Is the 2GB limit for FXT files still around? - page 3
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
Can other traders verify that the same thing is happening for them, i.e. that, at least on 64-bit systems, the 2GB FXT file USAGE limit appears to have disappeared.
Cheers all.
~DV
Isn't my very first post enough for you? No prob over 2GB on Win7 64 bit MT4 version 445.
For everyone's benefit, I have just finished the Win7, 64-bit, build 500 test on my laptop (a near identical test to the one I described earlier in this thread that failed on XP 32-bit). The .fxt in this case was 6,372,486kb and it FAILED to produce any results after getting about 60% of the way through the run. I don't have the 32bit test results to hand to see if one got any further than the other but IIRC it was in the same vicinity.
I am going to run a couple more tests (same test but narrower date ranges) to see where the cut-off is as regards the .fxt size.
Just had it confirmed that the same test on 32bit machine cut off at the same place as the 64bit machine and both were just over 6gb .fxt
I've now re-run my 64-bit tests with narrower date ranges. It would appear that the .fxt limit is now circa 4gb on both architectures, one test at 3,470,706kb .fxt size completed, one at 4,269,203kb crapped out just short (in the same place as the full 6gb run). 4gb = 4,194,304kb (in JEDEC).
It would be good if some of you were able to verify the same. Certainly DV's test was less than 4gb, which is why it will have been ok.
The thing is, I think I've run this test on previous versions of MT4 (though I'm not sure if it was pre build 445) and it ran to roughly the same point, so I'm not sure if we're any better off. Whatever the case, it would appear that on both 32bit and 64bit the limit is now 4gb for .fxt reading.
It would be good if some of you were able to verify the same. Certainly DV's test was less than 4gb, which is why it will have been ok.
The thing is, I think I've run this test on previous versions of MT4 (though I'm not sure if it was pre build 445) and it ran to roughly the same point, so I'm not sure if we're any better off. Whatever the case, it would appear that on both 32bit and 64bit the limit is now 4gb for .fxt reading.
I've already confirmed that a 64bit OS ( Win 7 ) will run past 2GB . . . I don't have a isolated 32bit OS only XP Mode on Win 7.
RaptorUK:
Certainly seems to be the case, my test run produced a 3.10 GB fxt file and the test ran from start to finish.
Hi Raptor. Sorry, I think you misunderstood. I've confirmed at my end that 4gb is now the limit on both 32 and 64 bit architecture. I don't think 32/64 makes any difference.
What I was hoping was that someone here would now confirm that they are also finding 4gb is the limit, since successful tests over 2gb but under 4gb (which I've corroborated) have been confirmed, but only I have tested beyond 4gb.
I've updated my service request in light of this btw. No answer in 8 days...
Hi Raptor. Sorry, I think you misunderstood. I've confirmed at my end that 4gb is now the limit on both 32 and 64 bit architecture. I don't think 32/64 makes any difference.
What I was hoping was that someone here would now confirm that they are also finding 4gb is the limit, since successful tests over 2gb but under 4gb (which I've corroborated) have been confirmed, but only I have tested beyond 4gb.
I've updated my service request in light of this btw. No answer in 8 days...
Ah I see, I'll see what data I have to make a 4+ GB fxt file and report back. I would assume that 4GB is the new limit but it's good to test and be sure.
Thanks gents. Good to be sure.
On a related note, if I run an optimisation, are there any .fxt or other related constraints? What I mean is, if I do a single run on a dataset and it creates a 1gb .fxt say, if I then want to run an optimisation on it and the permutations required to complete it are say 200 runs, what it is that is allowing 80 of them to complete ok (for example) but the rest to crap out? Is it still a .fxt limitation, something else, or a combination of things? If it's purely the .fxt, and it's 1gb on a single run, why it would do more than 4 runs before failing? Any advice?
Thanks gents. Good to be sure.
On a related note, if I run an optimisation, are there any .fxt or other related constraints? What I mean is, if I do a single run on a dataset and it creates a 1gb .fxt say, if I then want to run an optimisation on it and the permutations required to complete it are say 200 runs, what it is that is allowing 80 of them to complete ok (for example) but the rest to crap out? Is it still a .fxt limitation, something else, or a combination of things? If it's purely the .fxt, and it's 1gb on a single run, why it would do more than 4 runs before failing? Any advice?
I haven't ran any Optimizations for some time so I can't comment directly on the stability of the ST for Optimizations on recent builds, but I have run many Optimizations in the past and never had a premature end to the expected number of runs. I have seen it reported that the start date can go awry but I never tried to replicate this and it might be particular to using tick data and the TDS: https://www.mql5.com/en/forum/140857
What do you mean by "crap out" ? perhaps you hit a combination of parameters that show a divide by zero error in your code, have you checked for this ?