Hacker News Reader

Comment by almostnormal | Hacker News Reader

Devices sold should work to a certain expected level. Regulators could define (or even just judges decide in a specific case that can be applied more generally) what that level is. Losing bits randomly isn't within the range of what is expected. The more memory there is, the more likely it is going to happen. And at some point devices without ECC could simply be returned as faulty for a full refund. ECC would be required, without any explicit rule.

View on HN

That is already how it works now… you can already return defective ram, ECC or not.

I am happy with the point in the price/reliability/power usage tradeoff that non-ECC ram represents. Why force me to have fewer choices and use something that is worse for my needs in order to solve a theoretical problem that I don’t care about?

To me, it sounds like "if it doesn't work 100.00% of the time, we cannot trust it and should forbid it", but the reality is that neither the world (and even internet) nor electricity nor healthcare, are "100.00% available/reliable".

In general don't worry that for very important calculations, they are run a couple of times, at least just to verify.

>In general don't worry that for very important calculations, they are run a couple of times, at least just to verify.

Citation needed, as this is unheard of in software we trust regardless to do banking or sign legally binding documents with.

Well the trick is a process or your entire OS crashed, why? Was it the powersupply? Motherboard? CPU? DIMM? Which DIMM? Are you going to keep a complete PC of spares to replace piece by piece to see if the crash stops?

For such a minimal cost I'd rather have DIMM 3 died with a non-correctable error.

Too late to edit my original response but I also want to point out: I’d be fine mandating ECC in critical applications like medical devices.

Not really. Some errors are expected, are you going to mandate 2 GPUs because the Memory could fail and it could show artifacts on the screen? Electronics failure is always a possibility with a small but nonzero probability. If your RAM is faulty, you can already return it. That does not mean that every device has to have ECC to prevent very improbable bitflips.

Processors too can have implementation bugs (and had!), and bit flips, so even a perfect memory won't solve this edge scenario.

An excellent example of Whataboutism.

Measures are in the works there, such as the switch to a simpler and cleaner ISA.

The value of simplicity is often overlooked; RISC-V will lead to higher quality software and hardware.

But I digress, as this has no weight in whether or not lack of ECC is acceptable.

Reply on news.ycombinator.com