uluyol

15 ms·

121.
by uluyol 4y ago
Well there are definitely cases where you want to render frames and not view them in real time. For example, rendering the final cut of an animated movie.
122.
by uluyol 5y ago
I think it's also aimed at the C++ crowd where generics = static dispatch and interfaces = dynamic dispatch and people try to eek out the extra performance.
123.
by uluyol 5y ago
Some conferences are open access and the materials are freely available. Anything USENIX falls into this category. In my area (networking/distributed/operating systems) the other (non-USENIX) major venues are also open access (SOS
124.
Upcoming code review requirement changes(groups.google.com)
1 points by uluyol 5y ago|0 comments
125.
by uluyol 5y ago
Same thing happens with IO interfaces. https://lwn.net/Articles/805235/
126.
DaisyNFS is a formally verified NFS server(github.com)
2 points by uluyol 5y ago|0 comments
127.
by uluyol 5y ago
I used acme as my main text editor for a few years and a lot of what you say resonates with me. The biggest problem I had with amce wasn't the editor itself but the lack of easy to use integrations. Acme integrates well with external p
128.
by uluyol 5y ago
"Not worth it, at least right now" is a perfectly valid answer to this question. Before doing X, someone needs to estimate the benefits of doing it / the costs of not doing it. If there's no clear reason to do X, but the
129.
by uluyol 5y ago
Would something like this be accepted in open source projects that are not significantly driven by Google? E.g. would Linux accept code for drivers / architectures that are not available to the public? I'm genuinely curious.
130.
by uluyol 5y ago
The point is that the amount of silicon that is wasted when there is a defect is amplified more with larger chips. That assumption was more for ease of explanation, but perhaps I could have explained better.
131.
by uluyol 5y ago
One of the reasons, yes. Another benefit of chiplets is mixing and matching chips made on different processes or from different manufacturers. AMD uses an IO die manufactured on an older process at another manufacturer (14 or 16nm Global fo
132.
by uluyol 5y ago
See the next sentence: "Of course you can build chips that can tolerate a few defects, but the principle holds."
133.
by uluyol 5y ago
Cost increases super-linearly with size. One reason for this is defects: if a single defect ruins a whole chip, then for a constant number of defects per square inch, you'll get more usable square inches of silicon when you have small
134.
Picking Parameters(imperialviolet.org)
1 points by uluyol 5y ago|0 comments
135.
Announcement: Gcobol(lwn.net)
13 points by uluyol 5y ago|0 comments
136.
by uluyol 5y ago
Go is not quite doing this level of expansion: https://github.com/golang/proposal/blob/master/design/generi... In particular, a single expansion is shared for all pointer types.
137.
by uluyol 5y ago
I'm sure you'll see plenty of comments saying "Go generics lack support for X". Just ignore them — some people seem to have nothing better to do.
138.
by uluyol 5y ago
Can these code changes be automated with a tool? One of the strengths of Go and Java is the ease of writing tools. C++ on the other hand is a mess since it's hard to look through templates (concepts probably help) and macros (good luck
139.
by uluyol 5y ago
Or that they all employ similarly effective techniques.
140.
by uluyol 5y ago
What makes you think Russia is special in this regard?
141.
by uluyol 5y ago
This is also true in C++ templates, and C macros for that matter.
142.
Proposal: Arena: new package providing memory arenas(github.com)
6 points by uluyol 5y ago|0 comments
143.
by uluyol 5y ago
The answer is full disk encryption.
144.
by uluyol 5y ago
I've always enjoyed Ryan Smith's articles, but I think he's been busy with other duties since taking over. I have noticed that the quality of work coming from their junior writers improves over time. I mean just look at Ian.
145.
by uluyol 5y ago
They adapted after Anand and Brian Klug left, so hopefully they can do it again.
146.
by uluyol 5y ago
This feature is pretty niche and really only helps on high thread count workloads (relative to the number of cores). Otherwise you can use per-thread data and it'll be the same, since the number of threads is approximately (or less tha
147.
by uluyol 5y ago
The original use case (I believe) was tcmalloc. Tcmalloc maintained per-thread caches of free memory, but that's wasteful since it increases fragmentation but has no parallelism advantage compared to per-CPU buffers. OTOH, you can&#x27
148.
by uluyol 5y ago
I think most allocators don't care which core you call free on. It's already the case that you can malloc from any core and free on another. Special purpose allocators maybe. Agree that this mechanism overall is rather niche, but
149.
by uluyol 5y ago
Out of curiosity, what are some languages that can automatically switch layout between struct-of-arrays and array-of-structs?
150.
by uluyol 5y ago
Er, when it comes to concurrency, non-determinism is usually cheaper than determinism. As soon as you care about ordering, you almost always have to synchronize, and that has a cost. Austin Clements (of the Go runtime team) wrote a paper th

More ›