OCZ Colossus, dual JMicron?

[deputc26]

http://www.tomshardware.com/ne...ssus-SSD-1TB,8376.html
I was mildly shocked to hear that the Colossus is supposed to have dual JMicron controllers ala Apex/Titan.

Two possibilities here that I think are most likely:

1. It's JMicron next gen controller.
2. Tom's is wrong.

Seriously, OCZ would not bring an ultra expensive ($2200) product to market if it suffered from the well documented issues that plague JMicron's controller.

 

[TC91]

I'm pretty sure they have dual (or quad on the higher end ones) indillinx controllers but with a jmicron raid controller for the internal raid 0.

 

[deputc26]

Originally posted by: TC91
I'm pretty sure they have dual (or quad on the higher end ones) indillinx controllers but with a jmicron raid controller for the internal raid 0.

That makes a lot of sense, I believe you are correct.

"the 1 TB drive integrates two separate SSD components that are arranged in a RAID 0 configuration, driven by a JMicron controller."

This sentence is confusing, it reads as if the SSD itself is "driven" by the JMicron chip but really it is the Raid 0 that is driven by the JMicron Chip.



Thanks for the input TC91,

_Nate

 

[alcoholbob]

SATA2 bus will be saturated over and back. This seems to me a hardcore push for pure IOPS. If it comes out in 3-4 weeks then Intel will be #2 and can you say time for massive price cuts on the X25 series.

 

[taltamir]

maybe we should skip SATA3 and go straight to SATA4

 

[deputc26]

Originally posted by: Astrallite
SATA2 bus will be saturated over and back. This seems to me a hardcore push for pure IOPS. If it comes out in 3-4 weeks then Intel will be #2 and can you say time for massive price cuts on the X25 series.

I am not sure this will relegate Intel to number 2. Sequential reads are already pushing the limits of SATA II and this drive will almost certainly be interface limited in sequential writes as well with 3x+ the speed of Intel. But when it comes to small random reads and writes Intel is still 1.6-5x as fast as Indilinx. It will be close though and I would guess that it will get OCZ parity with Intel.

Also I am not sure how IOPS scale with RAID 0, I believe (please correct me if I'm wrong) latency increases slightly and IOPS increase ~60%.

 

[betaflame]

What I read was SATA->Jmicron->2x Indilinx Barefoot...

The Indilinx controllers have caches so it doesn't appear to be an issue. I really do hope they have Sata 6Gbit, because they will cap out with 2x 230MB/s drives internally.

 

[shabby]

Originally posted by: Astrallite
SATA2 bus will be saturated over and back. This seems to me a hardcore push for pure IOPS. If it comes out in 3-4 weeks then Intel will be #2 and can you say time for massive price cuts on the X25 series.

PcPerspective did some quick benches with the colossus, seems to be a hit and miss in some spots.

 

[capeconsultant]

Is RAID 0 really the way to go with SSD's? Or is it inherently less reliable than non RAID. Like with regular drives, with RAID, there is 3 pieces of hardware possible to fail as opposed to one with non RAID. Just checking. Seems like it would be a small bit less reliable.

 

[betaflame]

SSDs aren't mechanical, so they are inherently less likely to fail. Bad firmware or controllers or flash is still present, but those would probably result in a QC fail or DOA.

I've never heard of an SSD failing if it wasn't DOA.

 

[capeconsultant]

I guess it is just hard to wrap my mind around the fact that RAID 0 can be a good thing

 

[taltamir]

Originally posted by: capeconsultant
Is RAID 0 really the way to go with SSD's? Or is it inherently less reliable than non RAID. Like with regular drives, with RAID, there is 3 pieces of hardware possible to fail as opposed to one with non RAID. Just checking. Seems like it would be a small bit less reliable.

This is "raid0" like having 2-5 platters is RAID0 compared to 1 platter.
Make no mistake, a single platter drive IS more reliable than a two platter, more reliable than 3, and so on...
going from 4 to 5 platters seems to be the limit for HDD. With most sticking to 4 platters for reliability issues.

 

[fleshconsumed]

I'm thinking that internal RAID is reserved for higher capacity drives and lower capacity such as 128 and 256GB won't have it. Time will tell I suppose.

 

[Denithor]

Originally posted by: taltamir
maybe we should skip SATA3 and go straight to SATA4

I'm waiting for one of the SSD companies to pick up on the layout used by the Acard 9010 RAM disk. As in, dual SATA connectors on one drive for a "raid0" connection to the motherboard, essentially doubling the bandwidth to enable higher sequential reads/writes. Combine this concept with SATA 6.0 and you should have adequate bandwidth for these "internal raid0" drives.