It's close, but not quite. Getting it to do 'y = ax + by' like vector operations without superfluous reads/writes is tricky.
HN user
Fronzie
(Home) batteries are quickly becoming cheap and per-hour electricity rates can be implemented at a reasonable time. With that, the grid owner can influence the grid stability without having to build capacity or generation itself.
Do they require anti-cheat kernel drivers? If not, there's a good chance proton/steam to runs them.
With debian and KDE (both personal preference), but no snap or flatpak, it works wonderfully. Power/sleep-management has become better than a default windows install. All hardware, including the fingerprint sensor, just works.
There are EV's which have drum-brakes. That's only possible because the regenerative braking takes most of the load. Also, there are stories of brake-discs rusting out on EV's. Both point to less (friction)-brake use on an EV compared to an ICE.
Steinberg spectral Layers is one of the commercial ones. It sounds really good.
And the tin droplets are in vacuum, with contamination of the wafer being very critical to control.
There has been some testing within llama.cpp, which supports both Vulkan and ROCM-Blas. When it works, the latter is about 2x faster than the Vulkan version.
From what I've seen, AMD has done... none of this.
There are a number of pull-requests to ROCMblas for tuning various sizes of GEMV and GEMM operations. For example: https://github.com/ROCm/rocBLAS/pull/1532
ISA cards, like the soundblaster had the MIDI and joystick on the same connector. The joystick had up to 4 analog inputs, maybe that's what's meant here.
And ASML relies on Zeiss, which is as big a bottleneck as ASML and TSMC.
As far as I know, even with LTO, it requires -ffunction-sections -fdata-sections in order to strip out unused functions.
If there's a chain of dependencies (libraries depending on other libraries), a single process might end up with different versions of the same library in it's memory. That's not going to work, since the interface/API of the library is typically not versioned.
Currently, Russia is moving it's military away from the Finnish border. So NATO being on it's doorstep is apparently not a thread.
Yes, I know of friends and coworkers who have had to change career because they couldn't do their old job anymore. This after a year long recovery. They aren't happy about it, they really want to do their old job, if only they could. Then there are a few more who where home sick for 1 or 2 months. They all seem to have recovered.
In the Netherlands the same happened with the railway and healthcare coverage. The motivation was that market pressure would lead to increased efficiency. In practice, there's a tendency towards monopoly and a forest of product variants leading to more confusion and people buying sub-optimal. The net effect over here is negative, in my opinion.
In the 70's, the city of Utrecht in the Netherlands was renovated from car central to bike central. That only took a few years, bike paths aren't that expensive.
The last few years ASML's problem was not supply chains, it was a very high demand. Production was ramped up successfully, but not enough to meet that demand. With regards to Peak IC: atoms aren't getting smaller, Feynmans' "Plenty of room at the bottom" might indeed run it's end.
reflection is coherent, a single physical process. The light keeps it's phase and polarization. Re-radiation is incoherent, caused by multiple particle interactions (physical processes). So there's quite a difference in physics.
The first CD's were not that great, being effectively around 14-bit. Better D/A converters and noise-shaping during mastering greatly improved the sound quality. After that, the loudness wars destroyed it again.
Leaded gasoline has been a former experiment in population-wide reduction of intelligence and increase of aggressiveness. Supposedly that affected that 80's, but I am not aware of systematic studies on that.
The Netherlands has started burning more coal, so Germany can have more gas. Drilling more gas from Groningen can be done at a moments notice, and will be done before either Dutch or German houses are without heating
Transmission also has the client/server model, where the local UI can connect to a transmission running as daemon.
For me, https://tailscale.com/ gives me an easy-to-use personal network. It uses wireguard with custom configuration on top of it.
The motivations I have seen were:
- If radar/lidar fails, vision needs to work. If vision works, radar/lidar are not needed. - Sensor fusion is hard: outliers in radar data ended up hurting more than the non-outliers helped.
The second one sounds to me like the real reason: Sensor fusion is not trivial and if it doesn't work correctly, you'll get the worst of all worlds.
Their vaccine is not very effective against omicron, and they have far less hospital beds per capita than the west. So a 'flatten the curve' strategy will cause a lot more deads than it has in Europe and the USA.
The Eigen library (https://eigen.tuxfamily.org) has great attention to numerical robustness and accuracy in it's implementation. The documentation is good, but for the numerical discussions, you'd have to read the source code.
It is, the neural network part seems to be here: https://github.com/commaai/openpilot/tree/master/models
http://ceres-solver.org/ works well, in my experience. It does require code to be templated, which might increase computation time. It is possible to explicitly define derivatives for individual functions, this can help preventing numerical instabilities in corner cases.
"a bit of all" can be done immediately. A bit improved isolation, solar panels, wind farms, more working from home (to reduce travel), it all adds up.