Cluster for parallel processing of data from laptops. How to implement?

Question

Clusters

Good day to all. I was very interested in such a topic as parallel data processing. I heard about it for a long time, but the opportunity to realize this myself presented itself only now.
Specifically what interests me.
I'm interested in creating a cluster of several laptops ( HP Pavilion g7 and Lenovo ) and possibly PCs ( years of the 2000s ) of different companies and with different architectures (OS does not matter, you can change it) to improve game performance . In particular DayZ SA and ARMA. I also do visualization and modeling. Here, too, extra performance does not hurt.
And so:
1)Is it possible to implement such a cluster, from laptops and PCs, which will be able to run games at maximum speed without lags. It's the same with visualization.
2) If this is possible, what will be required for its implementation (what software, what equipment)? What are the nuances, aspects?
3) If this is not possible, please write how to make it possible?
ZY: I don’t have much money for a new super-efficient PC for several hundred thousand, but I want something like that :) I hope you can help with this issue.
Well, if I get everything, it will be a very useful practical experience for me.

Answer 1

For distribution of calculations - it is possible. With your wishes - no.
The bandwidth (and delays) of external interfaces is not enough for real-time data processing (this is about games).
Visualization and modeling can theoretically be broken into separate pieces and calculated on different devices, but here I can’t tell you about specific technologies.

Answer 2

It's unlikely to come out. The cluster, in fact, becomes beneficial when you can split the calculations. In this case, it is practically impossible, because. transferring from one computer to others will take an unreasonably long time.
PS You can completely modify existing computers for more power. As for the maximum speed - I don’t know, but the average, in theory, should pull

Answer 3

For games, no. For rendering - only if your render engine supports it. Popular V-Ray and Corona can. The profit, however, will be doubtful - just for example, i3-350 vs i3-6100 - the difference is ~ 4 times. accordingly, if the old is something very old, then the difference can be 10k1 even to i3.

Answer 4

If this is not possible, please write how to make it possible?
earn money for a new computer for igor and not **** Moscow :)
for "parallel data processing" you also need to write a program accordingly. eg MPI to use.

Answer 5

The only way I know how to seriously upgrade a laptop for games is to connect an external video card to the pci-e connector, if one is present. Moreover, since most often it is pci-e x1, the increase will be only for very weak laptops.
Distributed computing is possible and used. For example matlab, ansys, etc. support this possibility. If you write models yourself in one of the languages, then you implement this support yourself accordingly. I somehow had a similar task, but the model itself could not be parallelized, and it was not very large, I just run it on several computers with different input parameters.

Answer 6

Well, I have about the same set - that's just even tench distros with a difference of 3 years are already very different in libraries ... So only C and build statically. I generally scored, but then I remembered that I had a bunch of unfinished toys on the doom engine.