Opinion practice correspondence chessplayer:
Using a supercomputer - what is it? Myth or Reality?! The benefit or misleading?!
I during the three days involved in extreme tests 40 engines to the limit of equipment and user. Conducted different methods of analysis in the shell Aquarium. The final analysis of the work certainly deserves attention and further study of this option.
Supercomputer provides new opportunities, in fact represents the analytical laboratory, where much depends on the user. Noticeably wider opportunities disclosed Aquarium, its effective use.
Incredibly easy to make the connection of the engine, just the touch of a button and have all the required settings.
When using a supercomputer running engines are constantly visible in the aquarium, and it is important that the work and manage you have to yourself.
What good and that problems arise and how to avoid them?!
I during this test significantly deepen and expand their projects, find new opportunities in the individual opening variations. When analyzing the current tournament position is necessary to continue, while spending much less time. Acquired experience with a large number of engines in the Aquarium encouraged to seek more acceptable options, as well as to study possibilities not previously claimed. There is hardly a player that will negatively to treat this. Is it just lazy and envious.
Problems arise and need to know they to use the supercomputer.
1. Bad Internet. I with this directly encountered, and only he is not allowed to experience the rave emotions. I have a bad internet - modem connection. Although, emphasize, the overall impression is positive. There are questions about the use of modern equipment that was good bandwidth and high speed external network. This will not "crash" of engines and they are all currently available.
2. Problems Aquarium. And I had to decided arised troubles. But, most importantly, they proceeded due to poor internet. When an internal network connection to the Aquarium of such problems did not arise. Thus, it was found that the shell is not the main cause of failures in the supercomputer, namely bad internet culprit "of all evils."
3. The inability or rather unwillingness to improve of user. This reason may be, therefore, "pessimists" Aquarium advise first understand their intentions to work in the shell.
The correctness of the choice will often depend on the chess skills.
Who does not have problems under item №1 - can safely take advantage of opportunities!
Project organizers supercomputer make a connection to the Internet and "not very good" to its aquarium on 50 engines engine + IA. Besides, now experiencing a new way to connect to users with bad internet. So what is the future of this project.
Since my problems were of them will not leave, I would like to make your own notes on this subject.
The best option when working in a project with a lot of problems (I had 100-odd thousand). They can generate the most IDeA of exhibiting a large number of roots or send to interested parties from the IDeA database. This project can be somehow debut or very problematic position. But we must not forget that control the "quality" of work remains to be late and be sure to polish the analysis.
Not a bad option - working in two projects. They are easy to control, load engines is constant and failures should not be.
When working in several projects (tried to 5-10) need to constantly adjust the process. Bad if minimaxes intersect. And constant rearrangements-change also do not add to stability. There may be some "downtime" in the analysis. But it's better to let them be, than had bugs. But here, a large effect can be achieved, but the user is "in the soap." If the trees are small in size, the effect will be much higher, since minimaxes will pass very quickly.
When your Aquarium hang (by the way, in the background analysis continues!), it is desirable to shut the engines work, and then squeeze the projects - to restart the program. You can delete data IA (and the contents of folders infinite_analysis.ldb_data infinite_analysis $ 1.ldb_data).
Perhaps there is the influence of large trees, while during deceleration analysis are not talking to them. Recording goes on infinite_analysis file, and it really matter.
On your computer, it is necessary to leave one core is not busy.
Using more than 50 engines considerably hinders work.
I emphasize that in a good online mode can be a little different. Those problems that I listed just will not.
Using a supercomputer can almost all players - correspondence chessplayers, professional chessplayers, coaches and fans of analysis. The main thing - to set reasonable goals for this.
We hope for developers and programmers to improve the Aquarium shell and look forward to the new version!
Original text -
http://rybkaforum.net/cgi-bin/rybkaforu ... pid=560232