AMD or Intel as well as the memory brand - page 11

New comment
 
Mathemat >>:
Спасибо, four2one. Короче, число ядер для МТ4 не играет абсолютно никакой рояли :)

Totally agree, it's the amount of memory that counts and it's not for speed.

 
four2one >> :

>> Totally agree, more important is memory size and not speed.

It's not the amount of memory which matters but the speed of both CPU and memory bus...

Celeron is good because it has an 800MHz bus

 
keekkenen >> :

It's not the amount of memory which matters but the speed of both CPU and memory...

Celeron is good because it has an 800MHz bus


I'm not sure it was communicating with memory at all during script execution. All of the native (machine) code could have gone into the cache. That's why I'm talking about the EA. If the EA is optimized, SAR will be crucial. And so ... It proves what I've already said: i7 and Pentium have basically the same core.

 
Svinozavr >> :

I'm not sure if it was communicating with memory at all while the script was running. All the native (machine) code could have gone into the cache.

Funny... I doubt that the terminal works directly with the CPU cache, bypassing the memory where the script itself is registered...

 
keekkenen >> :

Funny... I doubt that the terminal works directly with the CPU cache, bypassing the memory, where the script itself is registered...

What's so cool about it? The processor loads code (machine commands) into its cache. That's what it's for. It can be loaded from memory or from wherever. If it gets all code, it is not communicating with memory any more, but takes instructions from its cache and kernel chops them. If it were to retrieve instructions from memory, it would be much slower.

Therefore, the more cache, the faster programs are usually executed. And such programs as test script, or rather native code generated from bytes of MT4 code can fit into my fucking 1 mb cache.

 
I mean that it doesn't communicate with memory... because by execution I mean not only running the script, but also loading it and returning the result...
 
keekkenen >> :
I mean that it doesn't communicate with memory... because by execution I mean not only running the script but also loading it and returning the result...

But I'm talking only about the process of execution!

As neither loading of script into cache, nor result return in our case has no effect on the speed. Loading code into cache from memory at once is very fast operation. But it is slow to pick it out one command at a time. This is what the idea of cache is based on. And I'm even silent when it comes to outputting data. What kind of output is there?

Because - once again!!! - this test is not representative! You need the stone to communicate with the memory. Quote history, for example, does not have to go into the cache.

 
Svinozavr >> :

And I'm exclusively about the process of execution!

Since neither loading of script into cache, nor output of result in our case has any effect on speed. Loading code into cache from memory at once is very fast operation. But it is slow to pick it out one command at a time. This is what the idea of cache is based on. And I'm even silent when it comes to outputting data. What kind of output is there?

Because - once again!!! - this test is not representative! You need the stone to communicate with the memory. The history of quotes, for example, doesn't have to go into the cache.

Well, let's enter: one of the operations of the test is to assign a variable by looping

you can divide it into asks, for example... ;)

start=GetTickCount();
for( i=0; i<1000000; i++) { tt=iOpen[ i];} 
test2=GetTickCount()- start;
 

or not, not by claws but by local time!

start=GetTickCount();
for( i=0; i<1000000; i++) { tt=TimeLocal();} 
test2=GetTickCount()- start;
It's understandable that it won't change much within a second or two, but the appeal will. ?
 
kombat >> :

Well, let's enter: one of the operations of the test is to assign a variable by looping

you can divide it into asks, for example... ;)

Well... >> you could. But why? Look, what's the problem with taking a standard Expert Advisor from MT4? We are interested in optimization, not abstract scripts. Save the history in the archive and post it together with the test EA, so that everyone would test it on the same one. We will talk about optimizable parameters in the Expert Advisor and their ranges. And all...

1...456789101112131415161718...94
New comment
Reason: