dooglius

49 ms·

721.
by dooglius 7y ago
In mice
722.
by dooglius 7y ago
Well, the scheduler has the same problem when the preemption timer goes off, doesn't it?
723.
by dooglius 7y ago
Well, it's a hint (but then so is `nice`), so it doesn't have any strict postcondition. The meaning of the hint would be that the process won't make good use of CPU cycles at present, and it would be preferable to yield to an
724.
by dooglius 7y ago
sched_yield does seem like it should have a pretty obvious meaning: do whatever you would do if the preemption timer went off right now. Whether it's a good idea or not, I don't see why it's considered "undefined".
725.
by dooglius 7y ago
Well, antioxidants prevent damage to cells, so while that's normally helpful, it directly works against things like Chemotherapy which are actively trying to kill cells. If I understand correctly, it looks like this is just evidence co
726.
by dooglius 7y ago
Generally RDMA implies kernel bypass as the point is to write into a process' address space efficiently, but you could conceivably implement RDMA in a way that goes through the kernel, e.g. by using a conventional NIC and doing a copy
727.
by dooglius 7y ago
It would be expensive as the DMA's MMU would now have to maintain a large TLB to be performant, and there would be a startup overhead as new page information is loaded.
728.
by dooglius 7y ago
Well, the first three words of the letter are "Dear President Trump"
729.
by dooglius 7y ago
It appears that it has been updated to use bittorrent as a temporary solution: https://colab.research.google.com/github/nickwalton/AIDungeo...
730.
by dooglius 7y ago
My understanding is that each user runs a separate Colab instance with its own hardware, it's not the kind of thing you could replicate easily with one server and a couple GPUs.
731.
by dooglius 7y ago
Granted, I have not tutored new students. I am making the assumption that someone encountering this would be familiar with basic algebraic manipulations, and the solutions to x^2=a. This is all that is necessary to justify completing the sq
732.
by dooglius 7y ago
I disagree that this proof is better pedagogically; it assumes the quadratic case of the Fundamental Theorem of Algebra (the correspondence between factorization and solutions), which at this stage would have to be taken on faith by student
733.
by dooglius 7y ago
Speaking from personal experience, I usually see several results for any search. Granted, there's a big selection bias there, but 20% seems way too small.
734.
by dooglius 7y ago
There is a huge amount of duplication there (i.e. books that have many scans), I wonder if it would be better to tackle that versus doing a straight backup.
735.
by dooglius 7y ago
I wish we could switch to a system like Tor's onion services where each URL has an embedded key, it would solve so many problems!
736.
by dooglius 7y ago
Disagree in this case, it's about exposing a simple abstraction, which may mean a simple implementation, or may mean a complex one, depending on the impedance mismatch with what's going on under the hood.
737.
by dooglius 7y ago
The problem is that userland has to be really clever because of overcommit... that's what the whole thread is about! The reason people want complex things like fsync and barriers is that the OOM killer has normalized the bad idea that
738.
by dooglius 7y ago
What do you think the "insoluble" problem is here?
739.
by dooglius 7y ago
How many students do teachers have on average? $37 per month definitely does not seem significant enough to explain the effect found.
740.
by dooglius 7y ago
According to the paper, many states are still using VAM so it bears repeating.
741.
by dooglius 7y ago
They could just carve out an exception for themselves
742.
by dooglius 7y ago
> It's not that learning anbn is not a toy problem. It is. But it's a toy problem for ILP. I don't see why the distinction you are making matters. > Neural nets can't solve it. They can, it takes additional trainin
743.
by dooglius 7y ago
The fact that logic programming is still looking at "languages" like a^nb^n while neural nets can generate at least somewhat sensible English only serves to underscore how utterly the field has failed. If you think that such simpl
744.
by dooglius 7y ago
True, but for reliable programs, generally a separate tool will be used to statically ensure this can't happen (e.g. checkstack.pl for the Linux kernel), and these existing tools wouldn't know to parse this a new alternative alloc
745.
by dooglius 7y ago
Except functions like qsort which are in the standard library wouldn't have such functions in the signature, so the code in the example still wouldn't work.
746.
by dooglius 7y ago
No, the problem is that such a function is impossible to implement; any function that allocates memory can fail when no memory is available, and if the call fails there is no recourse in order to proceed per the spec. The compiler could eas
747.
by dooglius 7y ago
If you like this kind of puzzle, check out http://faculty.uml.edu/jpropp/srat-Q.txt
748.
by dooglius 7y ago
Does anyone else find it ironic that this is hosted on GitHub?
749.
by dooglius 7y ago
The article mentions a companion paper about how this was used to speed up physics computation, so it sounds useful at least in that domain.
750.
by dooglius 7y ago
See also: https://en.m.wikipedia.org/wiki/F._D._C._Willard I don't see anything wrong with this if the parents want to share credit with their kids. In any case, there are many cases where authorship is shared wit

More ›