dgacmu

42 ms·

841.
by dgacmu 3y ago
The phrase you're looking for is "embodied energy", and the answer is first, that solar generates a net return on energy investment in under 10 years, and often in much much less than that, with a panel lifetime of over 30 ye
842.
by dgacmu 3y ago
It was one of the most insightful comments in this thread. Note also he was a director, not the author's manager. That's a very different thing - he has a view of this as a strategic business decision, not as a personnel issue. I
843.
by dgacmu 3y ago
It is a strange twist that Google, Microsoft, and Amazon are some of the best places to work on aspects of software that many people don't care about enough to pay for. A 1% improvement in the memory wasted by malloc is a cool thing th
844.
by dgacmu 3y ago
*you're (I'm a grammar bot)
845.
by dgacmu 3y ago
It's just the "click to favorite this repo" mechanism. It lets people track repos more easily ("see updates to started repos") and it's become a social gaming mechanism ("the most-starred deep learning cry
846.
by dgacmu 3y ago
Could be. I'm definitely a 90s-and-later unix kid. I wonder when -0 was added to `find` Edit: The quote is actually from the Unix haters handbook, published in 1994. I don't remember too many horrors with spaces in filenames from
847.
by dgacmu 3y ago
This is generally not true. Most unix utilities handle filenames with spaces just fine. It's mostly at the shell where people get mixed up about quoting spaces - so while it _appears_ that the problem is with the utility, it's rea
848.
by dgacmu 3y ago
I love that this discussion is rapidly headed to sticking an XO on the circuit suppling current and frequency to control the old digital clock that lacks an XO. Hackery at its finest.
849.
by dgacmu 3y ago
Ballpark: depends how you want to depreciate servers but under a penny. They're not charging for cost; they're charging ad revenue replacement for users using apps. It's pretty clear in the way the Apollo costs would be about
850.
by dgacmu 3y ago
In case anyone wants a quick tl;dr: as the length of a program (expressed as the number of states in a Turing machine) grows, the relative fraction of undecidable programs decreases. In other words, "most" (approaching "nearl
851.
by dgacmu 3y ago
Since this comment seems the be getting down voted, it might benefit from context: Activ8te is Niels Provos (author of this article)'s EDM pseudonym, where he's making music with cyber security themes: > To address this scarcit
852.
by dgacmu 3y ago
They're determining safety and tolerability, not efficacy. For this, they just need adequately compensated trial volunteers who take one of several different doses of the medicine and track/report symptoms, have some doctor visits
853.
by dgacmu 3y ago
That's not really a proof that we can't get a nice thing, though. SAT is NP-complete, but here we are talking about it. There may be ways to use parallelism at a higher level. And even if there isn't an alternative to unit pr
854.
by dgacmu 3y ago
I think it's helpful to categorize the things that go into an ML accelerator as those that are big picture architectural - things like memory bandwidth and sizes, support for big operations like transposition, etc., -- and those that a
855.
by dgacmu 3y ago
Mostly you need to be able to stash intermediate products computed in the forward phase so that you can access them in the backward phase. This requires more memory, more memory bandwidth, more transpose, and also, training usually operates
856.
by dgacmu 3y ago
Not really. This is needlessly cynical. The transition to college is a big one for many students, with a whole lot of "it depends". At a hard school, students are put in a situation where they might have to manage time and learn t
857.
by dgacmu 3y ago
There is no "you". There is a text stream that is being completed with maximum likelihood. One way to imagine it is that there are a lot of documents that have things like "if you are in a lightning storm, you should ...&quot
858.
by dgacmu 3y ago
You'd be surprised. There are some workloads - for me, it's geospatial data - where bzip2 clobbers all of the alternatives.
859.
by dgacmu 3y ago
It's popular in news articles to report the maximum possible sentence for a crime, but this isn't particularly informative. The federal sentencing guidelines usually give a more clear picture, and my -very much not a lawyer much l
860.
by dgacmu 3y ago
With many architectural caches, it's typically valuable to think of miss rate instead. A 5% miss rate is very high if the miss path is 20x the cost of the hit path - misses account for 1/2 of the execution time, so reducing that t
861.
by dgacmu 3y ago
You'd be surprised at the variation present in complex assignments implemented by non-expert students. It turns out that doing things in buggy and randomly wrong or nonstandard ways leads to things looking more unique than they would i
862.
by dgacmu 3y ago
We're a one car family for two working parents and two kids, living in Pittsburgh (in the city, not the burbs). We make it work but there's a bit of Uber involved sometimes, and some compromises. Generally worth it - cars are both
863.
by dgacmu 3y ago
Without weighing in on exactly if it is or isn't a "language", I think VHDL/verilog absolutely have a place in the list and are a fantastic thing for (some, particularly lower-level interested) programmers to learn preci
864.
by dgacmu 3y ago
To GP, I'm sorry your family isn't eligible yet. Pfizer has an in-progress vaccine for which they had good phase 3 results in both older adults and expectant mothers (looking at the outcomes for the infants once born). Their FDA a
865.
by dgacmu 3y ago
It _has_ been settled in court. Please read the article I linked, or one of dozens of others on the subject. It's been litigated for print, for photos, for magazines, for film. The distinction that you are looking for is between materi
866.
by dgacmu 3y ago
Could work. I don't know enough about the specifics of experiment design for sleep research. You could also try things like transparent masks + lies ("they filter specific wavelengths"), etc. Studying people is awfully hard.
867.
by dgacmu 3y ago
The placebo effect is really powerful. When you see a study like this, one of your first questions should be "how can I tell whether or not this is just placebo?"
868.
by dgacmu 3y ago
> but watching porn is not a right if any kind That's not really true, depending on how you define "porn". There are very strong first amendment rights around producing and consuming all sorts of potentially objectionable
869.
by dgacmu 3y ago
The README indicates: The base model checkpoint is licensed under the Creative Commons license (CC BY-SA-4.0). Under the license, you must give credit to Replit, provide a link to the license, and indicate if changes were made. You may do s
870.
by dgacmu 4y ago
Thank you. I was just yelling about that part. Science literacy is so lacking.

More ›