Apple Launches MacBook Pros with Thunderbolt, Quad Core CPUs, AMD GPUs

[Denarius]

I'll believe "TBolt in all the Mac range" if and when it happens. Apple has a long history of crippling lower end systems to push upsells to more profitable ones. I'd be surprised if it shows up in the MacBook, not surprised to see it limited to top Imac configs, and not surprised to see it absent from the MiniMac. (Some bean counter might decide that too many Mac Pro sales would be lost if the MiniMac became more useful.)

Well, it's certainly possible, but I would have thought that Intel would want a pretty solid commitment from Apple to push the technology across the board in exchange for a year's exclusivity.

By the way, I like the coined word "TBolt" - I'll stop typing ThunderPort and use TBolt instead.

Thank you! Was quite gratified to see others adopt it.

Does anyone know if TBolt is *required* to carry DisplayPort signals?

The simple and obvious solution for the Mac Pro would be to put TBolt on the mobo as a DisplayPort-less connector.

There's also the issue of a system with more than one TBolt port. Is one TBolt port the "primary" monitor connection? Can you plug your monitor into any one of them? Can you only plug your monitor into the primary? Can you plug multiple monitors into separate TBolt ports? Inquiring minds want to know.




In a recent post I predicted that the most popular TBolt accessory will be a hub that exposes eSATA and USB 3.0 ports....

Good to know that I am not the only one that has reached these conclusions. I really do doubt that Apple will put a displayport-less TBolt port on a Mac Pro, but it really does seem that such an option is Apple's only resort unless they want to jack up the price of graphics card options. TBolt works wonders for all-in-one systems, not very clear on upgradeable towers.

I agree with you Aiden. I doubt we'll see TBold on a MBA anytime soon. I think it'll find itself on the Mini though.

Thinking about this more, I was just looking at the TBolt chip on the motherboard of a MacBook Pro on ifixit. I'm no hardware engineer, but I can't see any evidence of any graphics chips nearby, then it occurred to me that PCIE is a serial technology so I'm wondering whether that may mean that TBolt will get the graphics information via the PCIe bus regardless of what graphics card is in there, thus a Mac Pro could use the ports on the graphics card itself or the TBolt port?

Bear in mind that I'm speculating on limited understanding so I'm more than happy for someone in the know to set me straight!

 

[AidenShaw]

Thinking about this more, I was just looking at the TBolt chip on the motherboard of a MacBook Pro on ifixit. I'm no hardware engineer, but I can't see any evidence of any graphics chips nearby, then it occurred to me that PCIE is a serial technology so I'm wondering whether that may mean that TBolt will get the graphics information via the PCIe bus regardless of what graphics card is in there, thus a Mac Pro could use the ports on the graphics card itself or the TBolt port?

On the 2010 MacBook Pro, the GPU (orange) is in the same place as the new one:

2010 MacBook Pro

click to enlarge​

The DisplayPort traces got to a connector in the same place on that motherboard without any PCIe stuff.

Signals can be carried on circuit board traces for quite a distance (especially uni-directional signals like a video signal). Designers try to minimize this, but some long lines are necessary. (Note that the CPU (red) and GPU (orange) are close - lots of traces, and the chipset (yellow) is also close to the CPU.)

2011 MacBook Pro

click to enlarge
GPU - orange
CPU - yellow
chipset - red
TBolt controller - blue​

 

[cmaier]

Signals can be carried on circuit board traces for quite a distance (especially uni-directional signals like a video signal).

Uni-directional signals are worse for long distances because of electromigration and because they create massive inductance.

 

[mjteix]

Thinking about this more, I was just looking at the TBolt chip on the motherboard of a MacBook Pro on ifixit. I'm no hardware engineer, but I can't see any evidence of any graphics chips nearby, then it occurred to me that PCIE is a serial technology so I'm wondering whether that may mean that TBolt will get the graphics information via the PCIe bus regardless of what graphics card is in there, thus a Mac Pro could use the ports on the graphics card itself or the TBolt port?

Bear in mind that I'm speculating on limited understanding so I'm more than happy for someone in the know to set me straight!

This may help, LR being the TBolt controller:

 

[Denarius]

On the 2010 MacBook Pro, the GPU (orange) is in the same place as the new one:

2010 MacBook Pro

click to enlarge​

The DisplayPort traces got to a connector in the same place on that motherboard without any PCIe stuff.

Signals can be carried on circuit board traces for quite a distance (especially uni-directional signals like a video signal). Designers try to minimize this, but some long lines are necessary. (Note that the CPU (red) and GPU (orange) are close - lots of traces, and the chipset (yellow) is also close to the CPU.)

2011 MacBook Pro

click to enlarge
GPU - orange
CPU - yellow
chipset - red
TBolt controller - blue​

This may help, LR being the TBolt controller:

Okay, so with a tower graphics card you won't have the option of the direct traces between the TBolt chip and the graphics card. Does seem to reinforce the idea of TBolt as a graphics card technology rather than a motherboard technology.

 

[Digital Skunk]

Thinking about this more, I was just looking at the TBolt chip on the motherboard of a MacBook Pro on ifixit. I'm no hardware engineer, but I can't see any evidence of any graphics chips nearby, then it occurred to me that PCIE is a serial technology so I'm wondering whether that may mean that TBolt will get the graphics information via the PCIe bus regardless of what graphics card is in there, thus a Mac Pro could use the ports on the graphics card itself or the TBolt port?

Bear in mind that I'm speculating on limited understanding so I'm more than happy for someone in the know to set me straight!

Understood, we all are taking a wild guess and it's going to be an exciting day to see how Apple will handle this question. I just can't see Apple putting in DisplayPort disabled TBolt ports. It'd be wonderful to have some data only ones to help solve the problem of hotswapping, but it doesn't scream Apple if you know what I mean.

Okay, so with a tower graphics card you won't have the option of the direct traces between the TBolt chip and the graphics card. Does seem to reinforce the idea of TBolt as a graphics card technology rather than a motherboard technology.

Yes, I have the same thought. I think Mac Pro users may start to see GPU prices increasing if they have to carry TBolt chips as well as all the other hardware.

 

[AidenShaw]

Uni-directional signals are worse for long distances because of electromigration and because they create massive inductance.

I didn't mean DC - just that the highest bandwidth signals in DisplayPort are the video signals and round-trip latency is not an issue - the video goes in one direction only.

 

[diamond.g]

Understood, we all are taking a wild guess and it's going to be an exciting day to see how Apple will handle this question. I just can't see Apple putting in DisplayPort disabled TBolt ports. It'd be wonderful to have some data only ones to help solve the problem of hotswapping, but it doesn't scream Apple if you know what I mean.



Yes, I have the same thought. I think Mac Pro users may start to see GPU prices increasing if they have to carry TBolt chips as well as all the other hardware.

Wait, I thought Intel claimed that ThunderBolt expansion cards wouldn't be possible? Wouldn't having Thunderbolt on a GPU basically be the equivalent of an expansion card?

 

[Denarius]

Wait, I thought Intel claimed that ThunderBolt expansion cards wouldn't be possible? Wouldn't having Thunderbolt on a GPU basically be the equivalent of an expansion card?

That's true. The whole thing's really quite confusing on the whole.

The one thing that did occur to me was that in a tower context, possibly a small multiplug on the PCIE graphics card with a cable to a corresponding multiplug at the TBolt end would fill in those direct traces between the GPU and the redriver/TBolt processor. I would have thought that would mean making TBolt compatible graphics cards pretty trivial as they'd just need a little multiplug to send the displayport info to the redriver and the TBolt chip unless I'm missing something.

Incidentally, has MacRumors now got McDonald's sponsorship?

 

[Digital Skunk]

Wait, I thought Intel claimed that ThunderBolt expansion cards wouldn't be possible? Wouldn't having Thunderbolt on a GPU basically be the equivalent of an expansion card?

Very good question. Let's add that to the endless list of questions we have about TBolt.

 

[chrmjenkins]

Uni-directional signals are worse for long distances because of electromigration and because they create massive inductance.

I assume due to unilateral current flow? I've only heard of it as a concern when driving a transistor too hard.

 

[Eidorian]

Wait, I thought Intel claimed that ThunderBolt expansion cards wouldn't be possible? Wouldn't having Thunderbolt on a GPU basically be the equivalent of an expansion card?

The biggest concern appears to be sharing bandwidth with the GPU. Not that you are not currently on the Macbook Pro...

It will be even more irrelevant when PCIe 3.0 rolls out.

 

[cmaier]

I assume due to unilateral current flow? I've only heard of it as a concern when driving a transistor too hard.

Yep. (And long straight wires are worse than bendy wires for electromigration, because bends act as stops).

 

[chrmjenkins]

The biggest concern appears to be sharing bandwidth with the GPU. Not that you are not currently on the Macbook Pro...

It will be even more irrelevant when PCIe 3.0 rolls out.

Really? Intel loves starving their platforms of bandwidth for PCIe et. al.

Yep. (And long straight wires are worse than bendy wires for electromigration, because bends act as stops).

Actually, they can reflect quite a bit of current depending on the miter/chamfer, frequency and a few other factors.

 

[cmaier]

Really? Intel loves starving their platforms of bandwidth for PCIe et. al.



Actually, they can reflect quite a bit of current depending on the miter/chamfer, frequency and a few other factors.

Yep. I was talking only about electromigration.

 

[diamond.g]

The biggest concern appears to be sharing bandwidth with the GPU. Not that you are not currently on the Macbook Pro...

It will be even more irrelevant when PCIe 3.0 rolls out.

GPU bandwidth should only be an issue when you are putting/pulling on/from system RAM. Other than that you shouldn't have a problem running on a 4x or 8x link. With PCIe 3 a 2x or 4x link should be fine. The main question I would have is power. The low powered GPUs should be fine with 10W pulled off the PCIe bus, but the ones that already require power from a 6/8 pin plug may not be as flexible. I have seen power starved GPU's do so fun stuff...

 

[takao]

having to wire the DP outputs from the graphics card back to the motherboard sure is going to be a problem for the towers especially for the price of the graphics card and also the time it takes to get them (months late as usual with apple's mac pro support)

on the PC side this will seriously hurt their adoption rates because i suspect the non-TBolt graphcis card will be always first out of the gate

from a design point it sure does look like a mess

 

[diamond.g]

having to wire the DP outputs from the graphics card back to the motherboard sure is going to be a problem for the towers especially for the price of the graphics card and also the time it takes to get them (months late as usual with apple's mac pro support)

on the PC side this will seriously hurt their adoption rates because i suspect the non-TBolt graphcis card will be always first out of the gate

from a design point it sure does look like a mess

According to the pictures posted, ThunderBolt seems to require a video signal. But do we know if that is 100% true? It seems to me that GPU's could be fine without having to go back through the ThunderBolt port as long as they didn't use a mDP connector.

By fine I mean not confusing.

 

[MagnusVonMagnum]

According to the pictures posted, ThunderBolt seems to require a video signal. But do we know if that is 100% true? It seems to me that GPU's could be fine without having to go back through the ThunderBolt port as long as they didn't use a mDP connector.

By fine I mean not confusing.

Is this "Thunderbolt" how LightPeak is going to be implemented on everyone's computer in the future or is this an Apple only thing? I wouldn't be 100% shocked to find Light Peak called BitStorm or something on PCs and using USB3 connectors in a year or so. Then Apple would be left with their Mini-Display Crap that virtually no one else uses....

I can't see ANY point to Thunderbolt + Mini-Display Port on a Mac Pro. The GPU is on a card. Thunderbolt would be on the motherboard. There is literally no point in mixing them, IMO. In fact, I think it's stupid to mix them on a MBP. It just leads to a situation where you have to unplug your external monitor from the computer and plug it into the back of say an external hard drive and then when you're done, unplug it and the hard drive and plug it back into the computer. WTF would want to go to all that bother for just a hard drive??? The display port deserves its own dedicated port! If they wanted to put the option of a *2nd* output on the Thunderbolt connector, I could understand that kind of flexibility. But it seems to me the ONLY thing they saved by combining the two together on the MBP is an inch or so of space on the case. If my 2008 MBP can have a full-size DVI connector on it and still have room for TWO separate 400/800 FW ports AND a USB port all on the same side of the computer, I think they could have managed a lousy 2nd Mini-Display port dedicated to video on that thing!

 

[Swift]

The Faith of Crazy People

We Apple fan boys and girls and men and women are usually disparaged as belonging to an "authoritarian personality cult". As if the advent of Windows as the business standard meant the end of creativity. The Apple hater has a hatred for a different business method from Windows, whose software has the profit margins and whose commodity hardware model never varies -- it's always whatever Windows supports. But Windows is unprotected, it's just software. So our software patents have gotten very aggressive in the age of Microsoft.

Apple charges for the software because that gives them a profit on that accounting item, called "Development."

What makes me an Apple fan is simply the idea that a lot of creative thinking is going on at Apple. What is a computer? What do you need from it? I think that the new iPad is likely the closest that the Jobs aesthetic has come to "the computer for the rest of us." Easier and easier. Faster. More intuitive. Touch. Beautiful.

The business-school ideology of the '80s brings forth Windows. Low-cost computers. Great increases in market share, but only Windows profited. Our software is setting the terms of commerce, keeping the lion's-share for itself. Windows on every desktop.

I really wish Apple was being competed with on their terms, not by Android's replaying of the Windows economic strategy.

I'd really like to see a challenger to Apple that succeeds as a design house capitalist as well: the idea is, from A to Z, you have control. I think it's only that way, by joining art to technology under one roof, that you could do more than make commodity hardware.

 

[Swift]

I just don't like how Apple charges premium prices for Made in China stuff... I don't hate Macs (I own an iMac) I just hate the pricing scheme (even though I can afford them).

True, but the only way to solve that is to invest a few tens of billions of their bank account to build innovative manufacturing facilities in our society. The truth is, now, we can't compete not because we don't have money, and not because of high-priced labor here; labor is a small percentage of the cost of Apple devices. Double that and the price goes up $50 on a MacBook Air. Or Apple swallows it.

It's not that they couldn't be made here, but we've got serious "bone loss" as an economy. Our past industrial might has been sold for scrap. Our steel is rust. In the 19th century, we went to the city from the farm. The 20th, to the suburbs from the city. Where to now? We've disinvested in what we call "infrastructure," and the structure itself is threatened. I'm not sure you can run a country like that. Not for long.