What happened to 64 bit CPU`s..?

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Now, just to clarify further, AMD only has 64 bit chips in the works with the Hammer codename. Since it seemed easy to deduce who I was responding to in the thread I didn't bother to quote it and opted to save bandwidth instead.

Sorry you seemed to have difficulty following the thread in this case.

Mark-

When all else fails, throw your computer out the window!!!
 
FUGGER.Intel lemming! cant handle the heat so fugger off.

Dont see the point even discussing anything with you, cant even say good about p4. I wanna buy one but you make a good arguement not 2!

get a life idiot.....
 
Who really cares, your babbling on about something you know nothing about. your correct about you wasting bandwidth talking about this subject.
 
No, their 64-bit instruction set offers no performance benefits. It merely allows access to a 64-bit address space using instructions based on the same decades-old design. The IA64 is a full revamp optimized with the latest EPIC technology and will enhance performance in addition to offering 64-bit memory addresses.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
Unfortunately I cannot release any specific information on that product at this time. Everyone will have to wait to see it. I can tell you that the 64-bit aspects will be IA64, just like the Itanium, and the 32-bit aspects will be IA32 compatible with better performance than that which you see in the Itanium today.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
Yes, The itanium is a whole new design, and will beat the sledgehammer in 64bit apps, but the sledgehammer is not meant to go head two head with itanium, it is designed to be able to run 64 bit programs(which need alot of memory addressing) while still maintaining the ability to run older 32 bit code.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
 
Who really cares, your babbling on about something you know nothing about
Click to expand...

lol
Right, FUGGER. And that's different from you how?



Thanks, Raystonn. I'll just have to wait and see, I guess.



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
 
"the sledgehammer is not meant to go head two head with itanium, it is designed to be able to run 64 bit programs(which need alot of memory addressing) while still maintaining the ability to run older 32 bit code"

Could you point out some organizations that would be willing to sacrifice performance in the 64-bit arena (an investment of thousands of dollars for the CPU) in order to be able to run older 32-bit (an investment of a couple hundred for a CPU) on the same machine? This seems like a poor business decision to me. If you are going to pay thousands for a shiny new 64-bit CPU, you'll want it to be the best performer won't you? Throw in another couple hundred bucks and you've got yourself a dedicated 32-bit machine to continue running the old applications at the best possible speeds. (Chances are everyone will already have these 32-bit machines when they buy a 64-bit one. Just keep using it.)

Could you point out some people (home users) that would be willing to spend thousands of dollars to expand their memory addressing space beyond the 4GB limit? (This is all the AMD chip offers with their 64-bit solution. There's no performance benefit as they use the same old design scaled out to 64-bits.) I don't see anyone at home actually using 4GB of memory yet. Does anyone else? In my opinion the home user would be very happy with a better designed, faster CPU with the same 4GB memory range (this would be a better, faster 32-bit CPU), for 1/15th of the price.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
A: how do we know what the sledgehammer will cost, amd is notorious for low prices.

B: you would have to buy an entire computer, not just a chip to do said 32 bit apps

c: the price difference between a sledgehammer and a whole 32bit computer may not be that large at al(especially with corporate pc prices)

It seems to make tons of sense to me, why make a chip which negates every bit of software there is currently, when you can have one, which can look forward and backwards.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
 
Ok, thanks for the update, Yangsing...

Matisaro, remember that IA64 and x86-64 are incompatible. There's no reason for a home user to buy x86-64, intending to upgrade to IA64.



<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
 
Agreed. As far as being an "upgrade path" to the future of 64-bit computing, x86-64 is just as much a dead end as the old Pentium 4 socket motherboards. It's equivolent to buying a VLB (Vesa Local Bus) motherboard and a few VLB cards because you just can't wait for PCI, only to find out a bit later that you just wasted your money. (You all remember VLB, right?)

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
x86-64 is just as much a dead end as the old Pentium 4 socket motherboards
Click to expand...
Most of Intel's products in the last couple of years. But at least you don't have to buy a new OS to upgrade to Northwood.

But perhaps x86-64 will triumph over IA64, and it will be a moot point.


There was someone around here a while ago who said he had Vesa slots...can't remember who.

<font color=blue>Quarter pounder inside</font color=blue>
<font color=red>Change the Sig of the Week!!!</font color=red>
 
I know, but a home user may buy x86-64 for applications which require additional addressing space, with ia64 if you have 64 bit apps you would like to run, you cannot have any kind of 32 bit apps, but when you have several apps you need to run in 64, you can with x86-64, thats all I'm saying.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
 
Nobody is even using the full potential of our current addressing space, which is 4GB of memory using near pointers. (Much more if we wanted to start using the selector registers again.) I don't foresee the need for anything above 4GB at home for a few years yet. How many consumer motherboards even accept 4GB of memory? I don't really see the point of buying into the extra memory space at this time; extra performance from a new design, yes. For consumers AMD's solution isn't needed and IA64 is still too expensive.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
The sledge hammer is not for home, even scientists use 32bit apps too! That is the point, itanium can do 64bit very well, but its [-peep-] at 32 bit, the people who buy 64 bit cpus need them for something, but perhaps they want 32 bit functionality without buying a whole other computer, hense the sledgehammer. 32bit functionality with 64bit compatability.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
 
Decent <A HREF="http://www.tomshardware.com/technews/technews-20010809.html#0116" target="_new">scientists</A> will not be willing to sacrifice performance in the 64-bit arena in order to be able to run older 32-bit apps on the same machine. If you are going to pay thousands for a shiny new 64-bit CPU, you'll want it to be the best performer won't you? These scientists are already working with proprietary chips, including the Alpha. They do not need x86 compatibility for their number crunching. If they do they already have plenty of 32-bit machines lying around for the job. Seriously though, scientists are looking for the most processing power, not the most compatibility. They're willing to pay top dollar for it as well. If the hammer is not targetted toward home users (they don't need the extra memory) and is not targetted toward the scientific community (it will not offer performance comparable to other 64-bit CPUs), then who's going to buy it? You? I don't see a very large market for that line of processors.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
I know 64-bit seems like overkill, but I can think up a few tricks that I could perform with 128 64-bit registers. SIMD does a fair amount of specific operations but its not nearly as programmable as such a processor. Your talking about double the throughput when it comes to any bit wise operations. Sure were also talking about twice as big cache hits, but that’s what a 4mb cache is for. From what I read in the Software Developers manual 32-bit operations don’t seem like they require much of a leap.

Remember optimize 5% of your code to get 80% returns, not a rule to live by but something to strive for.

Schmide
 
"I know 64-bit seems like overkill..."

If all your 64-bit CPU gives you is access to more memory and extending the same old set of registers to 64-bits wide, then it's very much overkill for the home market. That's not much of a performance increase. It's time to let a decades old design die. We've learned much over the years. We need to put it to use in a new design. IA64 does this.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
 
well, both the architectures have their pros and cons and its really hard to tell one better than other especially we havent yet seen the Hammer in flesh and blood.

now it costs $4000 for a 800 MHz Itanium and a whole Itanium system for not less than $10k, and going by AMD pricing tradition I expect it to cost less than $3000 for a Clawhammer running at 1 GHz. As regards the performance, its really left to the respective architecture and the code running on it. Itanium software will be optimised at compile time, hence the compiler as well as the software that compiles on them will be costly.

Hammer, since it is a mere extension of older 32 bit registers, the compilers will be lesser complex and the processor will be working to optimise the incoming code at runtime. while that is what current Pentium-IIIs and 4s and Athlons do. working with such code however complex, worked fine and contemprory compilers generated code that would be fairly near to VLIW code.

I have been a great fan of Mr. Micheal Abrash and his <font color=blue>Zen of Code Optimisation</font color=blue> (too bad it went out of print) for years and have studied a fair number of compilers and the code generated by them (taking que from a old issue of PC Tech Journal, I guess it was July 1988 issue which probabely was the last issue of this great magazine!) I do concede that the compilers and assembly code then was not complex as it is today, but it does give an insight as to whats happening inside.

So I would say that properly generated code alongwith the ability of the processor to schedule it well at runtime, should deliver performance comparable to a VLIW code. the difference is in the cost and complexity of the code as well as that of the processor it is running on.

64 bit systems and the software will take a while to be mainstream, and there will be a fairly long period when both 32 bit as well as 64 bit apps will be used equally. so in my opinion the AMD approach seems more logical than Intel's whereas the Intel IA-64 will see its fruits after a while after the IA-64 compiler, code - the OS and apps will mature and become commonplace.

anyway none of the 64 bit processors are targetted towards home, but the workstation and server segment is too complex to suddenly shift to a newer architecture, newer code and new applications. it will need to support 32 bit applications as efficiently for some time to come.

girish

<font color=blue>die-hard fans don't have heat-sinks!</font color=blue>
 
Heres one, say a scientific group wishes to build a large cluster, but one with 64 bit capability. However, they also have 32 bit apps they need done, which also require a cluster. Which would be the more logical choice, a cluster of 64bit sledgehammers, with say 30 computers, which can perform all of the needed apps, or a cluster of 20 sledgehammers and a cluster of 20 32 bit pc's.(adjust the numbers to price). Like I said, the sledgehammer may not cost thousands of dollars.(we dont know what its pricing will be), However I can deffintaly see a possible market for it.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
 
they are a scientific community, always short of funds.
in practice there is no choice really. but i would definitely go for the 64 bit cluster.

<font color=blue>die-hard fans don't have heat-sinks!</font color=blue>
 
Your post to Raystonn was eminently logical and complete, I agree fully. IMHO Current performance of 64 bit code from Itanium is disappointing, good FP but poor integer performance. I am not convinced that VLIW is a good alternative architecture, of course improved compilers will help (but current compilers for X86 are STILL relatively poor compared to Assembler, so what hope for IA64 in the next year or so). X86-64 is far more interesting to me as a transitional architecture, but if I want a proper 64 bit machine I would go for an Alpha before Intel kills them off.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom