.

What I have seen with RAID performance that I cannot comfortably explain is performance increases where the drive is operating with I/Os below the stripe size. As I understand RAID0 performance should be near identical to a single drive. Perhaps this simply means that there is some sort of gain from command queuing - while one drive is busy then so is the other getting a different file.

Here is the review I was talking about.

[One_Hertz]

^^^^

You can't really take that review as a basis of raid performance increases. The real world tasks they did are supposed to get good boosts from raid setups, but if they measured loading times of apps there really wouldn't be much difference, if any, when going to R0. Many reviewers have done comparisons like these in the past with a similar conclusion. That Tech Report article is one example. 4x X25-E array loads games slower in some cases than a single X25-E. Access time rises slightly when you stick drives into R0 and sequential rates go up. App loading is mainly dependent on access time/iops so sometimes you will even see raid hurting more than helping.

[Lu(ky]

OCZs spec for the drives was 32MB cache on the 30 and 60GB drives. It turns out that pre-production ones have arrived with the 32MB cache, but retail come with 64MB cache. Post from Tony here.

I'm not so sure about RAID improving seek times, however when I've looked at the overall performance 2 drives gives a noticable boost, 3 gives a bit of a boost, but 4 or more it really does tail off.

Well I got one 30GB Vertex now from Mwave and on the retail box it says 64MB cache I thought it was a misprint..

[saint-francis]

If you use a really small stripe size does it increase IOPS? I'm curious. My 3 x 30 GB Core v2 array is about to be deprecated by a Vertex. I might just try a 4k stripe and use it for storing the 1 game I play and things like the page file.

[DYM]

Are you all using the 128 alignment? I've read 32, 64 and 128 beeing recommended at the OCZ forums!?

Tried 128 (Vertex 120Gb) on my Lenovo T61 laptop and all it can manage is about 120 mb/s under Windows 7....

[Orangeblast]

Here are my results with 2 30GB in RAID0 on my EVGA X58 ICH10R controller.

I'm pleased with the results.

128K stripe with an offset of 128. I experimented with 1024 and 512 offsets, but 128 yielded the best results. My burst speed is not too shabby

[karateo]

@Orangeblast
can you please re-run the crystalmark with 1000mb?

[modboxx]

Need some advice,

I have a Rampage extreme2 and a Lsi 3 Gb/s, SAS, 8-port Adapter, i currently use this adapter with my two hp 15k sas drives in raid0. I want to upgrade to two 120gb vertex drives and put them in raid0. My question is should i use the on board controller from my rampage extreme or the lsi sas controller.

Here are the specs of my of card
http://www.lsi.com/storage_home/prod...2er/index.html

P.s Will two Vertex drives in raid0 be faster then two of these in raid0? http://www.newegg.com/Product/Produc...82E16822332008

Thanks for the help

[Orangeblast]

@Orangeblast
can you please re-run the crystalmark with 1000mb?

Sure...I'll try it right now.

EDIT: Done.