kbwt

9 ms·

31.
by kbwt 8y ago
Sure, but this patch is only useful for single-purpose, HFT-like workloads. Batched io_submit on sockets would be useful for any application sending many small packets to lots of clients, such as game servers.
32.
by kbwt 8y ago
There's io_submit which lets you batch IOs on any number of fds. It appears to lack socket support, but I don't see any reason why that couldn't be added.
33.
by kbwt 8y ago
I'm relatively sure ICOP still requires two syscalls per IO.
34.
by kbwt 8y ago
The only thing this really improves is latency, because you can always batch more I/O into a single syscall.
35.
by kbwt 8y ago
So you need to dedicate two cores, making them unavailable for general-purpose tasks in order to reduce the syscall overhead for a single application?
36.
by kbwt 8y ago
Worth more than the entirety of the VAT? I think not.
37.
by kbwt 8y ago
Steam certainly won't internationalize your game for you.
38.
by kbwt 8y ago
> One would imagine that there is some fixed cost for adding new titles and maintaining steam store. I would imagine that it's mostly automated, similar to the Google Play store. If anything, it's the developer bearing this cos
39.
by kbwt 8y ago
Can't something like Stripe + Braintree provide the same at a fraction of the cost?
40.
by kbwt 8y ago
> Any game needs to be 25% better Actually, 42.86% better with the 30% cut, and 33.33% better with the 25% cut, and 25% better with the 20% cut.
41.
by kbwt 8y ago
Also idealo.
42.
by kbwt 8y ago
Not OP but you could write a whole book on "what exactly does dx (and its variants) mean and why can you sometimes manipulate it algebraically?"
43.
by kbwt 8y ago
If you try to implement a class in C, you always end up with duplicated cleanup code between your "constructor" and your "destructor". Being able to jump from the constructor to the corresponding part of the destructor w
44.
by kbwt 8y ago
And yet this proposal still doesn't let you implement zero-overhead objects-as-coroutines. I'd like to be able to create a coroutine that initializes some state ("constructor"), yields back to the caller, and lets you ca
45.
by kbwt 8y ago
Well, yeah. If you are memory bandwidth-constrained it's a bad idea to go off-chip. But for ray-tracing, what does it really matter? We are already assuming that you will wait a full memory fetch cycle to get the next node's child
46.
by kbwt 8y ago
Citation needed. There is great tooling to pin down incorrect memory accesses when you are using the system allocator (valgrind, clang sanitizers). You're truly on your own if you access logically repurposed memory within a persistent
47.
by kbwt 8y ago
> The switching isn't free. It absolutely is, on current GPUs. Think of it like a larger-scale version of SMT (Intel's hyperthreading). GPUs are able to do this because they execute instructions in-order and do not need to trac
48.
by kbwt 8y ago
> It's annoying that 50% of the die is allocated to hardware that requires feature-specific implementations. That's the future. While we may be able to cram more transistors onto "7nm" chips, only a tiny fraction of t
49.
by kbwt 8y ago
But GPU cores are already king at latency hiding. They can run hundreds of threads doing pointer chasing, switching between them round-robin as the memory reads complete.
50.
by kbwt 8y ago
Interesting thought, but I'm not convinced. Constructing a robust BSP tree is very expensive, because you need exact rational arithmetic and the number of bits grows quite large. You wouldn't want to do that at runtime, so you cou
51.
by kbwt 8y ago
> Any situation where C would segfault immediately, Rust would also segfault immediately. For equivalent code, yes. The borrow checker can however compel you to write code that would not immediately panic in Rust, when the C code you wou
52.
by kbwt 8y ago
Thanks for the response. I wasn't trying to speculate on your intentions for publishing the blog post. > There's a world of difference between "logic error and/or panic" and "undefined behavior". Is is
53.
by kbwt 8y ago
I thought this was going to be a response to Jonathan Blow's video about how doing your own memory management is effectively turning off the borrow checker: https://www.youtube.com/watch?v=4t1K66dMhWk The takeaway bein
54.
by kbwt 8y ago
Rotation matrices by themselves do not suffer from gimbal lock. I think you meant Euler angle representations with the rotations always applied in a consistent order around the pitch/yaw/roll axes.
55.
by kbwt 8y ago
They would build a really nice app store and then it would sit empty because it is not preinstalled and the network effects of Google Play make it pretty much impossible to gain any traction for an alternative.
56.
by kbwt 8y ago
OT: Is asking for money to opt out of tracking and personalized ads actually GDPR-compliant?
57.
by kbwt 8y ago
> The planned obsolence - unreplaceable battery, glued screen, glassback etc mean phones break all the time. Don't forget planned insecurity two years after release through unpatched software vulnerabilities.
58.
by kbwt 8y ago
> most games today are built using Unity3D, Unreal Engine etc On PC: Maybe by number of games, but not by the number of players. (I count Fortnite as an outlier because it's technically not built on a third-party engine)
59.
by kbwt 8y ago
That's eerie, the same thing happened to me.
60.
by kbwt 8y ago
The native code (RedPhone remnants?) did not inspire confidence last I looked at it.

More ›