Intel's 34nm NAND SSDs launch in two weeks

[NapalmV5]

You're working awfully hard to stir the pot. New products always see price hikes from retailers, but Intel drives have never been a "pos."

oh really ??

http://www.xtremesystems.org/forums/...ts&pollid=1707

[deathman20]

oh really ??

http://www.xtremesystems.org/forums/...ts&pollid=1707

Prior to april Intel was like all other drives with the performance issue. If they never upgraded there firmware they are undeserving for the drive.

If Jmicron hasn't been put on that list or Samsung then there is issues since Intel and Indilinx are the ones that have really been pushing to fix the issues with updates on there firm wares at least. J-Micron would be the worst, and not sure where Samsung and Indilinx are similar but Intel is the best in terms of controllers.

[JohnZS]

I'm happy with my POS, it scores 7.9 in the Windows 7 HDD test and plays all my games fine oh and loads apps instantly and windows boots in next to no time... not bad for a POS, I dread to think what a half-decent drive could do

AudienceofOne
I was surprised to see Intel handing out samples this early, however we do not know how "immature" these samples are, it would not surprise me if we do not see them in the UK Retail until September.

US retail might get them a bit sooner, but to be honest I cannot see anything other than a "psedo launch" in two weeks..as in every review site would be salivating over the samples but nobody would be selling them...

Remember when the Core 2 Duo was "launched", it was a good 2 months before they were actually available after the likes of Toms, Anandtech et al drooled over them.

Hopefully we shall get some Trimmings for our X-25m's in a firmware update soon too

(Would be nice is sequential writes got a 20MB/s boost too, but I can't see Intel being so kind)

John

[Ao1]

^^ yeah you are probably right, hopefully some more info will come out soon. If it does come out soon hopefully they have got it right and not just rushed it out.

[JohnZS]

Very true, the good news is Intel have a reputation of not rushing anything out.... (well except processors now and then)

[Rhys]

Prior to april Intel was like all other drives with the performance issue. If they never upgraded there firmware they are undeserving for the drive.

If Jmicron hasn't been put on that list or Samsung then there is issues since Intel and Indilinx are the ones that have really been pushing to fix the issues with updates on there firm wares at least. J-Micron would be the worst, and not sure where Samsung and Indilinx are similar but Intel is the best in terms of controllers.

"If they never upgraded there firmware they are undeserving for the drive."
Sorry what does that mean?

From what has been discussed over at OCZ the samsung's are now the drives that suffer most, as Indilinx now has internal defrag like the x25 & the use of the Trim tool to boot.
However it seems Samsung has changed the firmware recently and that may be better than the previous FW at defragging the drive, the only problem is samsung hasn't publically released (as far as I know) a FW update for the people stuck on older FW.

[deathman20]

"If they never upgraded there firmware they are undeserving for the drive."
Sorry what does that mean?

They don't deserve to have it.

[Ao1]

oh really ??

http://www.xtremesystems.org/forums/...ts&pollid=1707

Nothing ends up on your sammy's cos its always somewhere in cache

[NapalmV5]

damn right..

[Ao1]

The Register lets slip a few more details. The new drives will be complient with ONFI 2.0 standards. ONFI 2.0 sets new speed limits - defined as 166 MB/s (mode. 4) and 200 MB/s (mode 5).

Looks like MLC versions of the drives will be available before SLC versions.

http://www.theregister.co.uk/2009/06...ron_34nm_nand/

http://onfi.org/wp-content/uploads/2...2_1webcast.pdf

ONFI 3.0 (in development) will reach speeds of up to 400 MB/s.

[One_Hertz]

No performance figures at all yet... And more importantly, no mention of whether TRIM is there or not. I am sure I could obtain one well before the public release, but I don't really want to pay for a cat in a bag. I'd probably go with an E or two again.

[Rhys]

I think the MLC's will be the way to go again, in this round especially if the MLC sequential writes get sorted out.

3x MLC in raid 0 should offer great performance as well as much higher capacity for your cash.

[Levish]

i can't imagine anyone considering the Intel drives a POS, overpriced sure (especially when compared to the GB/$ of the latest Samsung based drives)

[aamsel]

i can't imagine anyone considering the Intel drives a POS, overpriced sure (especially when compared to the GB/$ of the latest Samsung based drives)

Please inform me of the advantages of the Samsung drive other than cost/GB?
All reviews I have read comparing the Intel to the 2nd gen Samsung P series drives give a huge advantage to the Intel, claiming much better random writes.

I have not found a single review that gives the nod to the Samsungs over the X25-M's.

[Rhys]

Please inform me of the advantages of the Samsung drive other than cost/GB?
All reviews I have read comparing the Intel to the 2nd gen Samsung P series drives give a huge advantage to the Intel, claiming much better random writes.

I have not found a single review that gives the nod to the Samsungs over the X25-M's.

So you base a drives performance solely on random writes considering neither drive actually suffers stuttering issues?

You 'could' actually buy your self a nice cheap 128gb Sammy and kit it out with MFT and it would eat an x25's random writes (and general IOPS) for breakfast. It's 4k reads are also around 20% faster which I put more weight on them considering they generally outnumber 4k writes significantly.
You would also get significantly more sequential write performance as a bonus.

I'm guessing Intel use a similar technology to MFT on their drives because how else do they get 20mb/s 4k reads but 50mb/s writes when the sequential's are so poor?
If anything it should be the otherway around just like every other SSD.

My guess is writes are stored into the drives cache and do not get written to the drive until 8mb (Perhaps 4mb) worth of writes has been used up and then that 8mb is written 'sequentially' to the Nand while the remaining 8mb of cache is collecting writes ready to transfer another 8mb.

This would also help explain the good write amplification the x25 has also, since MFT pretty much solves write amplification and prolongs a drives life considerably also.

But of course I could be wrong.

[One_Hertz]

It's 4k reads are also around 20% faster which I put more weight on them considering they generally outnumber 4k writes significantly.

Man you gotta get with the times... the X25 4k reads, the true 4k reads not the linear garbage you are quoting, are A LOT more than any competition.

[Ourasi]

Man you gotta get with the times... the X25 4k reads, the true 4k reads not the linear garbage you are quoting, are A LOT more than any competition.

Exactly, it's about time the Queue Depth 1 numbers just crumble up and die already.. Useless and nonexistent, and people still live and die by those numbers, yikes...

The X25-E/M's real random read 4kb are close to ~3x better then the Indilinx/Samsung SSD's, and the writes are ~5x better on the M's and ~15x better on the E's.

The I/O Pattern most users see every day, is the IOmeter "Workstation" test, and this is how it looks:

Here you can easily see the difference between HDD's, mediocre SSD's, good SSD's, and the X25-M....

[aamsel]

Is it thought that, if these anticipated new Intel 34nm devices are soon introduced,
that we will be at end of life for the current X25-M models?

[Rhys]

Please can someone with a single x25-m run this http://www.ocztechnologyforum.com/fo...6&d=1238177389 Iometer config and post their results?

@ Ourasi
Do you know where I can get hold of that config?

[Ourasi]

Please can someone with a single x25-m run this http://www.ocztechnologyforum.com/fo...6&d=1238177389 Iometer config and post their results?

@ Ourasi
Do you know where I can get hold of that config?

Workstation Pattern is something you set yourself in IOmeter based on a predefined setup and Queue Depth 32-64 or above: http://ixbtlabs.com/articles/hddide2k1feb/iometer.html
Remember that starting an ordinary application, you will see Queue Depth 30-50. Queue Depth 1 as used in Crystal on the 4kb random test, is useless and worthless, as QD#1 does not exist on an OS level, even opening Calculator sees Queue Depth 4....

If you want a fast and easy test, use AS Benchmark, and test only 4kb 64thrd (Queue Depth 64) random read and write..: http://www.ocztechnologyforum.com/fo...ad.php?t=58059

[Mic]

Anyone heard any news regarding the 34nm drives yet? its been about 2 weeks now.

[suffocatio]

Q4 is relase date,other is total BS

[Mic]

Kinda suspected that since we havent seen any previews or words from Intel yet. To bad though, really looking forward to see some tests on these new drives. Would be nice to get one when windows 7 is released.

[suffocatio]

next is jmicron 612 www.bit-tech.net/news/hardware/2009/05/28/jmicron-612-ssd-controller-to-feature-cache/1

[vab206]

why does that press release say "instantly doubling the current storage volume of these devices and driving capacities to beyond 256 GBs in today's standard, smaller 1.8-inch form factor."