64 and 32-bit hosts forever?

[chk071]

And it can be very counterproductive to serve a minority, and waste time you could spend on the needs of the vast majority, as a developer.

BTW, as i mentioned, that was a wild guess. Probably it's even less than the figures i stated. I NEVER saw someone use Windows 32-bit, not on PC, not on a tablet. It just doesn't make any sense, unless your device is very low specced.

[jancivil]

VE Pro probably still has a 32-bit 'server', it did earlier this year, last time I used it. Which will run side-by-side with its 64-bit. OSX.12, I didn't need to enable it as 32-bit to run it. Last time I opened Kore2 like that it was possible, but it buggered the whole installation somehow.
Shoulda just run it in VE Pro 32-bit as a plugin.

[T-CM11]

And it can be very counterproductive to serve a minority, and waste time you could spend on the needs of the vast majority, as a developer.

And there's more to running a business than just plain productivity.

[jancivil]

It isn't just a minority, it's a minority that has shown it's not very interested in continuing to buy the majority of things which will be released. It's a normal business model to phase out what is essentially deprecated product. I preferred to use Amplitube 2 which is 32-bit but certain things of this vintage can't even be installed on OSes I wind up using to support other things - other far more compelling things - so it's history, I'd have to put up with AT4 . The upshot here is that people can continue refusing the future or really the present and keep their system locked down until that Pentium 3 running some older than SX version of Cubase no longer runs.
I've heard of people for whom that is far past where they'll go with their superbly running neolithic system...

[fluffy_little_something]

So, for making music on the computer, is there any other drawback except memory usage?

[aciddose]

Quick overview of 32-bit Intel x86 vs. 64-bit AMD64:

• 32-bit memory addresses are 32-bits long, while 64-bit addresses are twice the size. For inefficient, badly designed software which stores memory addresses or native format integers in large arrays of structures 64-bit platforms will require twice as much memory and twice as much bandwidth. This can result in reduced performance for such badly designed software and is the #1 argument (mostly nonsense) against 64-bit platforms.
• The AMD64 platform took Intel's x86 and slightly extended it. The Intel Itanium (IA-64) platform allowed for hundreds of registers (CPU storage locations, working memory) and a large dynamic register stack rather than using fixed format registers. AMD64 only slightly extended the x86 instead by doubling the length and number of most register types while maintaining their static format and allocation. This made AMD64 x86-compatible and allowed AMD64 to run 32-bit x86 code without a complex emulation layer - a major advantage and the sole reason it "won" the contest with IA-64.
• The AMD64 platform has a minimum specification for SSE2 and eliminates the x87 FPU and floating point stack. It also improves upon the x87 in several significant ways which either reduce or eliminate sources of computational error. This means the same equations computed on AMD64 using SSE2 instructions are often times going to produce more accurate results where those computed on x86's x87 FPU may be inaccurate or entirely invalid!
• Due to the way the 32-bit x86 addresses memory, any single process is limited to at most 4 GB of address space without using special extensions to address/map more. This makes the 32-bit x86 platform totally impractical for processing large amounts of data The AMD64 platform extends the addressing to allow a process to access far more memory with the maximum addressable space at 256 TB!

AMD64 (64-bit) is better than x86 (32-bit) across the board unless you're trying to run buggy, poorly written, unsupported old software that doesn't run correctly on a modern processor.

That about sums it up.

[chk071]

Quick overview of 32-bit Intel x86 vs. 64-bit AMD64:

• 32-bit memory addresses are 32-bits long, while 64-bit addresses are twice the size. For inefficient, badly designed software which stores memory addresses or native format integers in large arrays of structures 64-bit platforms will require twice as much memory and twice as much bandwidth. This can result in reduced performance for such badly designed software and is the #1 argument (mostly nonsense) against 64-bit platforms.
• The AMD64 platform took Intel's x86 and slightly extended it. The Intel Itanium (IA-64) platform allowed for hundreds of registers (CPU storage locations, working memory) and a large dynamic register stack rather than using fixed format registers. AMD64 only slightly extended the x86 instead by doubling the length and number of most register types while maintaining their static format and allocation. This made AMD64 x86-compatible and allowed AMD64 to run 32-bit x86 code without a complex emulation layer - a major advantage and the sole reason it "won" the contest with IA-64.
• The AMD64 platform has a minimum specification for SSE2 and eliminates the x87 FPU and floating point stack. It also improves upon the x87 in several significant ways which either reduce or eliminate sources of computational error. This means the same equations computed on AMD64 using SSE2 instructions are often times going to produce more accurate results where those computed on x86's x87 FPU may be inaccurate or entirely invalid!
• Due to the way the 32-bit x86 addresses memory, any single process is limited to at most 4 GB of address space without using special extensions to address/map more. This makes the 32-bit x86 platform totally impractical for processing large amounts of data The AMD64 platform extends the addressing to allow a process to access far more memory with the maximum addressable space at 256 TB!

AMD64 (64-bit) is better than x86 (32-bit) across the board unless you're trying to run buggy, poorly written, unsupported old software that doesn't run correctly on a modern processor.

That about sums it up.

The problem with all that is that noone, except programmers will be able to understand that. From what i gathered in your post, 64-bit is better. Because... something. Well, thanks, but, i highly doubt that fluffy will understand any of this. I surely don't.

[chk071]

For me, the most relevant arguments are that 32-bit will soon or later disappear, and less and less plugins will be developed. And that more and more hosts go 64-bit only, without a bridge.

[fmr]

Quick overview of 32-bit Intel x86 vs. 64-bit AMD64:

• .../...
  That about sums it up.

The problem with all that is that noone, except programmers will be able to understand that. From what i gathered in your post, 64-bit is better. Because... something. Well, thanks, but, i highly doubt that fluffy will understand any of this. I surely don't.

What I would like to know is if the Intel referred CPUs include ALL the current models, or if Intel did improve on the alleged flaws. Because, for what was written, I have no way to know which CPUs are being compared here.

[jancivil]

I'll never run any computer with AMD so it's quite a special argument which I don't need either since I run an OS that boots as x64 and have abandoned pretty much everything that is x86.
Developer jargon to impress, shrug.

[fluffy_little_something]

I suppose we all have 64-bit processors by now, be it Intel or AMD. And, on the consumer level, we use 64-bit operating systems.
So, basically I was wondering how much of the 64-bit potential is lost when still using 32-bit software on 64-bit Windows.

And from the perspective of developers, do they have to take into consideration any 32-bit limitations as they compile the 32 and 64-bit versions of their software such as DAWs?

[fluffy_little_something]

Quick overview of 32-bit Intel x86 vs. 64-bit AMD64:

• 32-bit memory addresses are 32-bits long, while 64-bit addresses are twice the size. For inefficient, badly designed software which stores memory addresses or native format integers in large arrays of structures 64-bit platforms will require twice as much memory and twice as much bandwidth. This can result in reduced performance for such badly designed software and is the #1 argument (mostly nonsense) against 64-bit platforms.
• The AMD64 platform took Intel's x86 and slightly extended it. The Intel Itanium (IA-64) platform allowed for hundreds of registers (CPU storage locations, working memory) and a large dynamic register stack rather than using fixed format registers. AMD64 only slightly extended the x86 instead by doubling the length and number of most register types while maintaining their static format and allocation. This made AMD64 x86-compatible and allowed AMD64 to run 32-bit x86 code without a complex emulation layer - a major advantage and the sole reason it "won" the contest with IA-64.
• The AMD64 platform has a minimum specification for SSE2 and eliminates the x87 FPU and floating point stack. It also improves upon the x87 in several significant ways which either reduce or eliminate sources of computational error. This means the same equations computed on AMD64 using SSE2 instructions are often times going to produce more accurate results where those computed on x86's x87 FPU may be inaccurate or entirely invalid!
• Due to the way the 32-bit x86 addresses memory, any single process is limited to at most 4 GB of address space without using special extensions to address/map more. This makes the 32-bit x86 platform totally impractical for processing large amounts of data The AMD64 platform extends the addressing to allow a process to access far more memory with the maximum addressable space at 256 TB!

AMD64 (64-bit) is better than x86 (32-bit) across the board unless you're trying to run buggy, poorly written, unsupported old software that doesn't run correctly on a modern processor.

That about sums it up.

Not sure what the point is in comparing 64-bit AMD to 32-bit Intel platforms.
I hope the choice of processor brand is not really relevant anymore with the latest generation.

[aciddose]

What you may not know is: current 64-bit intel CPUs are AMD64 format.

I don't believe any IA-64 models have been produced since several years ago (2011?)

See here: https://www.intel.com/content/www/us/en ... -8700.html

Notice they write: "Instruction Set 64-bit".

Technically the instruction set is called "AMD64". There is no such thing as a "64-bit" instruction set.

https://en.wikipedia.org/wiki/X86-64#cite_note-11

Various names are used for the instruction set. Prior to the launch, x86-64 and x86_64 were used, while upon the release AMD named it AMD64.[1] Intel initially used the names IA-32e and EM64T before finally settling on "Intel 64" for its implementation. Some in the industry, including Apple,[2][3][4] use x86-64 and x86_64, while others, notably Sun Microsystems[5] (now Oracle Corporation) and Microsoft,[6] use x64. The BSD family of OSs and several Linux distributions[7][8] use AMD64, as does Microsoft Windows internally.[9][10]

[fluffy_little_something]

And is all that of any importance when you write your Xhip?

[aciddose]

Yes. For example the "mod matrix" (technically: routing tables) use pointers directly which makes the 64-bit tables twice as long as the 32-bit tables. Does this affect performance? Yes, but not by much since the AMD64 instruction set makes up for the difference by being overall faster.

If I wrote a specially optimized version of the code I could easily beat the 32-bit version with the 64-bit version using identical code. Essentially each "voice" would be addressed once and the table would contain short (16-bit) offsets rather than whole native pointers. This would significantly improve performance both in terms of cache and memory access and increase the number of possible active voices by up to (I estimate) 20%.

So it is "important", assuming I wanted that extra optimization and didn't have 9001 other things to do that are more important.