No but they could lay an electrical line, burn the gas to generate power and stop polluting.

In the UK flaring in the North Sea has been banned for quite a while now. I would at this juncture note that they always use some of the gas to power the oil rigs and it is only the excess that is flared. For years now it has been piped back ashore and brunt to generate electricity. Not burning it for power generation is a pathetic excuse because they are not paying for the cost of the pollution from flaring it.

LTS is about keeping an old system going. If you have new hardware upgrade the version of Ubuntu that you are using. If the hardware is broken then eBay is your friend. If the old hardware no longer can cope with an increased workload then upgrade. However frankly suck that dam thing into a VM for crying out loud and your "hardware" issues go away.

No, it was set to "encourage" people to use exFAT instead when Microsoft was trying to extract license fees from everyone for its use.

Really there was me thinking the maximum size of a FAT32 filesystem was 2TB.

The problem at the moment is power. I mean a 2000GPU system consumes megawatts of power and cost millions of dollars a year to run and tens of millions of dollars to procure. I would say till it can be fitted into a single rack and run on a 63A 3-phase supply without expensive heat exchangers and water cooling looks (water cooled rear doors are OK) it is a novelty. My guess is I will be close to retirement before that happens at the earliest.

Fundamentally it would not take much effort to make the likes of slurm work well with GPU loads. The issue is the idiot users who have no history in shared user systems and expect to be able to treat them like their personal workstations.

Yes that is a solved problem for traditional compute workloads, that is my day job running an HPC system. For AI it is a freaking nightmare. The first problem is the researchers have no history in shared computing resources, so come bitching they cant install latest wacky framework using sudo I kid you not. They seem to think they can treat a shared multimillion-pound GPU cluster like their personal workstation. Secondly, existing job schedulers like slurm etc. are not great with GPUs. You can make it work but it is sub-optimal.

So basically the likes of Run:AI allow you to have the researchers bring whatever wacky container they are using and have that scheduled on a cluster, oh and provide a bunch of off the shelf containers for the most common frameworks people are using that are kept up to date.

Honestly, that was significant value, and the whole AI thing is utterly insane in the amount of money it requires. The procurement cost for a 2000 H100 GPU system is in the region of $80 million with a $10 million per year electricity bill. Oh and there is around at least a nine month lead time on delivery possibly a year. There is a reason NVIDIA has the valuation it has.

Really, Russia Today has been forced off UK television for repeated breaches of the rules. Though it might come as a shocker to an American but biased reporting on TV and Radio is illegal in the UK.

Let me introduce you to the IBM TS4500. You are looking at around $2 per camera for an entire years worth of footage. A single tape library could store a years footage from over 850,000 cameras. Tell me you don't know about long term enterprise storage without telling me don't know. Off the top of my head to have that replicated at two sites for good measure, all in less than $10 million including a couple PB of disk cache.

If you have an 8 way SXM box with H100s then they are *ALL* aircooled. Its fing crazy but Nvidia supply the SXM module and it comes with an integrated heatsink. I have had quotes from all the major vendors at the start of this year and it was surprise to us we assumed we would be able to get direct liquid cooled servers. You can for four way SXM but not eight way.

Military missiles already do this, including using terrain profile and inertial guidance. It is handy to have a GNSS, especially for the start coordinates but it is not necessary to achieve good precision.

The fact that it has to be "upgraded" to Rust 1.77 shows that they are busy hanging themselves already. That is the issue with all these fad languages is the constant maintenance cycle on existing code so you can continue to use it. Python being the worst of the worst in this regard.

Someone needs to write a tender for a sizable GPU system that mandates CUDA support to get AMD's attention.

That sounds complicated. The cheapest form of hydrogen is "white" or natural hydrogen.

https://en.wikipedia.org/wiki/...

https://www.theguardian.com/en...

Turns out the stuff is in the ground just waiting to be released by drilling for it. It first turned up by accident in 1921 and till now the result has been to simply plug the wells and move on.