B
B
beduin012022-03-24 13:11:19
Search Engine Optimization
beduin01, 2022-03-24 13:11:19

What server config is needed for ML?

It is required to write a server config for the needs of ML. Are there any typical examples of servers that can be taken as a basis?
Of the inputs, only a 100GB database, the data from which is supposed to be fed to ML algorithms. There are no more specifying parameters. You need at least something.

And it's not about the home config, but about the server.

Answer the question

In order to leave comments, you need to log in

5 answer(s)
O
Oleg, 2016-07-03
@politon

Several options:
1. View here https://developers.google.com/speed/pagespeed/insi... It is natural to make recommendations.
2. Change provider
3. Throw out hardware that slows down

X
xmoonlight, 2016-07-03
@xmoonlight

Convert to SPA, use GZIP and caching on both sides.

D
Dmitry, 2016-07-04
@soledar10

1. Optimize images
2. Remove rendering-blocking JavaScript and CSS from the top of the page
- in head - include one common style file
- down before </body>- move all scripts

D
Dmitry Dart, 2016-11-08
@gobananas

If images have transparency, then PNG, if a couple of colors, then GIF for the rest JPG
And so the list can be large: Use icon fonts, a minimum of pictures, although the fonts are also heavy. Do not produce CSS files, but make one, minify it. Sprites can be used instead of pictures. A special JS script so that images in the content are loaded only at the moment when they fall into the visible area of ​​the screens. In general, there are almost an infinite number of life hacks))

R
rPman, 2022-03-24
@rPman

There is not enough information to make a correct decision.
machine learning is, first of all, data processing, your 100GB can be ready-made data, in a format suitable for the algorithms used, or maybe data that still needs to be processed with complex capacious algorithms and the intermediate result should still be recorded, in an order of magnitude larger volume.
For example, it is easy to get intermediate data a hundred times larger from a video stream, on the other hand, if there is enough processor (not a video card), this operation can be carried out on the fly
. And yes, more memory, better processor, cooler video card .. the more the better, but in such conditions there is no way to save

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question