calaphos

Recompiling is not an option? Also seems weird that the binary only ships with a AVX512 codepath and not AVX2 as well.

I remember there being also an issue of linked MKL libraries shipped with various software (e.g. Matlab) not using the appropriate AVX2 codepaths on 'unsupported' AMD processors and isn't was falling back on slow non vectorized code.

The number seems extremely low, especially considering this is completely before taxes, health insurance and retirement ('Arbeitgeber brutto'). The actual net salary after everything is roughly half of that.

But that's not inherent to the mode of transportation. In germany (and generally in Europe) taking the train is often a bit faster (~20%) than by car - assuming no traffic. Of course this doesn't apply to rural regions, but between most cities trains run every 1h.

And the train network in germany isn't even very good.

Of course that 90mph top speed is essential for the quality of life of its driver - especially during the 80% of time it's stuck in traffic moving forward at 10% of that speed :)

For me it's the single best feature about VSCode.

Extremely useful for projects which have an environment that is a bit tricky to setup - so usually python. I can just run all my python projects on a workstation or even some powerful cluster node and never need to mess with a local python version on the windows laptop. A giant improvement over the rather lackluster remote mode of Pycharm which needs a local environment to be useful at all. It even supports a remote connection via a jumphost, e.g. for cluster nodes behind another login node.

On the other end the editor is significantly worse when compared to Pycharm. The refactoring is mostly unusable across files and more often breaks things than helps. No perfect tool I guess.

Not surprised, season 2 was rather lacking, especially after the amazing first season.

If you liked the story and idea I would recommend breading the Novel trilogy it's based on. Season 1 was fairly close to the first book while the second one was a really weird and IMHO badly done mix of the later two.

Another issue on the third generation of ICE is that the windows have a coating meant for blocking out heat which is also very good at blocking cell signal. The provided wifi is done via external antennas and multiple carriers and usually quit a bit better, however there are still long parts where it's unavailable as well. Also quite limited at 200MB of volume.

Deutsche Bank is not a state institution, neither is it partially or fully owned by a state entity (like Deutsche Bahn or VW). The national bank would be deutsche Bundesbank.

> While I say all of this, I should also point out that the top500 benchmark pretty much is not representative of most real-life workloads, and is largely based on your ability to solve the largest dense linear solve you possibly can - something almost no real application does.

They also publish the HPCG benchmark with sparse matrixes. And unsurprisingly an order of magnitude lower flops across the board. The Fujitsu chip scales a whole lot better than the usual Nvidia GPUs though.

Even more impressive than the linpack result (2.8x faster than the runner up) is the HPCG result at 4.6x the result of summit in second place.

That benchmark consists of more sparse matrixes which are a lot more realistic depiction of hpc workloads. Seems to scale a lot better with irregular access patterns than basically Nvidia GPUs on the other systems.

In my experience high DPI actually works somewhat decently and those fractional stuff can be fixed with don't scaling. At least if you only use one monitor. I personally have two I would use with 1.25x and no scaling which is just not possible with the x11 model.

Additionally having two monitors completely breaks vsync with very ugly tearing lines.

And Wayland on Nvidia has been very unsuccessful for me. Even when I got it to run the performance in e.g. firefox was absolutely terrible. Sadly Wayland is necessary if you want hardware video decoding.

I started working like that after being work from home started rather aprubt a in March. The work laptop is a very slow windows machine but with remote - ssh to the office workstation the local environment does not matter.

With some settings tweaks I could even convince it to do a remote session via a jumphost.

Afaik pythons success in that field is relatively recent, only really taking of ~2010. The language was already 20 years old by then. Compare that to rusts first stable release in 2015.

Building ecosystem just takes time.

Rust is very very new compared to Cpp. People have decades of experience with cpp, learned in in University. Why would you throw all of that overboard just to learn a fancy new language with a weird ownership system? Keep in mind that people often still write cpp98 and have a rely hard time getting used to newer cpp concepts and programming styles. And all of that is backwards compatible with their existing codebase.

With the impressive performances of generative AI I'd raises the interesting question of what people actually want when they take a photo. Is it preferable if it looks great, even if the details aren't actually true and imagined by a neural network?

There was some discussion about it a few years ago when one of huaweis phones had a 'moon mode' that afaik basically recognized the moon and pasted in a static picture. Complete fake, but actually a reasonable idea - it's not like the moon will look any different and that's what the people want after all.

The factor is not directly sensor size but rather focal length of the lens. But with smaller sensors comes a smaller field of view on the same focal length, such that you need to reduce focal length to keep it the same. I think for phone cameras the factor is ~10x compared to 35mm sensors.

In the end it's just an index linking papers usually hosted in arxiv with code usually in GitHub. It's more of a nice starting point for literature research rather than some content owning platform.

I use both extensively but settled for windows + WSL2 for now. In the end it's the little things:

- (fractional) scaling actually works quite well. - the shell works a big better. Feels like a good compromise between gnomes simplicity and KDEs over the top customisability to me. KDE has actually been the most stable out of the three though. - (Nvidia) driver support. Using Manjaro I had to reinstall the Nvidia driver every couple of months because a kernel update broke things. Again, not huge but annoying. - Working hardware video decoding in firefox. On Linux playing videos results in significantly higher CPU load, noise, slowed down browser, etc. Even on a fast machine.

With wsl2 I can now do all my development in that environment and don't need to live with windows slow file system, missing package manager, etc. The remote development features of VSCode with wsl (and other ssh hosts) are also simply amazing.

Turns out it is a lot more useful to (potentially) reuse the first stage than to carry a lot of additional weight to reuse the last stage.

SpaceX developments have a clearly defined goal with most engineering and production decisions made according to mission + cost constraints. The space shuttle development had dramatically changing requirements from multiple stakeholders (cheap launches, crossrange capability, recovery of military satellites, etc.) with production facilities based on jobs in voting districts. As impressive as the space shuttle was from an engineering perspective, from an economic and safety standpoint it was a complete desaster.

In a lot of places in the world fresh water is not an issue and a higher rate of usage actually helps with sewer systems and prevents them from clogging.

However recovering the waste heat from a shower, e.g. by running the drain water trough a heat exchanger with the freshwater would already help saving a lot of energy.

Here in Germany we commonly have negative electricity prices at exchanges and with a increasingly higher percentage of renewable energy is will only become more common.

It would be very easy to use the energy to heat up warm water in the already existing warm water boilers just by adding a cheap electric heating element to the usual gas burner. The necessary smart meters for billing are actually mandatory in new installations. However with current consumer electricity prices completely removed from the large scale exchanges and more expensive than ever I don't see that happening right now.

The energy a car uses is insanely large compared to e.g. a fridge. You could probably run it at least a month with a smallish 40kwh battery.

Even running everything as normal probably uses less than a kw on average so electricity outrages would have to last multiple days to be a problem.

How did these work? The article says you could ssh into them, so they all had their own RAM and ran their own individual os? I don't think the CPUs were multi socket capable.

How did they boot? Network? So the pcie connection was essentially just for networking?

The systemd problems remain in WSL2 as they same custom Microsoft init system from wsl1 is still used. There are workarounds however

Interestingly I have the opposite experience with the previous battlefield 3. There are still several active servers, all of them are generally fun to play with no weird rule sets. I remember in the earlier years of the game some servers had rules like no sniper and you were automatically kicked if using one but that was made clear in the server title.

I don't know about steel but carbon fiber has so far never succeeded as a material for cryogenic rockets or fuel tanks.

The proposed space shuttle successor was designed around carbon fiber fuel tanks / hull and engineering problems with them eventually killed the project.

The material struggles with low temperatures of cryogenic fuels (brittle), high temperatures of re-entry (just burns), oxidation in oxygen tanks. Also they apparently tend to sponatiously combust with rocket fuels.

Some reference: https://www.compositesworld.com/articles/an-update-on-compos...

It's certainly to the same type of people who would buy a new product for 5x the price instead.

I've had very good experiences with Linux on my laptop. Definitely more stable than windows, literally never had to fiddle with kernel modules or any sort of hardware support. Battery life is actually better than in windows.

You have to be a bit careful with the hardware, especially non Intel wifi/Bluetooth seems not well supported. And Nvidia graphics are a bit risky, but of the alternative is a MacBook that shouldn't matter as much

YMMW but I had a really good experience with 2x scaling on a 4k Laptop Screen in standard Ubuntu and gnome. There are still some apps which ignore dpi settings (zoom) or need to be forced to scale (often electron stuff, although you can usually just zoom with Crtl+ there). Otherwise awesome sharp text and fluid.

Fractional scaling seems to be quite a mess, at least in KDE it breaks a lot of layouts.