Veserv

43 ms·

541.
by Veserv 2y ago
Having seen some of their talks, I do not think they realize the underlying problem and understand its manifestation. They just see that their “fellows” churn out slow software without even understanding how to make things fast due to trade
542.
by Veserv 2y ago
No, it is quite reasonable with efficient implementation. 10-50% overhead or so depending on function size distribution (since it is small fixed overhead per call, smaller functions result in a greater fraction of overhead). 10x for just fu
543.
by Veserv 2y ago
“Civilization in decline” is overstating the problem. It is a common refrain that performance and security are poor because they are not priorities. The corollary being that the only reason things are slow and insecure is because we do not
544.
by Veserv 2y ago
I was talking about automatic instrumentation of every single function call by default. No manual instrumentation needed because everything is already instrumented. To be fair, you do still want some manual instrumentation to correlate
545.
by Veserv 2y ago
The textbook example of course being Sovaldi [1] the cure for Hepatitis C that was invented and then sold for ~84 K$ when introduced which massively undercut the prevailing treatments which cost ~40 K$ per year for a lifetime (40+ years of
546.
by Veserv 2y ago
Looks fine, but it does not look like there is a automatic full function entry/exit trace, just sampling. The real benefit is when you do not even need to insert manual instrumentation points, you just hit run and you get a full system
547.
by Veserv 2y ago
Is a non-network connected car model at risk of all cars being taken over by a economical sustained assault by highly trained adversaries in a foreign country? No. Is a network connected car vulnerable to such attacks? Yes. The product deci
548.
by Veserv 2y ago
You said: “If a single simple mistake has the potential to cause devastating liability” Liability is in proportion to actual harm. A single simple mistake can only incur devastating liability if it causes devastating harm (as in harm actual
549.
by Veserv 2y ago
Is that something that you expect to be attempted against a significant fraction of the products over the average expected lifetime of the product? If so, they should absolutely be required to make a product fit for its usage context that i
550.
by Veserv 2y ago
Ah yes, the Elon Musk special. The term you are looking for is “pigeon CEO”[1]: “he comes, shits all over us, and goes”. [1] https://electrek.co/2024/04/22/elon-musk-pigeon-ceo-former-t...
551.
by Veserv 2y ago
If a single simple mistake has the potential to cause devastating harm, then that is precisely the standard that should be demanded. If you do not want that, then you can work on something where your mistakes do not cause immense harm to ot
552.
by Veserv 2y ago
You can already do that by just closing your eyes and letting Jesus take the wheel. No, the point is doing so while maintaining safety. It is materially less safe to operate a ADAS while distracted than driving manually. Humans are exceptio
553.
by Veserv 2y ago
Exactly. Elon Musk said you have to be braindead [1] to prefer being banned in a country over honoring censorship requests by the government. But he immediately abandoned any pretense of that position due to his personal animus against the
554.
by Veserv 2y ago
The Breakthrough Devices Program [1] is just a special review program. Designation just means you get to go through a non-standard approval process because your technology is too different to be well-served by the normal review process. It
555.
by Veserv 2y ago
Every company, car or otherwise, has security grossly inadequate for the current and especially the near future threat landscape. Everybody knows everything is easily hacked and companies people tout as exemplars like Apple and Microsoft ju
556.
by Veserv 2y ago
At timestamp 23:40 in the video by Alex Pshenichkin from 2024-06-10 it says data ingestion comes via VMCALL interactions. As such a call is literal nonsense if you are not virtualized, any such call inherently means you are using a paravirt
557.
by Veserv 2y ago
The problem is that the filesystem primitives are garbage, so it is impossible to make something safe and reasonably performant. This is not a case of "speed at all costs" where huge footguns are added for marginal performance, th
558.
by Veserv 2y ago
Found the answer in a different post [1]. The hypervisor and virtual machines are single-core only. The talk also indicates that all I/O operations need to be manually rewritten to use the instrumented mechanism, so it demands a highly
559.
by Veserv 2y ago
Is the hypervisor multicore? How do you handle shared memory non-determinism? What is the runtime slowdown for shared memory multicore (lets say 16 cores if you need a concrete example) execution?
560.
by Veserv 2y ago
QUICv2 is not really a new standard. It explicitly exists merely to intentionally rearrange some fields to prevent standard hardcoding/ossification and exercise the version negotiation logic of implementations. It says so right in the
561.
by Veserv 2y ago
The system call to send multiple UDP packets in a single call has existed since Linux 3.0 over a decade ago[1]: sendmmsg(). [1] https://man7.org/linux/man-pages/man2/sendmmsg.2.html
562.
by Veserv 2y ago
Yes, if you have cheap threads at the required scale. But that is the entire problem as you also attest and agree that with the author and me that OS threads do not scale to the required level. Asynchronous, non-blocking primitives do s
563.
by Veserv 2y ago
No, it shows you did not understand the context or the point of the comment you originally responded to. The author wants both synchronous and asynchronous modes, but they complain it is painful to provide asynchronous modes using synchrono
564.
by Veserv 2y ago
No, the hardware is frequently inherently asynchronous. You write some memory and then the hardware consumes the prepared data asynchronously, in parallel, until it informs you in some manner that the operation is complete (usually eithe
565.
by Veserv 2y ago
One of the other aspects of this is that implementing a synchronous model on top of asynchronous primitives is absolutely trivial. You just wait until the asynchronous operation completes. Any program designed for asynchronous execution can
566.
by Veserv 2y ago
The things you are complaining about are things that are handled by higher level human-scale protocols. They can and are layered on top of the existing low-level hardware-scale protocols. You might think that those layers suck because they
567.
by Veserv 2y ago
The paradox is tolerance is just the paradox of imprecise language. Once you actually define what you mean by the word "tolerance", as in the acts being tolerated and what acts you are prohibited from doing to something being tole
568.
by Veserv 2y ago
I was using "application" level to mean the consumer of the raw packet stream which would generally be a transport protocol (i.e. TCP, UDP, QUIC, etc.), not a actual user application. Truncation is trivial to support at the transp
569.
by Veserv 2y ago
You can actually do better by just having network hardware truncate and forward instead of drop when encountering a packet that is too large. Then the receiver can detect every packet that went through a route with a small MTU and know, pre
570.
by Veserv 2y ago
Yet Elon Musk said that following government censorship requests is the right thing to do: https://www.reddit.com/r/TheMajorityReport/comments/13h0qon/...

More ›