"Unlike transistors, flash memory cells can no longer be made any smaller or faster without incurring reliability tradeoffs that negate the benefits of a process shrink."
It would be a real shocker to learn that flash memory is made of ... transistors...
And while it is true that flash memory becomes progressively unreliable as process shrinks, the same hindrance is intrinsic to all other circuitry as well. Maybe not as pronounced in detriment but apparent nonetheless.
Have also worked on embedded microcontroller products and agree with you. This article comes off as a puff piece/PR regurgitation.
However, I'm not sure I've seen this exact type of addressing mode in current QSPI flashes. Normally you are either in linear or wraparound mode. This, on the other hand, performs a single wraparound and then proceeds with linear access for the next lines.
This single, marginal improvement is really not all that amazing.
Nonsense. Of course Moore's law also applies to flash memory: The "law" is simply the observation that the complexity of circuits doubles about every two years; it's actually even easier to determine for flash because one simply needs to look at the available SSD sizes over time...
First off, I think the fact that an observation about integrated circuits made by Moore has been taken way too far over the intervening years to the point where it's widely misunderstood, taken far out of its original context, and doesn't say anything meaningful when invoked by people to support a discussion. I really would be a bit happier if we put it behind us in the same tech graveyard where the Bill Gates 640KB of RAM myth and the dollar sign-substituted Micro$oft also ought to be buried.
However, since we're already invoking...Moore's comment wasn't about complexity doubling, but about the number of transistors in an IC. Of course, with increased transistor counts there's an implied increase in complexity, but there are lots of ways to add transistors to an integrated circuit that won't make the design inherently more complex.
Eek, I'm sorry! That sounds kind of like sarcasm and I don't want you to feel upset. I was really just ranting about Moore's Law in general and not specifically about your misunderstanding. Please don't interpret my comment as something worth getting triggered about. I was just trying to clear up confusion.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
7 Comments
Back to Article
ddriver - Thursday, November 17, 2016 - link
"Unlike transistors, flash memory cells can no longer be made any smaller or faster without incurring reliability tradeoffs that negate the benefits of a process shrink."It would be a real shocker to learn that flash memory is made of ... transistors...
And while it is true that flash memory becomes progressively unreliable as process shrinks, the same hindrance is intrinsic to all other circuitry as well. Maybe not as pronounced in detriment but apparent nonetheless.
patrickjchase - Thursday, November 17, 2016 - link
What you describe is critical-word-first cache line fill. It's been around for ages. Nothing to see here.Also, I've worked on several embedded products that executed in place from flash. Nothing new there, either.
evancox10 - Sunday, November 20, 2016 - link
Have also worked on embedded microcontroller products and agree with you. This article comes off as a puff piece/PR regurgitation.However, I'm not sure I've seen this exact type of addressing mode in current QSPI flashes. Normally you are either in linear or wraparound mode. This, on the other hand, performs a single wraparound and then proceeds with linear access for the next lines.
This single, marginal improvement is really not all that amazing.
Daniel Egger - Thursday, November 17, 2016 - link
> There is no Moore's Law for flash memory.Nonsense. Of course Moore's law also applies to flash memory: The "law" is simply the observation that the complexity of circuits doubles about every two years; it's actually even easier to determine for flash because one simply needs to look at the available SSD sizes over time...
BrokenCrayons - Thursday, November 17, 2016 - link
First off, I think the fact that an observation about integrated circuits made by Moore has been taken way too far over the intervening years to the point where it's widely misunderstood, taken far out of its original context, and doesn't say anything meaningful when invoked by people to support a discussion. I really would be a bit happier if we put it behind us in the same tech graveyard where the Bill Gates 640KB of RAM myth and the dollar sign-substituted Micro$oft also ought to be buried.However, since we're already invoking...Moore's comment wasn't about complexity doubling, but about the number of transistors in an IC. Of course, with increased transistor counts there's an implied increase in complexity, but there are lots of ways to add transistors to an integrated circuit that won't make the design inherently more complex.
Daniel Egger - Thursday, November 17, 2016 - link
Thanks for rephrasing my point in your own words.BrokenCrayons - Friday, November 18, 2016 - link
Eek, I'm sorry! That sounds kind of like sarcasm and I don't want you to feel upset. I was really just ranting about Moore's Law in general and not specifically about your misunderstanding. Please don't interpret my comment as something worth getting triggered about. I was just trying to clear up confusion.