raggi

14 ms·

151.
by raggi 1y ago
> which seems too high for a system call anyways spectre & meltdown. > you get 2 us of processing in addition to a full system call to reach 4 Gb/s TCP has route binding, UDP does not (connect(2) helps one side, but not both
152.
by raggi 1y ago
I'm on the systems side, and I find some of what Chris and team are doing with Mojo pretty interesting and could be useful to eradicate a bunch of polyglot ffi mess across the board. I can't invest in it or even start discussions
153.
by raggi 1y ago
Please forgive my pushback but: No third party: Almost certainly as a user there are still third parties involved, this isn't (AFAICS and based on other discussions) a user facing chain (edit: correction, they do say the chain is publi
154.
by raggi 1y ago
Can you expand on "easier" and "faster" in easier/faster/better. I understand "better" in terms of transparency, shared standards for integration, and various other properties. I can less immediately
155.
by raggi 1y ago
They don’t have exactly the same coverage. UEFI provides a standard set of APIs, some optional, some mandatory. It also specifies a boot path via gpt and efi system partitions. A common secondary piece is ACPI, which isn’t strictly mandator
156.
by raggi 1y ago
what areas of u-boot do you find much simpler than edk2?
157.
by raggi 1y ago
There is a UEFI standard in some of Arms specs, but it’s only mandatory for the server class IP atm. Some soc vendors are starting to adopt it even in non-server class but it’s slow. My feedback to an Arm devrel a few years ago was exactly
158.
by raggi 1y ago
Customers not being able to tell that it happened was a goal of that deployment. It wasn’t anywhere near that late the numbers you’re saying was about the whole build cycle - board bringup for one of the two boards first commit in zircon wa
159.
by raggi 1y ago
fsync is rarely truly durable in the sense the article describes. it does help with loss ordering in a lot of cases, and flushes often do some work, but true durability, the idea that after fsync there will not be any rollback or tail loss
160.
by raggi 1y ago
Yeah, the various docs and presentations given by reversers on the details of nintendos os really showcase a lot of good design choices. An impressive amount of work for just trying to enforce a harder IP moat around entertainment software.
161.
by raggi 1y ago
Just overclock (more) the system that’s already in a severe struggle to meet power, thermal and fidelity budgets?
162.
by raggi 1y ago
For this use case a major one would be better models for carved up shared memory with safe/secure mappings in and out of specialized hardware like the gpu. Android uses binder for this and there are a good number of practical pains wit
163.
by raggi 1y ago
By forced I guess you’re referring to the room full of leads who all said yes, but then reported otherwise back down to their ics to avoid retribution. I caught early wind of this from folks being super rude in early on the ground discussio
164.
by raggi 1y ago
I think what op was thinking of was early harmonyos, seen people confusing those a lot. Harmony now ofc isn’t what
165.
by raggi 1y ago
Extremism and division aren’t a path to wisdom or to healthy cultures, but you do you.
166.
by raggi 1y ago
I guess while I’m here, please stop putting generated stuff in .config or even worse junk you downloaded, half of you without even checking it’s the right content or having any docs, ui or ux for cleaning up.
167.
by raggi 1y ago
The thing that annoys me more than the authors main complaint is when the XDG specs are ignored on windows and macOS (I.e. if XDG vars are set, respect them). I work on all platforms, I have decades old dot files that work on over 9 differe
168.
by raggi 1y ago
I used to do this in the 2000's era, there was a lot to love about it. At the time though the IE engines were far more complete and less buggy than others with various XML features.
169.
by raggi 1y ago
fwiw, the tailscale fork of Crawshaw’s library has a good number of allocation removals and other optimizations, but cgo is still expensive.
170.
by raggi 1y ago
No. I don't think "novelty" would be relevant in such a case. How much risk you have depends on many factors, including what you mean by "use". If you mean sell, and you're successful, you're at risk. That
171.
by raggi 1y ago
reddit is a user hostile company, have been forever, always will be. they take rights over your content, farm things about you, sell data, do invasive things in the mobile apps, use creepware cookies, etc. Excerpt from the user agreement:
172.
by raggi 1y ago
"an LLM" could imply an LLM of any size, for sufficiently small or focused training sets an LLM may not be transformative. There is some scale at which the volume and diversity of training data and intricacy of abstraction moves a
173.
by raggi 1y ago
In the US you can not generate copyrightable IP without substantial human contribution to the process. https://www.copyright.gov/ai/Copyright-and-Artificial-Intell...
174.
by raggi 1y ago
For a large LLM I think the science in the end will demonstrate that verbatim reproduction is not coming from verbatim recording, as the structure really isn’t setup that way in the models under question here. This is similar to the ruling
175.
by raggi 1y ago
I know, but the _implication_ is that it's extremely unsafe, I don't buy the implication - code gets executed.
176.
by raggi 1y ago
I love this implication that there's some valuable body of code out there that gets reviewed, compiled and never executed.
177.
by raggi 1y ago
These prices seem very low.
178.
by raggi 1y ago
LTS has had major breaking changes in various areas in recent times too, virtio was badly broken at one point this year, as was a commonly used netlink interface. Hat tip to the Arch kernel contributors who helped track this down and chase
179.
by raggi 1y ago
"which is ostensibly AES" in the 5% or less of deployments that turn that on Both of the systems are crap, when we were evaluating them for nationwide purchase we chose TETRA because of systemic safety features (like local DMO han
180.
by raggi 1y ago
This is a story of a bad LLM user (in this context), which is perhaps implied by “vibe coding” but folks should be aiming higher. Making people review slop is lazy, rude and disrespectful.

More ›