Optimization on a dual-processor computer - page 5

New comment
 
Nikolay Kuznetsov:

by the way, while researching the ryzen issue i found a pretty good deal on the ryzen 7 2700x... for the price of a 9 you can build a complete PC (without the external graphics card), yes there will be fewer cores/threads, but the cost per core/thread is lower.

Of course, if you need a number crusher for MT, then 2x processors on xeon will probably still be above competition on price/performance, but their future liquidity and using for other tasks, e.g. video processing, is under great question in view of limited set of processor commands and low (relatively modern CPUs) frequency...

PS also was getting closer to this topic from an EAs side, but while working on an old office computer somehow the topic has disappeared and the need in a heap of threads/core is not necessary anymore (or is it so far?).

The performance to price ratio of 3700 is higher than that of 2700.

The 3700 is 48% more expensive than the 2700 and, according to the screenshot, about 100% faster.

The result of the 3900X processor multiplied by 2/3, got about 3700...

Well, there's the cost of the motherboard
 
Although, wait, it might be possible to break it. I'll have to think about it again. I'm confused.
 
pribludilsa:
Although stop, maybe you can. Need to think about it again. I'm confused about this.

Of course you can. Confused.

There are two parameters:

  • "A" needs to go from 1 to 100,000 in increments of 1.
  • "B" needs to go from 1 to 50,000 in increments of 1.

The total sum is 5 000 000 000, which is unacceptable for the optimizer.

We divide the optimization into 10 parts:

  1. "A" from 1 to 100,000 in steps of 1, "B" from 1 to 5,000 in steps of 1: a total of 500,000,000 passes, fits
  2. "A" from 1 to 100,000 in increments of 1, "B" from 5001 to 10,000 in increments of 1: a total of 500,000,000 passes, fit
  3. Etc.

The results are merged into one file. Plot any graphs you want.

Maybe even the glued result can be slipped back to MT. But I am not sure.

 
Andrey Khatimlianskii:

Of course you can. Confused.

There are two parameters:

  • "A" needs to go from 1 to 100,000 in increments of 1.
  • "B" needs to go from 1 to 50,000 in increments of 1.

The total sum is 5 000 000 000, which is unacceptable for the optimizer.

We divide the optimization into 10 parts:

  1. "A" from 1 to 100,000 in steps of 1, "B" from 1 to 5,000 in steps of 1: a total of 500,000,000 passes, fits
  2. "A" from 1 to 100,000 in increments of 1, "B" from 5001 to 10,000 in increments of 1: a total of 500,000,000 passes, fit
  3. Etc.

The results are merged into one file. Plot any graphs you want.

Maybe even the glued result can be slipped back to MT. But I am not sure.

Thank you. It all worked out. It happens, you get lost in three pines. I'm usually an obsessive thinker, double-checking everything, it turns out I need to double-check even more.
 
Generally speaking, it's easier and cheaper. i bought 10 AMD 14. 2 cores3500, ddr3 8gb, cost 25$ , to them ssd 120 for 20$. Old monitors for 8$. Total cost $530. I rented an office for $300/month which includes an outlet. And everything is fine. 10$/month Internet. Who is interested in the address of the magic business centre can be in a private scribble. With mining tried to get in there they have a machine at each office. Failed. But in any case, if you need constant computing power is the solution. Territorially Moscow. PS furniture included in the lease.
 
Dmitriy Prigodich:
Generally easier and cheaper. bought 10 AMD 14. 2 cores3500, ddr3 8gb, cost 25$ , to them ssd 120 for 20$. Old monitors for 8$. Total cost $530. I rented an office for $300/month which includes an outlet. And everything is fine. 10$/month Internet. Who is interested in the address of the magic business centre can be in a private scribble. With mining tried to get in there they have a machine at each office. Failed. But in any case, if you need constant computing power is the solution. Territorially Moscow. PS furniture included in the lease.
What kind of computing power did you get? It may be good in terms of money, but in terms of convenience... Not for everyone.
 
Dmitriy Prigodich:
All in all easier and cheaper. bought a '14 AMD 10. 2 cores3500, ddr3 8gb, cost 25$ , to them ssd 120 for 20$. Old monitors for 8$. Cost a total of $530.

10 mums and memory are still to be counted. And the PSU, that's what you can't skimp on.

And SSDs are not needed, there is network booting and a network drive or server. And you don't need monitors/knobs/mice unless it's a workstation.

 
Use our calculation cloud - it's super cheap.

Some of our clients do massive and constant calculations in it, the equivalent of years of local multi-threaded calculations.

Soon we will release an upgrade to MQL5 with in-house vector/matrix/complex calculations, which will allow for fast machine learning without the use of python or tensorfoe.
 
Dmitriy Prigodich:
In general, it is easier and cheaper. I bought 10 AMD 14. 2 cores3500, ddr3 8gb, cost 25$ , to them ssd 120 for 20$. Old monitors for 8$. Total cost $530. I rented an office for $300/month which includes an outlet. And everything is fine. 10$/month Internet. Who is interested in the address of the magic business centre can be in a private scribble. With mining tried to get in there they have a machine at each office. Failed. But in any case, if you need constant computing power is the solution. Territorially Moscow. PS furniture included in the lease.

$3,600 a year office + $530 worth of old junk.

That's the dumbest solution you can come up with.

Is the old hardware gonna break? Or is it forever?

For this money (4100$) you can buy normal hardware under warranty and put it in your home.

It will be the same in terms of performance. And they'll deliver it to the floor for free, plus all sorts of cashback and bonuses.....

Your G hardware a year later no one for $ 100, and the normal ones may rise in price with the growth of the dollar or crypto.

 

the most reliable computer technology of all time

ad16d7d16ca407b4835d9e6490672398

12345678
New comment
Reason: