Writing x64-based apps. Is there really any point?

Obviously, it probably has some (or many) advantages over 32-bit that I'm clearly not aware of. So, what are they?

I just don't get it, so many things still aren't supported on X64 PC's. For example, on Internet Explorer 8 and 9 64-bit versions don't support Flash, and when I manage to get it working, it then UN-works, then brings up a message telling me that 64-bit IE's don't currently support flash or Flash isn't available on 64-bit browsers.

I have a 64-bit pc now with Windows 7, and am still writing 32-bit apps, and they all work perfectly (minus a few bugs here n there, which would appea开发者_开发问答r whether you're using 32/64-bit). Why should/would one want to develop for 64-bit systems? I don't see how they are any different and, if I were to learn more about developing for 64bit, where would you recommend I start?

The most commonly cited reason for 64-bit applications is access to more memory. Database servers, for one obvious example, can benefit tremendously when most (or all) the data they're working with is available in memory instead of being stored on disk.

You can also gain extra speed, especially for floating point-intensive applications (I've seen a 3x speed-up fairly routinely, though it also depends somewhat on the CPU).

Some other applications, however, gain little or even lose some by moving to 64-bits. The CPU still has the same bandwidth to memory, but all your pointers double in size so if you're using pointers a lot, it can end up a net loss.

It depends what you are doing.

If you're writing a standalone app that doesn't talk to anything else, isn't going to need a huge amount of memory and wouldn't benefit from the extra registers x64 provides then you won't get much (except bloated structure sizes :)) from making an x64 version.

OTOH, for code that runs in-process, x64 is kinda viral. The shell itself is 64-bit now so if you want to plug into it you have to be 64-bit as well. (Or at least provide an adapter which can talk to the 64-bit world.) As a result, it's often easier to compile everything as 64-bit so you don't have the hassle of marshalling calls between the two worlds.

(While still having a 32-bit build for 32-bit OS, of course.)

Edit: Forgot to say, it's also useful to target x64 if you want to present the "real" view of a machine. 64-bit Windows "lies" to 32-bit processes about various things for compatibility reasons. You can disable/bypass the lies but doing so without breaking things (e.g. 3rd party DLLs) can be tricky and it's best avoided.

64-bit software can address more than 4GB of memory (in reality the limit is ~3GB) directly and it uses additional hardware (extra registers, etc.) available on modern CPUs, thus improving performance. These are the two major reasons of the migration to 64 bit.

Normally you would develop cross-platform software and your compiler would take care of using all the 64-bit features.