Questions for CPU Architectures in the next few years...

[Buddha Bart]

Will 64 bit take over the desktop market in the relativley near future?

Will Intel drive EPIC down to the desktop?

If so, will it allow AMD to license/impliment IA-64?

If not, can AMD survive alone with x86, or could they possibly impliment and survive with their own ISA?

If not, will Intel see monopoly litigation, or will it be able to point to POWER as its new peer/competitor?

How will Microsoft align itself through this, will it live with Intel only, or try to sustain multiple architecture ports?

What ever happened to MAJC?

Do you see everyone eventually moving toward EPIC much like everyone incorporated RISC?

Do you think that compiler design could never really get the level of ILP nessisary out of common languages to make it worthwhile?

bart

 

[Shalmanese]

Is there currently any need for 64 bit on the desktop? AFAIK, the killer apps for 64 bit is huge memory address space databases and high precision scientific applications. I suppose when typcial RAM usage jumps above 4GB, we will move to 64 bit but I guess the only real use of 64 bit on the desktop would be encryption.

 

[sxr7171]

I think it's good that they make the move now, so it will be ready and all the kinks worked out before I suddenly need (want) a machine with 4GB RAM. I don't see this happening until 2006 or so for me, but I'd like to see things in their 2nd or 3rd generation by the time I put money into it. However, your point stands about there not being a need today, but every revolutionary technology has to be invested in much before it achieves crtitcal mass.

 

[rimshaker]

Go here and read ... good practical info about 64bit computing.

http://www.overclockers.com/articles709/

 

[zephyrprime]

From the looks of it, the next Sim City will probably take more than 4GB of ram! I'm just kidding but it's surprising how a mere game like Sim City uses as much RAM as it does. linky

 

[SuperTool]

Originally posted by: Buddha Bart
Will 64 bit take over the desktop market in the relativley near future?

Will Intel drive EPIC down to the desktop?

If so, will it allow AMD to license/impliment IA-64?

If not, can AMD survive alone with x86, or could they possibly impliment and survive with their own ISA?

If not, will Intel see monopoly litigation, or will it be able to point to POWER as its new peer/competitor?

How will Microsoft align itself through this, will it live with Intel only, or try to sustain multiple architecture ports?

What ever happened to MAJC?

Do you see everyone eventually moving toward EPIC much like everyone incorporated RISC?

Do you think that compiler design could never really get the level of ILP nessisary out of common languages to make it worthwhile?

bart

I think it a lot of it depends on that last question, and what the competition has up its sleeve.

 

[SuperTool]

And yes, I do think 64 bit is coming to the desktop.
Notice how MS is basing its new consumer OS's on the business OS's. If business moves to 64 bit Windows, consumer PC's will follow shortly.

 

[rimshaker]

Originally posted by: SuperTool
And yes, I do think 64 bit is coming to the desktop.
Notice how MS is basing its new consumer OS's on the business OS's. If business moves to 64 bit Windows, consumer PC's will follow shortly.

You guys didn't read that link from my post did you? Realistic 64bit computing is a LOOOOONG way off. You know why i say that? Cause even a paltry 850MHz PIII machine is still more than enough for like 95% of the population even right now.

 

[SuperTool]

Originally posted by: rimshaker

Originally posted by: SuperTool
And yes, I do think 64 bit is coming to the desktop.
Notice how MS is basing its new consumer OS's on the business OS's. If business moves to 64 bit Windows, consumer PC's will follow shortly.

You guys didn't read that link from my post did you? Realistic 64bit computing is a LOOOOONG way off. You know why i say that? Cause even a paltry 850MHz PIII machine is still more than enough for like 95% of the population even right now.

That's cus you don't link And yes, I read it. There are more things aside from technical need driving the marketplace. At some point there is a critical mass, and then you need to get with the program or be left behind.
What if MS says tomorrow: "Look, we are already writing an OS for x86-64, IA-64, and X86 for example, and we are really sick of having to port it to 32 bit, and having to test all the patches on 3 architectures, etc. So if you guys want the latest and greatest Windows, for some reason, you need to upgrade to this x86-64 or IA-64"

 

[borealiss]

"You guys didn't read that link from my post did you? Realistic 64bit computing is a LOOOOONG way off. You know why i say that? Cause even a paltry 850MHz PIII machine is still more than enough for like 95% of the population even right now."

this is mainly why amd is releasing opteron first, and why it is such a priority over athlon64. it will be a surefire win in the corporate sector, where systems easily cost 10k and up. here you get cheap 64 bit capability and huge memory addressing, as well as a good software base to boot for peanuts in comparison. having athlon64 released later on in the year is just planting the seed for 64 bit computing.

but with this in mind i doubt 64 bit computing will gain momentum on the desktop market for a while. imagine how long it took SSE, 3dnow, and mmx to gain momentum. now we're talking about doubling the precision/addressing space of the desktop. the applications aren't going to be there for a while. but while it is there, it certainly doesnt hurt having it, as more applications get ported over to 64bit capability. i can see this to be a major selling point for games.

as for the future of computing architecture, i'm guessing something along the lines of transmeta's crusoe technology will emerge, where cpus get fast enough to have a software frontend for a lot of the decoding that can be changed fairly easily. then you could target any market. if things go the way of itanium though, we're going to need more talented compiler writers, which i hope isn't the case. but i doubt EPIC will come down to the desktop level for a long long time. i think this architecture will be doomed for the highend unless they introduce some code-compatibility layer that simply makes it a competitive x86 solution.

currently, amd can't survive alone with just x86 i think. which is why they have their hands in many different areas of the market, nics, flash, embedded devices, etc...

MS will probably stick to 1 generalized ISA, and probably implement extensions to specific processors. they ported NT4.0 for alpha, but that went the way of the dodo.

 

[ztadmin]

64-bit computing isn't really all that far off when you consider that there are a number of x86-64 reading linux distributions, and MS is developing x86-64 windows. There's widespread industry support for the 64bit side of x86-64 already, without x86-64 capable processors even out, and the fact that it supports 32bit x86 apps makes it the perfect ISA. You may not need an Athlon 64 or Opteron to play Unreal Tournament or use MS Word, but professional rendering, large databases, etc... will be able to take full advantage of x86-64 very soon. Once the enterprise and enthusiast users hop on the bandwagon, the masses will follow shortly after, I think. I expect 64-bit computing to be commonplace by the end of 2005.