Fedora Linux May Further Demote i686 Support

[s_j_newbury]

Also, Linus on PAE.

Using x86_64, you get:
- Access to more registers, allowing compilers better optimization strategies
- x86_64 calling convention, which passes the first 6 arguments via registers instead of via the stack, which is much faster (similar to Microsoft's fastcall calling convention)
- Access to 64-bit address space per process, which enables your programs to access more than 4GiB of data at once. This also helps things like ASLR, since it makes attacks like heap spraying less effective.
- Assumption of SSE1 (or is it SSE2?), which enables better optimizations (though no one is preventing you from compiling everything with SSE1 support on x86)

Downfalls of x86_64:
- Can't run 16-bit code (not a huge deal nowadays) once processor has been put in to "long mode"
- 64-bit pointers take up twice the space as 32-bit pointers. This increases memory use a bit (can't find the benchmarks I saw, but IIRC it was under 5%) and decreases cache efficiency

Even using slightly more memory and filling up cache faster, x86_64 is generally 10-15% faster, due to the increased register size, better calling convention, and access to SSE registers for floating point (instead of using x87 fp).

Of course, this is doesn't matter in the slightest if your processor doesn't support x86_64. Pretty much all AMD and Intel processors in the last decade have supported x86_64: all Athlon64, Phenom, FX, APU processors; a select few Prescott P4 processors, all Pentium D processors, all Core-series processors starting with Core 2 Duo, etc... The big outliers are: early Sempron processors, most Prescott P4 and older processors, Pentium M, Core Solo/Duo, and some Atom processors manufactured up to at least 2008.

It's SSE2. SSE1 goes all the way back to the Intel PIII Katmai, and AMD Athlon XP. AMD processors didn't support SSE2 until amd64 chips were introduced.

[GreatEmerald]

I suppose this is an OK step for Fedora to take. Now I do have a 32-bit device, a 2011 Atom tablet, one of the last batch of 32-bit devices, so things like that are fairly unfortunate, but then Fedora is not Gentoo. And I do need to run Gentoo (or Windows) on the tablet, otherwise it would be too slow anyway.

Speaking of PAE, one thing I found out about it is that for some strange reason it's required to enable NX. My tablet has NX capability, but doesn't have 4 GiB RAM. So it's in this strange gap where it needs PAE for a different reason than what PAE was made to address to begin with.

[vsteel]

You can still run a current OS, it is just going forward where you won't be supported. it isn't like they are suddenly going to stop working. I think it is funny how people are OK with really old hardware but they can't stand old software.

[Scimmia]

Way to early to demote regular x86 (stop saying i686...).

so much hate from idiots

x86 != i686

You guys should really know something about linux target architectures before spouting off.

In this case, i686==x86.

[Passso]

You idiot, who said that x86 = i686?
You, naive idiot, to Africa this has about zero importance and only your cartoonish imagination of Africa can make such an argument. You also forgot to mention the homeless Americans who can't afford amd64 haha, idiot.

Homeless don t care about laptop as they cannot plug it. Their Samsung Galaxy from 2 years ago is enough

Did you see how the desktop and the kernel memory usage went high since 10 years? Even if you can install a recent 32 bits system on a Pentium III it will be unusable... now a simple Ubuntu + firefox with 10 tabs is near 2GB ram!

[Passso]

I can picture that. Maybe what I said was a bit harsh. But I do think 64bit adoption should have been heavily encouraged about 9 years ago.

The adoption has been encouraged but failed because the performance gain was 0 and there were technical workarounds for "big memories".

From a technical point of view it was a revolution but from a user aspect, AMD64 seemed useless.

Just like IPv4 / IPv6 but this is another subject...

[CrystalGamma]

Did you see how the desktop and the kernel memory usage went high since 10 years? Even if you can install a recent 32 bits system on a Pentium III it will be unusable... now a simple Ubuntu + firefox with 10 tabs is near 2GB ram!

10 years ago, browser tabs were just starting to show up. Also, instead of being an interactive document viewer, people now expect their browser to support a fairly complex vector graphics format (SVG), audio and video playback, as well as 2d (Canvas) and 3d (WebGL) rendering APIs. Back then, XMLHTTPRequest was hot new stuff! JS runtimes have also grown, because people expect more performance from their browser (and at least for GNOME, the JS engine is used for the desktop as well ...)

[duby229]

The adoption has been encouraged but failed because the performance gain was 0 and there were technical workarounds for "big memories".

From a technical point of view it was a revolution but from a user aspect, AMD64 seemed useless.

Just like IPv4 / IPv6 but this is another subject...

Well, if that's true then people are dumber than I thought. And because of that most games are still limited to only 32bits. It's so annoying. Even though it's 9 years too late, I guess I should be happy that it is finally happening. But it's hard to be happy when games needed 64bit binaries 9 years ago and most still don't.

[Passso]

Average people are not dumb, they just have a simple vision that is : what is the gain here? And just like children if the gain is not "right now" then they prefer "good old stuff".

Now we must remember toot that Intel + Microsoft were engaged for Itanium 64 and they made all to have AMD64 fail.

The funny thing is that Itanium64 failed for an even greater thing : performances "right now" were lower than i686

[s_j_newbury]

IA64 wasn't slower than i686. It was highly dependent on a high quality optimizing compiler; the technology for which wasn't really there at the time which did limit the gain from the architectural design. The i686 code performance was poor due to the x86 execution unit being weak, it was almost an afterthought: Hey, lets tack on a x86 emulator to make it backward compatible! They should never have done this.

There was also a lot of poor feeling towards the project due to the almost conspiratorial destruction of competing technologies of the time for which customers were not happy and felt betrayed/railroaded; DEC Alpha for example. All in all, it was a great example how not to introduce a new technology.