Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
...everybody had better quit whining about the pro line for a long time. Dual channel DDR 400..... mmmm..... 6.4GB/sec ......

Apple's pattern is to leapfrog the competition followed by anemic updates for the next couple years. If they are ever going to be more than a two-bit player, they have to stop this practice and come out with a continuing stream of killer hardware/apps.

And while many may point at Motorola as the cause for most of the performance issues, there has always been someone at Apple who signs the agreement with Motorola.

So no, I won't stop my whining until I see a sure sign that Apple is doing something to address this problem. A move away from Motorola would probably do it (for me). Even if Motorola comes out with a decent G5, I'd rather Apple packed their bags and made a move to IBM.
 
Re: Re: Dumb question

Originally posted by szark


The G4 is the main CPU, so it is definitely necessary...

Are you a PC user who is looking to switch?

Nope, I use a PowerMac G4 500mhz.

But I could definitely use a speed bump.
 
BTW, people keep calling this the "nforce2 motherboard". It's not a motherboard, it's a chipset. Nvidia doesn't make motherboards. Nvidia doesn't make graphics cards for that matter. They make chipsets. Other companies make the motherboards and graphics cards using Nvidia chipsets and Nvidia specs.

So, if Apple used it, Apple would be making the motherboards themselves -- just like they do with the graphics cards.
 
they may win me back yet...

If this is for real, I must give huge props to Apple. 1) For restoring my confidence in the most innovative computer mfr out there, and 2) for keeping this under such tight wraps for so long.

In just 6 short hours we'll know the truth! nForce on a Mac would be off the hook.
 
I remember reading in an article that Apple was licensing a technology from AMD that was used in dual processors. This was about 6 months or more. I could imagine that AMD might fab chips too. ahhh.. rumors.

This is not the article I was thinking of, but it's interesting even if it's old. I'm sure someone has posted this in the past. http://www.theinquirer.net/?article=2416

Another article, go to the bottom - rumor
http://www.newsfactor.com/perl/story/16589.html
 
hmmm. this is amazing and all, but something doesn't feel right, and i don't know what. being on the west coast i have some sources, and i haven't heard anything here. in any case, i hate to say this, but power macs probably WON'T be released till the 13th. Now whether that means introduced and released, or introduced tomorrow and released on the 13th, i don't know.

but people, now that we know about this chipset, all our hearts will be broken now if apple DOESN'T use this! so try and hold back the enthusiasm, i want it too, but let's just see....
 
nforce2 will not happen with mac

nforce2 does not support firewire 2. i dont foresee mac ever useing the nforce chipset unless they have decided to drop firewire for usb2.0

mabey mac might do somthing with a future AMD chipset. havnt seen any specs on it yet but im sure they have them ready and waiting for the hammer CPU's to come out.
 
Re: nforce2 will not happen with mac

Originally posted by Aris620
nforce2 does not support firewire 2. i dont foresee mac ever useing the nforce chipset unless they have decided to drop firewire for usb2.0

mabey mac might do somthing with a future AMD chipset. havnt seen any specs on it yet but im sure they have them ready and waiting for the hammer CPU's to come out.

there might be a treat (finally) for when the powermacs are released in mid august (if they do release them)
 
Re: nforce2 will not happen with mac

Originally posted by Aris620
nforce2 does not support firewire 2. i dont foresee mac ever useing the nforce chipset unless they have decided to drop firewire for usb2.0

mabey mac might do somthing with a future AMD chipset. havnt seen any specs on it yet but im sure they have them ready and waiting for the hammer CPU's to come out.

There's nothing stopping Apple from adding Firewire2 support externally from the nforce2 stuff.
 
I remember reading that the nForce2 was able to interleave 2 memory banks together for some dual channel, twice the memory bandwidth action somewhere. Didn't Apple come out with that first back in those old powermac 9500 towers?
 
This will never happen....

I'm sorry everyone, this won't ever happen. Nvidia will NOT produce a chipset that supports AMD and PowerPC.

The truth is, the architectures are radically different. They don't use the same instruction set, the aren't even the same Endian (that means the order of the data is reversed between i386 and PPC code). The processors are TOTALLY [excuse the pun] apples and oranges.

There is no way that AMD will produce a chipset that supports both processors. I don't believe it is even possible, irregardless of how extravagant the design is.

I know that some of you wish and hope so much that you will refuse to belive me. Let me go through this point by point so that you can, hopefully, see reason.

1) The processors are too different to run on the same chip set.
* Have you EVER seen one chipset that supported Intel and AMD processors? I can't think of any since the Pentium (1) days. Why can't the nForce handle Pentium 4's? Why can't KT333 support P4s? Surely it would make good business sense to use the same chipset in multiple motherboard designs, right?
If it is so hard to make a chipset that handles K7s and P4s, how are they going to do this with RADICALLY different chip architectures?

2) Chipsets are designed to recognize certain processor buses.
*Read above... Now why is it so hard to support K7s and P4s on the same chipset? Perhaps because the K7 bus is based off the Alpha bus and the P4 uses a totally different bus? Well guess what, PPC chips use yet a different bus spec. The current one is MPX, but they are moving to yet another bus.

3) Apple's inclusion in the Hypertransport consortium does not mean they can use nForce2.
*Hypertransport is a bus that the chipset uses. It is not, as far as I understand, the processor to chipset bus. I could be mistaken about this, but I believe Apple planned to create a hypertransport chipset, but to interface it to a G4 with the BookE bus spec.
This is what
The Hypertransport Consortium has to say about it...

"An example is the NVIDIA® nForceTM chipset that utilizes HyperTransport™_ technology to deliver up to a six-fold increase in bandwidth between the nForce Integrated Graphics Processor (IGP) and nForce Media and Communications Processor (MCP). In fact, the bandwidth supplied by HyperTransport™_ allowed NVIDIA to implement full Dolby Digital 5.1 3D audio processing and broadband networking in the MCP, something no other existing technology allowed them to do."

Notice this has nothing to do with the processor to chipset bus.

4) The so called proof of this rumor is a general statement by Nvidia that says, "You are going to see many more exciting introductions in [the Mac] space".
* Did anyone consider, for even one second, that Nvidia was talking about new Graphic Processors coming into the Mac space? Why does this have to refer to the nForce2 when it is (most likely) completely impossible to make a chipset that supports both the PPC and Athlon?
*The other claim is that, the fact that the integrated graphics core supports Apple's Endian-ness, the whole chipset will. Supposedly Nvida confirmed this. I have to doubt this. I believe that the big reason that other graphic chipsets won't work on the mac has to do with the way that Macs handle video at a low level. RGB datastructures are different on the PC and the Mac. It doesn't (at least directly) have anything to do with endian-ness. In fact, PCI is the wrong endian-ness for the Mac, but the mac chipset does a byte reorder on all data bound for PCI. (this is how I understand the process to work). If I understand Endian issues correctly, the other rumor is patently false, and it was generated by someone who doesn't understand hardware as well as I *think* I do. ;-)

OK, OK, I sure as hell don't know everything. I do know enough to realize this though. I've come to accept that there are only three steps in computer knowledge... 1) I don't know anything, 2) Boy, I sure know a lot, and the ultimate... 3) Boy, I never realized how much I didn't know. I feel I've reached the third level.
What I'm trying to say is, I don't design chipsets. I don't have a PhD in Engineering. I do, however, follow hardware closely... as part of my job, and as a hobby. I'm pretty sure I know (mostly) what I'm talking about.
.... take it for what it's worth. If nForce2 comes out for the mac next year, I'll be the first to admit I was totally wrong. However....I'd bet my admittedly small savings account that it won't happen.
🙂


.....ffakr.
 
Originally posted by topicolo
I remember reading that the nForce2 was able to interleave 2 memory banks together for some dual channel, twice the memory bandwidth action somewhere. Didn't Apple come out with that first back in those old powermac 9500 towers?

Yes, the 9500 could do this. I believe the 8500, 8600, 9500, and 9600 could all do this. If you installed matching DIMMs in alternate banks, they would interleave. I belive it was slots 1&5, 2&6...

Apple only generated about 10% performace increase by this though. The memory bus was slow, but the processors weren't hugely faster than the bus. The fastest machines had a side band L2 bus that helped alleviate bandwidth issues too.
I've seen synthetic memory tests on the new interleaved systems that were MUCH faster. That doesn't, of course, mean the same for real world performance. If you check reviews, real world apps didn't hugely benefit when the K7 went from SDR to DDR (effectively doubling bandwidth).

Apple didn't 'invent' this though... PC servers used interleaved memory (2 up to 8 way) for ages.
 
ffakr, You mentioned a few times there somthing along the lines that AMD would not produce a chipset for AMD and PowerPC. You're probably right. The thing is I don't believe anyone here has said that. I think you're misreading things. We're talking about nvidia and frankly I think it's very possible. Don't you think someone like nvidia looking to get into the chipset business and who caters to performance would want to grab most of the share of the Mac market with their chipsets if they could sell Apple on it? Handling the endien crap shouldn't be too difficult if that's even an issue. Who says it has to be the same chipset but merely one that will support PPC with all of the other features?

Frankly it makes alot of sense to me for Apple to look into this as they've been criticized for not being able to keep up with the industry and nvidia is attempting to lead it. Apple would then have a better foot in the door for going with other cpu's from the amd/intel world.
 
Actually, I don't think the rumor was ever clear as to whether it was one chipset or two chipsets. I'll re-read it but it came across as 'the nForce 2 chipset will support PPC and i386'.

Sure, Nvidia design a chipset that supported Athlon. They could make one that supported P4 (if they can license the P4 bus from Intel). They could support the PPC too... though that would take more R&D than the Athlon to P4 move would require.

Now lets look at this with reality in mind. Apple has around 3% market share. A great nForce2 chipset could hope to claim 5 to 10 percent of the PC market. 5 to 10 percent of the PC market would be >4 to ~9% of the market.
All companies have a finite ammount of R&D. Would Nvidia be better off making a P4 nForce2 or a PPC nForce2? Would nVidia be better off making a PPC nForce2 varient or fighting ATI in video R&D (ATI is currently top dog again)?
Plain and simple, companies [are supposed to] do what is best for the company. They put money, research, and effort where they do the most good.
The best move for nVidia is NOT to develop a varient of the nForce 2 for Apple, but rather to push farther in to the i386 desktop while pushing back into the lead of video chip design.

Other argument... if Apple could outsource the chipset at a cost where nVidia could make a profit, it would make sense for APPLE to keep that profit in house. Why pay someone else to do something you've been doing all along. The only reason would be if nVidia had the bulk of the R&D costs taken care of already. I'd assert that Apple's R&D expendetures already put them in a better position to finish the next chipset... nVidia would have to do too much to re-engineer the chipset for PPC. In short, Apple is closer than nVidia would be in any effort to re-engineer an x86 nForce2 over to PPC.

So... to arguments here:
1) Nvidia makes one chipset that works for a variety of VERY different processors, or...
2) Nvidia re-engineers their chipset over and over for increasingly smaller markets.

Neither of these sound likely to me. I think it is quite possible that nForce may support P4 in the future if they can license the P4 bus (remember VIA and Intel are in court over use of the p4 bus). I don't, however, see nVidia doing major redesign to accomodate a totally different processor family... especialy when the nForce 2 is targeted to all-in-one type systems (everything is integrated). That means, they would likely only get put into consumer Macs... which wouldn't happen if they performed better than Apple's own towers. Other option would be to put them in all of Apple's desktop machines. This would mean Apple would have to integrate a FW controller on the board, and they would have to ship redundant hardware on the tower... the Tower simply isn't a 'built in video' type of machine. It is supposed to be a pro machine, and nForce isn't about top of the line video... it is all about 'all in one' design, and yesterdays video using system RAM.

just my 2cents.
ffakr.
 
Now lets look at this with reality in mind. Apple has around 3% market share. A great nForce2 chipset could hope to claim 5 to 10 percent of the PC market. 5 to 10 percent of the PC market would be >4 to ~9% of the market.
All companies have a finite ammount of R&D. Would Nvidia be better off making a P4 nForce2 or a PPC nForce2?

AMD has a stake in the development of nForce ( HyperTransport ) and would not let the technology be adapted to an Intel chip. Apple, however, is a member of the consortium which governs Hyper Transport, and thus has a better chance of using this tech. Also, as stated before, nForce is a chip set. Apple should be able to incorporate it onto their own motherboards.

Another note, which I don't believe has been discussed here, the second series of nForce mobos will have the option of no onboard GPU.
 
Originally posted by Cappy


You're arguments for the most part are arguably valid but you keep throwing AMD's name in place of nvidia's thus potentially confusing people who may be reading this. AMD is *not* making this chipset. This is *nvidia*. You did this in your first post and you did it again. Please fix it!

totally my mistake, I'll edit that.
Sorry,
....ffakr

I think I got them all. Thanks for the heads up. I'll delete this after Cappy's post is deleted so no one get's confused.

....ffakr
 
Originally posted by jethro421


AMD has a stake in the development of nForce ( HyperTransport ) and would not let the technology be adapted to an Intel chip. Apple, however, is a member of the consortium which governs Hyper Transport, and thus has a better chance of using this tech. Also, as stated before, nForce is a chip set. Apple should be able to incorporate it onto their own motherboards.

As I stated, hypertransport is a bus that is employed by the chipset, internally. Hypertransport isn't used to interface with the Athlon (or the future PPC). This is how I understand Hypertransport to work and what I've read at The Hypertransport Consortium seems to confirm this.

This means, Nvidia (and AMD) are bullish on Hypertransport for their CHIPSETS, not for their processors. If you could license the P4 bus, it would be a lot simpler to use a Hypertransport chipset with a P4 than it would be to use it in a mac, where the processor architecture is radically different.

One more thing that occured to me. I was thinking about what types of interactions the chipset and the processor have with each other.
PPC and i386 are different in not only the language, but in the very basic way they express their data. CISC instructions are fundimentally different than RISC instructions. How does a CISC chipset throw an interrupt to a RISC processor? They handle instructions very differently.

Perhaps I'm getting over my head... as I've said, I don't design chipsets, I'm just a hardware geek who keeps up on this stuff (and a computer professional that runs an office in charge of an entire Universities computer repair ;-)
I'd love to hear corrections on any of my assumptions if I'm off base here.

Unless I hear otherwise [from a creditable source] I'll have to stick with my guns on this one. 🙂

.... just a stupid ffakr.
 
Originally posted by ffakr
This means, Nvidia (and AMD) are bullish on Hypertransport for their CHIPSETS, not for their processors. If you could license the P4 bus, it would be a lot simpler to use a Hypertransport chipset with a P4 than it would be to use it in a mac, where the processor architecture is radically different.

What are you pointing to specifically on the processor architecture being radically different in terms of use with this technology? I only ask this because the PPC can do both big and little endian. I realize there are other differences but for Hypertransport to be pushed into the fields they've talked about I find it hard to believe that it couldn't work well with various architectures with little work relatively speaking.
 
There is absolutely no technical reason why Nvidia couldn't design an nforce2 chipset that would work with whatever processor they wanted it to work with -- AMD, PowerPC, Intel, 68000, 6501, etc.

All it would take for Nvidia to make a chipset for Apple would be a contract with the signatures from the two companies on it.

This actually makes sense in another way. This would offload a LOT of the design work from Apple to another company with more experience in the areas where Apple is lacking. Rather than have Apple reinvent the wheel, they can adapt to one that someone else has made. Most every PC maker uses someone else's chipsets for a reason.
 
Originally posted by ffakr

Apple didn't 'invent' this though... PC servers used interleaved memory (2 up to 8 way) for ages.

And, of course, heavy memory interleaving in supercomputers, etc, goes waaaay back to the sixties.

One thing I didn't see mentioned about the nForce dual channel approach is that you only get the benefit of it if you're using a dedicated video card (and even then, at least on first generation nForce boards, the actual memory bandwidth vs theoretical was a little disappointing). If you're using the on-board video, one of those channels is pretty much dedicated to keeping the GPU fed (since the built-in GPU uses system memory).

Al
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.