Microsoft is advancing its 64-bit agenda with the addition of some new AMD Opteron computers.
Microsoft, AMD Evangelize 64-Bit
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
The future of computing (64 Bit Addressing)has arrived. It’s funny how Intel didn’t see this comming. Once the M$ apps reach the 64 platform, the rest of the venders will follow like good soilders. Even Apple has seen the light. It’s funny that Linux is ahead of them both.
The future has been here, AMD paper launched the “hammer” like 2 1/2 years ago, with the first release maybe 1 1/2 years ago. BTW, in most of the benchmarks I have seen AMD’s Athlon 64 runs 32 bit code faster, and native 64 bit applications require more RAM.
Considering it will be a long way off before I need more than 4 gig of RAM I’d say 64 bit processors are more of a marketing tool currently than a technological breakthrough.
Sorry I don’t share your enthusiasm, I just don’t see a reason to be excided about the technology for anything but server class systems.
At least for my needs, Athlon 64 has less to do with the “the future” than PCI express.
Maybe MS just thinks that if it can convince people that 64-bit is needed, they will be able to sell more licenses of Windows?
This move probably has nothing to do with the technical merits of moving to 64-bit.
4G+ will probably come sooner than expected. Anyway, it’s a good idea to prepare the software for 64 bits before it is actually needed.
See the step from 16 to 32 bits as a counter-example. Lots of crappy memory management extensions (XMS, EMS) were invented just because nobody planned ahead.
Don’t care about 64 bit. Someone wake me up when the engineers finally figure out a way to replace the hard drive with large capacity solid state technology.
If you want big and fast solid state storage, you could always get a RamSan: up to 1TB of storage, 2.5KW of power, 8 Fibre Channel connections. Just be prepared to pay for it.
“Someone wake me up when the engineers finally figure out a way to replace the hard drive with large capacity solid state technology.”
Keep an eye on Nantero:
http://www.nantero.com/
http://www.nantero.com/tech.html
I’m hoping it doesn’t go the same way as the FMD (Fluorescent Multi-layer Disc).
BTW Time passes quick, we will truelly be in the 64 bit era very soon.
“The future of computing (64 Bit Addressing)has arrived. It’s funny how Intel didn’t see this comming. Once the M$ apps reach the 64 platform, the rest of the venders will follow like good soilders. Even Apple has seen the light. It’s funny that Linux is ahead of them both.”
intel did see it but they want to make a proprietary platform that they could protect with patents and lock out the clone makers like AMD and maximize their profit margins but they failed to see the collapse in the economy and the desire for companies to find cheaper alternatives with smaller technology budgets. more people wanted the backwards compatibility of AMD64 and itanium wasnt optimized for that. intel’s management gambled and lost.
…and lock out the clone makers like AMD…
attila, you do know AMD and Intel have a cross licensing agreement right?
Also, Itanium is an entirely different animal than AMD64 and was never disigned as a dekstop solution.
Compatibility with x86 makes sense for more than just AMD.
Nice Itanium appologist view of history; Intels goal of Itanium was to de-throne the UNIX RISC vendors with Itanium. Itanium was meant to be the rock solid CPU, great performance, and massively lower than what the RISC vendors sell their systems for.
Fast forward to a place I like to call *reality*; there are no “low cost” Itanium systems; there are no affordable Itanium Workstations or servers. The Itanium, for all intensive purposes has failed to live up to Intels dream.
It could have worked, had Intel put the money were its mouth is, the fact remains, no one is going to purchase a platform where there are no bloody applications! neither for the high end workstation or servers!
Itanium IS a failure; AMD won. Get used to it, suck on a popsicle and get used to it. Bigger companies don’t always know best. The best ideas seem to come from small companies who have their back against the wall; AMD has proven that it can not only make cheap-chips for the working man, but compete at the high end of town.
“Fast forward to a place I like to call *reality*; there are no “low cost” Itanium systems; there are no affordable Itanium Workstations or servers.”
“Low cost” is a relative statement. You might be too young to remember when UNIX workstations – even the LOW END ones – started at $10,000USD and went up from there…
Compared to that, an Altix is downright cheap. What they AREN’T is “mass market” chips.
“Get used to it, suck on a popsicle and get used to it. Bigger companies don’t always know best. The best ideas seem to come from small companies who have their back against the wall; AMD has proven that it can not only make cheap-chips for the working man, but compete at the high end of town.”
If Itanium is a failure, so is POWER. They are both intended for the high-end market.
No, actually, POWER is a success. You have the PPC970 – POWER-Lite, and you have the big iron POWER cpu for the AIX big irons. No different to SUN and their UltraSPARC IIIi/IV for the high end of town, and UltraSPARC IIe/IIi catering for the low end workstation market (below $5k).
Both the POWER and UltraSPARC are *VERY* successful chips – be it that not as wide spread as AMD or Intel, but they’re still very much ahead of Itanium.
Btw, there are more POWER and SPARC cpus sold per quarter than Itanium sold in a year. That alone tells you something. Oh, thats in raw CPU shipments – so a 128 way machine would register 128 cpus sold.
“You have the PPC970 – POWER-Lite”
I’m not saying it isn’t, however POWER and PPC970 aren’t the same thing. POWER is a line of CPUs with very large caches that are aimed at the highend server/workstation. Just like Itanium. I2 chips are very fast. For the market they’re aimed at (HPC) that’s what matters.
“That alone tells you something”
If you want to use that argument, Apple is a complete failure.
“Low cost” is a relative statement. You might be too young to remember when UNIX workstations – even the LOW END ones – started at $10,000USD and went up from there…
How much things cost 10 years ago is completely irrelevant. Itanium is expensive in todays environment and what ever Intel may try and tell you, they know it is a failure.
I personally agree with the guy who said it was all because Microsoft needs to sell more Windows licenses, i.e. they’ll start saying now something like “The revolution is here! Dump your old XP in the bay, buy our shiny new Windows’64!” It’s a useful technology, of course, but probably only when it comes to breaking the memory limit. Also, as far as I know Intel has employed some 64-bit addressing extensions on their latest motherboards that deal with this problem…
“Itanium is expensive in todays environment and what ever Intel may try and tell you, they know it is a failure. ”
Again, compared to what? Xeon? Opteron? Yes. POWER? SPARC? MIPS? Not really, no.
“Also, as far as I know Intel has employed some 64-bit addressing extensions on their latest motherboards that deal with this problem…”
I think youre talking about intel’s PAE (Physical Addressing Extensions). It extended memory addressing from 32 to 36 bits. Im pretty sure it has been enabled in xeon chips for a while, not sure about desktop parts. PAE is slow (similar to DOS page flipping as I remember reading) and nowhere near as fast as native addressing.
JCS is a chutiya, who is trolling for the sake of trolling.
It’s here. People are buying Opteron boards, because of minute price difference to pure 32 bit system. The avalanche has started and nothing can stop it. We will have to port our applications once again. Last time Microsoft got a tremendous advantage from this process. I remember dumping Borland C++ in favor of Microsoft Development studio. A painful but inevitable move.
Looks like the transition will be much faster this time. It lasted several years before DOS and WIN31 were replaced by 32 bit OS. In fact this schizophrenic situation when everybody possessed 32 bit hardware but run only 16 bit OS gave birth to Linux. Nothing similar will definitely happen now. Debian even decided AMD64 port is not worth inclusion into the next stable release. Sad story.
“JCS is a chutiya, who is trolling for the sake of trolling.”
What’s a “chutiya”? BTW, learn what a “troll” is.
People are buying the Opteron boards due to the chip’s price/performance ratio more than “64bitness”.
“The avalanche has started and nothing can stop it. We will have to port our applications once again”
There’s little reason to build most applications as 64bit binaries at this point (this wasn’t true during the 16-32bit transition). On x86-64 – alone – they might be faster with the extra GPRs, but on other platforms they’re likely to be slower. There’s a reason why most commercial UNIX operating systems with 64-bit kernels still contain significant 32-bit software.
Do you guys forget, that 64-bit XP is free for existing XP owners?
[quote]”Also, as far as I know Intel has employed some 64-bit addressing extensions on their latest motherboards that deal with this problem…”
I think youre talking about intel’s PAE (Physical Addressing Extensions). It extended memory addressing from 32 to 36 bits. Im pretty sure it has been enabled in xeon chips for a while, not sure about desktop parts. PAE is slow (similar to DOS page flipping as I remember reading) and nowhere near as fast as native addressing.[/quote]
No, what he’s talking about is Intel’s EM64T extensions. These are almost-but-not-quite compatible with AMD’s AMD64 extensions. The EM64T technology is only available on the XEON line right now.
> It lasted several years before DOS and WIN31 were
> replaced by 32 bit OS.
Refer to Windows NT 3.1.
>In fact this schizophrenic situation when everybody
>possessed 32 bit hardware but run only 16 bit OS gave
>birth to Linux.
Well, businesses can leisurely move to 64bit ecosystems at their pace with the current X64 hardware.
>Debian even decided AMD64 port is not worth inclusion
>into the next stable release. Sad story.
Not a big downer in most corporate view point e.g. Red Hat, Novell SUSE, Sun Solaris 10.
A64s can be viewed as Intel Pentium IV clones i.e. mastering SSE2 and high FSBs (improvement over the old AXPs).
>attila, you do know AMD and Intel have a cross licensing >agreement right?
Does it cover IA-64?
>The revolution is here! Dump your old XP in the bay, buy
>our shiny new Windows’64!” It’s a useful technology, of >course, but probably only when it comes to breaking the >memory limit.
X64 includes register expansion to 16 GPRs and 16 SIMDs. PS; Register renaming tricks not factored.
It must be BIG. Lets forget that linux distros everywhere have been using it for well over a year already and Windows 2003 for X64 is still in BETA. And although the 64-bit addressing extensions are available, eventually. The OS still uses a 32-bit word.