Apple will switch to Intel Itanium?

ibme

Registered
Apple will switch to Intel Itanium?
Joun Dvorak prediction
PCMag 4/8/03 issue, just out

He didn't say that he had any inside information.
He came to his conclusion because:

1. January's Intel sales conference keynote speaker was Steve Jobs.
2. At the Macworld Expo, top Intel executives where in the front roe.
3. Pixar announced that it would become an Intel shop.

I hope it's true, Apple would be jumping to the front of the line.
 
anything is better than a G4

this guy is the ugliest man on earth. is that a serious picture? or did he wear a costume?
0,3363,i=1753,00.jpg
 
Hmm. Doubtful.

The Itanium is a bit of a let down for Intel. If a switch was coming to anything other than a PPC, it would probably be a mainstream x86 chip, not the Itanium. Xeon, maybe... P4, probably...

But Dvorak is wrong more than he's right, so I wouldn't put any stock in this. Yes, the facts he lays out are true (Pixar switching, SJ at the Intel conference, etc.), but they don't point to a switch.

On the contrary, the IBM 970 is a much safer bet. It's scheduled to be ready to ship in late summer, requires no retooling of the apps, and puts Apple back in the game of CPU performance.
 
Dvorak thinks Itanium for 4 reasons:

1. no evidence of Apple talking to AMD
2. Apple likes to make jassy announcements in which it claims to be the first - only servers are using Itanium at this time
3. less likelihood of OS X being ripped off by mormal PC users
4. Apple will have an ally in Intel, who will put its design team to work for Apple, knowing AMD isn't the picture.
 
Well, I know Steve Jobs is tightening up on having leaks in information on whats to happen next with Apple...I mean...NO ONE thought that a 17 inch PowerBook would come about...not until the night before they were announced. So Im sure that whatever happens will be a suprise to us all...(and Im also sure it'll be a very nice suprise at that:D :p) So Im gonna expect anything to happen...
 
I doubt this will happen, but who knows.

1. January's Intel sales conference keynote speaker was Steve Jobs.
2. At the Macworld Expo, top Intel executives where in the front roe.
3. Pixar announced that it would become an Intel shop.

Of those 3 only 1 is debatable in his theory. #2. The others are deals about Pixar. Has nothing to do with Apple there. Remember Steve is CEO of two companies that have nothing to do with either other. His purchase of Intel systems was for renderfarm machines at Pixar, to replace SparcStation servers. They will probably be all headless just crunchin away on frames.

Of the CPU theories I think IBM fits the mold better as a future CPU should they ditch Motorola. The PPC arcitecture of the CPU is more compatible with things as is. But, who knows for sure...
 
One "interesting" thing about this if true, is that though the Itanic may be a more powerful chip overall, it clocks considerably lower than the P4. So if they were to go with Itanic, they'd still have the "mhz myth" battle to fight.

Personally I don't think the whole Pixar/Intel thing means anything. Many cg houses are moving to Linux/x86 based render farms because they are cheaper and pretty much have the best bang for the buck than pretty much any other solution out there (including XServe). So that news in and of itself is a red herring.

Point 3 (less likelihood of OSX being "ripped off" by "normal" pc users) again is a non issue, since Apple would have to be brain dead to ship Itanic boxes that were basically beige boxes. They would most likely still ship proprietary boxes to continue to distinguish themselves from every other pc manuf out there, so the pirating issue is moot (until some pimply faced wise a** figures out a way to emulate the rom in software).

Point 1 is only meaningful if one assumes that Apple's switch over to x86 is inevitable, otherwise it's a big SO WHAT?
 
Originally posted by ibme
Apple will switch to Intel Itanium?
Joun Dvorak prediction
PCMag 4/8/03 issue, just out

He didn't say that he had any inside information.
He came to his conclusion because:

1. January's Intel sales conference keynote speaker was Steve Jobs.
2. At the Macworld Expo, top Intel executives where in the front roe.
3. Pixar announced that it would become an Intel shop.

I hope it's true, Apple would be jumping to the front of the line.


God I hope not. The Itanium (known in the industry as the "Itanic") is a colossal failure in the its targeted market (the high-end enterprise server business). Mostly because the first Itanium (Merced) was late in it's release (by several years), and when it was released performed on par, or less than it's competitors, AND required rewrite of all software, including the OS. Although Itanium2 may perform better (don't know), this would still require a substantial investment on Apple's part to build a system around the Itanium. For the desktop, I think this is a 0% probability. If Apple is trying to get into the enterprise server market, this would seem more likely.

My opinion is however, that Apple would be best served to stick with the Power line of CPU's from IBM. This seems like the most straightforward solution for them, but if the opportunity is there, then they might go that route.

If Apple were to come out with Itanium desktops, these things would be so expensive, they'd make the current pricess of top-of-the-line makes look like spare change! The Itanium CPU chip itself is in the $1000.00 range.
 
The picture included above is not Dvorak. That is J. Seymour. I don't know who gave the opinion above, apparently Dvorak according to to the post above, but here is Dvorak's picture


0,3363,i=1743,00.jpg
 
i know, i was looking for the article and came across that ugly bastard who looks like he's in a costume (afro?)

also HERE IS AN IDEA : Apple and Intel having some sort of business deal that invloves Intel providing help for IBM's PPC chips?

• Intel doesn't care about which side they are on as they are here to make money. Being on all desktop computers would be nice.
• IBm made a deal with AMD right? Why not R&D with Apple/IBM/Intel specifically for the PPC 970?

or if you guys can come up with something better, i'm no expert
 
Apple has too large an investment in PPC to just dump it for Intel. Things would have to get much worse for that to be a viable option.

The IBM 970 is poised to be the PPC savior. Motorola will continue to increase G4 chips, which will become the consumer chip, while the 970 becomes the Pro chip.

I also except Apple to adopt a "mhz rating" when the 970 debuts, similar to what AMD does with it's chips. Since a 1.8ghz IBM 970 will have the processing power comparable to a 3ghz P4 (not to mention it's quad core achitecure allowing for SMP on just one CPU), except Apple to develop another rating to shift the focus off the mhz.
 
Originally posted by serpicolugnut
...
I also except Apple to adopt a "mhz rating" when the 970 debuts, similar to what AMD does with it's chips. Since a 1.8ghz IBM 970 will have the processing power comparable to a 3ghz P4 (not to mention it's quad core achitecure allowing for SMP on just one CPU), except Apple to develop another rating to shift the focus off the mhz.

I don't know about this last point. Even AMD is giving up on the whole rating thing (at least in it's current form where the rating looks suspiciously like a mhz value). If Apple continues to make multi-processor/core machines, coming up with this "rating" would be specious at best.

Originally posted by Jack Hammer
also HERE IS AN IDEA : Apple and Intel having some sort of business deal that invloves Intel providing help for IBM's PPC chips?

• Intel doesn't care about which side they are on as they are here to make money. Being on all desktop computers would be nice.
• IBm made a deal with AMD right? Why not R&D with Apple/IBM/Intel specifically for the PPC 970?

or if you guys can come up with something better, i'm no expert

Don't hold your breath. Intel is making plenty of money with x86, why would they bother with PPC? They already effectively are on ALL desktops. Plus, the deal that IBM and AMD have is for IBM to help AMD, not the other way around, AND only for manufacturing. Remember, IBM is one of the largest semiconducter companies in the world (just the semicon part, not including the rest of IBM). I doubt that Intel could lend all that much to the party even if they wanted to.

Originally posted by
malexgreen

Mostly because the first Itanium (Merced) was late in it's release (by several years), and when it was released performed on par, or less than it's competitors, AND required rewrite of all software, including the OS. Although Itanium2 may perform better (don't know) ...

Well not quite here. Merced was VERY late, though by many accounts it was late because their x86 emulation sucked. Which brings us to the next point, it didn't require a rewrite since it is backwards compatable. Plus rewrite is kinda strong, since the compiler does most of the work, assuming you're rebuilding to say 64bit Windows, then you would have to do some tweaks, but certainly not a "rewrite". Also Itanic2 does perform well, with native code. Right now Itanics greatest competitor lives next door, P4 Xeon. Intels greatest problem has been it's success with ramping P4 clock speed up at such a quick rate. Hard to sell people on a 1.5ghz "replacement" for a 3ghz chip.
 
Hard to sell people on a 1.5ghz "replacement" for a 3ghz chip.

The irony is, even Intel is now actively promoting a lower mhz chip vs. one of their higher mhz chips. The new Centrino platform operates at mhz ratings of .9mhz to 1.6ghz. These chips are in direct competition with the PIII-M and the P4-M chipsets. The Centrino 1.6ghz chip is actually much faster than the 2.4ghz P4-M. Intel is already discounting the mhz myth for themselves.

Also, keep in mind the other benefits of the IBM 970. First of all, it is Altivec enabled. Second, it is a 64-Bit architecture, which doesn't necessarily mean an increase in speed, it is a next generation enhancement. Third, the 970 is a multi-core chip. Four cores on one CPU effectively means you can multithread your apps, and effectively get quad CPU performance.

The segment the 970 is aimed at, the Mac pros, are well aware of the mhz myth. Give them a 1.8ghz 970 that beats a 3.0ghz P4 in Photoshop, After Effects, Lightwave, Indesign use, etc - and the pros will line up to buy them.
 
sounds to me their should be 1 standardized test that is cross platform that they can all use to rate their chips. save everybody time and money.

like apple can say the g4 is legimately a "17893 score"
and intel can say they have a "25,000 score"
and the dual xeon servers can say they have a "50.000" score

or whatever...as long as it's easy for customers
 
Originally posted by Jack Hammer
sounds to me their should be 1 standardized test that is cross platform that they can all use to rate their chips. save everybody time and money.

like apple can say the g4 is legimately a "17893 score"
and intel can say they have a "25,000 score"
and the dual xeon servers can say they have a "50.000" score

or whatever...as long as it's easy for customers

Ah, but if were that simple, it would have already been done. There are several "standard" benchmarks out there (SPEC being one of the most well known). The problem is that writing any benchmark that provides meaningful results is very difficult. Each cpu has it's own strengths and weaknesses. Your benchmark would have to be written in such a way as not to lean to heavily towards one architecture or another. Plus synthetic benchmarks are just that, synthetic, they rarely ever provide meaningful results. As an example, the P4 with Rambus scores extrememly high on memory throughput benchmarks and certain integer benchmarks due to the higher speed rating of the memory and the memory bus. Though an Athlon with DDR will score lower, the Athlon in the real world will perform significantly better than that particular benchmark would have to believe.

Then if you try a "real world" benchmark (like running popular appications), you inevitably end up testing other parts of the system as well, the most obvious being the OS and compiler.
 
Originally posted by binaryDigit
Your benchmark would have to be written in such a way as not to lean to heavily towards one architecture or another.
Even then you'd be weighed, as strengths like AltiVec ARE quite special and SHOULDN'T be ignored.

I rather think the user does and will have to make his choice upon actually using the machine he or she wants to buy. There's nothing wrong with a hands-on test before actually buying something. Especially if it's above a certain money level, and computers, for me, usually are. Even though I trust Apple, I've always tried and used a PowerBook or iBook before buying them. Speed, actual and perceived, is only a part of the equation when buying a new computer system.
 
Originally posted by fryke
Even then you'd be weighed, as strengths like AltiVec ARE quite special and SHOULDN'T be ignored.

I rather think the user does and will have to make his choice upon actually using the machine he or she wants to buy. There's nothing wrong with a hands-on test before actually buying something. Especially if it's above a certain money level, and computers, for me, usually are. Even though I trust Apple, I've always tried and used a PowerBook or iBook before buying them. Speed, actual and perceived, is only a part of the equation when buying a new computer system.

Right, but an excellent example of "utilizing proprietary features" were early benchmarks that would utilize Intels SSE extensions, but not AMD's 3DNow. These showed the Intel chips totally trouncing the AMD in floating point performance, but obviously they were weighted towards the Intel. So your benchmark has to somehow "utilize" these things, but do it in such a way as to either be generic (which may not truely utilize the extension) or you are highly dependant on the benchmark writer(s) to be equally efficient writing SSE, 3DNow, AltiVec code.

I think that the original comment was about benchmarking cpu's specifically, not systems, to help avoid the whole which is faster 3ghz P4, 2.4ghz Athlon, 2.4ghz 970, 1.8ghz Itanic2, 1.4ghz Centrino, etc, etc.
 
I totally agree for th IBM 970...it's gonna be hopefully the new processor for next generation macs and will predictably beat any P4...
wait and see
 
Back
Top