@DONoticing I mean, if the decision was unpopular enough, we’d just enact a new constitutional amendment overriding the ruling.
It’s the same for the executive. Practically, who can stop them?
Checks and balances rely on everyone respecting the constitutional power structure.
Senate rejecting nominees, blocking nominees like Garland, impeaching judges, removing judges, impeaching even a Supreme Court justice (Chase), resizing the Court (5/6/7/9/10 seats), nine isn’t in the Constitution, court-packing, abolishing lower courts, Judiciary Act of 1802, jurisdiction-stripping, Exceptions Clause, Ex parte McCardle, overriding rulings by rewriting statutes, Lilly Ledbetter Fair Pay Act, RFRA overriding Employment Division v. Smith, controlling the judiciary’s budget, appropriations power, canceling a Supreme Court term (1802), president picking every nominee, presidential pardons wiping out convictions, executive refusing to enforce rulings, courts have no army, “neither force nor will but merely judgment,” constitutional amendments overriding decisions, 11th Amendment, 14th Amendment, 16th Amendment, 26th Amendment, states calling an amendment convention, good-behavior tenure, standing doctrine, mootness doctrine, political-question doctrine, self-imposed justiciability limits
@WenHop21505@incxlchud@DocBorelli@jarrichvdv Whether it would’ve moved back into its lane or not quite literally doesn’t matter
It pulling crap like this is critical, should be pointed out at such, and fixed with utmost priority. No normal person in this situation would feel safe or that the car was “doing the right thing”
@AgileJebrim@adamthesherwood Just need to get builders down here to take building envelopes, pressure management and attic ventilation more seriously. Too many moisture and IAQ issues in houses down here in the Texas coast.
@zeuxcg I saw that post, and immediately thought about how that doesn’t apply to something like meshoptimizer at all.
I’m immediately grabbing the newest meshoptimizer when a new version is released!
There is a break even point with solar, but nuclear is the best long term option IMO.
It’s funny you say that because the Texas power market is completely deregulated and hardly subsidized yet we still lead in renewables. The open market has decided it’s at least somewhat a worthwhile investment. If what you’re saying was true we should see next to zero wind/solar.
@AgileJebrim@FUZxxl Yeah obviously not ideal if you’re powering your heat pump with a mostly NG grid 😂
But Texas has a mix of renewables so may still be a net positive - also if you have your own solar panels it’s definitely more efficient.
The funny thing about this is natural gas can run into issues as well if it gets that cold, i.e the 2021 freeze where our infra wasn’t properly designed for the worst case scenario.
Heat pumps have come a long way since the 90s, and most include electrical resistance based backups.
Of course there’s always the question of where the electricity comes from to power the heat pump, as that determines overall efficiency. But if you can accomplish the same goal with less joules of energy consumed, why not?
This is a codebase-specific anomaly anyways. Most of these bugs come from untrusted data exchanges between user space and kernel code, a microkernel would fix that just as well so the Rust transition lacks uniqueness and would need a cost-benefit analysis against the alternative. But the whole C vs Rust debate is kind of a joke anyway, since neither language actually maxes out the hardware. If you actually care about CPU or GPU performance, you're using something like ISPC or GLSL, not arguing about memory safety in scalar languages.
In our codebase we only create new permutations if it's verified with profiler data that doing so would result in measurable gains.
Whereas a lot of other engines default to this "specialize everything" approach, which actually ends up pessimizing performance and hurting user experience.
@MCH2024@SebAaltonen The more unique PSOs you have, the more graphics pipeline binds you have to issue which means higher CPU utilization. Not the other way around.
Having divergence within a wave is a completely independent performance characteristic of register pressure. The two can be covariate but are not causal of one another.
Nobody is saying you should compile shaders at runtime. We're saying that the current toolset encourages thousands of individual PSO objects, and that should be avoided.
"the PSO explosion enables more optimal GPU utilization per draw call" Not if you are i-cache or CPU bind rate limited. Thousands of PSOs with few draw calls each will quickly turn into a CPU bottleneck. You are not solving the problem you think you are with this.
"smaller shaders or reducing PSO permutations is a regression" Back your claim up with empirical data in a well controlled environment. Otherwise, how are justifying building infrastructure.
@MCH2024@SebAaltonen Increasing the number of unique PSOs will only make this problem worse. Binding PSOs on the GPU still hasn't been widely adopted yet.
It absolutely can be scaled and the vast majority of graphics applications should not be shipping with thousands of PSO permutations. The occasions where you would need such are very limited.
The majority of PSO permutations can just use wave coherent branches which are basically free given you aren't running into register issues.