Intel Core 2 Quad on MBP and iMac. - Page 2

I hope it was not your intention to be so offensive. Why don't you debate the argument and stop calling people pathological?

Quad-core processors at 45nm consume too much power for use in laptops, in my opinion. With 32nm processors coming out in the next quarter, I hope Apple waits. If you disagree, that's fine, but it doesn't make me pathological.

Personally, I'm not particularly interested in any MacBook Pro. I'll buy the first MacBook Air with 4GB of RAM, which I expect will coincide with a speedbump to the Arrandale processor.


Arrandale will be much more power efficient than C2D + 9400M or Clarksfield. Ultimately, that's what's important in a laptop. For a given level of power consumption/heat dissipation, how much performance is available? By this metric, Arrandale beats the pants off Clarksfield.

The numbering does suggest it may be from the desktop class Core 2 Duos though, which are numbered Exxxx and yet the iMac uses laptop 1066MHz Ram.

The Ram suggests that it's a laptop class processor in which case the cost of it should be much higher than the desktop processors. Although you have to consider how Apple manage to amass their $25 billion cash reserves:

cash reserves:
2003 - $4.6b
2005 - $8.7b
2008 - $19.4b
2009 - $25b

10 million Macs per year 100 million ipods making about $3-5billion per year.

So 10m * Mprofit + 100m * iprofit = 3b

The ipod is said to be about 25% of the total so we'll say 10m Macs = 2b profit - this means that they make $200 profit average on each Mac?

Like I say, their processors can't be as cheap as $350 in a $2200 model if they only make that amount of profit. They would obviously sell more of the lower models, which make less money but I think they have to be laptop-class CPUs and in the $550-850 price range (accounting for the $300 difference between models).

When we all stop behaving in such manner I'd love to. I find it ridiculous that in discussions that so many Apple users are stuck in this "speculative" mode from Apple's PoV. I love speculating as well but I can separate my discussion and argue about what's in my best needs versus what's in Apple's best needs.

The 1.66Ghz Clarksfield Quad-core is 35 Watts TDP. That is the same as the current Core2 Duo 3.06 processor available in the 15" and 17" as CPU upgrades.

Apple's casing is pure aluminum, if their engineers cannot coax another 10 Watts out of that case they should hang it up.

Arrandale is a "value" processor. It's going to be in laptops under $1000. This year EVERY PC vendor selling laptops will have Clarksfield options in their lineup. So the whole "Apple's going to wait for Arrandale" commentary is quite odd because I'd like to see how Apple intends to deliver "Pro" laptops with processors that are being used in the $699 laptop specials.

Clarksfield's TDP has yet to be confirmed.
Since Apple will announce i7 generation on february 2010 they'll likely use high-end 32nm Arrandales for the whole MBP line. They'll add 32nm Quad cores (Clarksfield successor) on October 2010 with some other minor updates (IMO)

I really doubt Apple will go with 45W Quads just to be on par with other manufactures. And I'd prefer dual core in MBP due to power consumption and usually higher clock speeds with lower price.
It seems logical to me.

High end Arrandale is almost an oxymoron. It's integrated nature is suited towards general purpose high volume computing. I still don't get how Apple is supposed to market an upscale premium product based and a "value" processor from Intel.

The real issue is whether Clarksfield and Arrandale offer Turbo Boost Mode.

If Clarksfield does but Arrandale doesn't that means that Quad Core 1.66Ghz Clarksfield chip runs at 2.4Ghz for dual core and potentially 2.8Ghz for a single core all within the same 35 Watt TDP envelope.

I'm guessing that Arrandale will not have this because it's aimed at the value segment and contains an ondie GPU.

Apple has an aluminum chassis and there are other methods for cooling the system down like improved heat piping or better fan systems.

The problem with your premise is that it's entirely based upon one foundation and that foundation is maximum TDP. If Apple re-engineers the notebooks and they manage 45 Watt TDP safely then that opens up the potential for

1.66Ghz Clarksfield (est 35 W TDP)
1.73 Ghz Clarksfield (est 45 W TDP)

So the question is "what's the harder task?"

Convincing consumers to pay a premium for a "value" procesor or engineering the MBP to support 45 W TDP processors ( which is only 10 W TDP over what they support now with the 3.06Ghz C2D procs)?

It something to consider with these new class of processors. Just because that chips power usage goes up does not imply that total system power profile goes up. There is a resonable expectation that the total system power will go down. This due to more functional blocks located on the system processor.

As to chips coming from Intel I can't imagine that Apple is happy right now with intels line up. I just can't imagine them accepting an embedded GPU after all the OpenCL success they are having with Nvidia. We can all hope, along with Apple, that Intel will make a reasonably well performing and bug free GPU but we all know the track record here. I wouldn't doubt if behind the scenes Apple isn't trying to strong arm an agreement between Intel and Nvidia.

In the end if they can't work something out with Intel there is always AMD which frankly seems to have a better lineup of processors to suit Apples mentality.


Dave

When Apple added the 13" Macbook Unibody to the "Pro" lineup they also did another interesting thing. They created a IGP 15" MBP and reduced the pricing by 300 dollars.

This is a wise move by Apple.

1. It signals that IGP graphics "can" come in a Pro Notebook.

2. It allows them to target markets that aren't dependent on discrete graphics

3. It will allow them to toggle between IGP and discrete graphics on future versions.

Which is nice because"

Arrandale will likely become the IGP Pro notebook choice for a reduced price.

Clarksfield will be the higher end version with discrete graphics.

Nvidia is going to get pushed out of the IGP biz but they should clean up on the discrete GPU biz with 200 series GPU mated to Clarksfield.

apple going to ANY INTEL BASED on board video is a bad sign ati and nvidia have done good on board video at the same price or less then intel POS GMA video.

If the market accepts an IGP as a viable solution. You seem to have things backwards as the signals go from the market to the manufacture. Remember things like the cube and AIR, Apple has influence but they can't change the fact that people have standards. Sometimes those standards aren't rational but hey they are the customer.

All that aside from what I can see right now the 9400M is the best IGP going. Given that I'm not sure it is good enough for me. Unlike the dual core quad core debate at least they are giving the customer a choice.

yes it does. Further more it lowers the power demand resulting in an even longer battery lifetime. Battery lifetimes may be a bigger issue than people give credit to. It is obvious Apple is marketing this feature hard.

Eventually descreet graphics will go away except for the extreme high end. It isn't really even debatable anymore.

God I hope not!!!!!!! I'd rather see Apple cross over to AMD. One thing you should never do as a manufacture is to reward your suppliers for their own stupidity. Intels little war with Nvidia is frankly disgusting an a notable abuse of power.

We have to be careful about terms here. Clarksfield could have what amounts to integrated graphics, it just that what is in that integrated chip is different from todays integrated system chips. Of course if Intel doesn't license the bus we might not get anything from alternative vendors like Nvidia.

This remains to be seen. Frankly I see intels behaviour here as something that will draw regulatory oversite and maybe even legal action. If not in the US certainly in Europe. It will be interesting to see how this unfolds.

Even on Clarksfield though I think Nvidia and intel are at odds. The place to put the GPU is on the DMI bus which intel is apparently not licensing. Unfortunately this subject is a bit opaque so I don't have a good idea of the depth of the problem. Nor do we know how influential Apple can be here.

Apple has various ways to apply influence and one of those is to buy from AMD. Honestly I think they would be a better choice for some of Apple machines. An AMD based Mini would rock and maybe even be cheaper. If nothing else it would send a message to intel to get their act together.



Dave

I challenge your assertion that Arrandale will be a low-end processor. Increasing integration is not the enemy of high-end computing. Quite the opposite is true. Increasing integration is one of the drivers of increased performance. Putting the GPU and the memory controller on the same chip package with the CPU increases the communication bandwidth and reduces latency. The next step will be including the GPU and the memory controller on the same die with the CPU.

If your argument were true, then "high-end" computers would be built with discreet transistors rather than ICs.

Arrandale won't be an low-end CPU.

Both Arrandale and Clarksfield will support Turbo Boost.

Apple have to use IGP in all notebooks for better battery life option. Would they choose G 210M or Arrandale's iGP?

What about triple graphics solution?
1. One in Arrandale (lowest power consumption)
2. Second in chipset (like 9400M)
3. Third as discrete powerful one.

What about disabling 2 of 4 Clarksfield cores and some clock speed limitation in power optimised mode?

And until quad cores go 32nm they won't be as power efficient as Arrandale.

I agree with this. Intel need to quit their petty arguments and start thinking about the consumers. Their behavior over USB3 and not licensing their processors for use with Nvidia chips is highly objectionable and the worst thing to do would be to buy Intel's chips and drop Nvidia's IGPs. I can't see it happening to be honest. I think Apple will have to add an Nvidia GPU along with Arrandale.


I think Turbo Boost won't be that impressive. Intel say that the processor may only allow something like a couple of frequency steps when other cores are idle and it depends on temperature and power draw. 266MHz increase is better than none at all but only about 15-20% speed boost.

But Arrandale won't be lower end as you said. It will have higher clock speeds than the quads. Clarksfield will be 1.6-2GHz but if Arrandale matches today's dual cores, it will be 2.5-3GHz. Still slower overall but more battery life and still 4 threads.


Sometimes higher end GPUs can be better if they can support hardware encoding/decoding as well as OpenCL. It will be beneficial to have something like a G210M if Apple can shunt some processing off the CPU onto the GPU, which may do it more efficiently while allowing the Arrandale IGP to render the interface.

I think 3 chip solutions would be overkill. On the low end laptops, Arrandale + 210M and higher end, Arrandale + GT 230/240M. In the Mini, same as the lower end laptops and in the iMac, 1.6 quad core Clarksfield with GT240M in the lower models and 2GHz quad + GTS 260M in the higher ones.

The laptops have been refreshed so in September, an iMac refresh will have to come to keep the sales ticking over. There's nothing to go into the Mini or Mac Pro until 2010.

This is certainly true if you are at the right point in the process shrink cycle where you can add the additional logic without compromise. That and you need competitive IP. This is what makes one leary about Intels plans, they simply don't have a track record with respect to graphics processors that is acceptable.

So yeah you are right in a perfect world it might actually would result in better performance. Unfortunately Intel is less than perfect in the world of graphics processing.

actually it is highly debatable if it is wise to put that GPU on die with everything else. I can make a very good arguement that the GPU would be the last thing you would want on die. For one thing demands for incresed GPU performance continues to move forward, locking the GPU into a high integration chip means slowing down that growth in performance. Second, it greatly reduces options where the customer needs them the most. It is not like you see a lot of customers saying they need to change their north bridge because it doesn't perform well with the last piece of software installed.

While an on die GPU might make technical sense in some cases today, I do not believe this move on intels part has anything at all to do with technical merits. Instead it is a calculated move to damage nvidia.

I think that is going a bit overboard, though I can't read the original posters mind I think it is a matter of the processes not being ready for GPU integration yet. Especially if it means dual core over quad core. Personally I think what we need to do here is say wait a minute is this really a step forward. It is not in my estimation.


Dave

In this day and age a dual core CPU is low end! Turbo boost is nothing more than a marketing gimmick.

I would hope that the choose the better performing option. Frankly if it wasn't for Intels history with GPUs we might not be having this discussion. The point is we don't want to see a regression in performance simply to save power. Given the right mix of tech we ought to be able to save power and increase performance over todays systems.

talk about wasted power. In any event this highlights my point that now is not really a good time to be integrating a GPU onto the CPU die. In many cases it will end up being a waste. This mainly due to the wide ranging performance needs of the end user.

Ah but they are more power efficient. If you double the cores your power usage dies not double. The processor may use more power but it is also more efficient.



Dave

From my experience given the same technological process and clock speed doubling core number will result in close to 2x power consumption.

Turbo Boost isn't a gimmick with Quad cores (but probably is with Dual cores) since most demanding applications don't use more than 2 cores effectively.


What do you think about iGP + 210M for 13" and iGP + 250M for 15" and 17"?
Arrandale's iGP will be used for better battery life still supporting OpenCL and hardware HD video decoding.

You'll still able to use better graphics on both machines.

Decreased graphic performance in battery saving mode is something I can live with if it result in lower power consumption. Even 4500M HD's performance is ok for battery life mode (but it's not power efficient as it doesn't run significantly cooler than 9400M)

Apple has the best battery tech in the laptop biz right now. It's absurd that they would forgo on Quad Core for battery life when they have a distinct advantage.

I've seen no confirmation of Turbo Boost in Arrandale or Clarksfield for that matter so I'll wait until Intel releases official specs.

Arrandle is in fact not low end but the volume/mainstream chipset. As I've said before they will be in low cost notebooks far cheaper than the 1200 and up Macbook Pro series. No one has answered my question.

"how does Apple market a Pro laptop using the same chip that's in a $700 home notebook?"

According to Wikipedia Arrandale is est at 18/25/35 Watts TDP. The low end Clarksfield is est at 35 Watt TDP. Other than the potentially slower GPU I don't know where you all think Arrandale is going to make a significant impact. Only the lowest end Arrandale is under 20 watts.

It's a lateral move as far as consumption considering your graphics probably won't match the Nvidia GPU.

You'll probably have second GPU in 13" as well so why complaining about graphic's performance?

And TDP is for both CPU and iGP. Given clock speed difference Arrandale will result in better overall efficiency.

Why do you say Arrandale to be cheap? What's the difference between future and current Quad cores and Dual cores product placement in mobile market?

Again, no matter how large current batteries are, Apple probably won't go with 45W CPU. At least I hope so.

And Intel won't restrict their line of high-end mobile CPUs with hot Clarksfields.

They'll use models with higher clocks and more cache. Just like now.

Turbo Boost mode makes lower clocked Quad-cores feasible in a notebook.

It neatly addresses the marketing issue that arises when consumers say

"my last notebook was a 2.5Ghz ...I'm not paying for a 1.66Ghz downgrade"

Today is different than a decade ago. You go to the airport to fly out there are AC stations
for phones or whatever device you need to charge.

There's AC jacks everywhere (if you can beat the rush). Our environments have changed so that battery life is rarely an issue for people unless they have a long flight without access to AC.

Let's address another issue. Do we really think Apple spent all the time and energy in developing Grand Central Dispatch only to fail to deliver QC iMacs and Macbook Pro? Sure you'll see some improvement with dual core but really quad cores and above are going to see very nice benefits and GCD makes it pretty easy to build in concurrent support for many applications. I expect that in a year we'll have a lot of developers who will have opted in and it'll be evident how important QC Macs are.

If it takes something like Turbo Boost to make quad core possible in notebooks then Intel is doing something wrong. I really can't see the usefullness of something like this and a lot of frustration for the user.


So don't market GHz! It's been shown again & again that it doesn't correlate from processor to processor implementation.

This will have to be adressed anyway because of the significant performance per clock difference. GHz is the old approach to sneaky marketing, they will have to find other deceptive practices.

I'm not sure what the battery life issue has to do with this discussion. My point is that Apple needs the option of quad cores on it's laptops and iMacs. It is something they could use now but I understand the wait for the next gen. If you want battery life the solution is easy, buy a low end machine.

They are already failing. Any IMac sold today is not going to be able to leverage SL the way it could if it had even this generation quad cores. GCD is a wonderful thing if there are cores to despatch threads to.

We are in agreement then. SL really needs more than two cores.

Exactly! But developers are only part of the issue, the extra cores and threads should lead to smoother operation when multiple programs are running.

Will the transition to quad cores be as noticable as the move to dual was? That depends upon just how good SL is but I suspect quads will be seen as worthwhile fairly quickly. If any of the leaks about SL and Apples pro software are true, quads might be seen as mandatory right off the bat.

Dave

generally I'm of the impression that it is less than 2X. That versus two processor chips. The reasoning is that you eliminate a whole set of bus transcievers. Bus interfacing eats lots of power.

Now adding two more cores to a dual core processor is a different story but you still have a fixed number of IO pins to drive. This is one of the reasons Intel implemented two DRAM channels on it's low end or mobile chips as opposed to three it has on some of it's other new products.

I wouldn't be surprised if Apple turned the feature off. It flies in the face of all the effort they put into SL to use all of those cores. Besides your arguement isn't even true today, some of the most demanding apps use four cores very effectively. Frankly Turbo boost is a feature for workloads and software of the past.

Well i hope Apple has nothing to do with Arrandale. If they are stuck with it then I suspect that IGP will be the only choice on the 13". On the other machines Intel needs to license the DMI bus in my estimation. This would allow for next generation graphics processors. Frankly this is where I see Intel really trying it's best to exert excessive control over the market. People want better performing machines and to see intel publically squash attempts to move to the next gen is disheartening to say the least.

I truely doubt that on the 13" machine. As to the others it depends upon what you mean by better. Better as in a standalone GPU or better as in a new generation.

Frankly the only way I could see IGP, on Arrandale, as being useful is if it is OpenCL capable and actually works well in that mode. Well works well and is bug free.

To put it mildly I don't gave any confidence in Intel actually making a good GPU. The could shock us which might change my mind but right now it is wait and see. Since this is a whole new generation of machines simply matching 9400M performance is NOT GOOD ENOUGH! To be successful Intel really needs to triple performance over the 9400M. Yeah that is a high mark but again this is a whole new generation of machines that hopefully will remain around for awhile.



Dave

Why? That's what PCI Express is for.

Because I believe you could offer much higher performance in a slightly integrated GPU. PCI Express would already be there for many a PCs needs so very little would have to be integrated onto the DMI based integrated GPU. This would keep the costs of the iGPU down while preserving much of the DMI bandwidth for graphics.

Obviously this is not a solution for all systems but for things like the low end of Apples desktop line up and most of the portables it would be fine. More that fine it should be very economical and at the same time high performance.

I have to wonder if Intels fight with Nvidia doesn't revolve around using DMI in this way. The difference in performance could be enough to screw up Intels marketing and positioning of it's new processors. In otherwords Intel is being bad again.

What do you think?

Dave

I did a little research and it seems clear that Nvidia does intend to release a chipset for Intel's Lynnfield/Clarksfield type processors (the ones people are calling Core i5):


Then again, Intel is suing them over that. Intel has a history of suing other companies for making chipsets for its processors.

It's true that development cycles for GPUs are about twice as fast as for CPUs. However, Intel might (I'm not sure they will) update the on-die GPUs twice in each tick or tock cycle. I'm also skeptical that choices and market segmentation would be limited. Intel have demonstrated an ability to offer a very wide range of processors and can be expected to continue to do so.


On-die integration of the GPU is driven by both technical reasons and anti-competitive ambitions.


I think that's an over-the-top statement. Dual-core CPUs still dominate the desktop market. I can see 45nm Clarksfield processors making serious four-core inroads into the desktop market, but the laptop market will wait for 32nm.


There are many metrics of performance. I would be happy to sacrifice an iota of speed for a windfall in battery life.


That's what I think of putting four 45nm cores in a laptop.


Why would Apple sacrifice their battery life advantage by putting four 45nm cores in a laptop?


The overwhelming majority of Apple customers have no idea which CPU they're buying. All they look at is the clock speed, if they compare CPUs at all.


So, you think it's better to use a low-end Clarksfield rather than a high-end Arrandale?


Why? The masses of consumers don't know what a core is, so they won't be making purchasing decisions based on the number of cores. In most applications, a 3GHz Arrandale will outperform a 1.6GHz Clarksfield, so why do Apple need four 45nm cores in a laptop?


I don't want a low-end machine. I reject the notion that discreet GPUs and number of cores are the be-all and end-all of performance.


I very much doubt that. Some applications e.g. large transactional databases benefit a lot from additional cores. Most do not.


Certainly not. There are far more possibilities to take good advantage of a second core than a third and fourth.


Seen by whom? Not by typical consumers.

In this case I'd have to say the primary reason is anti competitive behavior. Mostly because The choice to integrate the GPU first doesn't make a lot of sense from the standpoint of the user or Systems Integrator. Frankly it ought to be close to last to go on the die.

it's not over the top at all. Right now we are in the same mind set as when the dual cores started to come on line. People thought we would never need that, but it didn't take long for the OS and apps to catch up. People quickly realized that dual cores make for a far better experience. SL will have the same impact.

Again that is what the low end machines are for. I don't have a problem with Apple offering low end hardware, the problem is at the other end.

That is fine for you, not everybody so limits themselves. Besides I thought intel was skipping 45nm to go to 32nm. It is only a waste if you can't leverage the hardware.

For the simply reason that there is demand for it. Besides Apples battery life advantage is only part of the equation and it doesn't mean we impact the bottom end of the line up at all. Solving one need doesn't mean that you give up servicing others.

I don't buy that at all. Especially considering Apples professional clients. The GPU & CPU can have a huge impact on their ability to compete.

Are you sure you got that right?

Let's just say you are making a generalization based on old solutions. In any event to put it simply more cores allows your system to better service threads and processes. In part the answer reflect what one thinks about SL and how successful Apple will be with the exploitation of parallel hardware.

That isn't what you said above. You said you would give up performance for battery life. In any event when it comes to laptops which do you think will give you a longer battery lifetime, two fast cores or four running at half the clock rate or so? This given that the systems are balanced for maxiumum power.

Again I think you are either living in the past or are uninformed. There are many apps that already leverage more than two cores. This without the advantage of SL and the new tech there.

That is simply garbage. All a user needs to do is to have two active processes running and he is likely to have multiple threads going for each process. This without even trying.

You can't just look at a single app and say quad cores aren't worth it because the app isn't highly threaded. A modern PC has a lot more going on at any one time than a single simple process.

You don't get it, there is no such thing as a typical consumer.

Dave

wizard69
Let's think real. Apple probably won't use hot and expensive 45nm quads in notebooks because their low clocks and negative impact on battery life make them hard to advertise as a breakthrough. Not because of uselessness.

However, Arrandale's iGP isn't something to rave about. I doubt it'll support OpenCL (it probably won't) and it's HD decoding capabilities remain to be seen as well. Overall, I'm not impressed with perspective of not having a fully functional (Open CL, HD, light 3D) GPU with relatively low power consumption (that's what 9400M exactly is)

Well extremely hot no. I'm not even going to assume that they will implement 45nm parts as they may simply wait for Intel to deliver 32nm parts.

In any event this association with short battery life and quad cores just isn't a given. The fact is they can and most likely will be ran at a slower clock rate. This can result in better performance at a lower power point. It was one of the arguments for going to dual core in laptops so early in the game, dual core simply delivered better performance for the watt expended.

I don't know the specifics of Arrandales iGP but its Intel and their track record sucks. If it doesn't support OpenCL then I would think the whole chip would be dead upon arrival at Apple.

Exactly! Why give up on the progress that Apple has made in partnership with Nvidia? The last thing Apple needs to do is to regress just because Intel has a processor some adore.


Dave

We'll probably know by the end of the year who is right. Some, including myself, expect the first quad-core Apple laptop will wait for availability of 32nm processors. Some others expect Apple to use 45nm quad-core processors in laptops. Time will tell.

I do expect to see 45nm quad-core CPUs in the iMac, but not in the Mac Mini.

.
Apple's current 3.06 Ghz C2D option utilizes a 35 watt TDP proc (T9900) the entry level Clarksfield 1.66Ghz is estimated to have this same 35 watt TDP. So how does a 1.66Ghz Clarksfied use more battery power than a 35 watt C2D? I think logical progression would have Apple utilizing hardware that supports their latest OS. Offering Snow Leopard with Grand Central Dispatch and OpenCL and then standardizing your notebooks (which are Apple's best seller numerically) on Dual Core would be absurd.

We cannot talk about expense because we don't know what Arrandale costs and we've only got estimates for Clarksfield. You know that Clarksfield will likely have Turbo Boost mode and I believe you're just being disengenous here.

My thoughts are Apple will use Clarksfield on the high end and Arrandale (early 2010) on the lower end Macbook Pro models. They will definitely use Clarksfield at some level and people doing audio or video production who need laptops for the portability will buy these because many are hooked to AC %75 of the time. Battery life is important to most but where it ranks on the scale depends on if you're talking about a road warrior or someone that occasionally need to travel.

The problem with Clarksfield, at least the upcoming version, is that:

1- it is very expensive: $350 1.66GHz, $560 1.73GHz, $1,050 2.00GHz, for similar prices, you can get a P9700 (2.80GHz, 25W, $348) and a T9900 (3.06GHz, 35W, $530). Offering the quad 2.00GHz will be a +$500 option over the most expensive MBPs today.

Just in terms of prices, Clarksfield cannot be used on all models of the current MB(P)s:
$1199/1499 13" MBP ($209 cpus) - no Clarksfield to fill this need
$1699/1999 15" MBP ($209-241 cpus) - no Clarksfield to fill this need
$2299 15"/$2499 17" MBP C2D 2.80 ($316 cpu) could be replace with the quad 1.66GHz
$2599 15"/$2799 17" MBP C2D 3.06 ($530 cpu) could be replace with the quad 1.73GHz
If Apple uses the quad 2.00 model, prices will be at least $3099 and $3299 (and probably $3299/3499).

2- While SL is supposed to take advantage of multiple cores, gpu, etc., those are not a requirement, SL doesn't need quad-core cpus or multiple gpus to run.

3- Arrandale will have 2 cores + Hyperthreading, that means 4 threads at once, it's better than the C2D we have today. Clocks and prices are supposed to be similar to those of the current C2D. Of course, Arrandale IGP won't be better than the 9400M, but that's where the 16x PCI Express lanes are useful. Since the 13" is now a MBP, I can see Apple using the newly announced 200M series of nvidia GPUs on all MBPs early next year (210M on the 13/15", 230M on the 15/17", 250M as a BTO option)

Depending on the outcome of the Intel/nvidia feud on chipsets, it is possible that nvidia could come up with a version of the 9400M (or whatever) that connects thru DMI, but has a 9400M iGP (or whatever) on the chipset.

While it would be great to have some quad-core notebooks, I don't think Apple will cut the prices again (like they have done recently) to be able to offer a full-line of quad-core notebooks. And if they offer them only on the 15"/17" models, the move of the 13" to MBP status would have been useless.


Of course, they are: Arrandale will replace the current SL, P, and T series of the penryn C2D:
- 18W Arrandale could be used for the MBA
- 25W Arrandale could be used for the Mac mini, 13/15" MBP
- 35W Arrandale could be used for the 15/17" MBP
And later Arrandale will replace the other series (SU, U, etc.)

I am not that worry about the notebooks regarding the quad-core adoption, they can wait for the 32nm versions or later to move to quad-core, I wonder more about the fate of the iMac, because Apple didn't move to quad-core yet. Lynnfield will be 95W, Clarksfield is too expensive for the iMac, that still uses inexpensive custom-hybrid 55W cpus, 65W Lynnfield will be released early 2010, but yet is 65W too high a TDP for Apple's liking?

All we know is that they didn't use the 65W desktop quads that have been available for +6 months along with nvidia 9300/9400 desktop chipsets. It is possible that Apple would use custom cpus again for the next version of the iMac (propably 2.00/2.13/2.26/2.40GHz models, with a TDP lower than 55W) this fall.

Now, don't get me wrong, I'd love for Apple to offer nehalem quads on most Macs, but I think it is unlikely for the next updates (especially the MBP) for reason of prices and/or TDP of Intel cpus.

I don't think Apple has the luxury of waiting.

People running Final Cut Pro on MBP 17 or Logic Studio on MBP 15 aren't as price sensitive as the parents sending little Freshman Johnny to college.

Arrandale is fine for low end uses but for clients that bill by the hour a dual core laptop is a buzzkill. Every PC vendor will have quad core laptops available.

The people stating "Apple doesn't need Quad Core" aren't the people doing video encodes that take hours or running plugins across 50 tracks.

When you run your business off your computers processing power, spending hundreds of dollars more for an increase in productivity is a value add. Apple may not utilize quad core but with the lack of Blu-ray options and the silence with udates to their Pro apps I'm already seeing Pros preparing to move. It's their business..their not going to sit and wait for Apple.

For Pro usage a high price is not always a killer. Even then Apple seldom implements top end hardware on thier base systems. So the top end could simply be a build to order option.

You are making an assumption here that the prices are final at Intel and that Apple will pay these sorts of prices. Obviously Apple gets a significant discount and even has enough pull for custom versions.

On top of all this we need to remember that the coming chips through a monkey wrench into the power disapation issue. Remember that at least part of the power that the new chips expell is what would have been seen in other chips in previous generations. So even though a Clarksfield or what ever, might have high power relative to old platforms it does not mean that total system power is higher. It is not inconcievable to see a solid drop in overall power draw.

That is a poor arguement as we can easily say Leopard doesn't need a dual core chip. The problem is the community has pretty much come to the conclusion that dual core is a requirement. SL doesn't need quad cores either, in one sense but I think the market will come to the same conclusion in that SL can really rock on four cores.

The usefullness of hyperthreading varies widely with the software on older implementations. Plus we don't really know how successful Intel has been with this new version of Hyperthreading. So while I don't doubt it helps in the majority of cases it is not the same thing as adding more cores. As to Arrandales if the iGP isn't usefull for video or OpenCL then the extra logic is just a waste.

well hopefully we will see something like that. The problem is the smart place to connect the chips is over DMI. Hopefully Nvidia and Intel can resold their issues here.

I can't believe that Apple isn't actively involved in trying to resolve this dispute. The ability to use DMI for graphics is key to building low cost high performance machines.

Honestly I don't understand your reasoning. There is no need for Apple give up on dual core for the low end. It is more important that we have the option to fit the processor to our usage.

If Apple doesn't have a quad core iMac by the time SL hits people will be up in arms.

Well atleast one model Mac! On iMac though i suspect that they will stay with mobile chipsets. Not that I want to see that mind you, just that it helps Apple look green.


Dave

Sure but will a more expensive quad 1.6GHz be much better value than a dual 2.8GHz? On the desktop end we have a quad 2.66 vs 2.8 and it's much more clean cut.

If you do time sensitive long encoding, why would you even buy a laptop? Plus, you don't really charge clients for encoding time as it's not down-time - you can leave it overnight. If you did charge for it, a quad 1.6GHz vs dual 2.8GHz saves you about 15% of your encoding time assuming the software works at its best.

I believe that PC manufacturers will have quad laptops but not all of them. They will have Celerons too. The people who buy quad core laptops will probably be the typical Alienware buyers who think they need a laptop with the highest spec and then use it to post benchmark scores online.

Gamers probably won't buy it because low-clock quads will perform worse than high-clock dual cores.

To be honest, I'm not sure why anyone would buy a laptop with Clarksfield.

The best application I actually see would be in the Mini for running as a server. An 8-thread CPU in something the size of a Mini with 4GB Ram would make a very nice little dedicated server and a quad 1.6GHz would be a marked improvement over the dual 2GHz.

"People running Final Cut Pro on MBP 17 or Logic Studio on MBP 15" are using dual-core cpu right now, until those apps and others like Pro Tools, AVID, etc. are rewritten to make "better" use of multiple cores/gpus, most users will probably choose faster cpus vs more cores with that kind of gap between the respective clocks, for similar prices.

I'm not a FCP user, but I've been using Pro Tools for 10 years, and I can tell you that faster (clock) is still better than slower (more) cores - Quad 2.93 vs Dual-quad 2.26 Mac Pro, 6GB RAM, Leopard, latest (8.0.x) versions of PT LE/HD. A hundred tracks and about 3 plug-ins per track.

And if you didn't know, almost every PC vendor already offers quad-core notebooks, that doesn't mean they are successful (as models).

Again, I'm not the one saying "Apple doesn't need Quad Core", I just think that Clarksfield is not the answer (for Apple) for the MBP or the iMac. I've been "asking" for a quad-core iMac (or a small headless desktop) since the 65W quads from Intel's S series have been planned (almost 2 years ago)!!! And you know that.


No I am not. I never said that Apple was paying those prices. Apple is currently using cpus on the MBP that retail for $209-$530, Clarksfield cpus will retail for $350-$1,050, there's a huge difference.

in the $350-550 retail segment, you can choose:
- either a 1.66 quad at +35W or a 2.80 dual-core at 25W
- either a 1.73 quad at x5W or a 3.06 dual-core at 35W

The 2.00 quad at x5W has no equivalent in price/tdp...

I have no problem with Apple offering quads as an option on any Mac, I just don't want to prices to go up just because they are using quads instead of dual-cores.


Now, you're the one making assumptions. Who is the market?


This "new version of Hyperthreading" was launched with Core i7 cpus, 9 months ago, the Xeons on the Mac Pro/XServe have it, if it was useless we'd know it by now.


What I've been saying is that Clarksfield is too expensive, I have no problem with it as an option but the increase of power (if any) is yet unknown if you compare cpus at the same price: DC 2.80 vs QC 1.66, DC 3.06 vs QC 1.73 (with or without SL).

And if Apple was to release some models with quads this fall, they would need a brand new motherboard (55 series chipset or a new chipset from nvidia), that would increase cost of producing both dual-core and quad-core models. All I'm telling you is that, if Apple was to release a quad-core 15" MBP it will cost you at least $2299 (1.66GHz), $2599 (1.73GHz) or $3x99 (2.00GHz) - add $200 for the 17" version.

IMO, they can probably update the MBPs to Arrandale early next year with similar clock speeds + HT, and offer some quad models mid-2010 when quads will be made on a 32nm process (speedbump and lower power requierements) and both use the same chipset (Intel or nvidia).


Only in some forums and in your dreams
Most people don't care what's inside, they just want the latest.
Most of the customers don't even know that Intel will release Clarksfield cpus this fall.

Like I said, I wonder what Apple will use on the next iMacs, but it won't be 95W Lynnfield cpus, nor $350-$1,050 Clarkfield cpus. 65W Lynnfield cpus will be available early 2010, only.

*** Good post Marvin

Yes I believe so because of the potential flexiblity of Turbo Boost. If I need the system to stay cool and handle a lot of i/o from applications then it hums along balancing these across the 4 cores. If I need a lot of processing power for one or two apps then Turbo Boost will shut off some cores and boost the clock of the active cores. This gives me the best of both worlds. High megahertz for single threaded apps and lower megahertz but more breadth for multi threaded apps or heavier multitasking.

I know a lot of musicians and video pros that need to have a computer at the location so the Macbook Pro is the choice here. They may have a Mac Pro as well but they still need decent power on the road.

Quad vs Octo core

http://www.barefeats.com/octopro1.html

I await the tests comparing Dual core vs Quad core when the Dual has a higher megahertz. I think the slower Quads are going to do nicely on Snow Leopard. In the end there's no substitute for having more pipeline stages cranking out more IPC (instructions per clock).

Well, looks like nVidia lowered idle (2D) power consumption within it's 200M series. So, GT240 (GTS250) while being middle-end solution may be as cool (temperature wise) as entry-level 9400M when not doing lots of computation (3D, OpenCL action etc.)

That's why battery life difference between GT 240M (GTS 250M) and G210M in 2D mode may be less than what we have now with 9600M GT and 9400M combination.

Reduced necessity in G210M to complement it's stronger brother will justify going to Arrandale iGP + performance GPU only combination in higher-end notebooks.

Lower end machines (13" and 1700$ 15"), however, should have G210M instead of GT240 because Arrandale's iGP is only good for power saving.

Definitely plausible though I wonder if it's a bit "Hacky" from a drivers standpoint.

What do you mean?

Both 210M in 13" and 240M in 15/17" will be discrete GPUs with GDDR3 connected via PCIe x16.

What some people here seem to overlook is that a 32nm Nehalem processor with any number of cores will necessarily and dramatically outperform a 45nm Nehalem processor with any number of cores if the clock rates are set such that the TDP is the same.