I can't speak to how much RAM AmiBroker uses as I am just starting to learn how to use it, but I really wouldn't worry about it. I have developed software for almost 20 years, and based on that experience, I would say that RAM is so inexpensive now, that it would pay just to load up your system with as much RAM as the motherboard/chipset can use. The caveat there is 32 bit Windows won't see more than about 3.5 GB of RAM. If I were buying a new system, I'd get one with 64 bit Windows installed, and thus all the RAM the motherboard/chipset can handle (it will be a while before there are motherboards that can accommodate anywhere near the maximum a 64 bit OS can handle in its address space).
That way, the impact on the different processes you're running have on each other will be reduced to a minimum (a minimum specific to the other components in your system). I would suggest, though, that unless your machine will be dedicated to running AmiBroker only, it is not practical to try to optimize a machine for it, simply because everything else your machine is doing can impact it behaviour. Do you get real time quotes? Do you browse the web, or at least favourite sites, for information relevant to your use of AmiBroker? Literally anything you do can impact the performance of AmiBroker or any other application you run. The next big things to consider, apart from your processor itself, are your caches (on the chip: L2, for example), and your buss speed: how fast your processor can fetch from, and write to, memory. I find it a bit surreal that you would worry about processes that take 3 minutes the first time, and a minute subsequently. Would getting a result 10 seconds sooner really make a difference? But my experience is with whole ecosystem environmental risk applications, where a comprehensive assessment model may take weeks to run, and the engineers that have to use them would dance a jig if they could get the same results in a week. They'd probably faint if they could get those same results in an hour. My colleagues would advise me not to worry about little things, as long as we can get the information we need in time to make appropriate decisions and act in a timely manner. To answer your last question: yes there is a way to determine the impact of each system component on the performance of a specific application, even taking other applications that may be executing at the same time. But that is so time consuming and expensive to do right, it isn't worth the effort: to do it alone would take long enough that your machine would be obsolete by the time you're done, and to have a team do it would cost several times the cost of just getting the fastest available processor on the fastest available motherboard with as much RAM as it can take. It would make for an interesting thesis topic for a graduate student in computer science, but it is impracticable for most real world users. It is smarter and much less expensive just to make sure you have the fastest processor, with the biggest on-board caches, the fastest bus, and all the RAM that will fit, that you can afford, and even buy several to share the load. In my work, we don't go beyond making sure we have a good, stable machine to do the job, and then we spend a lot of time on writing high performance code (that involves extensive profiling to find where the bottlenecks are - a process similar to, but more focused than, the work required to give you the kind of answer you're looking for). HTH Cheers, Ted On Mon, Nov 16, 2009 at 9:41 AM, jmdeacon <[email protected]> wrote: > > > I am considering buying a new computer and was wondering if there was a way > to figure out how much Ram I am using. I am not a programmer by profession > and have self taught myself a fair amount, but sometimes my assumptions > about computers etc are just plain wrong, so let me no if my thinking is > correct. > > I noticed when I run an optimization, the first pass will take about 3 > minutes and then each subsequent step takes about 1 minute. I noticed a > while ago that when I installed more ram, the first step time did not change > as much as the additional step times. I am assuming that the amount of ram > is part of this. If the amount of ram is more than the amount of data loaded > in to be tested, then I will minimize the time for each additional step > after the first pass. > > If I am correct of this, is there a way to find out how much ram that > threshold is on my system? > > Thanks, > > John > > >
