Intel X25-M 160Gb G3 Prototype

[Gilgamesh]

In China. I don't know is fake or not:
http://en.expreview.com/2010/10/28/i...bao/11208.html










"Now X25-M G3 with capacities of 160/300/600GB are available,prices range from Yuan2400 to Yuan8000,which equal to US$359/US$1197"


Not so impressive 4K read speed.

[lowfat]

If 160GB is going to be $359 this definitely isn't going to be a game changer like they seem to think.

[sin0822]

what have they improved?

[tdream]

Sequential write?

[irev210]

If 160GB is going to be $359 this definitely isn't going to be a game changer like they seem to think.

No, I think you will see 160GB for ~250

Impressive they are about to launch 25nm NAND SSDs.


I thought the refresh was about bringing size up/price down.

[zalbard]

what have they improved?

MTBF (durability) and read + write IOPS.

[iddqd]

Well, since it's China, it's most likely fake.

[tajoh111]

Even 259 dollar drives 160gb will be really not all that great. In canada, you can find 120gb sandforce drives for 220 dollars which from previews, looks like they perform better than these new intel drives. Considering the saving of using 25nm memory, this drive should be cheaper(the price of opposing companies 80gb drives). The new sandforce stuff coming out is going to make intel SSD look archaic in comparison.

[Flying duck]

hmmm... unimpressive, would rather have a sandforce based drive...

[bamtan2]

they can make the ssds as fast as they want. do you think they'll make faster ssds? or do you think they'll make ssds which are still super fast, but way cheaper, and make way more money?

[Telperion]

Extremely disappointing if this is the actual performance and pricing.

[Flying duck]

they can make the ssds as fast as they want. do you think they'll make faster ssds? or do you think they'll make ssds which are still super fast, but way cheaper, and make way more money?

They should be able to do both, guess Intel isn't as innovative in this area as their competitors now.

This new Intel drive won't be hardly any faster than my G2, in the desktop realm who cares about IOPS any more, when our CPU can't take advantage of them?
Sequential's is where it's at, and is the only place where higher performance can be noticed, yes the G3 is has improved, bit it hasn't improved nearly enough.

looks to me like the current snadforce controller and the upcoming controller will dominate the market, might even pick me up a PCI'e raid solution like this .

[odditory]

in the desktop realm who cares about IOPS any more, when our CPU can't take advantage of them? Sequential's is where it's at, and is the only place where higher performance can be noticed

:facepalm:

[ripken204]

Sequential's is where it's at, and is the only place where higher performance can be noticed, yes the G3 is has improved, bit it hasn't improved nearly enough.

haha sure, b/c on desktops we frequently transfer large files around on our SSDs

[Flying duck]

:facepalm:

How's that?

Aside from server workloads, extra IOPS over what the current gen offers is a waste of time.

[Flying duck]

haha sure, b/c on desktops we frequently transfer large files around on our SSDs

Maybe some don't, but I for instance edit super high-res PSD's so ram quickly runs out and then I'm at the mercy of my page file, and guess where that sits?

Edit: would also be well utilised for scratch-disks

[josh1980]

Yeah, except the operating system will NOT be accessing the page file sequentially.. it'll be random.. guess what those IOPS are good for?... random.

Your logic is completely backwards my friend.

If your pagefile is being kept on an SSD and you need to have quick access to it, you SHOULD be caring about the 4k IOPS speed....

[Computurd]

gah for some reason i find this to be a little fishy. notice they blanked out the 512b results and parts of the atto screen unexceptionably? why would they do that? is this a knockoff already?

[ripken204]

Maybe some don't, but I for instance edit super high-res PSD's so ram quickly runs out and then I'm at the mercy of my page file, and guess where that sits?

Edit: would also be well utilised for scratch-disks

for the few people who actually do that then fine, but my point is that the vast majority of people shouldnt really care about sustained transfer rates.

[RVWinkle]

Wasn't the next gen Intel SSDs supposed to be 100 and 200 gig?

[iddqd]

gah for some reason i find this to be a little fishy. notice they blanked out the 512b results and parts of the atto screen unexceptionably? why would they do that? is this a knockoff already?

What do you mean "already"? They've had plenty of time to make a knockoff.

They usually have a knockoff iphone ~2 weeks after an official one is launched.

[Cooper]

Well, since it's China, it's most likely fake.

EVERYTHING is made in China. How can it be fake ?
Intel never promised a ground-breaking performance with G3. They just trying to polish existing tech and lower the costs with move to 25nm.

[Ao1]

According to the Anandtech review 23nm is still being refined to get the best performance. This and the desire to ramp up production so that supply is not limited when the product is launched are allegedly the reasons that Intel has delayed until February.

This is clearly an early production model and somehow it would seem that some slipped out on to the black market. The pricing therefore has nothing to do with Intel and the performance is unlikely going to be anything like the final version.

[Flying duck]

Yeah, except the operating system will NOT be accessing the page file sequentially.. it'll be random.. guess what those IOPS are good for?... random.

Sorry Josh, your wrong, firstly you need to understand that desktop workloads have low QD's so more IOPS is useless to them. What can help however is improved access time, sequential speed & finally the speed of your CPU.

If your pagefile is being kept on an SSD and you need to have quick access to it, you SHOULD be caring about the 4k IOPS speed....

Access time and IOPS are not the same, this is where I believe your logic is suffering.
Also if you had any real world experience of using different SSD's with much different IOPS spec's like the old Indilinx based drives Vs the current gen, you would know that the IOPS bandwagon is BS, and is being perpetuated incorrectly because of the J-micron fiasco.

To help you understand it easier, IOPS are like multiple threads/cores in a CPU, and access time is similar to the single threaded performance of a CPU.

Desktop data workloads have low queue-depths (no more than about 5 or 6), so adding 1000's or cores when you already have 1000's of cores (IOPS), is not going to help any if you can't physically use them, on the other hand if you increased the performance per core (access time), those 5 or 6 concurrent tasks will be accessed sooner due to lower latency, and performance increases will be realised, and higher sequential speed will act as a kind of 'Turbo Boost'.

Decreasing access time, usually increases IOPS as a side affect, but the two are not the same, and this is where I think the confusion lays.

My logic is completely backwards my friend.

Fixed

[Flying duck]

for the few people who actually do that then fine, but my point is that the vast majority of people shouldnt really care about sustained transfer rates.

And they shouldn't care about IOPS either...