Intel 330 Series SSD (6Gb/s)

[Anvil]

A new series (330) is showing up in a few stores.

It is using 25nm NAND.
It's listed in 60,120,180GB capacities.

The 180GB is listed with the following specs.

Sequential Read/Write 500/450MB/s
Random Read/Write (4K) 52000/42000

Smaller capacities are listed with lower performance on 4K and the 60GB is also a bit slower on seq.writes (400MB/s)

Availability : end of April.

So, what controller could this be based on?

edit:

Link to one of the stores listing the new series. (Dustin, "Scandiavian")

[johnw]

A 60GB SSD with spec of 400MB/s on sequential writes does not leave much doubt on whether the controller is doing compression.

[Christopher]

That is actually pretty unfortunate. I saw this a few days ago, and I would assume they're using asynchronous NAND. In any event, I would very much like to see Intel's 6gbps controller in an Intel SSD and not a Hitachi.

[johnw]

I saw this a few days ago, and I would assume they're using asynchronous NAND.

What evidence do you base this assumption on?

[Christopher]

I don't know for certain, but that's probably just my disappointment talking. The thought of Intel degenerating into another SF partner pushing out synchronous and asynchronous drives is mildly appalling, though not surprising.

Since I have a hard time imagining Intel using asynchronous NAND in their own drives, I'm willing to entertain the possibility that they've just tuned the 330 series for better steady state performance at the expense of FOB speed.

520 60GB - 550MBs read 475MBs write 15,000/23,000

330 60GB - 500MBs read 400MBs write 12,000/22,500


520 120GB - 550MBs read 500MBs write 25,000/40,000

330 120GB - 500MBs read 450MBs write 22,500/33,000

[johnw]

520 60GB - 550MBs read 475MBs write 15,000/23,000

330 60GB - 500MBs read 400MBs write 12,000/22,500


520 120GB - 550MBs read 500MBs write 25,000/40,000

330 120GB - 500MBs read 450MBs write 22,500/33,000

I would have been satisfied if you had just said you thought the 330 specs look similar to other Sandforce SSDs with async flash (and slower than the 520 with sync flash)! But thanks for the detailed response!

[Christopher]

I would have been satisfied if you had just said you thought the 330 specs look similar to other Sandforce SSDs with async flash (and slower than the 520 with sync flash)! But thanks for the detailed response!

That's just it... they don't exactly look like other async 2281s. The closest 2281/async SSD I could think of, the Kingston V+200 which uses 25nm Intel async, looks like this:

Kingston V+200 60GB:
535 read 460 write 12,000/47,000

Well, here's hoping that Intel has just become serious about steady state performance and isn't just putting out a 520 with throttled FW or a drive with asynchronous flash.

[canthearu]

Not sure what is so sad about an Intel SF drive with Async NAND. Would be faster then the current 320.

As long as Intel really has nailed down and removed all the major problems with the SF firmware, they should be exploiting it as hard as they can and for all it is worth.

Of course, a new, awesome controller from Intel themselves would be better, or maybe a drive based on the new marvell controller, but there is only so much performance you can extract from cheap Async NAND, and the sandforce controller does a reasonable job of doing that.

Don't forget that awesome fast drives like the Samsung 830 and Mushkin Chronos Deluxe not only have a good controller, but are using the fastest toggle NANDs available to drive them.

[johnw]

Not sure what is so sad about an Intel SF drive with Async NAND.

Because it would be slower than an Intel drive with sync flash and a decent controller, which is what I might hope for given Intel's past leadership in SSDs. But it appears Intel has given up on innovating in the SSD field.

[Christopher]

Personally, I don't really think the world really needs another 2281 with asynchronous NAND, but that's just me.

And while it is true that toggle NAND kicks ass, I just don't think Intel really needs to be the one slapping the SF controller on a PCB laced with asynchronous flash - there are enough people doing that now. It's not because async is slow, it's because I object to Intel turning into just another SF partner.

[Computurd]

But it appears Intel has given up on innovating in the SSD field.

their stated intentions have always been to further the adoption of flash, not dominate the market.

[canthearu]

Because it would be slower than an Intel drive with sync flash and a decent controller, which is what I might hope for given Intel's past leadership in SSDs. But it appears Intel has given up on innovating in the SSD field.

The 3xx is the mainstream series of SSD. The whole point is to be cheaper, not more awesome.

[canthearu]

Personally, I don't really think the world really needs another 2281 with asynchronous NAND, but that's just me.

And while it is true that toggle NAND kicks ass, I just don't think Intel really needs to be the one slapping the SF controller on a PCB laced with asynchronous flash - there are enough people doing that now. It's not because async is slow, it's because I object to Intel turning into just another SF partner.

I disagree. Cheap and reliable drives, especially from Intel, are just what the market needs right now after everyone has been spending the last couple of years panicking people with controller and firmware problems from poorly tested drives. Hopefully in several months, my endurance test of the 520 will end, and we will see just how far an Intel driven Sandforce drive will go.

[johnw]

The 3xx is the mainstream series of SSD. The whole point is to be cheaper, not more awesome.

No, the goal is to be cheap AND awesome.

[Christopher]

I'm not saying it's 100% for sure SF+async, I just think it's a bad trend to be sticking a SF in the 300 series. If Intel wanted to stick with a 3rd party controller, I wish it had been Marvell anyway. A low cost Marvell powered 330 would be righteous IMHO. But Intel isn't paying me for my opinion, so it's inconsequential what I think.

I just hope this doesn't mean we're never going to see the next 10 channel in a consumer drive.

[johnw]

I'm not saying it's 100% for sure SF+async, I just think it's a bad trend to be sticking a SF in the 300 series. If Intel wanted to stick with a 3rd party controller, I wish it had been Marvell anyway. A low cost Marvell powered 330 would be righteous IMHO.

My guess is it won't be async flash, but I can see why you think it might be.

Anyway, I would prefer a new SATA 6Gbps Intel controller, second choice Marvell, and Sandforce way down the list. But I guess Marvell is out, probably some quiet agreement with Micron, since Micron is already all over the IMFT flash / Marvell controller SSD market. And for some reason Intel does not seem interested in making SSD controllers. I just don't understand that, it would seem like a good product for Intel, but the bigwigs at Intel don't seem to think so.

[canthearu]

Anyway, I would prefer a new SATA 6Gbps Intel controller, second choice Marvell, and Sandforce way down the list. But I guess Marvell is out, probably some quiet agreement with Micron, since Micron is already all over the IMFT flash / Marvell controller SSD market. And for some reason Intel does not seem interested in making SSD controllers. I just don't understand that, it would seem like a good product for Intel, but the bigwigs at Intel don't seem to think so.

Intel probably realised that if they developed a 6gb controller, it would end up looking a lot like either the current sandforce or marvell controllers. Sure, there is some room for improvement, but it probably wouldn't even enough, or the results economical enough, to make it worth the cost of development.

Pure speculation, but one of the reasons why intel haven't done much with the marvell controller may be because it too difficult to develop the timed workload feature. The 510 was suspiciously missing this feature, while all it's other drives have had it.

[Christopher]

Whether the 330 uses asynchronous or synchronous NAND isn't relevant; it's the principal.

I'm not saying that there couldn't be some redeeming quality of the 330, just that I'm disappointed.

EDIT
Now that I think about it, I believe I know exactly what the difference between the 330 and 520 is. And it's not the NAND...

[Highendtoys]

It could be less quanity but higher density flash. My money is on async!

[Anvil]

Unless they have more capacities (e.g. 240, 360, 480 or more) I'm not seeing the 330 as a replacement for the 320 series.

The 320 series have been great drives for me, I had expected a high capacity drive, meaning one equal to or greater than the 600GB drive they have been selling for almost a year.

Maybe there's more to come.

[Christopher]

I'm sure there will be higher capacity versions in the future. I've been using my 320 again for it's PLP recently.

Remember the data corruption I had with my system? It caused an issue that with my Octane that initially appeared as though the drive had died... I found out that the system instability was being caused by my GPU. Something about leaving the RDC to the endurance rig on for prolonged periods of time would crash the system (due to the GPU locking up for some reason), so I switched to the 320 to try and mitigate any more corruption issues if possible (on the system drive).

[Anvil]

Drivers (and software) are a constant threat to any system

One of my 24/7 systems have developed a nasty habit of rebooting and not being able to detect the system drive, I've tried finding the source of the issue but it's almost impossible.
(making it a non 24/7 system )

[Gilgamesh]

French review here (330 180GB):
http://www.cowcotland.com/articles/1...es-180-go.html

same controller and nand 25nm Intel of Intel 520 (perhaps nand 3k as Kingston HyperX 3k?)
Different firmware, and 3 years warranty vs 5.

Performance 330 vs 520:



in Europe, 35-40 euro less for 330 vs 520 (180GB).

[Highendtoys]

It appears to use six channels instead of eight. It does still use sync flash. I will have one here this week.

[Computurd]

yeah the reviews will be out soon