I want to build a PC to work with MT5, what do you advise and why? - page 19

 

Laptop, well for statistics

CPUGroupingDescriptionIPPRPassesRevenueCreatedActivity
Intel Core i5 M 450 @ 2.40GHz, 3766MB, build 2007
---***.***.***.***9900.002019.03.272019.03.27
 
So far the figures are really strange, the spread is large, but the treadripper is way ahead. I wish more Intel processors were available. Although the difference between two identical FX is the same as the difference between the best FX and treadripper. Maybe of course the ranking is non-linear, but then the results are even stranger.
 

You just have to ask Renat if PR is even a good guide to choosing a processor, who knows what it takes into account

We should make a single-core tester using different calculations and conversions

 

Intel's is not plugging in yet.

 
 

And here are the optimisation results, in order to estimate the real difference in performance

You can see that my base PC FX-8350 which got PR 101 is 20-25% faster than the other FX-8350 which got PR125 rating.

At the same time the Phenom II 960T is already 35% slower with a better PR rating of 119.

The optimization time took 1 hour 23 minutes, it was not to confuse the number of passes, just in each pass of 1000 frames - Mathematical Computation mode. Agents started almost synchronously, well, a maximum of 0.3 passes difference.

On this PR rating depends, in general, the income and expenditure of cloud users, so would really like to get comments @Renat Fatkhullin.

 
Aleksey Vyazmikin:

And here are the optimisation results, in order to estimate the real difference in performance

You can see that my base PC FX-8350 which got PR 101 is 20-25% faster than the other FX-8350 which got PR125 rating.

At the same time the Phenom II 960T is already 35% slower with a better PR rating of 119.

The optimization time took 1 hour 23 minutes, it was not to confuse the number of passes, just in each pass of 1000 frames - Mathematical Computation mode. Agents started almost synchronously, well, a maximum of 0.3 passes difference.

On this PR rating depends, in general, the income and expenditure of claud users, so would really like to get a comment @Renat Fatkhullin.

it turns out PR results are totally inadequate for evaluating real performance.

 
Maxim Romanov:

the PR results are totally inadequate for assessing real performance.

If memory serves me correctly, Renat said a long time ago that PR can change after processing real jobs. Maybe then it would be more adequate, but how to get these jobs with such PR is a mystery.

 

We continue to wonder, of all the majority of agents, Cloud has finally given the assignment - to the weakest!

Active in the last 24 hours
CPUGroupingDescriptionIPPRPassesRevenueCreatedActivity

Intel Core i5 M 450 @ 2.40GHz, 3766MB, build 2007
---***.***.***.***10120.002019.03.272019.03.28

At the same time raised his PR.

 
ThreadRipper 1920X - PR 140-143