Intel 510 series SSD

[johnw]

John IO have offended you somehow. Apologies. As I said, lets revisit this in a month or so. I haven't anywhere to hide.

More outrage than offense. No apology is necessary, but I hope you will raise your standards of evidence in the future. I think it is irresponsible for any serious publication, web or otherwise, to print such groundless speculation, even with a warning. Even Fudzilla does not print things with such thin evidence.

[alfaunits]

Considering how high the MB/s is on those graphs, it doesn't look like heavy at all.

Forget about the synthetic benchmarks for a while and focus on the real-world stuff.

This is the the Heavy Workload benchmark

Attachment 112612

Attachment 112613

Attachment 112614

Attachment 112615

Attachment 112616

Attachment 112617

The result is pretty clear, and it's the same with the Light Workload.

[flamenko]

More outrage than offense. No apology is necessary, but I hope you will raise your standards of evidence in the future. I think it is irresponsible for any serious publication, web or otherwise, to print such groundless speculation, even with a warning. Even Fudzilla does not print things with such thin evidence.

John... you seem like a really smart guy.... kind of. Rather than ranting as such which you seem to be accustomed to, do a bit of a background and ask the question of why I would ever put such an article out there unless.....

The problem we have anymore is, not only do we need everything handed to us on a plate, but also, we want to be spoon fed.

[Anvil]

Toms hardware has published an Review of the 510

Link

edit:
BTW, in the conclusion they are stating stat the next Intel SSD (the X25-M refresh) known as the 320 will be using an Intel controller

--

[flamenko]

I might suggest that the key to 25nm NAND being used in SSDs is really the write cycle which is significantly lower, keyed with the fact that SandForce write amplification is under 1/1. Accompanied by Durawrite technology, this has the ability to significantly stretch the overall life of the drive. Can present controllers do that?

I have dug into the 320 theory and think that the Sandforce SF-2100 is ideal for exactly that "should Intel not be able to extend the life cycle of that controller" with any expected SSD release. It also answer the question some may have been asking as to why the heck would Sandforce even come out with two 2100 series processors capable of SATA II only.

I also looked at the Intel article with the prototype Intel SATA6 controller and found it interesting but brought me right to the question of why the Marvell is in the 510 right off then? Wouldn't Intel be comfortable with waiting a bit longer until they can release that on a stable setup? Coincidentally, that strongly resembles the RAID chip inside present OCZ Revo drives.

Dont you love when people step out on such a limb as that? Fm Toms: 'That drive will employ Intel’s own controller.'

Started playing with my Intel 510 and will have the review out this weekend. The background on the Sandforce/Intel mix has swamped my time though.

I dont know if the site will allow it and if not, please delete but I have an incredible draw up for a very valuable Intel 975 Extreme coincidentally. I dont know of any tech site having a larger valued draw and know that that particular item may be key in builds of members here.

Maybe I should have started a Intel/Sandforce release Thread on its own... I didnt think it would take over this but there are alot of angles to consider, one of which I just detailed.

[Anvil]

The SF-2XXX controller is no doubt a great match for 25nm NAND.

I'm not sure that we are the right people to answer all questions but there really is more to it than 6Gb/s ports, there is no magical switch that makes millions of 3Gb/s ports suddenly work in 6Gb/s mode. Meaning: It will take years before all users have upgraded to 6Gb/s capable computers.
Ergo, 3Gb/s devices will be in demand for a long time and that includes SSDs.

But, do you really think that Micron would be releasing the C400/M4 if there weren't any other controllers that could cope with 25nm NAND?

There has been a lot of confusing things happening at Intels SSD department lately (mostly delays) but I don't think that SF is the answer to all problems.
I'm pretty confident that Intel can design a new controller for 25nm NAND and smaller , after all they have more info on 25nm IMFT NAND than any other except for Micron.

Well, lets wait and see, 1 month left or so...

[flamenko]

The key to the Crucial is that they, and others are becoming more transparent in expected SSD life. I believe the average calculation is 40GB a day for 5 years which is still a heck of a feat for ANY consumer. Marvell can then account for lifespan with the consumer and maybe even enterprise as we now realize that servers do much less writing than originally believed.

They key would then be to be able to supply a enterprise or 'enthusiast' level ssd with a significantly longer lifespan which Sandforce seems to have done. As much as we like to kick them in the stomach for their success, their progress in just over a year have been amazing and this can be seen with their new controller.

Many are not aware but it is absolutely impossible for any company not using compression to match their performance. It cannot be done simply because parallel read/write performance (as with the new controllers) can only be accomplished with a write amplification of under 1/1. This has only been accomplished by Sandforce who then steps it up even further with 256 bit AES encryption.

Believe me, I am not pwned by Sandforce at all and, in fact, heard it when I let out articles advising the consumer (and some enterprise clients) that their SF-1200 and SF-1500 were exactly the same controller physically, as are the new SF-2200 and SF-2500 controllers. This was followed by later articles of Sandforces releasing of firmware that significantly upgraded the performance of their drive without announcement. This was only seen in ssd refresh and new drives and not avail in a public firmware update (to my knowledge).

My interest is key in ability of the controller and I really think that Intel got caught which was evident in that roadmap. Why would you put out that drive when others were stepping up significantly? There were no SATA 3 designs on the table then by Intel. Intel then had the same stroke of genius as did many others and invested in both Marvell and Sandforce. The key here is, IMHO, the Sandforce drive is most compatible with the Intel 25nm chips and thats when the introduction will be....right along with the rest.

Just a rant of some senile old guy eheh and by the way...I just laid out my next Intel/Sandforce SSD article here without holding back.

[johnw]

John... you seem like a really smart guy.... kind of. Rather than ranting as such which you seem to be accustomed to, do a bit of a background and ask the question of why I would ever put such an article out there unless.....

The problem we have anymore is, not only do we need everything handed to us on a plate, but also, we want to be spoon fed.

This is why web journalism has such a bad name. Could you have handled it worse? No.

If you have a confidential source, you should mention that you have a confidential source in the article.

I asked you twice if you had more evidence than you presented. But you did not even respond to that.

[johnw]

Just a rant of some senile old guy eheh and by the way...I just laid out my next Intel/Sandforce SSD article here without holding back.

You can say that again. I'm sure the article will be similarly incoherent and poorly substantiated.