Cherryville - SSD 520

[johnw]

No idea, only that SandForce and custom firmware will be used. But it was hinted Intel had more in store then the other companies using SF-2281. Considering it is Intel I would assume there is a power-less feature but that's nothing I've heard of.

How can your inside source not know that? If he knows what controller Intel is using, and how long they have been working on the Sandforce firmware, it would seem he should know about power-loss protection. Maybe your source is not as reliable as you would have us believe.

[Smartidiot89]

How can your inside source not know that? If he knows what controller Intel is using, and how long they have been working on the Sandforce firmware, it would seem he should know about power-loss protection. Maybe your source is not as reliable as you would have us believe.

You really are after to discredit me, aren't you? It's not someone I speak with on a regular basis, I was only told about the controller and work on the firmware and I didn't bother to ask more since it's someone I respect very much. If you are after to discredit me - take it by PM. I will not give out any granular details in an open forum thread that could reveal the persons identity.

EDIT: On the other hand I am not in the mood so just forget everything I said and move along, it's probably all just wrong precisely like you want it to be This will be the last thread I visit when more information on Cherryville comes up. Why I and my patience even bothered with you is also a mystery.

[johnw]

You really are after to discredit me, aren't you? It's not someone I speak with on a regular basis, I was only told about the controller and work on the firmware and I didn't bother to ask more since it's someone I respect very much. If you are after to discredit me - take it by PM.

No need to get excited. Strong statements ("confirmed") require solid evidence. I am not trying to "discredit" you, but rather trying to determine if there is any solid evidence to back up your strong statements. So far, I am not convinced.

[Ao1]

Below are the specs for an OCZ SF drive. I have cherry picked the fastest performance available over the spectrum of different capacity drives.

OCZ SF
0 fill:

Max Read: 550MB/s
Max Write: 520MB/s
Random Read 4KB: 50K IOPS
Random Write: 4KB: 50K IOPS (Logical Block Address (LBA) range: 85% of total drive capacity)
Random Write: 4KB: 85K IOPS (Logical Block Address (LBA) range: 8GB)

Non compressible (AS-SSD; 64 thread for 4K specs):

Max Read: 510MB/s
Max Write: 280MB/s
Random Read 4KB: 56K IOPS
Random Write: 4KB: 56K IOPS

Here are the specs for the 520 (from Anvils post #35). Again I have cherry picked the fastest specs. It is unclear how Intel established their specs, but I think they typically use a 8GB LBA range.

Intel 520

• Max Read: 550 MB/s
• Max Write: 520 MB/s
• Read IOPS: up to 85K
• Random Write 4KB: 70K

So capacities match, as do sequential speeds, but the 520 read IOPs are significantly higher. Write IOPS on the other hand are lower.


The other thing that Intel state on the pdf that Anvil linked is that:

“IOPs and endurance specifications are for client-based workloads”.

It would be a bit disingenuous of Intel to base their specs on 0 fill and then say the IOPS and endurance specs are based on client-based works loads, so based on specs only I’m not convinced yet, although there does appear to be an overwhelming amount of circumstantial evidence that it will be SF based. Guess we will soon see.