Have they done rack scale nvlink other than Grace? Usually I see racks of systems with GPUs, but the NVLink terminates within each server, rather than going between servers.

NVLink nowadays even in a single system (as of Blackwell) looks a lot more like infiniband, but still seeing Blackwells usually as GPUs in x86 boxes with NVLink staying inside. Multi-server fabric seems to be RDMA over ethernet as the favored selection for now. Maybe I'm missing some segment, but I can at least say externally switched NVLink is absolutely not 'just as well as GPUs'.

Intel still has a large share of the datacenter market, whether they really deserve it or not. Hence they actually have competition. Versus their flailing around accelerators which both sucked and never got traction. Just like I went looking for a new laptop and in some segments, the vendors only did Intel even as AMD also has everything better in that market too.

Well, they actually have had that available for a bit of time with Grace Hopper. I don't have hard data, but anecdotally it feels like this and rack scale NVLink haven't had the uptake nVidia presumed. So nVidia is *trying* to compete but hitting headwinds even in their darling AI segment. Given nVidia failed to wholly acquire ARM and ARM being an uphill battle in the conservative datacenter market, I could see an x86 equivalent of the 'Grace' strategy being attempted in collaboration with Intel. nVidia gets locked in integration advantage and Intel gets the scraps of the x86 half.

Quite, but AMD is *miles* closer than Intel was to being a realistic threat on this front.

I was kind of surprised too, since growing up my household would make a 600 mile round trip like 3 or 4 times a year. The last 3 years I've been in that 16% too, though for quite a few years before that I admittedly was in the 84%.

Think I've read a fair amount suggesting that the 'family road trip' has declined over the years.

It's the basis of how people get elected and promoted.

If a bunch of people are marketing a 747 as a vehicle like some alien ship that can pause in mid air, go side to side, and backwards and underwater and into space, then it's worth pointing out it's closer to being a mechanical bird than an alien ship.

Based on how I feel when I have to incur such a trip, I wouldn't be surprised if making those trips routinely would have long term bad effects on your health.

Given where the timezones are, certainly not 'most' people. Yes, you can cross a time zone in less than 25 miles if you happen to live within 25 miles, this doesn't support your stance of "most americans spend at least a day timezone shifted every year", since that's a pretty specific circumstance that doesn't apply to most people.

Even for them, I wonder what percentage of those trips introduce inconsistency in their schedule. If they work in one timezone, then they would consistently be living according to that schedule, even if they technically sleep in another.

Personally, if I am stuck with a trip that goes more than a time zone over, I just hate the shift.

Shifting the time is a PITA that is pretty jarring in a way most people don't enjoy and it seems like it may be outright unhealthy.

This is incorrect.

61 percent of the population does not take a "long distance" trip in a year.

Incidentally, this defines "long distance" as "50 miles". Of the "long trips", 58% of those are less than 125 miles away. So only 16% of people travel over 125 miles away in a given year. Less than 125 miles is relatively unlikely to cross a time zone. Growing up my family would regularly make 300 mile trips but still not cross a timezone.

I'll confess to not having pushed my luck performance wise, but at least feature wise I've been satisfied with KDE/Wayland with Fedora 42 and proprietary nVidia drivers. There were some hiccups before but I can't recall exactly when things seemed to get fine.

Pretty valid point. If *anything*, if AOL had executed a tad more successfully, then we might not even have had widespread adoption of the internet. We'd be all complaining about how AOL has a monopoly, but how else could you imagine a global online network functioning except inside a monopoly? Weirdos would be bringing up that crazy Internet thing that came out of ARPAnet and everyone would laugh about how that would have not possibly worked...

I think if AOL had established 'AOL for University' and 'AOL for Business' technology deployments to businesses and campuses, maybe by around 1992/1993 or so, they would have had a good chance of heading off the explosion of the 'friendly' internet as realized popularly by Netscape. Early 90s internet left a lot of the less technical crowd scratching their head and not seeing where things could go, but could get what AOL was putting down.

Specifically, they cherry picked 2022/2023 and pretended those numbers were good examples of "normal" hiring. Looking at the chart, it's clear they had a huge hiring boom, enough to overcome the prior 5 years of demographic shift. This is consistent with the general hiring boom in tech that came about then, just before LLM hype launched into the stratosphere.

They talked as though 2024 was a precipitous drop, but as you say, it was just a return to 2021 levels.

Without AI, we probably would see similar employment trends in tech and note it as a "correction". With LLM in the mix, it becomes hard to say how much is genuine shift to LLM to take care of things or LLM as a rationalization to get rid of the tech workforce the companies probably didn't need to hire up so much in the first place. Can certainly say which option generates more clicks though...

Note this was mostly a simple demographic observation being written about, *not* about relative popularity of university among the populace.

It's not that there are the same number of high school students but fewer want university, it's just that not nearly as many people were born.

Since the housing crash, domestic stability has eluded so much of the population that you would count on to have children.

So particularly the cost management is certainly something to watch, but your deeper problem is just that society is failing to instill confidence in the people that they can support themselves and children.

I looked at some houses, and the Opendoor ones were just sad travesties.

What was likely nice wood grain cabinetry just blasted with paint. Just sprayed on and painted all the doors shut. Same for handrails, which felt horrible to touch. Nice grain patterns replaced with light beige wall paint. Looking deeper, they never fixed anything that I would have considered important, just made things worse with new paint without regard for the thing being painted. I think they were more valuable before they had it screwed over.

Changing a feather to a leaf seems a weird thing to consider harmful. A leaf is supremely uncontroversial and it's not like the feather was somehow core to why anyone should, even in theory, care about the ASF.

I don't know but *suspect* the people that were concerned would have been sufficiently satisfied by removing "Apache" and ignoring the feather, hence my theory that it's probably more reaction than was strictly called for.

I'm not exactly sure about the 'real' problem in this front. In my opinion the closest thing to a 'real' problem is that the foundation hasn't really had a specific meaning in a couple of decades.

I could see HTTP/3 as a bit more of a tricky thing for Apache. Other servers largely declined to have 'in-server' extensions and they get more freedom with how they treat network sockets.

Apache has a lot more things that are implemented as fairly intrusive extensions, and I could imagine a change from TCP to UDP being a more difficult thing to navigate.

If you have need of some of those, HTTP/3 is probably a broader problem for you anyway. If you don't need those extensions, then switching to something like nginx isn't a huge burden, and the default performance in nginx tends to be better than apache except for some of those select extensions.

But the ASF barely cares about Apache. It was the kindling to spark a 'foundation' when 'LAMP' was all the rage, but now it has next to nothing to do with anything they bother to think about and only remains as a residual brand from their heyday of the 90s to early 2000s.