AMD or Intel as well as the memory brand - page 30
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Everyone seems to be sick and tired of waiting for the next "grail" to be optimised.
A very average and inexpensive system is sufficient for normal operation. I would value stability more than performance.
And in general... I'm not sure that optimizing on history is the right solution. The way
... I'm not sure that the way MT4 has done it is adjusting to the history. And all the "backward" and "forward" tests are self-defeating
using a primitive methodology. That's why most strategies that are successful on history,
don't work for long periods of time. But of course, this is my subjective private opinion. And I don't in any way
I would not, nor could I, in any way impose it on others.
You know, I have a couple of EAs with auto-optimisation - pretty stable. But it's not on all algorithms, of course. That's why a couple.))
The general principle is the following: the Expert Advisor runs backward/forwardward through a certain number of bars (or by timer) and thus gets new inputs.
This is not new and has even been described in articles here. Only there the standard MT tester was run, while I have a tester (very narrowly-specialised) in the Expert Advisor itself. But the gist is the same.
However, I do not have much of a history of watching real money on MTs and forex, but judging by the spots, it has a right to exist.
And they will not succeed for at least another year or two, we will have to catch up. The 45nm process has only recently been adopted, not to mention the 32nm to which Intel is about to transition.
But the four2one config, bastard, spoils the whole picture of Intel's triumph. Maybe it's just the memory, which can't be spoiled there? But somehow that explanation doesn't fit, as 4 gigs is quite enough for such a test...
It's not ruining anything. It's just not properly tested. The frequency should be checked at maximum load time.
If you measure Nehalem this way, you can set it to 1.6G.
And for Athlon 4200 during the test it should be something like 2-2.5G. And the 4200-AMD figure normalized performance to some conditional P4 and as if to say-our brick is about the same as P4 with 4.2G.
The frequency ratio is around 2.
By the way, the Nehalem 3.33 has a real frequency in the test (which is what is shown) over 3.4G. It's not overclocking, it's turbo mod working.
When some of the cores are not loaded, the i7 can increase the frequency as the CPU as a whole consumes much less.
Moreover, the lower the original frequency, the greater the gain. I.e. 2.93G very often runs at 3.2G. 3.33G->3.44G.
And in IDLE - the frequency decreases significantly. And the power consumption drops to almost 0
And for Athlon 4200 during the test it should be something like 2-2.5G.
In short, BLACK_BOX test is conditionally questioned (I've marked it in red in the table; AMD fans, don't think this is racial segregation and discrimination). A 4200+ stone couldn't run at 1 GHz, even with AMD's preeminence at the time.
P.S. It remains to figure out the actual test data of four2one :)
P.P.S. It seems to be written in web-site, that 4200+ has 2.2 GHz. Let's put it in the table.
Uh-huh. With the advent of Intel's TurboBoost, comparing frequencies has become invalid. To be correct it has to be switched off to fix the frequency. But only i7 8xx,9xx and 5xx have it. The rest of the stones are "honest".
Realistically, it's 5-10% in frequency, and in terms of performance the gain is even 1.5/two times less. And besides, (it seems to me) the real frequency was displayed.
Another test - two teeminals "at the same time"
Realistically, it's 5-10% in terms of frequency, and in terms of performance, the gain is even one-and-a-half/twice as much. And besides, (it seems to me) the real frequency was displayed.
The 9xx range has only two multiplier steps, the newer 7xx and 8xx have more, as the TDP is initially lower. On our tasks, the gain is almost directly proportional to this legal overclock. Catching the real frequency is a non-trivial task - sometimes overclocking lasts for milliseconds. Take a look here .
>> one more test - two teeminals "at the same time".
Clicked on two mice "at the same time" too?))
Uh-huh. With the advent of Intel's TurboBoost, comparing frequencies has become invalid. To be correct it has to be switched off to fix the frequency. But it's only for i7 8xx,9xx and 5xx. The rest of the stones are "honest".
The last thing we need here is i7. We will take away all trump cards from AMD fans...
The last thing we need here is an i7. We'll take all trump cards from AMD fans...
Price aside, maybe...
AMD Phenom™ II X4 965 Black Edition | 3.4GHz | Socket AM3 | 6MB | BOX ~ RUB 10564 + MB to 6800
Intel® Core™ Extreme Edition i7-975 | 3.30GHz | Socket 1366 | 8MB | BOX ~ 36691 RUB + MB from 5800 RUB