Was looking at the new itanium 2 and was wondering if it would be able to run normal sw for an os or would it have to be made for the itanium. Basically what im asking is if its an x-86 arch chip. If so wouldnt it make an uber gaming rig? Id imagine it would have out of this world performance.
Can an itanium 2 run normal sw?
- Thread starter mrgoblin
- Start date
Actually, the Itanium does run 32bit software, and NOT in emulation. But yes, it is slow compared to a P4.
And keep in mind that it is not a desktop cpu... It's meant to run high end server applications (and it does it very well.) But there is no reason to even consider it for desktop use.
And keep in mind that it is not a desktop cpu... It's meant to run high end server applications (and it does it very well.) But there is no reason to even consider it for desktop use.
No, it does emulate x86 software (except it emulates in hardware). and yes, its real slow.
Intel has said the plan on disabling/removing the hardware emulation from future versions of the chip, and doing x86 emulation in software because it has been shown to be significantly faster with minimal hit to the rest of the system.
And using these as a gaming rig...thats just nuts.
Intel has said the plan on disabling/removing the hardware emulation from future versions of the chip, and doing x86 emulation in software because it has been shown to be significantly faster with minimal hit to the rest of the system.
And using these as a gaming rig...thats just nuts.
An Intel PR person said that a 1Ghz Itanium II would be the equivalent of a 1.5Ghz XeonMP inx86 mode. Those numbers are rather ambiguous, but you'd have to be a complete retard to run an Itanium in x86 mode.
This depends on your definition of emulate. If you want to take into account hardware emulation, then all modern IA32-compatible designs are emulating x86. The Pentium II, the Pentium III, the Athlon, the Pentium 4, etc. All of these take in long CISC instructions and translate them into small internal ops.
And it depends on your definition of "real slow" as well. I use an Itanium 2 system under my desk, and it's more than fast enough for the IA32 code that I want to run. I run my simulations on IA64 optimized code and then occassionally process data using IA32 applications. It works fine. And it's nice to have the option to run IA32 applications at all. In the past all of my workstations have either been Sun Solaris, HP PA-RISC or IBM AIX operating system based systems. If I wanted, say, the latest version of, for example, Mozilla, I would often have to compile it myself and then deal with various porting problems all simply to have a recent browser. Now I just go grab the latest binary for IA32 - if the IA64 binary isn't available - and just run it.
Has anyone who is saying these negatives actually used an Itanium 2 system? I have one under my desk and it's actually slightly smaller than the Xeon and PA-RISC based machines that I have used in the past. It doesn't seem especially heavier than my computer at home - and it's certainly lighter than some of the IBM workstations that I used years ago. The air that comes out of it is warm, but it's not noticeable in our office environment. At least I have never paid any attention to it. Look at the pictures of a workstation in this .PDF for the ZX6000 from HP - it's not an especially large box.
Of course, I am biased. I spent years working on the Itanium 2 microprocessor, but I don't think that this has somehow blinded me completely. They make nice workstations, and if I wasn't involved in their development, I think that I'd still think that they are nice.
I've used them, and I really like them for IA64 based code. I didn't like the performance with 32bit code all, but I would use it again for 64bit...
as for the emulation, the x86 hardware emulation xlates into IA64 code, then down to the internal ops, which would be a solid case for emulation. If it xlated down to internal ops directly, I'd say it had a IA32 unit built in
as for the emulation, the x86 hardware emulation xlates into IA64 code, then down to the internal ops, which would be a solid case for emulation. If it xlated down to internal ops directly, I'd say it had a IA32 unit built in
pm or Wingz, correct me if Im wrong, but wasn't there a switch made from "hardware emulation" or whatever you wanna call it, to a more traditional "software emulation", ala FX!32 in Maddison?
Or is that just something that's rumored for upcoming IA-64 CPU's?
IIRC I read about this over at Ace's, where someone mentioned that this was actually faster than the hardware that Merced/McKinely has.
Oh and by the way, those zx6000's are real freking sexy, I don't like them quite as much as the Blade 2x00 family, but they're still damn nice
Or is that just something that's rumored for upcoming IA-64 CPU's?
IIRC I read about this over at Ace's, where someone mentioned that this was actually faster than the hardware that Merced/McKinely has.
Oh and by the way, those zx6000's are real freking sexy, I don't like them quite as much as the Blade 2x00 family, but they're still damn nice
I've also read this from several different sources, and was like I elluded to before.
Intel has said it will be more efficient to use software, and phase the hardware emulator off the die
Originally posted by: rjain
As long as you don't pay the electricity bill, that is.
It's not as high power as you might think. The 3.2 GHz P4 has a TDP of 82 watts, compared to 107 W for the 1.5 GHz / 6MB Itanium 2, 91 W for the 1.4 GHz model, and 55 W for the new low-voltage 1 GHz / 1.5 MB model.
Madison has the IA32 engine; the software layer is going to debut in the first service pack for Windows Server 2003.
If you're allowed, could you just briefly describe how the IA-32 compatibility in Madison works, as compared to McKinely/Merced?
From the little I've read about it, it sounded like software binary translation, much like FX!32, but from your comment, it sounds more like the x86->uOP conversion used in P6/P7/etc?
I guess I could dig up some Intel whitepapers, but they tend to be a bit...overly technical.
Considdering the closest I've gotten to low level hardware functions streches to IA-32 assembly "Hello World!" type programs, they rarely make much sense to me
From the little I've read about it, it sounded like software binary translation, much like FX!32, but from your comment, it sounds more like the x86->uOP conversion used in P6/P7/etc?
I guess I could dig up some Intel whitepapers, but they tend to be a bit...overly technical.
Considdering the closest I've gotten to low level hardware functions streches to IA-32 assembly "Hello World!" type programs, they rarely make much sense to me
Hrm...I'm not too sure how much of the details I can discuss. Madison has the same IA32 engine as McKinley (which was redesigned from that of Merced). It takes the place of the IA64 front end, and is responsible for fetching and decoding x86 instructions into IA64 instructions (among other things).
So Madison retains the IA32 engine, but the upcoming software translation layer can take the place of it.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter