I watched Night Crossing in my german class in high school. I remember it being intense.

For the 80s it was intense yes. Watching it now that same tension feels milder but I guess that's because every single TV show now has to have constant explosions, car crashes etc in it.

There is actually gunfire in it and a teenager dies in the beginning but it still feels less intense due to the 80s pace IMO.

Slightly off topic, but they want to move from solid to react? Isn't that the reverse of the newest trend? Would be interesting to know more.

Most likely LLMs are better at writing react

I've always liked the idea of intelligence in the autonomous ships of the Revelation Space universe. Little agents reporting to progressively more intelligent and higher level ones.

That's essentially all life from the sub-cellular level on up

"Your neighbor is trying to recline, outbid them to stop them..."

One verification can to you, sir, for this chuckle.

Thank you sir for the one shiny object.

Don't give them ideas!

At first I thought this was a reference to https://en.wikipedia.org/wiki/Leonard_v._Pepsico,_Inc.

Hm... I don't know that I buy your argument, since just as you point out, traditional jets are already very optimized. One would assume there's less slack to pick up.

Fuel is a huge component of the cost of operating an airline, sometimes the largest component. LNG is a much cheaper fuel, so I can see it being adopted for mainstream aviation eventually. Existing jets could technically be converted, though the conservative nature of aviation would demand many years of testing before use on commercial flights.

It's also a pathway to incremental decarbonizing of aviation. LNG releases less CO2 per unit energy than oil, and methane can be produced biologically or synthetically which offers a path to total (net) decarbonization.

> Even if fossil LNG is used, it releases less CO2 per unit energy.

However released methane has a significantly worse greenhouse effect than CO2 (80x over 20 years, 28 over 100, 8 over 500 — this decreases because methane has an atmospheric lifetime of 12 years and decays to CO2). So leakage in the LNG chain is a massive problem.

Right, but are leakage rates high enough to make this a concern? Every methane molecule leaked is a methane molecule not burnt, so there's already a strong profit maximisation incentive to leak as little as possible (even before considering loftier goals like workplace safety or externally imposed regulation).

> Right, but are leakage rates high enough to make this a concern?

According to a recent IEA report, 30% of the LMG supply chain’s greenhouse impact is methane leaks.

> there's already a strong profit maximisation incentive to leak as little as possible

That runs against the stronger profit maximisation incentive of doing as little maintenance and being as cheap as possible.

A major difference is: there is an economic incentive to not leak methane since a leak is wasted fuel, while the economic incentive for CO2 is to make more of it.

> A major difference is: there is an economic incentive to not leak methane since a leak is wasted fuel

That economic incentive only goes so far given the entire point of the discussion: LNG is cheap. Per the IEA's recent "Assessing Emissions from LNG Supply and Abatement Options":

> Our analysis estimates total GHG emissions from the LNG supply chain are around 350 million tonnes of carbon dioxide equivalent (Mt CO2-eq) (this excludes emissions from combustion of the natural gas at the point of use). Around 70% of this is in the form of CO2 emissions which are either combusted or vented, and the remaining 30% is methane that escapes, unburnt, into the atmosphere.

> ...

> Globally, the average GHG emissions intensity of delivered LNG is just under 20 g CO2-eq/MJ, compared with an average of 12 g CO2/MJ for natural gas supply overall.

Jet engines don’t release the methane, they burn it, and they’re very efficient. And jets don’t leak fuel, that would be very hazardous.

LNG is not a nice liquid at room temp, so more use means more in transit and more complicated supply chains, meaning more leaks.

Traditional jets have a long inventory and regulation cycle but for example retrofitting a A320 to LNG appears to save 20%:

https://repository.tudelft.nl/record/uuid:63b89022-ac68-426d...

Which still puts it behind the 787 let alone the generation that comes next.. But you aren't going to succeed at making any new inventory without every possible efficiency improvement to drive sales and retirement of older inventory.

I assume these must be really expensive? Otherwise it seems like a great way to improve throughput on low concurrency tasks.

At least in older CPUs the caches were SRAM (static RAM). It is complicated but requires no refreshing. DRAM is basically just a capacitor per bit and capacitors leak so you constantly have to refresh the entire memory space. When the CPU sends a request to RAM, the memory controller might be too busy refreshing the soon to decay parts to actually respond right away. And if I recall correctly when you read from DRAM you destroy what was there so the process is to read it, then write it back, then send the answer to the CPU which is just a lot of steps. But the price and die size difference is huge so we use GB or TB levels of DRAM and MB levels of SRAM.

Wouldn't this bound the overall memory bandwidth, not the per core bandwidth? I've sort of assumed that just providing more line fill buffers wouldn't be sufficient, and that the number of LFB is chosen in tandem with a number of other things, but I'm not sure what the other things are (that is, just increasing the # of LFB might not be meaningful without also increasing XYZ).

I can’t speak to that. Last time I looked at this stuff was when I was taking an electrical engineering class and we were talking about constructing RAM out of flip flops.

bus wires. you can route only so many of them on a motherboard.

it's why GPUs have their memory chips in a circle around the GPU chip.

Wouldn't this be the limiting factor moreso for overall throughput, not per core? I believe with Zen 4 for instance it goes through a central memory controller.

sure, but you can have more memory controllers (I/O dies), like threadripper

If you're curious, you might also be interested in Cauchy-Reed Solomon coding. This converts Galois field operations into XORs by treating elements of GF(2^n) as bit matrices. The advantage then is that instead of doing Galois field operations, you can just xor things for much better performance. The canonical paper is https://web.eecs.utk.edu/~jplank/plank/papers/CS-05-569.pdf.

https://www.usenix.org/system/files/fast19-zhou.pdf is a more modern paper that goes into some related problems of trying to reduce the number of XOR operations needed to encode data.

That was a fun little rabbit-hole with a nice 20x (not backward-compatible with my existing memory layout) speedup. Thank you!

They blew up the air defences and reportedly had help from a CIA informant, but there's nothing to indicate that it was a coup.

True but you don't need advanced defense to take out slow moving helicopters, the fact that nobody used manpads is extremely suspect. Also in syria the russians did token airstrikes while jolani's forces blitzed through the countryside.

It was done at night with stealth helicopters, and over 150 planes in the air. Not sure it's necessarily easy to take out US military helicopters in that environment. They move pretty fast.

Sure, but bulky chinnok helicopters flying low to the ground and barely getting shot at? Smelling an inside job honestly, especially with rumors of trump wanting have Venezuela's current VP ascend to the presidency instead of the other investor lady.

That was only to save face. It was part of a negotiated exit.

2-3 years max in a federal country club prison, minimum security. Then it's off to Switzerland or Dubai with his ill gotten gains. It is rather sad to see people having a personal stake in this. It's a big club, and you ain't in it.

Ehhh. I think if you're trying to show the overall costs of something to someone that conclusion makes sense, but interactive flame graphs are the best way imo to look into things. Especially making use of sandwich views, which allow you to pivot the flame graph around some function to see callers and callees by cost.

Edit: I'll keep this up to share my embarrassment, but I missed entirely that the article was about disk space. I admit I only looked at the pictures haha.