A computer for testing and optimising the robot (Expert Advisor). - page 7

New comment
 
Dmitiry Ananiev #:

the setup is described above. The speed is just like on a dedik. You just need to find a good VPN. I just outlined the method. If you like it, invest in the latest processors and memory. But if you buy the latest model of processor and memory and a shitload of memory just for optimisations, it's a senseless waste of money.

If 3-4 rubles for an hour of 8cores+12Gb, then 100K rubles - three years of uninterrupted optimisation on the power of 8cores+12Gb.

The latest CPU+RAM is three times more powerful. I.e. they will be able to do the same number of calculations during a year of continuous optimisation.

It turns out that 100K rubles is the rent of the most advanced hardware for a year of continuous optimisation.

How much does this hardware cost?

 

If you approach it competently, the CPU came out to me in China about 30, + 18 memory top 64gb, and the board 10tr at the promo, and sold the Risen 5700x and everything else to it.

In DNS the CPU cost 47 + 27 memory + 21 board, a bit expensive.

 
lynxntech #:

If you approach it competently, the CPU came out to me in China about 30, + 18 memory top 64gb, and the board 10tr on promotion, while selling a Risen 5700x and everything else to it.

In DNS the CPU cost 47 + 27 memory + 21 board, a bit expensive.

Then renting is much cheaper, if, indeed, the top hardware is three times more powerful than the rented 8cores+12Gb.

 
fxsaber #:

Then renting is much cheaper if, indeed, the top hardware is three times as powerful as the 8cores+12Gb rented.

cheaper maybe, but this is your own2 home computer, and you don't count the water from the tap, even though it is cheap.

Besides, you don't buy everything from scratch, but gradually.

 
lynxntech #:

cheaper maybe, but it's your2 home comp, not counting how much water comes out of the tap even though it's cheap.

Besides, you don't buy everything from scratch, you buy it gradually.

My home comp is i3 8300. Ram 16 gb. When running optimisation on my cores for 20 minutes of optimisation the rented cores run about 20% more tasks per core than on my home PC. In this case the distribution of tasks is the same, i.e. start - all cores 8 tasks, then if 2 then all cores 2.
I for example 4 cores of my i3 is not enough. so I will just change the stone to i7 9b or 10-th generation. I was thinking of taking an i7 or i9 and DDR5 memory. But I won't optimise that much. and for the rest - there's definitely no point. Invest better in a wide flat monitor. I got myself one - I can't get enough of it. It's practically a double monitor. And the chart to leaf conveniently and in 1/3 of the screen code, in 2/3 - terminal.
Speed of deployment of servers -2-4 minutes. put Radmin and configure and metatester, taking into account copying another minute let 3. The service allows you to deploy several virtual servers at once. And everywhere you can run software in parallel. You can deploy 92 cores and 196 memory at once. But it is much more expensive. Standard sliced servers are much cheaper. I won't give the service itself so as not to get banned. But it is not a problem to find an hourly VPS in Yandex.

 
I don't know what any of these people are talking about. Server rental is cheaper? It's pretty clear without counting that it's a lot more expensive. I even got curious, so I went to check it out. No, not cheaper, but much more expensive. If we are talking about the cheapest segment, on the secondary market the comp will start to pay off in the fourth month, or even earlier. Of course, there are very big reservations on software and tasks, if you need something one-time, that of course rent is cheaper. But if you need something permanently, it is more profitable to buy, and it has always been.
 
Have they fixed the scheduler behaviour for the latest Intel processors? There is a wrong load distribution over cores. If we put optimisation in the tester in the background, for example, switch to the browser, the scheduler throws heavy calculations that we want to perform faster onto energy-efficient cores. Well, this is obviously not what we want from new processors.
 
fxsaber #:

It turns out that 100K rubles is the rent of the most tricked-out hardware for a year of continuous optimisation.

How much does this hardware cost?

To make the comparison correct, you should also take into account the power consumption in case of your hardware.
Let the power consumption be 200 W, and the price of KW*hour is about 4 rubles.
Then for a year of continuous work your PC will wind up 7000 rubles.

 
G1G2G3 Server rental is cheaper? You can tell right off the top of your head that it's MUCH more expensive.

Of course it is. Otherwise, what's the point of renting? At a loss to yourself?
Here we need to compare differently: if you need computing power on an ad hoc basis, say 1-2 times a month for 8-15 hours, then buying a PC with top hardware is clearly not economically justified.
If the computing power will be used in a 24x7 mode, then buying your own PC makes perfect sense. I.e. there is no unambiguous answer - it all depends on the tasks.

 

What you are all talking about here, I have already tested 8 years ago.

And I think that it makes no sense to buy a powerful computer for testing and optimisation.

I advise you to buy a desktop PC (you can also buy a laptop) somewhere around $1500, with a 2K or 4K monitor from 30"-35", and use Genetic Optimisation in MQL5 Cloud.

This is the most optimal option.

You can see in my profile how much I have spent on M QL5 Cloud for 8 years.

But you get fast results, especially when optimisation is performed in real ticks mode.

12345678910111213
New comment