[PaperNotes]

What's the consensus on this? There are no more Merom updates for this year.

[Simon]

Intel's ahead of the original schedule. The quad-core Xeon Penryns ("Harpertown") are expected to arrive already next quarter. The dual-core desktop version ("Wolfdale") won't arrive till Q108. Penryn-based mobile CPUs are expected in Q108. The Santa Rosa plattform will be updated to support Penryn. Later on in Q2 the updated plattform ("Montevina") will arrive with the new "Cantiga" chipset based on Bearlake.

DailyTech - Intel Leaks 45nm Xeon Clock Frequencies
DailyTech - Mobile Intel "Penryn" Core 2 Duos Revealed
DailyTech - Intel Prices "Penryn" Xeons

These guys originally even speculated Apple might announce Penryn MPs and iMacs today, but in the meantime they claim their 'source' might not be that reliable...

[PaperNotes]

I don't believe much of anything from TUAW or Apple Insider because they have a history of making things up to get visits and hits to get some Adsense earnings.

Doubt very much the new iMacs will have anything different from the current MacBook Pros. I hope that Intel stays ahead of schedule and can release mobile Penryns before Christmas. It would be sad not to have another update to the MBP for 5-6 months because I really want to buy a 17" but just want all the power they can offer for gaming. Many Vista games now list crazy recommended specs. Call of Duty 4 and Colin McRae Dirt make me want the maximum specs I can get. There's also 8700M GT available now and I hear an 8800M in the works.

[DKeithA]

I hope so. As soon as the MBPs are updated with Penryn I'm ordering one.

[mfbernstein]

Wow, updated only a month ago, and already looking to the next update?

As with Santa Rosa, despite the hype Penryn will be at most an incremental upgrade. A few more mhz, a larger cache, an expanded instruction set, and slightly better power usage etc.

The more interesting question is what Apple will do to the rest of the system.

[mduell]

Originally Posted by mfbernstein 

Wow, updated only a month ago, and already looking to the next update?

As with Santa Rosa, despite the hype Penryn will be at most an incremental upgrade. A few more mhz, a larger cache, an expanded instruction set, and slightly better power usage etc.

The more interesting question is what Apple will do to the rest of the system.

Penryn is already showing +10-30% performance improvements at the same clockrate. But I think what it's really going to do is replace duals with quads in the top half of the mobile market (i.e. MBP price range), like the recent Xeon/Core 2 price drop did for the workstation/desktop market.

[mfbernstein]

Originally Posted by mduell 

Penryn is already showing +10-30% performance improvements at the same clockrate. But I think what it's really going to do is replace duals with quads in the top half of the mobile market (i.e. MBP price range), like the recent Xeon/Core 2 price drop did for the workstation/desktop market.

Thanks for the link. Still, this seems to be about on par with the gains from moving to Merom from Yonah. I agree that quad core at the higher-end is likely, but for most workloads, I'm not sure how much of a real gain that'll bring

[PaperNotes]

It is obvious the next MacBook Pro will have the same keyboard as the new wireless keyboard! Wouldn't it be sweet if the keyboard and trackpad was a removable wireless unit that recharges when docked and the screen could slide forward in wireless mode?

[fuzzball963]

Hey all,

I haven't seen anything but does anyone know when the roadmap includes a quad core processor that could be included in the Macbook Pro?

I have the original Core Duo machine and it's working nicely but I'd love a quad or (gasp) Octo core .

[PaperNotes]

I read on various sites end of 2008 for quad-core notebooks. To fit one in the MacBook Pro design I don't know if it will be possible to put the fastest quads.

[mduell]

Originally Posted by fuzzball963 

I haven't seen anything but does anyone know when the roadmap includes a quad core processor that could be included in the Macbook Pro?

Sometime in 2008; possibly in the first quarter, but if not I'd guess fourth quarter.

[Simon]

On this recent Penryn roadmap I noticed Intel dropped the mentioning of a quad-core mobile Penryn.



There will certainly be quad-core Penryn-based desktop CPUs (the Penryn replacement for Kentsfield), but I'm surprised they stopped mentioning it for the mobile segment.

Maybe a quad-core mobile Penryn will arrive later as a XE version (like the 2.8 GHz Merom XE). In any event, if they make such a Penryn XE by taking two Penryn dual-cores and putting them on one MCM (their previous approach to quad cores) it means the TDP will scale almost linearly. If you then end up with a quad-core 'mobile' Penryn, you're looking a >44W TDP CPU for bulky and/or loud desktop replacement notebooks. Not something you'll find in a 1" thin quiet MBP. The iMac on the other hand is certainly a candidate. Just like it's now getting Merom XE which is also not found in the MBP.

[mduell]

Originally Posted by Simon 

Maybe a quad-core mobile Penryn will arrive later as a XE version (like the 2.8 GHz Merom XE). In any event, if they make such a Penryn XE by taking two Penryn dual-cores and putting them on one MCM (their previous approach to quad cores) it means the TDP will scale almost linearly. If you then end up with a quad-core 'mobile' Penryn, you're looking a >44W TDP CPU for bulky and/or loud desktop replacement notebooks.

No, it doesn't. The current C2Q are 95W TDP compared to the same clockrate C2D at 65W. That's not even a 50% increase.

It's unfortunate they dropped the mobile quad from that pitch; sounds more like Q4 than Q1.

[Simon]

You didn't read my post. The process had to be changed to get a lower combined TDP. If you take two core pairs and slap em together on one MCM you will get a nearly linear TDP increase. What Intel did with the C2Q is refine the process to lower the TDP of every core with respect to the original dual-core design. That takes time and a lot of effort. Definitely something that won't happen right around the Penryn launch.

Even if you assume Penryn will have a TDP as low as 28W (which is extremely optimistic), adding 50% (assuming they can further refine the process again) you're getting 42W for the quad. Just like there's no X7900 (TDP 44W) in the MBP now, they're won't be such a quad in the MBP then. You're kidding yourself if you believe a >40W TDP CPU will work out well in the 1" thin MBP with its 60 Wh battery. It's gonna be another while till a core design arrives that's suitable to be used as a quad-core in a MBP. Expecting quad-core MBPs in Jan is wishful thinking at best. The facts tell a very different story.

[mduell]

Originally Posted by Simon 

You didn't read my post. The process had to be changed to get a lower combined TDP. If you take two core pairs and slap em together on one MCM you will get a nearly linear TDP increase. What Intel did with the C2Q is refine the process to lower the TDP of every core with respect to the original dual-core design. That takes time and a lot of effort.

I don't see the mention of the stepping revision (the process feature size stayed the same) in your previous post; but even before the changes it's still not linear. The original C2Q release, B3 core stepping, was at 105W before the optimizations for the 95W G0 stepping. The B2 stepping duals are 65W. That's only a 60% increase in power consumption.
I believe the reason is that running the FSB takes quite a bit of power (note the significantly lower FSB on the mobile chips than the desktop chips), and with the quads you don't need to duplicate that since both dual cores share one FSB.

[PaperNotes]

Originally Posted by mduell 

I believe the reason is that running the FSB takes quite a bit of power (note the significantly lower FSB on the mobile chips than the desktop chips), and with the quads you don't need to duplicate that since both dual cores share one FSB.

Will still have to wait til end of 2008 for quad-core at the least. You know what Apple is like. They underclock mobile graphics chips by more than they should. They are all about boasting about battery life and having thin laptops more than they are about powerful desktop replacements. As long as we get our dual-core Penryn soon with 17" OLED.

[Simon]

No, not really. The process size is not the point. It's the process revision. Once you start optimizing against losses you get the power savings that allow you to multiply dies and go onto the MCM. Intel has never before taken a dual-core die and multiplied it right away. They usually go through two to three revisions of the process until they drop the losses to the point where they start multiplying. This would be different if they started out with a quad-core layout, but that hasn't been the case for obvious reasons and also for Penryn it won't be that way. IIRC AMD has done this before though.

The real issue is that quads have been scratched from the Penryn mobile roadmap. IOW even if Intel intended to do a quad-core mobile solution later on, they will do the two dual-cores on one MCM approach they have taken previously with Core 2. That means again waiting for two to three process revisions, i.e. 4-6 months delay with respect to the dual-core launch at the very least. If at all, that is. And then we'll be likely looking at an XE version with a substantially higher TDP than the dual-core. Taking into account the previous MBP history and design choices like GPU downclocking, it's pretty much certain we won't see a quad-core Penryn in the MBP.

[mduell]

Originally Posted by Simon 

No, not really. The process size is not the point. It's the process revision. Once you start optimizing against losses you get the power savings that allow you to multiply dies and go onto the MCM. Intel has never before taken a dual-core die and multiplied it right away. They usually go through two to three revisions of the process until they drop the losses to the point where they start multiplying. This would be different if they started out with a quad-core layout, but that hasn't been the case for obvious reasons and also for Penryn it won't be that way. IIRC AMD has done this before though.

The real issue is that quads have been scratched from the Penryn mobile roadmap. IOW even if Intel intended to do a quad-core mobile solution later on, they will do the two dual-cores on one MCM approach they have taken previously with Core 2. That means again waiting for two to three process revisions, i.e. 4-6 months delay with respect to the dual-core launch at the very least. If at all, that is. And then we'll be likely looking at an XE version with a substantially higher TDP than the dual-core. Taking into account the previous MBP history and design choices like GPU downclocking, it's pretty much certain we won't see a quad-core Penryn in the MBP.

I know, and Intel uses the stepping to identify what you call process revisions, which is why I referred to them. It's confusing to call it process revision when that's generally used for something else (process feature size).
Nehalem will be the first non-MCM quad, so I don't think anyone is expecting Penryn to be a non-MCM quad. AMD's "native" quads still aren't on the market; the best they can offer today is the joke called 4x4.

It only took 3 months to get from Conroe to Kentsfield; now that they'd done the MCM approach twice (once with P4, once with C2), I can't imagine it would take longer with Penryn. Probably a 50-60% TDP jump (high 20s to low 40s), but not the 100% you're suggesting.

[Simon]

Agreed, once we go to Nehalem we could be getting quad-cores on a single die right form the start.

With Penryn it's a different story. We'll be seeing a delay from the moment Intel starts delivering the mobile Penryn to the point where they can ship double dual-cores on one MCM - if they actually decide to do so at all!. Judging from the recently released roadmap they are at least not talking about it anymore.

Re: power increase I still think there's some misunderstanding here so let me try to clarify. I never said we're looking at a 100% TDP increase for a double dual-core Penryn on one MCM. What I said was that the TDP scales linearly with the number of cores. And probably this is where the confusion comes from. There's naturally an overhead (for example due to the FSB power you mention) that you pay for only once, but that's not a contradiction to linear scaling. The point is once you subtract this overhead you get a "base TDP" required by the two cores alone and - if you don't go through any further revision - this base TDP will multiply with the number of dies you slap onto the MCM. Or to put it in a more mathematical sense f(x) = ax+b is perfectly linear ("linear" does not require b=0) but this b is where you pay for the FSB power and other overhead. Doubling the number of cores will not double the power, but it will double the base power requirement (the "ax" term). Actually I also wrote "nearly linear" because of course there are a few small changes you can make when you go onto the MCM that allow you to drop the base power somewhat, IOW the "a" changes slightly with the number of cores on the MCM, but compared to the overhead you're talking about I consider that contribution fairly small. Hence "nearly linear".

The TDP I suggested for a quad-core version of the mobile Penryn is >44W which is certainly less than a 100% increase over the dual-core mobil Penryn. OTOH it's also so much that it looks very similar to the current X7900 which didn't find its way into the MBP. Therefore I'd say we will probably have to wait until Nehalem to get a quad-core MBP. The iMac will likely be earlier though. Other options would be a redesign of the current MBP to allow for a higher CPU TDP, a different GPU with a lower TDP, or added power savings due to Monetvina/Cantiga. We'll see about that in Q208.

[Simon]

New goodies form Intel.

I saw this slide just this morning:


So it looks like a quad-core mobile Penryn is back on the radar. That's the good news. The bad news is that the launch taget is 2H08 together with Montevina and the TDP is 45W (as expected). That means we won't see quad-core mobile CPUs for another year and it looks like it will be a XE version of Penryn geared towards desktop replacement notebooks. The 45W TDP (like the current 2.8 GHz X7900) means it won't go into the current MBP - it could find its way into the iMac though.

Other than that it's like the 'regular' Penryn (for the Montevina refresh): 1066 MHz FSB, 45 nm, socket P, 12 MB L2 cache (two dual-core Penryns dies on one MCM)

[Simon]

dp, sorry

[mduell]

It will be interesting to see how the process shrink and stepping revisions go... 2.4Ghz may be possible, but I think 2.13Ghz is more likely.

[The Placid Casual]

With the Macbook line they are so far behind the curve it is scary. I can't see anyone buying one in good conscience if you do a side by side comparison with some of the PC (Dell XPS m1330, Vaio FZ, Acer Aspire 5920 etc)...

Before they get a mobile quad, they need to stop relying on the 'Apple' factor, and keep up with the pack.

[wubrew]

Originally Posted by The Placid Casual 

With the Macbook line they are so far behind the curve it is scary. I can't see anyone buying one in good conscience if you do a side by side comparison with some of the PC (Dell XPS m1330, Vaio FZ, Acer Aspire 5920 etc)...

Before they get a mobile quad, they need to stop relying on the 'Apple' factor, and keep up with the pack.

No kidding and the top of the line MB 2.16 is not even given a v-card option. Considering the $$$.
On the positive side, I really like all you folks putting it together and laying it out there. Very educaional very satisfying.