OWC Releases 960GB Mercury Electra MAX 3G SSD
by Kristian Vättö on June 25, 2012 12:30 PM ESTOWC has released a 960GB version of their Mercury Electra 3G SSD series. The drive uses two SandForce SF-2181 controllers, configured in RAID 0 using Silicon Image's RAID controller. There is actually 1024GiB of NAND flash on the board, 512GiB per controller, but as is usual with SF-2281 64GiB in total is dedicated to RAISE and another 71GB is for spare area. Here are the full specs.
| OWC Mercury Electra MAX 3G (960GB) Specifications | |
| Raw NAND Capacity | 1024GiB |
| Formatted Capacity | 894GiB |
| NAND | 2Xnm Asynchronous MLC NAND |
| Controller | 2x SandForce SF-2181 |
| RAID Controller | Silicon Image SteelVine Sil5923CNU |
| Sequential Read | 254MB/s |
| Sequential Write | 250MB/s |
| Price | $1,270 |
OWC has not published any random read/write figures so analyzing performance at this point is rather difficult. Basically, we are looking at two SandForce SSDs in RAID 0 but throughput is handicapped by the SATA 3Gb/s interface. OWC told us that there are some architectural limitations which is why the drive uses SATA 3Gb/s instead of 6Gb/s.
If you have been following the SSD market, you have probably noticed that 2.5" consumer SSDs top out at 512GB. The only other 1TB SSD is OCZ's Octane, but unfortunately we don't know what OCZ has done to achieve such capacity (and I couldn't even find a single review of the 1TB model). Once Intel's 128Gb 20nm MLC NAND die hits the market sometime in 2013, 1TB SSDs in 2.5" form factor should become more common as special tricks like RAIDing two controllers will no longer be needed.
I'm very happy to see OWC making unique products but I'm not too keen on the idea of RAID 0. The risk of failure is twice as high because if one controller goes bad, all data will be lost. Furthermore, unless OWC has done something else, TRIM support will not be available. In this case performance gains are limited because especially sequential read and write speeds will be handicapped by the SATA 3Gb/s bus.
The pricing, on the other hand, is actually not too bad. The 1TB OCZ Octane costs $2,500, so you could buy two 960GB Mercury Electras for the price of one 1TB Octane. Sure, it's still noticeably more than what smaller SSDs cost in terms of price per GB, but there is extra expense coming from the second controller and there isn't much competition either. The 960GB Mercury Electra definitely has its niche since there are people who want an SSD but need more than 512GB. Desktop users can always put smaller SSDs in RAID 0 but laptop users are often limited to one 2.5" drive, which is where the market for 960GB Mercury Electra is.
We have already asked OWC for a review sample, so stay tuned for our review.
Source: OWC
Facebook
Twitter
RSS
20 Comments
View All Comments
Fujikoma - Monday, June 25, 2012 - link
"The pricing, on the other hand, is actually not too bad."Are you serious. I just paid $345 for a 512 Gig crucial SSD this weekend. There's a few hundred dollars wrapped up in premium that doesn't belong there.
I still feel like I'm getting soaked on it, but it's an acceptable drubbing.
imaheadcase - Monday, June 25, 2012 - link
It is cheap, considering that simular size ranges less than a year ago could easily be close to $5k range.Also, just because SSD prices are cheap, does not mean every size is going to be cheap. Consider that my 246 Crucial M4 i bought is less than 6 months old..i paid almost $500. So $345 you paid is SUPER cheap for 512.
mlcloud - Monday, June 25, 2012 - link
6 months ago, that wouldn't have been a terribly unusual price. However, many SSDs now have broken the $0.75/GB barrier, and even the M4 is usually found at a dollar pr GB. Samsung's 830 has fallen slightly below a dollar per gig at the right times and stores.Not saying Fujikoma is justified in his comments, just explaining why he feels that way. I think this SSD is a niche product for those who might need high capacities with high random reads and writes but not so much straight up bandwidth (hence sata 3 gb/s), and it is priced appropriately as a niche product.
Malphas - Monday, June 25, 2012 - link
"It is cheap, considering that simular size ranges less than a year ago could easily be close to $5k range."Yeah, but consumer electronics don't work that way otherwise you could sell a netbook for a few thousand dollars and say it was cheap compared to an IBM mainframe from the 70's with a similar level of processing power. The fact is flash prices have dropped massively in the last 12 months. Things are cheap/expensive based on today's prices, not based on what you paid 12 or even 6 months ago.
Klinky1984 - Monday, June 25, 2012 - link
How do you figure out something is cheaper without historical price points to compare to?Cheap flash prices have no translated to lower costs across the board for all sizes, especially the larger sizes. Only recently has 512GB started showing signs of big price drops. Try to find a 960GB that it significantly cheaper than this one.http://www.anandtech.com/show/6038/owc-releases-96... Also the price is probably MSRP, not street pricing.
Malphas - Tuesday, June 26, 2012 - link
"How do you figure out something is cheaper without historical price points to compare to?"By comparing it to current products obviously. Like I said judging the value of consumer electronics by comparing it to historic price points doesn't work for obvious reasons.
ShieTar - Wednesday, June 27, 2012 - link
You are just discussing semantics though, most solutions at similar sizes are still close to the 5k$ range. You can find anything between 2$/GB and 7$/GB for TB-sized SSDs, but nothing anywhere near the 1.3$/GB for the above product.Pessimism - Monday, June 25, 2012 - link
Unacceptable and a total deal breaker in a mid-2012 performance product. Why even bother?!danjw - Monday, June 25, 2012 - link
It isn't necessarily a RAID 0. With the SATA interface gating the speed, RAID 0 doesn't really have a benefit. While JBOD would reduce the data lost if one controller were to fail.ShieTar - Wednesday, June 27, 2012 - link
RAID 0 can still handle twice as many random reads / writes, even with the older SATA interface. Though if that is the reason for the RAID setupt, it is somewhat surprising that they would not publish IOPS results for it.