Nand, Controllers and associated components are Not created equal.

[L&LD]

More and more USB sticks and microSD cards are being made with dubious components — data recovery firm uncovers no-name, low-quality NAND inside many devices

Nameless NAND chips everywhere.

www.tomshardware.com

Spend within a right-sized budget, always buy quality, and ignore the temptation to claim 'best price!' on things of importance, and you'll more than likely (only) buy once.

And yes, our digital tools are important.

 

[TheLostSwede]

This is not news though. All the big manufacturers test and sort chips based on quality/performance and the leftovers get sold to dubious companies in xina. As an example, Kingston has been doing this for at least 15+ years. They have a really cool factory though and all their USB drives are laser engraved with the serial number, so it should be possible to check that it's a legit drive.

As far as NAND in devices goes, it's obviously impossible to check if it's from a reputable brand as a consumer. All the projects I've been involved in, has only ever used flash from the likes of Winbond, Samsung Electronics, Kingston or the like.

 

[L&LD]

Yes, not news to people who know.

But a lot of people here push specs, price, and 'don't pay more' for what is supposed to be 'exactly the same' things.

Yet, they're not 'exactly the same'. Are they?

With the included link, it's not just 'me' saying this. Maybe those types of posts will stop now. Or at least be taken with a sea worth of salt.

 

[sfx2000]

This is not news though. All the big manufacturers test and sort chips based on quality/performance and the leftovers get sold to dubious companies in xina.

And there's a huge amount of binning - one might find a 512GB die in a 32GB device - not an ounce of flash is wasted at the factory...

SK-Hynix (now Skyhigh), Winbond, Micon, Samsung, Kioxa (ex-Toshiba) - as a device manf and a confirmed supply chain, these are all good brands when one is making devices...

there's still a huge problem though with counterfeits and out-right fraud - it's an industry problem there, that's for sure...

 

[L&LD]

And that 512GB die in a 32GB device will operate at 2001 levels of performance because there is no parallelism or any Controller that can use a single chip at full specs.

 

[sfx2000]

And that 512GB die in a 32GB device will operate at 2001 levels of performance because there is no parallelism or any Controller that can use a single chip at full specs.

Actually, for speed on a device it doesn't matter - there's the PHY controller, e.g. USB, SATA, NVMe that the computer sees, and that abstracts out the controller on the NAND die itself - that controller defines the size and speed capability of that specific chip on the circuit board...

I've see clever fraud at that low of a level - it's not supposed to happen, but it does...

 

[sfx2000]

And yes, our digital tools are important.

Here - we're absolutely aligned...

 

[L&LD]

How can it not matter?

You need two or more physical nand chips to enable maximum speeds (Controller and firmware dependant). Apple has proved that to a disastrous conclusion on their latest few models. Including RAM bandwidth too on the later ones.

 

[sfx2000]

You need two or more physical nand chips to enable maximum speeds (Controller and firmware dependant). Apple has proved that to a disastrous conclusion on their latest few models. Including RAM bandwidth too on the later ones.

point taken on the the M2 entry level Macs - single NAND device is a bit limited - 2.1GBps for a Macbook Pro M2 13" 256 is half the speed of the 512GB version...

2.1GB per second isn't that bad...

RAM performance for the M1/M2/M3's have been on par with Intel/AMD over on Windows land...

Anyways - let's keep things on topic and not turn this into Win vs Mac - different tools for different purposes - not every nail needs a hammer to complete its task.

Agree to disagree?

 

[L&LD]

Nothing to agree or disagree about. The truth stands on its own!