ruby – How to efficiently process CPU intensive tasks on the background server

I created an audio processing web application using Rails. The user uploads a song to the website. The song is broken down into individual elements and then modified and recombined.

I am using an open source command line tool that is called from the rail controller.

My problem: the processing takes between 2 and 3 minutes and consumes a lot of memory. The browser is in the loading stage for 2 to 3 minutes. And this is only for 1 request from 1 user. I am using the Amazon ec2 t1.large instance simply is not enough.

I am planning to use background processing, but I don't want several requests to work at the same time.

I want the first request to take 3 minutes. Second request to take 3 + 3 minutes. Third request to take 3 + 3 + 3 minutes. Then the website does not fall.

In addition, the audio file is around 40 mb. Is it a good idea to use amazon s3 bucket? Or simply increase the space on the hard drive and store audio files on my server?

The reason why I don't want to use the amazon s3 bucket is because I don't want to transfer each file again from my ec2 instance to the s3 bucket, since it will be added to the latency.

Eric Matthes ex 7.8 intensive Python course

I have trouble doing an exercise in the book & # 39; Intensive course by Eric Matthes Python & # 39 ;. The exercises are as follows:

7.8 – Snack Bar: Create a list called sandwich_orders and fill it with
The names of several sandwiches. Then create an empty list called
snacks. Scroll through the list of sandwich orders with a loop and
show a message for each order, for example, I have prepared your
Tuna sandwich. As each sandwich is prepared, transfer it
to the list of ready sandwiches. After all the sandwiches are
ready, show a message that lists each sandwich prepared.

sandwich_orders = ('frango', 'peru', 'queijo', 'presunto')
finished_sandwiches = ()

    while finished_sandwiches:
        sandwiches = sandwich_orders.pop()
        finished_sandwiches.append(sandwiches)
        for sandwich in finished_sandwiches:
            print('Preparei seu sanduiche de ' + sandwich)

When I execute the file in vscode I answer this:

PS C:UsersPedro MoraesDesktopstuffExerciciospython> & "C:/Users/Pedro Moraes/AppData/Local/Programs/Python/Python38-32/python.exe" "c:/Users/Pedro Moraes/Desktop/stuff/Exercicios/python/lanchonete.py"

Do not "print" anything on the terminal.

That I have to do?

architecture – System design for intensive computing software

I have a use case that will take data from the front end and transform it with several parameters. Then a random effects model is executed in each transformation and multiple models are constructed

For example: the data has 5 independent variables. For each independent variable, about 3 consecutive transformations must be performed and each one has about 10 possible transformation values. The whole space or combination will be (10 ^ 3) ^ 5. But we will only give a smaller subset of values ​​for transformation and modeling, think of it as something similar to the grid search. Therefore, around 10k models will be built per user.

Now the problem I am facing is how to scale this system to multiple users. The AWS server with 64 gigabytes of RAM and 8 cores is significantly stuck with only two users (since the current system is horrible with a monolithic design).

I am a newbie in most of this, so the solution I can find is:

Divide the server into 3, one for application, preprocessing and modeling. This would separate the load and we can assign instances of intensive computation to modeling alone. If it is an acceptable solution, I would love to know how to scale at low cost by design instead of adding more ec2s and cores (have a docker image for each flask server).

How would you do with the modeling server? Do you treat it as a function and send you API calls with the relevant data and make models one by one or in some type of batches? I hope the WSGI (gunicorn) server can be configured to have multiple workers so that it resolves to run multiple models in parallel.

And mongo db is used to save the results. Now, mongo db apparently writes one at a time, so it seems that there is a bottleneck there (if this fact is incorrect, let me try to get the source)

Would lambda or sagemaker be useful for automatic scaling of the modeling part?

Also, let me know of any design or factual errors here.

Thanks in advance!!

dnd 5e: What is the highest damage production for consistent damage production that does not require intensive use of resources for a level 20 Demon Warlock? (Pact of the chain)

I currently have a third level demon sorcerer (chain pact) and I am trying to optimize my construction strategy at level 20 to obtain as much damage as possible to obtain constant damage that does not require intensive resources.

Including the level progression to achieve the final construction would be a very welcome addition to the answers.

The following are the limitations for compilation:

  • No multi-class
  • no wish ring, lucky sword, Vecna's eye, Vecna's hand, book of vile darkness, tomb of exalted works, covered with many things
  • no events or effects of extremely low probability
  • no arcana discovered
  • no benefits or potions
  • without help from others
  • no preparation time

The following are allowed:

  • Any magic item that is not listed above as a limitation
  • Specific benefits of Graz & # 39; zt (MTF)

What is the intensive course for Hong Kong pop songs? – Advertising, offers – TopGoldForum

What is the intensive course for Hong Kong pop songs? It is the accelerated course of improvisation of original pop music of 20 popular ECMUSIC.
from

What is included in the intensive course? What role does it play? Let's take a look:
from

1, the support of most students! More than one hundred students witnessed successfully! Impromptu hands-free game skills to ensure lifelong use
from

2, individualized teaching, designed for adults, without any musical basis
from

3, the creation of 20 unique rooms in Hong Kong will not delay learning progress, saving 80% of the money and the average time of the practice course
from

4, the first church can already be learned from popular music, and there is an exclusive patent "EC spectrum" for you to choose. If you don't learn, you can quickly learn to sing a song that you like.
from

5, 20 lessons, improvisation, autocanto and autocanto, and can be used for 95% popular songs
from

6, with a success rate of up to 80% in practice, you can sing yourself anytime, anywhere

The first stage: the first pop song, the improvised knot, the intensive course, sings and sings immediately without difficulty
from

Second stage: change the application
from

The third stage: improvised game + own singing

In addition to the left rose with the introduction of ECMUSIC, it will bring four makeshift knots to the accelerated course to learn to play one's own song / solo, and use the utmost sincerity to impress the other half to create an unforgettable surprise, which will definitely touch Your other half. Cash, what are you waiting for? Let's try it!

An intensive course on the essentials of human resources management

https://www.udemy.com/course/essentials-of-human-resource-management/

HRM
New
0.0 (0 ratings)
4 students enrolled
Created by Syed Bilal Hassan Shah
Last update 9/2019
English

[F4LT] An intensive course on the essentials of human resources management

https://www.udemy.com/course/essentials-of-human-resource-management/

HRM
New
0.0 (0 ratings)
4 students enrolled
Created by Syed Bilal Hassan Shah
Last update 9/2019
English

architecture: what tools should I use for long-term intensive 3D simulation?

I know that my project is not a game, but many of the techniques that I will have to use to develop this application will involve many of the tools used in advanced 3D and light rendering.

Specifically, I would like to create an application that can load multiple hardware profiles for CRT. These profiles will be considerably extensive with parameters for region configurations (for example, NTSC-US, PAL, etc.), screen mask specifications, electron beam length, vertical or horizontal retention configurations and many other options. In addition, the application will also take several parameters, such as the exposure time for each frame that generates the CRT output, as well as the direction of the camera, the refraction of the atmosphere, etc. In other words, this will be a fairly complex project, which simulates the internal physics of CRTs and external light emission.

I know this is not a game, but since my intention is to allow a high degree of historical preservation of analog equipment through digital simulation, I need some guidance on what tools to use to write this type of application.

To make this question appropriate for this site, I specifically ask you if you would like to create a 3D game / project, what tools would you use to develop a project similar to mine that could be compilable, upgradeable and useful for decades? come?

Logic: interpretation of the types of intensive identity and the Tarskian universe in the syntactic / classificatory category of a theory of dependent types

Leave $ mathbb {T} $ denotes an intensional MLTT with all the usual type builders. Also, suppose you have a universe $ U $ to the Tarski (defined by nLab here).

Remember the syntactic category $ C ( mathbb {T}) $, objects being well-formed contexts in $ mathbb {T} $ and morphisms are the so-called context morphisms, that is, equivalence classes of terms sequences $ ((f_1, ldots, f_m)) $ such that

begin {array} {c} {x_ {1}: A_ {1}, ldots, x_ {n}: A_ {n} vdash f_ {1}: B_ {1}} \ { vdots} {x_ {1}: A_ {1}, ldots, x_ {n}: A_ {n} vdash f_ {m}: B_ {m} left (f_ {1}, ldots, f_ {m- 1} right)} end {array}

are derivable in $ mathbb {T} $ where $ (f_i) = (g_i) $ when
$$ x_ {1}: A_ {1}, ldots, x_ {n}: A_ {n} vdash f_ {i} equiv g_ {i}: B_ {i} left (f_ {1}, ldots f_ {i-1} right)
$$

is derivable for each $ 1 leq i leq m $. The composition is given by substitution.

There are two ways I am confused about how I am supposed to model this category. $ mathbb {T} $.

  1. I have seen that types of extensional identity can be interpreted as diagonal morphisms in $ C ( mathbb {T}) $. Does this also work for intensive identity types? On the one hand, I have read that intensive identity types are supposed to be interpreted as path space objects, which seem to require a weak factorization system. On the other hand, example 1.2.5 in the article The simple model of univalent fundamentals (after Voevodsky) (arxiv link) states that $ C ( mathbb {T}) $ It has an "obvious" structure for identity types. What would this be? Perhaps it refers to a way of factoring diagonal morphisms?

    (You may need to consider $ C ( mathbb {T}) $ as a $ ( infty, 1) $-category with path space objects that represent intensional identity types to obtain an attached pair $ (Syn ({-}), Lang ({-})) $. But I guess the authors of that article only care about getting a solid interpretation of the syntax of $ mathbb {T} $.)

  2. The same example states that $ C ( mathbb {T}) $ It has an obvious structure for $ U $. I have read that a universe to the Russell should correspond to an object classifier in $ C ( mathbb {T}) $. Is this the case of a universe to the Tarski too? If so, how does this look exactly in $ C ( mathbb {T}) $?

Any clarification on these issues would be greatly appreciated.

networks: most bandwidth intensive workloads with a r610 computer and windows 10

I am in a bandwidth competition with some friends. Who consumes more bandwidth at home earns a free NIC. Therefore, I am looking for the most bandwidth intensive task I could do with a r610 and Windows 10 computer. The r610 has 24 gb of ddr3 ram, 2 SATA drives (I can update the drives to win this competition) 2 xeon cpus , and is virtualized in XenServer. The guest operating system is Windows Server 2019. On the other end I have a Windows 10 Pro computer with a Toshiba 256g. NVMe ssd, an I3 and 24 gb of ddr4 ram. I will connect the 2 computers with 40g infiniband. I hope to eat 10 concerts at least. Any ideas?