Crucial vs. OWC SSD for a Mac

[khs]

I've been tempted to get my first SSD, but the lack of TRIM support in OS X and the seeming lack of information regarding how well garbage collection works has given me pause.

I've been looking at the OWC Mercury Extreme Pro and the Crucial C300, and both drives appear to support garbage collection, but I can't find any useful information past that.

Does anyone have any additional information or first-hand experience with either of these drives? The last thing I want to do is drop the $650-700 on a drive just to have write speeds drop like a rock when the garbage collection doesn't work as well as advertised.

Thanks!

 

[poofyhairguy]

The best drive in my experience for OSX are OCZ Vertex drives. I have three of these and when I avoid filling them they keep near their maximum performance, despite OSX lacking Trim support, over the months I have owned them. My previous non OCZ SSD with a different form of garbage collection slows down within a couple days of a clean install.

In fact OCZ actually sells a Mac version, which is the exact same thing as the non-Mac version (hardware/firmware wise). They just added a Mac version to say "hey it works with Macs!" and to bilk Mac users who will pay extra for something with "Mac" in the name.

I don't know anything about the drives you just talked about, but personally because of the lack of Trim support I won't touch anything but OCZ drives for my Macs and Hackintoshes....

 

[flamenko]

Haven't got alot to say about the Crucial as they are still trying to work through issues with firmware updates but there hasnt been much (if any) negative with respect to the OWC drive. I reviewed the OWC Extreme Pro RE and if I can't break it....well eheheheh.

But seriously if money is not an obstacle I might consider the RE as it has 28% over provisioning and is still pretty much the only drive that makes the claim of no performance degradation as the drive fills and with age...and tests haven't shown any reason not to believe their claim.

I got better readings out of the drive full than I did with just the OS and empty.

With respect to the OCZ, like the writer says above, when he avoids filling them... To me this is a pretty big issue that not many companies (if any except OWC) has really tackled. I wish i could get my hands on a few newer Sandforce drives with larger overprovisioning to see if they have the same characteristics and tests when full as this drive.

Here ya go...full... and review is on the site for your assistance:

 

[kevinqian]

The OWC is based on Sandforce SF1200. It does not have active GC, none of the SF derivatives do. They have their own algorithm to handle clean up. But OCZ strongly recommends against heavy benchmarking these new SF1200 drives as it really kills the performance without an active GC mechanism.

If you really care about GC, then the older generation OCZ Vertex/Agility may be a better option as the Indilinx drives have active GC. But I would say the Sandforce drives are good enough with their own proprietary method.

If you're thinking about OWC Mercury extreme, then you should also consider OCZ Vertex 2, OCZ Agility 2, Corsair Force, and Patriot Inferno. They are all functionally equivalent clones of the Sandforce SF1200 controller. I would put more weight behind the OCZ drives as they are very active in firmware updates/support. Just visit these vendor's support forums to get an idea of how active they engage with their customers, and how helpful support really is. OCZ is on top of this list, by a long shot.

 

[flamenko]

I guess, with respect to firmware updates, it comes down to whether they outline a drive that may have been released to early or maybe shouldn't have been released at all.

Here is an interesting article with respect to the C300 firmware problems:

http://www.hexus.net/content/item.php?item=24756

 

[khs]

Thanks for the replies. I guess I should reconsider the C300, considering the firmware problems it's been having.

The OWC is based on Sandforce SF1200. It does not have active GC, none of the SF derivatives do. They have their own algorithm to handle clean up. But OCZ strongly recommends against heavy benchmarking these new SF1200 drives as it really kills the performance without an active GC mechanism.

I may be reading into the grammar too closely here, but are you saying that different SF1200-based drives have different cleanup algorithms? So if I were to get the OCZ Vertex 2, it's behavior would be different from the SF1200-based OWC drive?

If you really care about GC, then the older generation OCZ Vertex/Agility may be a better option as the Indilinx drives have active GC. But I would say the Sandforce drives are good enough with their own proprietary method.

It's not that I care about GC in particular, I just don't want to have performance take a massive hit after a couple months and have to reformat at it or whatever to get it back to where it was. It seems to me most SSD reviews now assume that TRIM is running to keep things in shape. Am I wrong about this?

But seriously if money is not an obstacle I might consider the RE as it has 28% over provisioning and is still pretty much the only drive that makes the claim of no performance degradation as the drive fills and with age...and tests haven't shown any reason not to believe their claim.

Based on Anand's article here -- http://www.anandtech.com/show/3690/...dforce-more-capacity-at-no-performance-loss/1 -- it seems that the reduced overprovisioning doesn't necessarily mean taking a huge performance hit over time either. Seems strange, since the 200GB Vertex 2 is about $80 more than the 240GB once.

 

[kevinqian]

all SF based drives are almost exactly the same. same tech, same algorithms. You should be alright w/o trim, might take a slight performance hit but should level out over time.

The new 60/120/240 sized drives are new and on promo right now so they are probably cheaper than the 50/100/200 versions. Buy whatever is the cheapest, soon (in a few months) OCZ will release a fw that will allow switching between 50<>60gb spare areas. They are identical hardware wise, just diff firmware for now.

 

[flamenko]

RE:

This is from Anands article:

SandForce's controllers are dynamic: they'll use all free (untouched or TRIMed) space on the drive as spare area. To measure the difference in performance between a drive with 28&#37; spare area and one with 13% we must first fill the drives to their capacity. The point being to leave the controller with nothing but its spare area to use the moment we start writing to it. Unfortunately this is easier said than done with a SandForce drive.

With that in mind....find any OCZ or other Sandforce drive except the OWC and see if you can put the two Crystal scores when full side by side. I looked hard but could not. I am still looking for another drive to boast no performance loss when seasoned and when full.. Its a hard search and the Vertex 2 hasnt jumped up at all.

Maybe its not the 28% overprovisioning and its the OWC specific Duraclass technology in the firmware....but the fact is I cannot find another ssd that someone has filled and published a Crystal score with...only the OWC.

Also the article was written before the release of the OWC RE drive of which I speak. I would love to see it incorporated in those tests.

 

[kevinqian]

Are you saying OWC has a different fw implementation than the rest of the SF drives? Sandforce does not disclose it's technology to its vendors and any vendor that releases firmware are essentially packaging what SF provided to them. It's a similar situation with the Indilinx drives.

Doesn't Anand have a review of an OWC SF SSD? Im still not getting what you are trying to say. You saying OWC doesn't have any speed degradations after use?

 

[flamenko]

Yes.... OWC claimed even before my testing that this specific drive which was released 10 May (Extreme Pro RE) does not slow with age or when being filled. The Crystal score I posted above shows the result. I cannot find any posts equal to this anywhere.

The firmware differs from manufacturer to manufacturer I would believe. If you head to the advertising site of the OWC drive, you will find firmware traits which are not mentioned with other SandForce drives...

Here is the link:

http://eshop.macsales.com/shop/inte...xtreme_SSD_Sandforce/Solid_State_Raid_Edition

I am not trying to sell these SSDs, but rather, am trying to find the equal. In all reviews of this specific drive, their has yet to be a negative found....

So...IMHO...is this better than the equal OCZ? If the OCZ slows when filled to capacity as the owner above describes, then this must be a step up. Surely we should be seeking drives that we can fill to the expected capacity rather than settling with the industry published norm that all drives will slow when reaching capacity.

If you think about it...whats the better buy??? A 160Gb Intel that you can only fill to 100Gb or a equally priced OWC that can be filled to 100Gb and absolutely blows the Intel away.

 

[kevinqian]

I can't seem to find OWC's website with strictly datasheet specs. The one you linked to is more marketing oriented. All those points they list are a rehash of every Sandforce SSD feature set out there. They don't even list the max IOPS number which would give clue to which SF controller/firmware is being used (SF1200 or SF1500). For all intents and purposes, I will assume the OWC RE is just a SF1200 controller with the high IOPS firmware, which is exactly the same as the OCZ Vertex 2.

Do not confuse the Vertex I with the Vertex 2. The above user is referring to the older Vertex I, with Indilinx controller. The Vertex 2, using SF1200 has 28&#37; overprovisioning, and would therefore provide equal performance to all the other SF based drives. It also has the benefit of 50k IOPS random writes.

You seem to be benchmarking CDM with 50mb test runs. I would suggest that you run CDM with 1000mb test runs. You will see quite a bit different results (read: much worse).

 

[flamenko]

Ive got it at all size test runs. I had to do it with the 50mb test run for it to be full. It was consistent at every stop from the 60&#37; mark and up (60/70/80/90/95/99) and I only reduced to 50 when I had to.

I am not confusing the Vertex 1 and 2 whatsoever. What I am looking for is a similar Crystal for the Vertex 2. I only started with this avenue because even the writer above stated he has to watch the vertex 2's he has because they slow after reaching a point. This did not. In fact, unlike Vertex, OWC will publish and standby the fact....

Can I ask ANYONE with a vertex 2 to do a Crystal score when it is completely full as the one I posted above? I am dying to see the result....ok maybe not dying but I am curious as to why the industry shies away from these tests and the thought that we should EXPECT a drive to perform the same full as when first installed.

 

[kevinqian]

the above writer said he has a Vertex. No mention of 2.

 

[kevinqian]

Ive got it at all size test runs. I had to do it with the 50mb test run for it to be full. It was consistent at every stop from the 60% mark and up (60/70/80/90/95/99) and I only reduced to 50 when I had to.

I am not confusing the Vertex 1 and 2 whatsoever. What I am looking for is a similar Crystal for the Vertex 2. I only started with this avenue because even the writer above stated he has to watch the vertex 2's he has because they slow after reaching a point. This did not. In fact, unlike Vertex, OWC will publish and standby the fact....

Can I ask ANYONE with a vertex 2 to do a Crystal score when it is completely full as the one I posted above? I am dying to see the result....ok maybe not dying but I am curious as to why the industry shies away from these tests and the thought that we should EXPECT a drive to perform the same full as when first installed.

Until someone does actual test another SF SSD full in the exact manner you described, I think it's pretty safe to assume all SF SSDs will perform almost identical. Any differences would be statistically insiginificant.

By the way, have you figured out what the major difference is between the Mercury and Mercury RE? I see warranty diff of 3yr vs. 5yr and some capacity differences. Maybe that's the only thing separating the two, the overprovisioning? I'd love to see a more technical datasheet of the two.

 

[flamenko]

The major thing is the over-provisioning. I spoke with OWC about this and they are identical except for that; I will be receiving the non-RE version soon enough to try out hopefully and would like to see its results full as well.

I understand your point with respect to all SF drives performing almost identical which brings me back to the question as to why no other company is not touting that there is no performance decrease as the drive fills. Its an important factor. I also question why the OWC has not gained better popularity since its testing from all has had great results without any hiccups whatsoever. I wonder whether many think right off that because it is an OWC it is specific to Macs regardless of whether TRIM is a characteristic of the drive which it is.

Have u the Vertex 2? Download the new Crystal, fill the drive completely and give Crystal a run remembering to ppost the result. I have asked many but nobody has been kind enough to do the test yet.

 

[kevinqian]

OWC benchs well, but so does every other SF SSD. So yes, they are great, but in the grand scheme of things, they are just run of the mill. OCZ has the benefit of better support forum for enthusiasts who purchase these products and better marketing. Really that's all that's separating these SF SSD makers.... the intangibles.

I do have a Vertex 2 and i'm not about to torture test my brand new drive. After following OCZ's forums, the consensus is that when you benchmark them hard, they tend to wear out very quickly. You should perhaps contact a professional reviewer with no attachment to their review samples.

 

[flamenko]

"Really that's all that's separating these SF SSD makers.... the intangibles."

The SSDs are separated by their firmware for the most part. The firmware may have a taste of the oem manufacturer but we have seen times over where significantly different performance can be achieved through the same controllers as a result. The firmware also allows different characteristics specific to each SSD.

A Vertex 2 is not a OWC is not a Torque is not a Crucial. If they were all equal Anand himself wouldn't need to bench them side by side.

"OCZ has the benefit of better support forum for enthusiasts"

This definitely does not make a drive as OCZ has also fallen prey to things such as releasing drives too soon (Vertex II's bricking initially) as well as they were nice enough to sell many consumers their product containing the infamous JMicron 602 controller. There are also many that would say that, with respect to support between those on the forums, thats the best OCZ has done to stand by their product. The web is littered with bad experiences ranging from dreaded MIRs to followup support.

I personally credit OCZ for being a pioneer and forging ahead where others have not amidst the hard times (MTron etc).

I was a bit miffed earlier at the suggestion, as well, that frequent firmware updates show the company is on the ball in updating their drives. I view it as the company fixing a substandard SSD, for the most part, and trying to get it right. Wouldn't it be ideal to get an SSD that didn't require a firmware update after you just bought it?

Its good to know you have a Vertex II as you obviously do have an attachment to the drive. Its unfortunate though that you would not trust the product to do the job that its supposed to do. A simple test of filling and then doing a Crystal score doesn't in any way constitute a torture test by any means. I have done a few reviews on a few drives and have even on my own purchased drive.

There is no attachment to any but I guess a lingering question as to why everyone is so afraid to fill and test a drive. I can't be the only Intel owner (or other SSD) who is dissappointed in the fact that I cannot use the entire 160Gb of storage I paid for without suffering performance degradation.

We seem to have become a consumer that thinks that because Anand identified it, thats gotta be. Perhaps there is a reason why the other SandForce SSD manufacturers aren't touting that their drives maintain performance even when full. To me, this would be a huge selling point especially when you add that the degradation is also not visible as the SSD is seasoned over time.

As for me...this drive goes back shortly but I gotta say that having tortured it for the past month plus, I am pretty impressed that someone will stand up for their drives and further, that the drives stand up for themselves through ALL of the reviewers tests conducted on them...

As I am the biggest anti-Mac guy you could find, I never would have believed I would speak so highly of a Mac oriented product... How can I discredit it if nobody will stand up with similar results from the sister products?

 

[kevinqian]

http://www.anandtech.com/show/2829/12

Read that page and you'll understand why i say all SSD vendors practically use the same exact firmware provided by the controller manufacturer (Indilinx). The same case applies to all Sandforce drives. Vendors simply test and repackage the firmware for their specific SKUs. If you do some fw hacking, im sure you can flash an OCZ fw onto another vendor's SSD.

As we all know, SSD is such a new tech that it is inevitable for firmware updates. Intel isn't even immune to this situation. As an enthusiast, I'm actually glad they provide such frequent updates because it shows they are constantly tweaking and improving various aspects of the drives. Stagnant technology never gets this kind of attention.

 

[kevinqian]

http://benchmarkreviews.com/index.p...sk=view&id=531&Itemid=60&limit=1&limitstart=5

Here's an interesting read about Sandforce firmwares. Seems like OWC uses the SF1200 controller with the regular low-IOPS firmware.

 

[poofyhairguy]

Can I ask ANYONE with a vertex 2 to do a Crystal score when it is completely full as the one I posted above?

The Crystal benchmark isn't on OSX (which is what the question is about). I use Xbench to test in OSX.

Good news is that the newer drive I bought with newer firmware holds its performance while being more full than the older one. I have pile of these drives, and they (well, larger ones probably) is what I was recommending for OSX:


http://www.newegg.com/Product/Produc...-393-_-Product