|04-19-2012, 07:43 AM||#1|
Join Date Nov 2004
Corsair Performance Pro 128 GB SSD Review
There has been a new article posted.
Title: Corsair Performance Pro 128 GB SSD Review
Here is a snippet:
"There are great performance enhancements when a system is upgraded to a solid state drive from a mechanical hard drive. A large number of SSDs are currently on the market, making the choice of which o..."
Comments on this article are welcome.
Hardware Secrets Team
|04-19-2012, 11:46 PM||#2|
Join Date Jan 2012
The article was good. But whenever a comparison is made between a SF controller and a non SF controller it's best to do two tests:
one with compressible data
two with in-compressible data.
At least the type of data used in this tests have to be specified, as SandForce drives perform much better with compressible data while others don't fare as good.
Also there is no indication about write amplification. SF drives have very low write amplification levels thus increasing the life space of the SSDs.
|04-30-2012, 04:54 PM||#3|
Join Date Apr 2012
My drive is probably from December 2011 if not earlier, well in the timeframe when Corsair was still sending drives to reviewers that have photos displaying them having 512 MB cache and 32 nm.
(Note: I haven't yet confirmed my device to actually be 24 nm and 256 MB, I'm looking at a way to do so without voiding warranty but if that's not possible then I may just return the device)
Prior to this review I saw only one mention of this Corsair Performance Pro 128 GB having 24 nm chips and 256 MB memory (This review does not currently mention this, but your photos prove it after googling the NAND chip numbers and datasheet for the cache RAM):
which was linked from
Where a guy from Belarus has similar concerns as I do, having bought this on the widely disseminated information about the components used. (Google shows ~100-200k hits talking about this drive and 32 nm with exact phrase searches, just 15k hits about 24 nm)
The Corsair employees "Yellowbeard" response is borderline insulting,
"If you have issues with a review site and what information and pictures they used, you should contact the review site and ask them about it."
That's the Corsair answer to a question which can be phrased as:
"Are you shipping to retail 128 GB Performance Pro's with 24 or 32 nm NAND and with 256 or 512 MB memory?"
How the beepin beep would a REVIEW SITE KNOW WHAT THE beep CORSAIR SHIPS TO TRADE (retail) after shipping different things to reviewers and then switching up components without telling anyone?
Alternative and much more unlikely theory is that Corsair is too busy counting the money to actually check what their suppliers are putting into devices with their logo and "Warranty void if broken" sticker there to ensure no one looks what they get.
While I'm not affected by a previous issue of these things not working with some RAID cards while being advertised as great for RAID, they handled that just as poorly in the forum.
( http://forum.corsair.com/v3/showthread.php?t=102375 )
( http://forum.corsair.com/v3/showthread.php?t=106229 )
( http://forum.corsair.com/v3/showthread.php?t=105110 )
Now there's a third issue popping up that is being confirmed by new thread on Corsair forum. I noticed this in january but thought nothing of it - after running AS SSD few times in early tests, I noticed the total unformatted disk size dropping with 0.01 GB. That could cause interesting issues I suspect. BIOS (eg. Gigabyte) and some OS can write stuff on the last sectors. I haven't looked at what would happen if there was data in the last sectors and the drive got smaller suddenly.. And hoes does that affect RAID setups?
(edit: I can't find the thread now but I have screenshots from AS SSD where within 8 hours between the runs the drive had dropped from 119,24 GB to 119,23 GB, I didn't do anything out of the ordinary between there and it hasn't dropped since)
Lastly, took me a while to pinpoint when I got this drive was that when the system is FSB overclocked (c2d, ich9r legacy ide mode), it didn't like as much overclock when the SSD is attached, it got very unstable until I dropped the system OC by 200 Mhz.
All that said, I think it's still one of the better drives out there (because there's been a lot of mediocre and even more buggy things) but IF WE ASSUME all Performance Pro's to be 24 nm, then there's NO REASON TO BUY Corsair ! Plextor has the same 24 nm hardware and performance but with 5 year warranty instead of 3! Or you could look at Intel 520/Vertex4 (Marvell chip also) or Samsung. I think those are the primary alternatives right now, major differences are probably in GC behaviour and suitability for 3rd party (LSI etc) RAID controllers, both of which few reviews have looked in detail.
Any company that ships one thing to reviewers and another to retail at the same freaking time can/should .. Well I let you finish that sentence.
There are some forum threads suggesting Toshiba's 24 nm Toggle MLC NAND has equal (5k) P/E cycles as 32 nm, but I consider that total unsubstantiated rubbish/BS since the datasheets to confirm that are nowhere I could find. If you have a link that's on Toshiba server that says so and talks what kind of over-provisioning and error correction they have to achieve that, then great but it's best considered BS until such appears. (The over-provisioning details would probably need to be divulged by Corsair and can they be trusted now?)
All information so far points that 24 nm is cheaper to make and less reliable. Only extensive independently done endurance testing with statistically significant amount of chips in well planned conditions using carefully thought out test methodology might convince me otherwise. Google might be able to afford such test in their data center.
PS. I couldn't copy text from the review page in IE or Chrome. Had to go to Firefox to copy those long numbers on the chips from the review to check what the mentioned chips were. The RAM is a Nanya 2 Gbit DDR3 SDRAM (datasheet NTC-DDR3-2Gb-B-Device-R11.PDF). 2 Gbit turns out to be 256 MB no matter what "configuration" the chip is in.
Frankly this pisses me off as I spent a ton of time (literally several weeks worth of full day reading over years) researching what SSD to buy and get excited about. So unless the wrong is righted**, I'll spend similar time doing whatever it takes to get Corsair and the review sites attention about this.
** (say, double warranty to 6 years for everyone now that they have that same 24 nm memory there as Plextor which has 5 years warranty, and if there are both 24 & 32 nm in retail channel then allow us to open up the screws without voiding the warranty to check and then either refund or compensation)
Last edited by avfx; 04-30-2012 at 05:34 PM. Reason: Clarifying the info about stability & add links about raid issue
|05-01-2012, 09:33 AM||#4|
Join Date Oct 2004
I forwarded your message to Corsair, and here is the answer they gave us:
|05-01-2012, 02:23 PM||#5|
Join Date Apr 2012
Says that even if the chip label says 128,256,512 in it, they are all 2 Gbit 256 MB DDR3 SDRAM just in different access configuration. Some reviewers have thought the number on the label eg. "128" means 128 MB but it does not.
like photos here:
And of course, how is the Plextor with identical chips good up to 5 years and Corsair is just up to 3 years? I think based on these facts, it's best if everyone return their Performance Pro and get a Plextor with 5 year warranty if they assumed they were getting 32 nm that would last longer. And there's still people even on HardOCP forum saying that they believe these have 32 nm chips. PR job well done Corsair, well done indeed.
And not just in HardForum:
"The Corsair has the better NAND chips isn't it? 32nm toggle vs 25nm sync? I know they share the same PCB and controller tho...
PP has 34nm Toshiba Toggle Nand, while the M3/M3P have 24nm Toshiba Toggle Nand..."
PEOPLE HAVE BOUGHT CORSAIR BASED ON 32 NM CHIPS! Corsair PR has let Reviewers let their readers believe these will have 32 NM!
Further, Corsair told investors their margins were down recently due to increased "low margin" SSD sales. Well I guess now we know why they silently switched to 24 nm.
"Plextor M3 <--24nm Toshiba Toggle
Crucial M4 <--25nm Micron Sync NAND
Corsair Performance Pro/Plextor M2P <-----32nm Toshiba Toggle"
Try this google search:
"performance pro" "34nm" site:hardforum.com
"2xnm nand hits around 3,000 write cycles, not 5000 and its really going to depend on your usage. If you're that worried about it, I'd go with Corsair's Performance Pro (34nm toggle nand)"
People are bit confused with 32 and 34 nm, but that 32 or 34 doesn't matter as much as a drop for 24 nm.
I knew that prices will drop by 50-100 eur this year when switch to 2x nm occurs. That's why I was excited to get (and pay ~220 eur) Performance Pro since it was one of the few which appeared to be 32 nm.
Last edited by avfx; 05-01-2012 at 02:49 PM.
|05-02-2012, 08:17 AM||#6|
Join Date Oct 2004
I will only comment on the part you are talking specifically about us: