News - Intel admits to 'bricking' firmware for 34nm SSDs

[HEXUS]

A fix is on the way, promises the chip giant.

Read more.

[nightkhaos]

W00t!

I'd be more excited if I actually had an SSD, but hey... whatever.

[goodl]

Glad I held off for user reports on this one before updating mine

[Perfectionist]

The update's totally manually right? E.g. If someone went and bought one and plugged it into a PC connected to the internet and installed any drivers, it's not going to brick itself or anything?

BTW I'd be really interested in hearing whether people think Corsair or Intel 256GB drives are better for gaming. Someone said indilinx are better, but Intel doesn't have that I believe, and other people told me Intel is better. So confusing. I did some research looking up benchmarks, but inconclusive due to different settings by different reviewers

[aidanjt]

I've yet to (thankfully) see an automated firmware update procedure. So yeah, it's manual.

[Ross1]

they could at least say what they did to replicate the fault... there are some of us (moi) who upgraded their firmware and havent noticed problems so far, but are kinda worried about may happen. If they could tell me something i should avoid, it would help ease those worries.

[aidanjt]

From what I can tell, if you haven't seen it after the first reboot after the firmware update has been applied (or second reboot after running the firmware update programme, should I say), then you should be alright. If you're really worried, you should probably downgrade the firmware (obviously make sure you back up anything of importance, first).

[zoomee]

"64-bit Windows 7 systems"

Thats incorrect - OS has nothing to do with this update as a number of linux users reported the same issue.

In actual fact the updater runs its own version of FreeDos from a bootable CD (which u burn using the ISO intel supply).

[Irien]

BTW I'd be really interested in hearing whether people think Corsair or Intel 256GB drives are better for gaming. Someone said indilinx are better, but Intel doesn't have that I believe, and other people told me Intel is better. So confusing. I did some research looking up benchmarks, but inconclusive due to different settings by different reviewers

I wasn't aware that Intel did a 256Gb drive (at least not at a price that is remotely affordable!

Right now, I'm only aware of five options in the "worth having, but not stupid price" 256Gb market:

Samsung PB22 oem (I think) 256Gb (available at Scan and Overclockers)
Corsair P256 (rebadged Samsung)
Corsar X256 (indilinx)
OCZ Vertex (indilinx)
Crucial M225 (indilinx)

Having had a Vertex brick on me (128Gb, 1.3 firmware), I was going to go for a Samsung drive (but then Scan went out of stock, sigh), on the basis that they seem to have fewer "clever features" (less to go wrong) but similar overall performance. Those drives are definately a good choice.

However, due to stock levels (and pricing) I have ultimately gone for the Crucial. It seems to be very similar to the Vertex (Win7 benchmarks it a bit higher than an OCZ Agility - haven't retested a vertex recently). I installed the latest (trim) firmware before installing the OS.

Have no idea if it will brick like the Vertex (hoping not!), but the price was quite reasonable (for a 256Gb SSD). Performance seems nice, and things are very snappy.

(Edit) Samsung-based SSDs don't currently have trim-supporting firmware, but perform nicely across the board regardless (see Hexus article). They have built-in garbage collection routines to keep performance up. The indilinx drives all have firmware updates for trim, AFAIK, although they may be in different states of beta.

[aceuk]

Irien. I don't know which article you are referring to, but SSDs which use Samsung controllers (like the Corsair P256 and OCZ Summit) are a very poor choice and should be avoided at all costs. Not only for the extremely bad random write performance but also the fact that you can't upgrade the firmware yourself.

http://www.anandtech.com/storage/sho...px?i=3631&p=19

From the article linked above...

The largest SSD maker in the world is Samsung. Samsung makes the drives offered by Apple in its entire MacBook/MacBook Pro lineup. Samsung makes the drives you get if you order a Lenovo X300. In fact, if you're buying any major OEM system with an SSD in it, Samsung makes that drive.

It's just too bad that those drives aren’t very good.

I talked to some of the vendors who ship Samsung RBB based SSDs and got some sales data. They simply can’t give these drives away. The Indilinx based drives outsell those based on the Samsung RBB controller by over 40:1. If end users are smart enough to choose Indilinx and Intel, why aren't companies like Apple and Lenovo?

Don't ever opt for the SSD upgrade from any of these OEMs if you've got the option of buying your own Indilinx or Intel drive and swapping it in there. If you don't know how, post in our forums; someone will help you out.

There’s currently no way for an end user to flash the firmware on any of these Samsung drives. To make matters worse, there’s no way for companies like OCZ or Corsair to upgrade the firmware on these drives either. If you want a new firmware on the drive, it has to go back to Samsung. I can’t even begin to point out how ridiculous this is.

The only real choices is either an Indilinx-based SSD or one of the Intel X25-Ms.

[Irien]

Aceuk - I'd tend to agree (I did mention half of your point in my post!), were it not for having experienced a bricked indilinx drive. On looking at forums (I know, bad source), I've seen that it is not uncommon across OCZ, Corsair and Crucial, not helped by the odd dodgy firmware. This makes upgradable firmware both a blessing and a curse, although I suspect that we'll see more rigourous testing from now on, since practically every player has issued a "dubious" firmware now!

As for the samsung performance, I think one needs to be a bit considerate about the needs of users. Having used Samsung-based drives in rigs, I've seen uniformally good performance for the things my users need/want, but it *will* depend on the user. For the vast majority, emphasis on reads (consequetive and random) will generally massively outweigh write (arguably espeically random write). Couple that with the reliability, and that'd be why Lenovo etc use them. Now, obviously we'd like to see better random write, but the drives suffer none of the stutter or other problems seen with jmicron controllers, and generally receive "thumbs up" in other reviews (even older anandtech ones!).

Right now, it is a case of buyer-beware, I think. indilinx drives *do* seem good, but with vendor-forums reporting issues (performance drops after firmware updates etc), some might prefer the tried-and-tested-if-not-super-performing nature of the Samsung drives. Each to his own, I think?

[Perfectionist]

http://anandtech.com/storage/showdoc.aspx?i=3631

[aceuk]

The main problem with SSDs at the moment is the large variation in performance (particularly new vs used) between all the various makes/models, not to mention the firmware bugs - like you say. It's definitely not as easy as buying a standard HDD.

When you are spending £100s of your hard-earned money on an SSD I'm sure most people want to get good value for money. Why buy a 128GB Samsung PB22-J when you could have a Crucial M225 for the same price? The latter will be even faster still - which is surely a good thing? If the Samsung drives were substantially cheaper than their competitors drives, that would be a different matter entirely, but they are not.

I will be happy as soon as you can buy any SSD from any manufacturer and not have to worry about the performance or other issues. The massive price-gouging that OEMs do with SSDs needs to stop too.

[MrRobin]

Any more news as to when this new update will be released from Intel?

[djfarmo]

Any more news as to when this new update will be released from Intel?

Bump

[djfarmo]

Any more news as to when this new update will be released from Intel?

By the end of the month, apparently:

• http://www.computerworld.com/s/artic...grade_for_SSDs
• http://communities.intel.com/thread/...5COMS?tstart=0