This isn't normal, this isn't how technology has worked.
Google doesn't report people to the police for the private (non-CSAM) contents of searches or emails.
You could definitely go through people's shit before, but Big Tech generally had serious prohibitions and a stronger presumption of privacy about this sort of thing.
CSAM was an exception where they theoretically had the evidence of the actual crime, rather than writings alluding to them. And even that has been problematic.
Frankly, the charge here is total bullshit and should be thrown out. The law is meant to prevent people sending threats to others, not keeping notes. If this charge sticks, the law should be changed.
1. Better memory locality by knowing what you load and when, exactly.
2. The ability to "cheat" on calling conventions.
3. The ability for techniques like threaded code, and in general, better cache-awareness.
4. Less mov's.
5. Guaranteeing no spilling in important loops.
6. Compilers don't do well with flags registers and you can't read/write them in high level languages. You're hoping your `if (result < a) { carry = 1; }` becomes a direct flag test. Especially important in bignum, you can't really utilise adcx/adox directly from high-level code.
7. Hot/cold layout without PGO. Yes PGO is good but sometimes you know better and PGO isn't very suitable for "configurable" code.
9. Exploiting uninitialised memory for classic party tricks like not initialising a buffer fully (let's say you have a library function with a return buffer. You don't want dynamic allocations for some reason. You can simulate this with a pointer return into a let's say a static 4KB buffer and a count return, you only initialise it until the count. Caller has the responsibility not to overread.)
An optimizer is trying to balance between compile time, runtime speed, and code size, and most optimizations will win you one axis at the cost of one or both of the other axes. (The rare optimizations that win on all 3 are already all implemented in the compiler.) Compiler developer time is also a scarce resource; I know of so many more optimizations I could implement, but without demonstrable code that would actually benefit, it's not a good use of my time to implement them. Compilers tune this balance by making lots of heuristic decisions, and these heuristics are tuned by large benchmarks, which often times involve a lot of flat code profiles (i.e., no code is worth spending a lot of time really nailing down the best code layout).
One of the advantages of hand-written assembly is that you get to opt out of the compiler heuristics and commit to being able to spend the time to optimize the one bit of code that you know is really important for runtime as perfectly as you want, instead of relying on the compiler to get it close enough to perfect before it exhausts its budget of caring about optimizing it.
You can usually hint the compiler, but the best idea is to provide it with more data. PGO with real-world data and LTO for whole-program analysis enable it to do much better decisions.
"We’d like to initialise our Vec in parallel, otherwise we’d have to wait for the main thread to fill the entire Vec with a placeholder value only to then have our threads overwrite those placeholder values."
Talk about overengineering :P Multithreaded vector initialisation instead of just...skipping it?
With hand-written assembly, you can (with effort and care) ensure that the code runs in constant time and doesn’t leak any information through side channels (such as which memory or cache addresses it accesses). That’s important for most encryption code. It’s difficult to ensure constant time execution in pure Rust (or C, or most high level languages in general) because you can’t tell if some future compiler optimization will break your attempts at constant-time code.
I'm so far downstream from the source language that I know of none specific to Rust. And note someone else's specific answer to your question above that has nothing Rust specific either.
I had this same thought and think this is a generally interesting direction, but I think we're in a bit of a weird spot where the compute heavy stuff is on GPUs already and most infra stuff is not compute bound (it's often I/O bound or memory bound in some way).
It doesn't help that FPGAs are not made at the same scale as CPUs so don't benefit from the economies of scale.
I'm super curious if you have thoughts on specific pieces of software that would be economically better because I've thought about this in my niche and sort of come to the conclusion that it won't help.
I do think things like SIMD in CPUs will get more use and maybe we will get more difficult to program for CPU features, but I haven't found a use case where off the shelf FPGA components would help with typical software.
I’m looking at realtime mechanical processes, like shaping extrusion beads from a clay 3d printer. Clay is heterogenous and pressure takes time, so hand tuning is never just right. But put an fpga with vision processing? Seems promising, with millisecond-level latency that I’d never get pushing to a remote system for processing.
Unclear that this sort of thing wouldn't already be quite well served by the existing gpu/npu hardware optimized for neural nets. If you're doing traditional CV, you can run it on the CPU just fine.
Google is not being dumb or making some 4D strategic move by not going after coding capabilities.
Occam's Razor is overwhelmingly that they just don't have the organisational capability to capture this market. If they did then they absolutely would have.
Occam’s Razor could also simply be that OpenAI and Anthropic are massively overvalued.
If OpenAI or Anthropic go away tomorrow, people find a new model and move on. If Google goes away tomorrow, people’s lives would be severely disrupted, and in the case of Gmail / Drive / auth access, even temporarily collapse.
Oh, for sure. I think this is definitely failure to execute and not a strategic choice. Their "acquisition" of Windsurf was a clear desire to be in this market.
Seeing mathematicians such as Terry Tao being unhappy with open problems being solved makes me sort of question the usefulness of any of this pure mathematics. If we're not happy that the problems are being solved, why care about this field at all?
Pure mathematics, almost by definition, doesn't typically argue the field is always or even often "useful" (for some other purpose or application).
But, as mathematicians learn and push forward, occasionally something like elliptic curves will emerge as having useful applications, making all that previously "pointless" specialized knowledge newly valuable.
Or advances in physics, that suddenly have a need for a specific mathematical underpinning to develop a theoretical framework. Like how Einstein benefited from Minkowski's work on hyperboloids to create a coherent mathematical description of spacetime.
It was the AI labs themselves not mathematicians who were happy to conflate proofs for open math problems with some kind of tangible technological advancement in the real world. They would surely prefer to be able to claim a cure for cancer vs. a math problem but that loop requires a lot more time/money/test tubes/etc and they need headlines now not in a decade.
And so, thanks to OpenAI/Anthropic, we're now in a world where thousands of crypto bots on X breathlessly hype up each new problem being solved that previously wouldn't have any got any attention beyond academia and passionate fans of math.
Hopefully this won't lead to a trough of disillusionment as more people start to feel like you, with mathematicians getting the blame for inflating the value of their work even though the hype was coming entirely from the labs not them.
His issue is more nuanced than that. Most of the value was in humans reaching new insights or new math during failed attempts to solve these problems, whereas AI is basically "too efficient" in beelining to the goal and discards potential new insights reached along the way. I assume this is solvable.
Surely it is. Ask it to keep a list of all the promising sub paths, reprompt the collections of agents again on these after the main problem has been addressed.
Or even release a list of them and let others investigate.
Well, if in future we do end up with a magical tool that can solve any formal mathematical problem on a whim, we really won’t need field of mathematics anymore as it is today.
There would be no need to deliver new mathematical insights by solving problems. You would just have a magical math problem solving machine and that’s it.
What do you mean “we won’t need mathematics as it is today”?
To further human understanding is itself a goal that single-handedly justifies our efforts.
Jumping straight to the “answer” and therefore missing both the understanding of the actual problem, and any useful discoveries along the way is a waste at best, and actively harmful at worst.
I'm not the one you replied to, but I think I get their point. If you have a machine that can solve math problems at the push of a button, you don't really need to amass discoveries anymore just in case someone might need it years down the line.
Someone in the thread mentioned Minkowski’s math work being instrumental to Einstein’s physics. If Einstein had had a math machine that can spit out the result at will, he wouldn't have needed Minkowski.
Nevertheless, I also see the opposing point: if Minkowski hadn't already published his results, it's possible Einstein would not have even had the inspiration to derive general relativity from it. I think this is the human aspect most critics are worried about.
Physics alone is more than enough to “further human understanding”. All current mathematicians can move to other sciences, closest being fields in physics, and it will all continue to progress just fine.
He's not unhappy with it being solved, but the solution is less important than the learning you have to do to arrive at the solution. If they're just chucking compute at it and publishing the answer and hiding the path to get there, it sort of negates the whole point of posing such problems to begin with.
Here's the point: when people solve problems, they come together and create a community to eventually use the new knowledge in positive ways, including inspiring younger mathematicians by sharing insights. The human element is key and it's not just about solving problems. People only think that because we've been conditioned by computers to value answers more than how we got to them.
But if AI can solve any problem and existing mathematicians just use AI to solve problems for the sake of solving them, the community itself with wither and so will the interest in mathematics and over a longer period of time, it will just become soul-less and uninteresting and the entire community powered by the fire of fascination will simply die.
> People only think that because we've been conditioned by computers to value answers more than how we got to them.
By computers? My dude, it's not computers that condition us that way. It's exams in school. Give the right answers or you're officially labeled a failure. It's waaay older than computers.
Where did you even read that Tao is unhappy with “open problems being solved”? There was no indication of that in his Bluesky thread.
Why would you go out of your way to make a case of something being not useful when, ironically, so much advancement in human history has come from the discipline?
Your motive is more worrying than your straw man argument.
You can just ask ChatGPT these questions these days, but it was a fun read:
fly.io depends on Firecracker microVMs, so you should be concerned if you're running untrusted code, but not concerned about other people interacting with you wrt v8.
Cloudflare depends on v8 isolates, which depend on the trustworthiness of the v8 sandbox, which is a non-trivial hurdle, but not an insurmountable one for attackers with the resources to have a v8 bug in the first place.
Google doesn't report people to the police for the private (non-CSAM) contents of searches or emails.
You could definitely go through people's shit before, but Big Tech generally had serious prohibitions and a stronger presumption of privacy about this sort of thing.
CSAM was an exception where they theoretically had the evidence of the actual crime, rather than writings alluding to them. And even that has been problematic.
Frankly, the charge here is total bullshit and should be thrown out. The law is meant to prevent people sending threats to others, not keeping notes. If this charge sticks, the law should be changed.
reply