News for 2011-02-10

Want to talk about one of the articles in SPCR? Here's the forum for you.
Post Reply
dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

News for 2011-02-10

Post by dhanson865 » Thu Feb 10, 2011 9:55 am

http://www.silentpcreview.com/news-2011-02-10.htm


* Flashplayer 10.2 supports ATI/AMD Video card acceleration (including AMD’s low-power C-Series and E-Series processors)
* 25nm flash forces Sandforce drives to reduce usable capacity by forcing an increase in flash dedicated to RAISE functionality.
* 25nm flash version of OCZ Vertex 2 has slower write speeds and is not labeled any differently than 34nm version
* Dustin at Anandtech.com covers the Puget Systems Serenity SPCR Edition
Last edited by dhanson865 on Thu Feb 10, 2011 2:08 pm, edited 2 times in total.

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: News for 2011-02-10

Post by CA_Steve » Thu Feb 10, 2011 12:58 pm

Another benefit of Flash 10.2 if you run dual monitors: flash no longer drops full screen mode when the browser window is not the active window....at least when using Hulu and the pop out window.

iconiK
Posts: 5
Joined: Sun Dec 05, 2010 12:30 pm

Re: News for 2011-02-10

Post by iconiK » Thu Feb 10, 2011 1:18 pm

So what's the point of purchasing an OCZ Vertex 2 25nm instead of Corsair Force, when OCZ didn't pass the lower NAND costs to the customer (just $10 less for 120GB) AND when the capacity is 4GB lower?

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Re: News for 2011-02-10

Post by dhanson865 » Thu Feb 10, 2011 2:12 pm

iconiK wrote:So what's the point of purchasing an OCZ Vertex 2 25nm instead of Corsair Force, when OCZ didn't pass the lower NAND costs to the customer (just $10 less for 120GB) AND when the capacity is 4GB lower?
Presumably the same capacity issue will appear on Corsair drives as well when they switch to 25nm (if they haven't already). I'm not saying you should buy OCZ (I'd actually be looking at Intel, Micron, and all the C400 lookalikes or at least waiting for them to come out before I decided).

The issue is how the Sandforce controller handles flash. If Corsair or any other company switches to 25nm flash with a Sandforce controller the same space will be taken for RAISE and presumably there will be a performance hit on write speed.

Oh and the marketing choice. If Corsair switches to 25nm and calls it a new product or has a revision number (some way for the customer to know which version they are buying) then you can give Kudos to Corsair and complain about OCZ.

I really think this sort of thing is bad whether it's Western Digital and 320GB traditional hard drives with either one 320GB platter or two 160GB platters both with the same model number on all the retail web sites or if it's OCZ Sandforce SSDs with 34nm or 25nm flash both with the same model number and product name. Change the name to Vertex 2.25 or Vertex 2-25 or Vertex 2+ or any variation you like but don't sell a functionally different product under the same model number and marketing name.

mkk
Posts: 687
Joined: Sun Sep 05, 2004 1:51 pm
Location: Gefle, Sweden
Contact:

Re: News for 2011-02-10

Post by mkk » Thu Feb 10, 2011 6:27 pm

Yeah that's the problem. It begs the question what kind of quality assurance OCZ has if they did not realise that the change would be too significant to be able to act like it's still the same product. In practice the drive might be just fine (minus the lesser storage space) but that's not good enough.

iconiK
Posts: 5
Joined: Sun Dec 05, 2010 12:30 pm

Re: News for 2011-02-10

Post by iconiK » Thu Feb 10, 2011 8:31 pm

I'm interested to see what Intel will do once they jump to 25nm NAND. OCZ said that Intel was/is intentionally limiting the write speed so the jump to 25nm would be unnoticeable; I think that's total crap, and doesn't speak high of a company when one of their employees directly blames a competitor just to appear as "the good guy".

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Re: News for 2011-02-10

Post by dhanson865 » Fri Feb 11, 2011 7:54 am

iconiK wrote:I'm interested to see what Intel will do once they jump to 25nm NAND. OCZ said that Intel was/is intentionally limiting the write speed so the jump to 25nm would be unnoticeable; I think that's total crap, and doesn't speak high of a company when one of their employees directly blames a competitor just to appear as "the good guy".
fwiw the Intel stance was that they were limiting write speeds on the 34nm drives to prevent drives from wearing out before the warranty period expired.

I remember them saying there was a daily counter (or by hours or some similar metric) and if you wrote too much to the drive within a certain window the drive would throttle writes defensively. I wish I had the foresight to document a URL or save a quote locally so I could reference the actual statements.

I remember that the number was high enough that casual users would never hit the limit. But when the final TRIM firmware came out they upped the limit on the 160GB drive (kept the limit the same on the 80GB and 40GB).

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Re: News for 2011-02-10

Post by MoJo » Fri Feb 11, 2011 8:29 am

Surely this is bait-and-switch fraud! You can't advertise a product as a certain spec and then silently make it worse and expect to get away with it.

Their numbers don't add up either. If the maximum program/erase cycles decreased from 8000 to 5000 that is nearly a 40% reduction, but the amount of spare capacity has not increased by anything like that much. Therefore the drive will fail sooner.

If they won't accept returns on this basis then someone needs to make a website you can register your SSD purchase date on. Then when the warranty is due to expire in say 2 months it sends you an email reminder so you can back up your data before it gets trashed and then spend a month hammering it with writes until it fails and OCZ have to replace it.

dragonfire
Posts: 62
Joined: Tue Nov 27, 2007 12:23 pm
Location: Massachusetts, USA

Re: News for 2011-02-10

Post by dragonfire » Fri Feb 11, 2011 12:07 pm

Curious whether any one else saw the Anandtech review of the Puget Serenity system. I was mildly upset to see it compared only against a selection of overclocked SLI rigs. As one would expect, it came in last on the performance rankings for almost every table. The kicker is the reviewer says up front that he does not have the tools to measure noise levels, particularly for such a quiet pc.

So I'm left wondering what the point of the review was. It should be no surprise to anyone that an OCed rig that draws 300+ Watts at idle will outperform a silence optimized build that idles at 60 Watts...

Worker control
Posts: 126
Joined: Wed Nov 16, 2005 12:16 pm

Re: News for 2011-02-10

Post by Worker control » Fri Feb 11, 2011 3:02 pm

dragonfire wrote:Curious whether any one else saw the Anandtech review of the Puget Serenity system.
Yes, I saw that review. The worst part was that he slammed Puget for overcharging, but couldn't be bothered to do the math carefully.

The good thing about it was that it got SPCR some press.

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Re: News for 2011-02-10

Post by dhanson865 » Fri Feb 11, 2011 4:31 pm

MoJo wrote:Their numbers don't add up either. If the maximum program/erase cycles decreased from 8000 to 5000 that is nearly a 40% reduction, but the amount of spare capacity has not increased by anything like that much. Therefore the drive will fail sooner.
No. At least not if RAISE space reduces write amplification as free space on an Intel SSD does. (I don't know, someone from Sandforce or OCZ would have to confirm this)

Write Amplification is affected by the amount of spare capacity in an SSD. Increasing the spare capacity reduces write amplification in a non linear fashion increasing drive lifetime exponentially.

http://intelstudios.edgesuite.net/idf/2 ... S003/f.htm was a good source but I can't get the video to play since the 2010 presentations went up. If it works for you check it out.

If not I'll quote from another thread where I discussed content from IDF.
http://intelstudios.edgesuite.net/idf/2 ... S003/f.htm around the 16 minute mark (slide 19) they took 160GB X-25M Gen 2 drives and formatted them differently and ran them to first failure in an accelerated write scenario a single user wouldn't encounter

Code: Select all

Formatted    TB Written
Capacity     before failure
160GB         29
144GB         68
128GB        104
_96GB        150
The thing is write amplification means more writes internally than what is reported by the OS so extra free space significantly increases the life expectancy of the flash. The only thing Micron or any other SSD manufacturer has to do to offset 25nm flash lifetime issues is put more in (it offsets the cost advantage but it's worth it because the benefit isn't linear).

from 19 minutes page 21

* Write workload impacts WAF (Write Amplification Factor) and thus endurance

* Increasing transfer sizes of random write workloads generally decreases WAF and increases endurance

* (I'll paraphrase the 3rd bullet) Endurance varies by use and no one can definitively say how long a drive will last. Actual quote "Only way to really forecast what an SSDs endurance will be is to measure it under the targeted workload"

They made it clearer in the 2009 IDF but the presentation is no longer available on the Intel website so I had to pick from data in the 2010 IDF.
Now does the increase in spare area on the 25nm OCZ Vertex 2 exactly offset the decrease in lifetime of 25nm flash as compared to 34nm? I doubt it.

I don't know if it increases or decreases the lifetime of the drive but I'm willing to say that the life expectancy of the drives due to flash wearing out should change significantly. This could actually be one thing that was improved. Longevity of the flash is either a wash, an improvement no one cares about, or such a minor loss it won't matter to the average user.

The capacity decreased no question about it. The write performance decreased some percentage that needs to be further tested and fully documented. These are the two things to complain about.

And to be clear if Sandforce drives benefit longevity wise from increased RAISE space I'm actually glad that the usable capacity reduced and was dedicated to RAISE. I just wish they had made the new version have a different model # / name.

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Re: News for 2011-02-10

Post by dhanson865 » Fri Feb 11, 2011 4:43 pm

dragonfire wrote:So I'm left wondering what the point of the review was. It should be no surprise to anyone that an OCed rig that draws 300+ Watts at idle will outperform a silence optimized build that idles at 60 Watts...
I think the point of the review is for Puget systems to get more publicity. Anyone that cares about noise will be impressed by a gaming system so quiet Anandtech.com staff couldn't measure the noise.

Sure it says more about Anandtech.com sound measuring equipment/procedures but at least there was that quote about the Puget Systems guys wishing they could get it below 10dba instead of leaving it at 11 or whatever ridiculously low noise level it runs at.

I think the final word sums it up. I can't afford a Puget Systems Silent PC but I'd consider myself very lucky and happy to have one given to me.

Arbutus
Posts: 234
Joined: Wed Jun 02, 2010 9:15 pm
Location: Calgary, Alberta, Canada

Re: News for 2011-02-10

Post by Arbutus » Sat Feb 12, 2011 12:37 pm

RE: "how the Sandforce controller handles flash"

There could also be an increased allotment of space for reallocating the higher expected occurrence of bad blocks developing over the expected lifespan.

Also, it could be that Sandforce is attempting to compensate for the decreased reliability and reduced erase-write cycles of 25nm NAND vs 34 nm NAND by increasing the ECC capability. The increased the ECC capability is gained by changing the data field and ECC field sizes and increases the 'ECC overhead' and consumes more space. These and other elements of the Sandforce controller system allows drives to be made with lower grade, and thus cheaper, NAND chips which can then be sold with a higher profit margin.

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Re: News for 2011-02-10

Post by MoJo » Mon Feb 14, 2011 10:36 am

dhanson865 wrote:

Code: Select all

Formatted    TB Written
Capacity     before failure
160GB         29
144GB         68
128GB        104
_96GB        150
The 160GB drive is a bit worrying. On a machine which used TrueCrypt for the system partition you can see exactly how much data is read or written to the drive. In my experience with Windows XP x64 between 1.5 and 2GB per day was normal, and that is only using it in the evenings. It seems that Windows writes a hell of a lot to disk.

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Re: News for 2011-02-10

Post by dhanson865 » Mon Feb 14, 2011 11:50 am

MoJo wrote:
dhanson865 wrote:

Code: Select all

Formatted    TB Written
Capacity     before failure
160GB         29
144GB         68
128GB        104
_96GB        150
The 160GB drive is a bit worrying. On a machine which used TrueCrypt for the system partition you can see exactly how much data is read or written to the drive. In my experience with Windows XP x64 between 1.5 and 2GB per day was normal, and that is only using it in the evenings. It seems that Windows writes a hell of a lot to disk.
Just to be clear all 4 numbers are from a 160GB SSD, just the formatted capacity changed.

FWIW the tests were accelerated testing in a scenario where writes were done at the highest possible speed non stop. This is considerably more demanding than your 2GB per day scenario.
http://www.anandtech.com/show/2614/4 wrote:Intel wanted to represent its SSD lifespan as a function of the amount of data written per day, so Intel met with a number of OEMs and collectively they came up with a target figure: 20GB per day. OEMs wanted assurances that a user could write 20GB of data per day to these drives and still have them last, guaranteed, for five years. Intel had no problems with that.

Intel went one step further and delivered 5x what the OEMs requested. Thus Intel will guarantee that you can write 100GB of data to one of its MLC SSDs every day, for the next five years, and your data will remain intact. The drives only ship with a 3 year warranty but I suspect that there'd be some recourse if you could prove that Intel's 100GB/day promise was false.

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Re: News for 2011-02-10

Post by MoJo » Mon Feb 14, 2011 11:55 am

Thanks, that is very useful.

As for the five year thing in the UK the law says that goods must last a "reasonable length of time". For computers that is generally recognised as being 5-6 years. Therefore if your SSD failed in the 5th year you could go back to the place of purchase and either have it replaced or receive a partial refund based on the time you have had it.

Post Reply