aey
aey on Hacker News
Would the sky also be filled with infinite black holes?
All that helps, but if you look at the die you are going to see a massive 12MB L2 with a bunch of routing to all the L1s. That’s where the majority of the cost is going vs any other chip.
It’s an L2, amd zen has 2x16mb L3s, with a wacky point of unification. L3 is muuuuch slower than L2.
Apple’s L2 is huge and is the PoU for all the cores. I wouldn’t be surprised if it has direct routing to all the L1s since they are so small. 5nm doesn’t hurt.
Because of the 12MB L2 cache as the point of unification. 865 snapdragon has a 4mb L3. Same as my recent intel MacBook.
Apple can use at least as much more die as amd/qcom/intel’s profit margin.
A lot easier to lug my 16lb road carbon bike up and down stairs then a 30lb commuter
Google android app search is just as bad. That’s because “store search” is a totally different problem than indexing the web.
Even indexing the web is a different problem now than in 2000, users expect the search results that are SEO optimized for google.
Nvidia buys ARM, Apple builds its own GPU, and the dance goes on.
These articles on HN always remind me how much more work crypto has to do in terms of education to the general technical public. Crypto Twitter is a recursive bubble.
2015 is the best MacBook! Touchbar is terrible, but I applaud them for trying. They are trying imagine a perfect touch interface.
The butterfly keyboard is such a classic Apple thing todo.
To push form well beyond the limits of current manufacturing capabilities just because of an ideology.
How many big companies try something that ambitious?
I wrote a Makefile that applied a bunch of imagemagick filters to thousands of photos. Married with two kids now :)
Posted by
aey on Saturday August 08, 2020 @07:02PM
from the ask-hn dept.
Hey folks! Solana (https://github.com/solana-labs/solana) is hiring for a couple rare skill sets.* low latency UDP networking* berkeley packet filter* llvm/GPU/SIMD acceleration* gossip, epidemic protocols, CRDTsIf this sounds interesting please email me at [email protected]. We have been remote and open source from day one.
> It's entirely hard to believe this considering the obscene amounts of profits that US companies make by just "designing in california". Our whole era is enabled and defined by cheap chinese labor.
Labor is fungible, if it wasn't in China it would be somewhere else.
Despite all this, US tech firms still have a harder time operating in China.
> A meta point I'd like to add is that currently 10 % of the earths population in the "Westosphere" controls 60 % of the worlds wealth.
True wealth is really hard to measure. Stocks and dollars and euros are not wealth. They are mediums of exchange no different then dogecoin.
When I am hungry, I can’t eat them directly, I can only trade them for food. And the amount of food I will get varies based on supply and demand.
Money printer go brrrr, means invest in housing. It’s the only investment that gives you 500k of tax free profit.
Also, you know your city, the local market and what the possibilities are much deeper than any corporation.
Pivot to the pacific was an Obama initiative that was loooong in the making.
It doesn’t matter who drives the train, the tracks are already laid.
P != NP will be proven before we have a breakeven fusion.
(BD -> Eng Stress) ^ pre PMF
BD needs to generate enough sales to keep the engineers stressed out. When you don't have that, you end up with politics. OR the company is pre PMF, and everyone needs to operate in that mode.
Actually both! Arithmetic coding works over any kind of predictor.
Compression is AI. It’s never going to be “all” figured out.
Bitcoin is a fascinating thing. People think this is a bug, but I think it's a feature. The cost of physical settlement and risk is measurable and transparent to everyone involved.
If you think that a bank transfer has less risk, you are not looking in all the right places :)
Not quite. You can bribe the miners with your old private key if you act fast enough :)
We actually build a whole public blockchain around VDFs.
Although we are using it as a clock in the network before consensus to do the same kind of optimization tricks that cellular networks do like Time Division Multiple Access.
Check it out,
https://solanabeach.io, a global permissionless cluster synchronized in 400ms :)
D was closed source when it was in the spotlight and missed its opportunity to gain mind share.
Is there a big runtime that’s linked to support all the language features? Like what is the fully statically linked size of
int main(void) {
return 0;
}
If it’s just providing free/malloc symbols, that’s wonderful!
Super cool! Any idea how big the runtime is? Is it easy to run a libc free version? I honestly feel like the embedded/firmware slice of the stack desperately needs a next gen language.
Afaik Germany is not part of 5 eyes or 9 eyes. There seems to be a tiny bit of ww2 era resentment there.
Rust has been really great at incremental language upgrades because of clippy. Stuff get deprecated, clippy issues warnings, buy the time support is dropped your code has been patched.
It feels like we are in some new weird software engineering world where the code is only alive while its actively maintained and worked on. The libraries that are shipped in the OS seem as constant and fixed in time as x86, basically completely abstracted by the constantly evolving software on top of it.