Hacker News Reader

Comment by pbhjpbhj | Hacker News Reader

This doesn't make sense:

>Toshiba Memory and Western Digital on Friday disclosed that an unexpected power outage in the Yokkaichi province in Japan on June 15 affected the manufacturing facilities that are jointly operated. //

Surely that's not the reason, it would have to be "and local [backup] power failed, and the failovers for that failed too"??

Toshiba manufacture generators too, it's not like they'd need to go far to get backup power designed for them.

There must be more to this? (Which explains why people are assuming it's suspicious, I guess; and this site is making 35% of global NAND output).

FWIW, I hadn't realised that it takes ~2months to process a wafer in to a chip.

View on HN

Semiconductor fabs require MASSIVE amounts of electric power. In fact, to the first approximation, 5-10% of the cost of a silicon wafer is purely the cost of the electricity consumption. Source: worked in finance dept managing fab spend

There had to be a problem with backup systems.

In the US in certain industries, you have to do quarterly disaster recovery testing.

I am surprised they don’t do something like that here. The losses would certainly warrant it.

Yeah I was surprised. Don't they have a ups for this kind of thing?

Yes. My Google-fu is not up to finding them but I happen to be familiar with this company's products and here is one. https://www.energy-xprt.com/products/purewave-ups-systems-55...

One application of this kind of product is chip fabs because they are so sensitive to power disruptions.

Whether Toshiba/WD had this type of system and if so, why it didn't prevent loss of product was not mentioned in the linked article. I have heard that there is a glut in chips for SSDs so a reason to cut production can't be ruled out. However it seems like Toshiba/WD would pay the price for this outage while their competitors would reap the benefits (unless the competition agreed to somehow share the cost.)

I believe the memory industry is also known for price fixing, so it's not out of the question. The oversupply is ridiculous though; the last time I shopped for an nvme they were $1000 for 500gb and the other day I bought 1TB for $500.

1TB TLC was $300-$500 3 years ago, where are you shopping, the expensive store?

The other day I bought 2 TB for $189 (Intel 660P). Actually I bought 5 of them, because they're so cheap in Taiwan right now. I was really pleased to discover that they fit my spare MacBook Air using an adaptor, and 10.14 Mojave supports TRIM.

Amazon was selling 2TB NVMe from Intel for under $200 just last week!

Yeah Intel 660p (QLC, but with some SLC cache and a decent controller) is going for under $100 per TB. In my (limited) experience it makes a solid non-24/7 workstation drive (e.g., I use it for local object storage for large builds).

> The oversupply is ridiculous though;

No it's not? Isn't that basically capitalism working as intended (in this rare case)?

Yeah I was surprised. Don't they have a ups for this kind of thing?

One would think. Though, the tech industry is much like any other industry, and I imagine a conversation like this:

Engineer: We need to install backup generators in case the grid goes down.

Middle manager: Can you do it without stopping production?

Engineer: No.

Middle manager: Screw it. It's the next guy's problem.

At this sort of level, all power systems can be replaced or redesigned without the load being switched off...

Reply on news.ycombinator.com