thesz

52 ms·

721.
by thesz 4y ago
I play Serious Sam to better focus on other tasks. I do so because I cannot walk long hours as I did before for exactly same purpose. If you ask me, I'd better walk. But I can't. PS Walking releases brain-derived neurotrophic fact
722.
by thesz 4y ago
You should credit your parents with allowing you to have so much free time and good nutrition.
723.
by thesz 4y ago
These compute units are usually sliced - they can perform either four FP32 multiples or one FP64 multiply on the same die part. This trick was done as long ago as PA-RISC was developed, from what I remember it was HP who introduced sliced A
724.
by thesz 4y ago
Your comment is full of good and deep analysis and it uncovers a deep flaw in the Copilot: Copilot helps create new code . All the problems and confusions mentioned above is due to this concrete inherent property of the Copilot. If Copilot
725.
by thesz 4y ago
It is extremely relevant for ML. I am familiarizing myself with recurrent neural networks and getting them trained online is a pain - I get NaNs all the time except for very small learning rates that actually prevent my networks to learn an
726.
by thesz 4y ago
I was not able to finish Blindsight. It is too pretentious for me, it ignores what state of art was at the time. Let me quote an old comment [1] of mine: Do you know that current 1sqkm radio telescopes produce data that can't be proce
727.
by thesz 4y ago
Define productivity first. Also, even if one does not read books (or especially papers), he still will ask friends and colleagues about programming-related things. And those friends and colleagues may be avid readers of books (or especially
728.
by thesz 4y ago
I think you are talking about second Prey game, not the first. First Prey game was unusual first person shooter, very interesting one. I haven't played second game.
729.
by thesz 4y ago
Speeding up databases, for one example.
730.
by thesz 4y ago
There are protocols to merge and squashing was required in two of four github-based teams I've worked with. The choice is done for you. It is not yours or theirs often enough.
731.
by thesz 4y ago
When commit is squashed and rebased upon main branch there is no pieces to bisect. Having private branches also does not always work. For example, there was a purge of unnecessary branches in our main repository. If I use a fork of main rep
732.
by thesz 4y ago
The sad reality is that testing and reviewing will only get us to defect density of one defect per 1000 lines of code changed - a nice round number for a clarity of discussion (and we are lucky here, see below). E.g., if patch changes or ad
733.
by thesz 4y ago
They actually right from my perspective. It so happen that I usually do commit squashed changes to many files over code base with long and heavy legacy. I am afraid that squashed changes lose their utility for bisection search for a bug. Ye
734.
by thesz 4y ago
Python is fairly distant from actual math. Haskell, for one example, is much closer, the definitions in Haskell are very math-y: let this = that in what; f x = z where z = g x y. Even list comprehension in Haskell is much more like the actu
735.
by thesz 4y ago
Languages are part of mathematics. Please read this fascinating book: https://dickgrune.com/Books/PTAPG_1st_Edition/ Parsing, generating phrases, understanding (analysis), etc, are all studied in applied math.
736.
by thesz 4y ago
One of the attacks on PBFT protocol is to split network in half, send block to one half, do not send anything to other half and the proposer wents offline and restore connectivity between the rest of the nodes. In that case, without any sta
737.
by thesz 4y ago
As a Russian programmer, worked in companies big and small, I cannot say this is even remotely right. All companies I worked with really cared about cleanness of the origin of the code.
738.
by thesz 4y ago
You mentioned only the rewrite of history, but there is at least one other attack on a network - a coordinated refusal of blocks with transaction(s) attacker deem ungood. This way, some person(s) will be denied the right to transact. A dire
739.
by thesz 4y ago
The hardness works both ways - these with stake can decline to process what they deem ungood and there's nothing one can do about it. You see "hardness of Sybil attack", I see "rule of unelected comittee". The very
740.
by thesz 4y ago
According to Louie Simmons, strength athlete at top shape can lose about 20% of 1RM weight in two weeks of inactivity. Regular people also lose their shape, at slower, but increasiingle faster rate, as they gain shape. It is enough to lose
741.
by thesz 4y ago
C++'s throw lets you postpone handling of erroneous condition. You throw, then sort. You can postpone sorting of causes indefinitely, basically, up to the level where all of context needed is lost. In C, you cannot throw. You have to h
742.
by thesz 4y ago
Ballistic table calculators can be used by insurgents/guerilla fighters and even by regimes you or me personally oppose to. Do we need to hold ballistic table calculators' developers accountable? What if these calculators' de
743.
by thesz 4y ago
And the idea that compression is intelligence also should be mentioned: http://www.hutter1.net/ai/uaibook.htm
744.
by thesz 4y ago
Okay, skip lists are built upon singly-linked lists, but there are still need to lock more than one object to modify. The need to lock multiply objects at once is the potential source of deadlocks.
745.
by thesz 4y ago
Double linked lists would help navigate quicker. But let's discuss why even skip list on even single linked lists is not that easy. The problem with double-linked lists modified in parallel is that one should lock at least two places w
746.
by thesz 4y ago
Please consider COLA: http://people.seas.harvard.edu/~minilek/cs229r/fall13/lec/le... A hierarchy of sorted arrays, with merging when needed. Single access time is O(logN), scan is theoretically the fast
747.
by thesz 4y ago
If you can turn your keys into byte/bit sequences, and most of the time you can, you can use crit bit trees or binary Patricia trees. These are fast structures and support ordered traversal. They also do not rebalance and do not requir
748.
by thesz 4y ago
Oh, you touched my favorite topic of whole dataset training. Take a look at [1] and go straight to the page 8, figure 2(b). [1] http://proceedings.mlr.press/v48/taylor16.pdf The paper talks about whole dataset training
749.
by thesz 4y ago
https://news.ycombinator.com/item?id=19473910 It talks about CSS layout algorithms, but TeX also does layout. The result: what you ask is very hard. The single letter change can result in changes throughout entire document.
750.
by thesz 4y ago
Code from yalsat (stochastic SAT solver) [1] made me learn something two years ago. I can declare an array of some elements and make access to elements statically typed. Same with maps, sets and others. [1] https://github.com&#x2

More ›