ori_b

28 ms·

601.
by ori_b 3y ago
https://arxiv.org/abs/quant-ph/0301141
602.
by ori_b 3y ago
Why is there a pipeline at all? Give me a framebuffer, a place to upload some CUDA or equivalent, and some dumb data pipe between the host program and GPU program.
603.
by ori_b 3y ago
That's because they modularized and stopped doing "katamari releases". The last X server release was in March 2023.
604.
by ori_b 3y ago
Software rendering. In parallel on the GPU.
605.
by ori_b 3y ago
Yes, but that's nothing new -- we've had computers that were used to solve simulations faster than digital comuters could by setting up electronic (or hydraulic, or pneumatic) parts that simulated the terms of the differential equ
606.
by ori_b 3y ago
Indeed. Like facebook chat was XMPP based. If the rumors are even true, let's see how long until the rug pull.
607.
by ori_b 3y ago
Does Google's implementation of quantum computing help with this sort of scenario, or is it a really fancy way of breaking the bridge?
608.
by ori_b 3y ago
Hm. That seems to be comparing against a 2013 era leveldb, which at the time also used mmap. (It's since switched the default for performance reasons) It's also strange to me that there's no transition in performance when the
609.
by ori_b 3y ago
Do you have benchmarks of lmdb when the working set is much larger than memory? I couldn't find any. In my experience -- and in line with the article -- mmap works fine with small working sets. It seems that most benchmarks of lmdb hav
610.
by ori_b 3y ago
No, huge pages wouldn't help. They would change when the TLB gets flushed, but the flushes would still be there.
611.
by ori_b 3y ago
The downside, of course, is that each program owns one instance of the hardware. Applications don't share the network card. This isn't a general purpose solution. That may be acceptable for your purposes, or it may not.
612.
by ori_b 3y ago
That's incorrect. (Well. Technically it's kind of true. You can run a container based on a newer kernel, it'll just crash when you use any new features)
613.
by ori_b 3y ago
Yeah, it's shame those vital arteries of goods are always so choked up with personal cars.
614.
by ori_b 3y ago
It can have a name in the file system: /dev/shm
615.
by ori_b 3y ago
The feature of plan 9 that makes it all hang together is that all features interact through a single, network transparent, universal layer. Fixing this in Linux means removing everything that doesn't fit in this interposable, redirecta
616.
by ori_b 3y ago
That's... More or less how it works in plan 9? http://man.9front.org/3/mnt
617.
by ori_b 3y ago
Plan 9 does this. The resulting interfaces are indeed easier to use, forward over the network, and redirect or emulate for testing.
618.
by ori_b 3y ago
I'm pretty sure your examples behave differently from the example programs. For example, 'console.log' will pretty-print the data structure, rather than a single space separated line, the duplicate counter only reads from one
619.
by ori_b 3y ago
Can you be specific about what improved? After peeking at the repo, the biggest improvement seems to have been reindenting it.
620.
by ori_b 3y ago
Try libressl ( https://github.com/libressl/portable )
621.
by ori_b 3y ago
> LLVM and Clang are also about having a compiler stack that provides the information you need for a modern graphical IDE. Is that why all GNU utilities shipped with OSX were also stuck at the GPLv2 versions and never upgraded?
622.
by ori_b 3y ago
And the peripherals are far more interesting from a license perspective.
623.
by ori_b 3y ago
Which come with well documented peripherals? The CPU is boring -- x86 instruction encoding is a bit ugly, but it works. It's all the stuff hanging off them that make the difference, and as far as I'm aware, risc-v hasn't done
624.
by ori_b 3y ago
That's a shame; people should complain more in general. Especially if it comes with action.
625.
by ori_b 3y ago
In the case of ChatGPT, the citations serve a different purpose. ChatGPT should be cited as a warning that the text needs additional investigation.
626.
by ori_b 3y ago
This is LLMs destroying the web we used to know. I would be willing to bet that the driving force behind the decision was to make it less trivial for LLMs to say "the data was already there under an open license, so we legally underc
627.
by ori_b 3y ago
It isn't; the first thing production optimizers do is throw it away. It supposedly makes verification of SSA properties marginally faster, but that's not the bottleneck.
628.
by ori_b 3y ago
Though, wasm has a high level structure that makes it a clunky compilation target. The bytecodes above improve on wasm in that regard.
629.
by ori_b 3y ago
It's likely that this is the result of training it to avoid bullshitting. It gave confident garbage, and they're trying to stem the flow. This likely leads to less certain responses in complex tasks.
630.
by ori_b 3y ago
This seems pretty inherently difficult to port or to run safely. Things like fuetxes will only work directly on Linux and OpenBSD, for example. From what I recall, OSX has some undocumented futex-like-but-not-quite ulock, and FreeBSD has ye

More ›