The good news is that Nvidia did test the hardware, and left the "defective" masks right there in the fuses for us to read. If you have a card that has been fuse-segmented well below its defective part count, you have a product segmentation product on your hands.
I guess they didn't decide this was worth it till they needed it to train an LLM, huh? Not like everyone (inside and outside of AWS) has needed such a test harness for years...
Diesel fuel has to be periodically filtered ("polished") by passing it through cans of dessicant. It removes accumulated water and prevents bacterial growth. There are also biocide additives but they just _extend_ the periods between maintenance.
That's also why it's not a great idea to use diesel for your backup home generator. It will go bad sooner than you expect, and then you're left with drums of gunk that you can't dispose and can't use.
And gasoline won't help you either. It goes bad even faster.
Propane tanks are the best option they can sit in storage indefinitely.
Or pull an AT&T in Nashville and depend on natural gas for your backup generators.
Which doesn’t quite work so well when an explosion cuts your electrics and the fire department cuts the gas.
If you do still decide to depend on another energy grid for an energy grid failure, at least have roll-up generator hookups so you’re not spending a few days figuring out how to wire that up…
They definitely don't have AI use cases that exceed a $20/month subscription. Hard to justify the amount of RAM needed for local AI for a non-developer's level of AI usage
Because fabs take years to spin up, decades to become competitive, and are a constant financial drain throughout. If (with great prescience) they started planning their own fab at the beginning of the RAM crisis, we should see Apple in-house RAM sometime around... maybe 2040?
I feel that with all the trillions Apple have lying around, they should've been thinking about their own fabs already when designing the first Apple Silicon chips.
> they should've been thinking about their own fabs already when designing the first Apple Silicon chips
Why, though? They had a sufficient mountain of cash to outbid everyone else for TSMC capacity, and no realistic hope of catching up to TSMC's process on their own fab
Well, outbidding everyone else seems to have worked up to a point, hasn't it. Of course, we could say that we're living in unprecedented times with regards to the demands currently being placed on fabs, but I'd say that I would've been worrying about TSMC being a fragile bottleneck for more than one reason, and for more than a few years now.
Also, I'm not exactly sure we should be claiming what is and what isn't realistic for a trillion-dollar company. People could (and probably did) argue that they had no hope of catching up to Intel/AMD, and yet here we are, with Apple arguably having surpassed both, at least from an energy efficiency point of view.
Apple has alternatives to TSMC. Intel and Samsung both have comparable process nodes, albeit at a pretty high cost and relatively lower yield. Apple would prefer to avoid paying competitors like that, but in many cases they have no other option. The iPhone Duo display, for instance, is sold to Apple by Samsung at a considerable per-unit markup. It was that, or no folding phone.
Realistically speaking, "catch up to TSMC" is something that every single EUVL fab has failed to do despite tens of billions of dollars in investment. Spending money is the easy part, developing high-yield processes over a timetable of decades is the hard part. I think Nvidia and Apple both agree that paying TSMC to build American fabs is a smarter use of their money if they just want excess fab capacity.
It has weird interactions with some modern browser features. For example domain isolation doesn't apply between subdomains of the same TLD, so john.smith.name and drew.smith.name don't get isolation in naive implementations.
The public suffix list is a stupid hack that should be replaced by formal protocols allowing boundaries to be defined by either DNS entries, or a file on the web server.
I personally think the public suffix list should be replaced by zero sharing even amongst subdomains, and if you want to talk between x.y.example.org and example.org you can use CORS just like any other cross-domain request.
Do it for every new TLD. Do it for every new domain in an existing TLD. Add an opt-in for security, then start pushing for the security measure, then get browsers to start carrot-and-stick enforcing it...
The real problem is that we'd have to add a wildcard to the public suffix list. That's it.
The public suffix list is the big list of all the URL suffixes that should not imply shared ownership of all their subdomains. For example the line
com
says "XYZ.com" and "ABC.com" should not be assumed to be controlled by the same entity.
Problem is, all 1st level domains registered to the .name TLD should be added to the list, and that's a lot of domains. The correct answer, IMO, would be to add a wildcard
*.name
but that would involve updating all the code that relies on the public suffix list to support said wildcard.
The "real" problem is the public suffix list. All domains should be considered public, that is, a sub domain gains no trust from it's parent domain.
There is a good argument that the public suffix list should be distributed via dns, and it is easy to propose a scheme for it. Two I know of are. A specific public suffix record marking a domain public or private. The presence of a soa record which already transfers the authority of a domain, should be used to determine if that sub domain is public or private.
But private unless in some third party list is the worst way to handle it.
*.name isn't a good solution either, because the vast majority of registrations are at the second level.
If you want to do it properly, you really do need to include all of the second level domains that have more than one third level registrant. And update that periodically as domains expire... there might be three registrants under .example.name today, but if they all (or all but one) stop paying, example.name is no longer a public suffix and it should be removed.
Having "no cookies shared between subdomains" as a restriction on the entire TLD isn't completely beyond the pale. There are other TLDs that have restrictions, like HTTPS being required for .dev
It's not 22000 registrants for .name, it's 22000 registrants of third level domains. There's lots more second level registrations in .name; it is massively more popular to register a second level domain, but it wasn't always an option.
It definitely is. .k8s..com regularly confuses what password I want to use if I have different authentication per ingress. More commonly if I run something else at ..com firefox and chrome are both happy to offer up my password for ..com.
yes, but other locality domains are fairly static and well-known
.co.uk wouldn't have the same problem because everyone knows that combo is a TLD in and of itself. with a seemingly infinite possible combinations of blah.name that could be a TLD itself, or, might just be a subdomain of .name, it gets really weird
Verisign claimed approximately 22,000 third level domains [1] in July 2026. I don't see an estimate of the number of shared second level domains, but it's finite. The current public suffix list includes 10414 names [2]; so at maximum, the list would approximately triple in size --- not ideal, but feasible.
Edit: another post [3] claims approximately 2000 second levels names with multiple registrations; that's only a 20% increase in size for the existing list, which is again a lot, but feasible. There's some more specific data in an issue for the suffix list [4], that shows there were nine second level domains with multiple registered names underneath them, one of which might all be the same registrant (domain.name)...
Nine would be totally doable, if the request came from Verisign...
22,000 now, but there was no limit and more importantly no rhyme or reason or other way to tell what was a third level registration and what was a subdomain on a normal second level registration
Verisign could have published a list of every third level registration, it wasn't an impossible problem to solve, but it is different than static regional TLDs like .co.uk or ..us
It's kind of hard for US techies to not have exposure to a company that accounts for >5% of the S&P 500.
reply