Intel x86-64. **Updated 2/17** It IS AMD64 compatible...out next quarter

[Lonyo]

Originally posted by: JeremiahTheGreat
Intel's offering won't hit mainstream for a while will it? (non-server)

AMD still has the upperhand, for now at least..

AMD doesn't really have the upperhand though. Server environments are the only real place the extensions are in use, and Intel has almost get there (when Nocona is released), the desktop market doesn't matter at all right now since there is no software suport available. When the OS hits the market, if Intel don't have a chip out then, AMD will have the upper hand, but for now it's fairly even in terms of mainstream - it's just server that AMD are ahead in.

 

[grant2]

Originally posted by: WingznutDo you disagree? Do you feel the average consumer NEEDS 64-bit on the desktop? Do you feel they NEED more than 4gb of RAM?

Btw, Nocona isn't a desktop chip. As of right now, Intel does not have any plans on bringing 64-bit to the desktop real soon.

Unless I'm really mislead, prescott is a desktop processor and its 64-bit version will be out in '04?

Your CTO's quote is funny is because it contradicts what your company had (apparently) been planning for for years. He says it won't be needed until 06/07... but intel figured it might be needed in 04.

Not that I'm knocking intel.... if you can't beat 'em, join 'em. Smart move.

BTW, a poster on aces raised an interesting point... if your desktop has 1gb memory, then it's possible you have a 2gb swap file. All of a sudden you're addressing 3gb of ram which is, from a rumour I heard, the effective limit on windows x86 for some technical reason.

So do desktops need 64-bit? Someone who's looking at upgrading to 1.5gb ram + 3gb swap (4.5gb total) might say yes.

 

[NFS4]

Originally posted by: Wingznut

Originally posted by: PetNorth
Sorry, but you're wrong ;-)

An example:

Wolfram Research Optimizes Mathematica for Linux on AMD64

As a reflection of Wolfram Research's rigorous software engineering standards, Mathematica 5 is among the first technical computing platforms specifically optimized for the AMD64 architecture--and it delivers impressive performance. The optimized Mathematica port outperforms a regular Linux version of Mathematica on AMD64 systems by up to 50 percent in typical scientific and technical calculations.

I believe this is more than tremendous ;-)

Regards.

Let's take a quote from the article you linked... "Especially when dealing with large data sets, greater than the 4-GB limit addressed by 32-bit processors, Mathematica 5 will help deliver exceptional application performance."

The ability to address more than 4gb of RAM is what makes the difference. And RaynorWolfcastle and dullard are correct... The AMD optimizations also have a hand in the difference.

Btw, Mathmatica isn't an average user's desktop program. It's primarily used at the workstation level.


Look, I didn't say that there were ZERO applications that would benefit from 64-bit. I'm just saying that the desktop user that uses them will be few and far in between.

I also find it a bit amusing that you find a single example of where 64-bit is advantageous, Accord finds one where it is not, and you bash his example. Trust me (not that I expect you to), there will be far more desktop applications that don't benefit from 64-bits.

Most desktop users don't need a 3.4GHz processor either. But when AMD is forking over $200 A64 3000+ processors and $275 A64 3200+ processors that kick ass in 32-bit and perform in 64-bit as well, why wouldn't the average user want to go with it...especially if they plan on keeping their systems for a few years?

 

[grant2]

No reply... is wingznut convinced by our iron-clad logic.... or just so annoyed by us he wants to move onto better threads?

 

[Pariah]

Originally posted by: grant2

Originally posted by: WingznutDo you disagree? Do you feel the average consumer NEEDS 64-bit on the desktop? Do you feel they NEED more than 4gb of RAM?

Btw, Nocona isn't a desktop chip. As of right now, Intel does not have any plans on bringing 64-bit to the desktop real soon.

Unless I'm really mislead, prescott is a desktop processor and its 64-bit version will be out in '04?

Your CTO's quote is funny is because it contradicts what your company had (apparently) been planning for for years. He says it won't be needed until 06/07... but intel figured it might be needed in 04.

I don't see why this is so hard to understand for some of you. Intel said they didn't think it would be needed for a couple of years. That does NOT mean, Intel wasn't planning on releasing a product before then. When was that quote made? A few months ago? You think the CTO had no idea Intel was well along in the development process for these CPU's when he made the statement? It's not like he said that and then realized he was wrong, and told his engineers to hurry up and make an x86-64 CPU they hadn't started on. As a company you don't wait to release a product until after it is needed, you always release before to give time for the market to adopt it and work out the kinks before it becomes the mainstream standard.

 

[Wingznut]

Originally posted by: grant2
No reply... is wingznut convinced by our iron-clad logic.... or just so annoyed by us he wants to move onto better threads?

Nah.

The problem is that we aren't on the same page. Let me do a better job explaining my point...

When you go from 3.2ghz to 3.4ghz, there is a performance increase (unless the app you are running has a different bottleneck... i.e. High resolution gaming.) I didn't mean "The desktop doesn't need 64-bit" as in that the desktop doesn't need more power.

When you re-code an application from 32-bit to 64-bit, you aren't going to see a performance increase because of the larger range of integer numbers for the processor to use in calculations. You may see an increase because of the added registers, or application optimization for that particular cpu. But again, this is not because of 64-bit.

Now, there is one significant benefit to 64-bit... The ability to address greater than 4gb of RAM. This is pretty much the only significant benefit (imho) for desktop applications. Considering that 6gb of RAM costs in the neighborhood of $3000, and that there really aren't any apps that benefit significantly from that much RAM, it's going to be a while before the desktop is ready for 64-bit. Since greater than 4gb of RAM isn't needed on the desktop, this is what I was referring to.

Check the AnandTech FAQ, "The myths and realities of 64-bit computing" for a real good explanation of how 64-bit works differently than 32-bit.

 

[ElFenix]

the desktop may not need it, but plenty of workstations do.

 

[grant2]

Maybe someone can actually comment on the virtual memory idea.?

 

[JYDog]

AMD has done a wonderful effort in this battle(x86-64 vs Merced(Itanium)). Though for certain, part of their success is because of Microsoft. MS has benefited well also(AMD provided them a rather painless(and accelerated) 64-bit path migration with Athlon 64 for the desktop), but their(MS) weight helped cemented the issue in AMD's favor. This partnership, in itself, also sped the adoption of 64-bit(on the desktop). Intel, in all this, had to almost bum along, almost expecting the eventual outcome but just couldn't be seen as the number two to AMD. The result, we get 64-bit a few years earlier than anticipated.

 

[MadRat]

Lets face it, Intel had PAE (also referred to as PSE36) all the way back in the Pentium for virtual addressing up to 64GB, so the 4GB limit was never a realistic limit for them. ESMA has allowed Pentiums since the P!!! Xeon to physically address up to the 64GB limits of memory addressing more efficiently, versus PAE's 3-6% CPU hit, but again well past the so-called 4GB limit.

Is it safe to say that the registers were long overdue? If they could have addressed the extra registers using 32-bit or 36-bit addressing then perhaps it should have been done along time ago. Plus AMD64 allows us to chug past the practical 3.7GB limit (4GB minus the system overhead) of memory allocated to each process. Neither PAE or ESMA allowed passing a 3.7GB of memory per process no matter how much fine tuning was done.

 

[Eug]

<a target=new class=ftalternatingbarlinklarge href="http://www.mdronline.com/watch/watch_abstract.asp?Volname=Issue%20%23118&SID=1137&on=T&SourceID=00000377000000000000">AMD and Intel Harmonize on 64
</a>

MPR compared all the new instructions, modified instructions, deleted instructions, and modifications to the register files?including control registers, system registers, and registers visible to application programs. We also compared the memory-addressing schemes and many other architectural features, such as data-addressing modes, context-switching behavior, interrupt handling, and support for existing 16- and 32-bit x86 execution modes. In every case, we found Intel had patterned its 64-bit x86 architecture after AMD64 in almost every detail.

However, we also found a few differences that could make some software written for one 64-bit architecture incompatible with the other architecture. Some of these differences may be resolved in future 64-bit x86 processors, or even in future steppings of x86 processors already announced or on the market. In other cases, software can easily adapt to the differences by executing slightly different code, after first probing the CPU to learn which 64-bit extensions it supports.

 

[jhu]

i can't wait until itanium hits the desktop

 

[buleyb]

why is that? I personally agree that Itanium is a great chip, but there are plenty of features that make it ill-suited for the desktop. Did you have some better detail to what it is you can't wait for?

 

[jhu]

not really. i don't even need a computer right now