“While the theoretical speed advantage and expanded resources of 64-bit computing are enticing to those in need of maximum performance, the road to a perfect AMD64 desktop, workstation or server machine is long and treacherous. What operating system will you use? Is there enough 64-bit software available? In this article we’ll explore some of the advantages and pitfalls of going totally 64-bit in a 32-bit world.” Read more at NewsForge.
64-bit Linux and BSD are maturing steadily
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I wish the article would list any differences between the two, or explicitly say they were the same or whatnot.
They almost always prefix every statement with “On AMD64” or whatever, leading to my doubt about the Intel equivs.
Yep, and I think that x86-64 would be a more logical and broader name for the architecture.
AMD64 is historically correct but now as Intel – and later maybe other too – seem to be following the same AMD64 route more or less, speaking of AMD only can be confusing. AMD is the leader, but still people should start to mention something about the new Intel equivalents to AMD64 too and their compatibility when writing about “AMD64”.
How many other vendors have had their X86_64 implementations on the market more than half a year (and mostly as engineering samples)?
AMD’s 64-bit CPUs have been longer than enough on market. (Wasn’t Linux and *BSD kernels ported to AMD64 long before actual processor with some kind of emulator, can’t remember).
AMD is the oldest x86_64 platform. Intel is just coming and it probably will take at least year to grow to the stage where AMD is now.
Yeah, the vendors have embraced Intel’s implementation faster than AMD’s, but anyway it’s almost brand new!
>> “There are parts bound to remain 32-bit, namely bootloaders”
Wasn’t it so that bootloaders need to be 16 bit programs since those are started in real mode ?
So does that mean x86-64 doesn’t support 8 and 16 bit software.
Goodbye (sniff) good ole pal MS-DOS
Fortunately we have dosbox.
Well, the BIOS boots the first sector in real mode like it has always done. GRUB starts your kernel in protected mode, and I doubt there’d be any advantage in it starting in long mode since the switch is trivial (once you have a page directory set up).
That’s only the beginning, though — the next step to maximum performance is to eliminate all 32-bit binaries from your system.
Please feel free to correct me if i’m wrong but i get the idea that source based distro’s pay off here.Installed Gentoo 2005 AMD64 not so long ago,i still have to find any 32-bit whatever 🙂
Gentoo’s Portage tree is hit-and-miss with programs that are able to compile and run. While you can get a 64-bit Mozilla, the MPlayer plug-in package is still masked and marked as unstable. There are many similar examples of masked packages, like OpenOffice.org and the Azureus BitTorrent client. Fortunately, most of the masked programs have 32-bit binary builds available in Portage, so you don’t have to go without them.
More or less true.Editing “/etc/portage/package.keywords” now and than works fine for me.While most packages are in “testing” phase i’ve experienced no drawbacks and/or stabillity issues (yet).Latest kvirc,mythtv,freevo and a couple of other “~” masked packages seem to run quite well actually.
You’re forgetting that AMD is using plently of Intel technology, MMX, SSE, SSE2 and soon, SSE3, the fact that Intel is supporting one of AMD’s ‘technology enhancements’ means sh*t all. Infact its great thing that two different companies are supporting a common 64bit architecture.
Theres far too much AMD-fanboyness regarding 64bit extensions, can’t we just move along now?
IA32 was engineered by Intel, and it stands for Intel Architecture 32-bit. AMD built upon that with AMD64, but the way they did it was unique. AMD engineered the ADM64 architecture (regardless of its roots) and so they get to name it. Having been the tagalong to Intel for so many years, it doesn’t surprise me that they’d want to include their company name in the architecture tag. I’d say that they earned it.
-Jem
They mentioned Fedora Core and Gentoo as being the only available 64-bit “community distributions”, but Ubuntu has a complete 64-bit version as well.
As I’ve pointed out before earlier (It’s less of a problem now than it was a year ago) if your 64bit software breaks in some cases you’re stuck waiting on a fix which there may be in the 32bit world but if you are using a “User” install and your os support does not provide a updated RPM then you’re stuck waiting on someone to recomple a package for you to install.
All of the open stuff works great. However I have had trouble with some of the random proprietary crap. Where you really see this is in drivers and plugins. See, if it is an entire program like OpenOffice (which doesn’t build a 64 bit version yet) then they simply package up the 32 bit version and you cannot tell the difference. However I had to use the less featureful open source ATI drivers because ATI hadn’t released 64 bit drivers until January. Same goes for ndiswrap and wireless support, I had to sit around with none until they finally released the Win64 drivers so that I had something to wrap. I still don’t have a working flash plugin and none of the proprietary video codecs work because they are all win32 libraries.
Michael
Looks like a 2.0 version of OpenOffice will be 64-bit clean, and other applications will soon follow
Not true. OpenOffice is about the only one that isn’t already 64 bit clean. And besides, it’s still open source so the distributions have wrapped it up nicely such that the typical user wouldn’t care either way.
Michael
could amd have added more registers? than the 16/16 gp/sse? if so why didn’t they do it? since they have rename registers, why not offer another 16 FPU registers?
“could amd have added more registers? than the 16/16 gp/sse?”
Yes.
“if so why didn’t they do it?”
That would mean that the processor had two intruction sets and would need two sets of decoders complicating the front end of the pipeline. That is a problem with a solution but AMD for one reason or another decided to do a more straight-forward extension (possibly after pressure from Intel). We know that they were considering a more complicated redesign for at least the FPU, called “technical floating point”.
“since they have rename registers, why not offer another 16 FPU registers?”
As the FPU is stack based an register extension would be more complicated.
The author clearly doesn’t know what he’s talking about. Sun ships the largest volume of AMD64 servers, when compared to HP or IBM. Sun ships all of their boxes with 64-bit Solaris as the default.
Also, re the comment about AMD’s FPU:
That is a problem with a solution but AMD for one reason or another decided to do a more straight-forward extension (possibly after pressure from Intel).
It is probably more likely that AMD had other priorities in their chip design. Regarding pressure from Intel, it should be clear that AMD is now the one applying the competitive pressure instead of being on the recieveing end from Intel. Keep in mind, that it was AMD’s Opteron chip that forced Intel to shoehorn EMT64 into their Nocona chip. More importantly, Opteron has been outperforming any of Intel’s recent offerings, largely because AMD decided to re-design their processor core, instead of trying to tack 64-bit instructions on instead.
The dual Opteron w2100z ships with Solaris 9 x86 by default (as recently as two weeks ago, anyway), so what you’re saying about 64-bit operating systems shipping with 64-bit Sun machines is factually incorrect. Your statement about sales volume is irrelevant to the article’s topic.