Using 32-bit software, rather than 64-bit version

>using 32-bit software, rather than 64-bit version

Other urls found in this thread:

phoronix.com/scan.php?page=news_item&px=Soft-FP64-GPU-Work-XDC2016
twitter.com/NSFWRedditGif

What is even the difference

32 bits

>using the 64 bit version because more bits when the 32 bit version performs better

More bits brah

More virtual memory and performance improvements.

More virtual memory also makes ASLR more effective so security improvements as well.

this is why the x32 ABI exists. sadly, it never became popular

When is this ever the case?

If you have low RAM but a 64-bit CPU is the only thing I can think of.

duh, there's less bits going from your ram to the CPU therefore it goes faster

Not accounting for different compiler optimizations, what is the overhead of running a 32 bit binary in compatibility mode on an AMD64 cpu? Is there any?

>a 32 bit binary

... or better yet, let's consider a whole 32 bit OS so we can pull stuff like WoW64 out of the equation.

On Win64 32bit code runs at near native speed, something like 95% native or better

Running 32bit applications in a 32bit OS(assuming same arch) is native speed.

It's only when you mix ISAs that conversions have to be done but if the conversion is from low to high like with 32bit code running on 64bit OS the conversions are very fast and not extremely costly.

High to low conversions like 64bit running on a 32bit OS would require emulation which would be very costly performance-wise so there's little to no point in that.

>the conversions are very fast
the conversions *can* be very fast rather

It's all dependent on how it's implemented.

Why is the Windows ecosystem so extremely slow to adopt 64-bit?
My Linux distribution was 100% 64-bit in the early 2000s.
Many Windows users STILL use 32-bit browsers. It's ridiculous.

I thought 64 bit is slower but allows for more advanced operations

since computer reads in chunks of 32 bits, vs 64, it can be twice as quick reading, but half as accurate (depending on implementation) since you only have 32 unique bits rather than 64.

it's not slow at adopting, it just excels at backwards compatibility.

AMD64 has twice as many general purpose registers as i386 which allows compilers to produce more optimal code.

Thanks for the answers.

In general, 64bit will be somewhat faster. Not just because it has more registers but also because it means applications can use newer instruction set extensions without worry of breaking compatibility. The main advantage is the huge increase in virtual memory though, it prevents out of memory errors and helps with security mitigation techniques.

Lots of third party, mature, unmantained and irreplaceable software. And what says, so there's less sense of pressure.

Just last week I noticed that Office 2016 comes in 32 bit as the recommended version (and it's not because of "people still running 32 bit OS and not knowing it, let's be safe").

>tempest 2000 soundtrack

>>assuming I'm using hw that supports 64 bit.

32bit office is recommended because most corporate plugins for office are still 32bit only.

>b-but muh few megabytes of saved memory

>Why is the Windows ecosystem so extremely slow to adopt 64-bit?
A lot of small company internal software depends on third party libraries or integration of software that never gets updated. Rather then spend money on newer licenses or software, they are going to run their integration with ancient versions of Crystal Reports, Adobe Acrobat, or some other library off into the sunset.

When you're such a terrible programmer that you hit memory bottlenecks from transferring 2x as much pointer data.

>Visual Studio

>High to low conversions like 64bit running on a 32bit OS would require emulation which would be very costly performance-wise so there's little to no point in that.

Actually I'll correct myself, sometimes (probably very rarely) there is a point to conversions like this.

Here's one example of this that is being worked on right now:
phoronix.com/scan.php?page=news_item&px=Soft-FP64-GPU-Work-XDC2016

Basically some older graphics processors have support for basically all of the instructions of OpenGL 4.X except FP64. Basically no games require FP64 support though so there's little reason that these processors couldn't run games that require OpenGL 4.X. This emulation brings very few downsides while increasing the compatibility with existing software.

I remember in some magazine they said how one of the reasons it failed was that it looked like a toilet seat. I've never been able to unsee it.

>needing more than 8 bits to run your FLOSS fizzbuzz apps

If you add the CD drive and a cartridge to it then it looks like a toilet. Especially with the lip popped up.

Do you use Steam? If so, I have some news for you...

That would explain the Chicago Bears then.

If you have a PC from 1995 or later, it always reads RAM in 64bit chunks.

>a PC with a 32 bit processor reads RAM in 64 bit chunks

Really?

How is that even possible?

Isn't RAM always stored in bytes?

>assuming windows. . .

64-bit irfanview literally cannot open Unicode filenames