Not exactly the first of kind, but first complete with important periferals inputs and outputs. The first was "adapter" that you could equip with most of ExpressCard for any periferals you needed.
It is worth to take a look at this almost year old piece:
Glad to see more TB products out there, but how about a TB external GPU—or slot in which to install one? Whatever card the drivers can handle! (Meaning, I suppose, whatever OS X expects a Mac Pro to handle.)
I can’t give up Air portability, but TB has held out the promise of greater GPU power when I’m at home at my desk. All promise... no reality, so far. A GPU made for Mac Pro would be FAR better than you’d expect an ultrathin laptop to be able to offer. Come on, TB, let me have my cake and eat it too! Plus then I could choose to keep the same (home-base-only) GPU when getting a new Air.
Realize tha the Mac Book Pro Retina can already drive multiple Monitors.
There will NEVER be a TB External GPU. There isn't much of a market for it. It already costs a lot of money to develop a TB product. So to waste it because there are few people that will buy it.
The best chance of having an external GPU is to have an expansion chassis with full-size PCI slots. The Magma Thunderbolt Expansion box has THREE full-size PCI Slots. BUT, the PCI GPU card manufacturer needs to write Thunderbolt drivers for the cards. Tough luck with that.
I think cost is due to low volume. There's a limited market for these things; really only MacBook Air owners who have bought in the last 1.5 years.
Anything else (MacBook Pro or desktop) doesn't need this device in the same way. There's also not going to be a lot of demand on the PC side either, I wouldn't think.
But I agree that the peripheral selection today is, at best, disappointing. But I'm not surprised. Firewire, which in the USB 1.0 days, had real advantages, never caught on much outside the digital video world (It was tough, for example, to find firewire card readers). Unless Thunderbolt becomes a consumer-grade technology present on basic PCs, then this will remain a niche technolgy with high prices, much like firewire and SCSI before that.
No
Nobody promised GPU support over TB except for a few poorly informed pundits on the web. The idea has no merit at this time.
I think it has merit. Eventually I expect desktop displays to Retina, but I don't think a MBA or an older MBP could easily run them. Having a display with a built-in video card (or external) would help make that an option. Perhaps it's not a big market but it might be big enough that it's worth vendors looking into. Maybe CES will be favourable to Thunderbolt.
Realize tha the Mac Book Pro Retina can already drive multiple Monitors.
There will NEVER be a TB External GPU. There isn't much of a market for it. It already costs a lot of money to develop a TB product. So to waste it because there are few people that will buy it.
The best chance of having an external GPU is to have an expansion chassis with full-size PCI slots. The Magma Thunderbolt Expansion box has THREE full-size PCI Slots. BUT, the PCI GPU card manufacturer needs to write Thunderbolt drivers for the cards. Tough luck with that.
The beauty about Thunderbolt comes from its hardware nature of being drivers independents. TB really is an external PCIe 4X BUS where almost any PCI cards in a TB chassis will work without any changes in it's drivers. This is also why Thunderbolt has to be built-in on the controller's motherboard and can't be add to non-thunderbolt ready PC via a PCI card.
This being said, a PCIe 4X is not enough bandwidth for high performance video card, you won't see external video solution suitable for gaming anytime soon.
Well equipped for the video semi-pro market. I don't understand Matrox thinking FireWire isn't necessary. They are missing they very market they specialize in, video editing. Who doesn't have some FireWire gear in this field?
It's not just video. There is a major part of the audio recording interface market that's between USB2 and PCI cards who live and die by FW. USB3 maybe be great but it's generally been ignored by recording gear manufacturers, and FW is certainly gone from new computers but is still what these interfaces are being made with, even new ones showing at NAMM. It's not a matter of getting new drives. I'm not about to sell my two year old interfaces which work great and haven't even been replaced withThunderbolt or USB3, except for a few pieces from companies, which I personally have yet to see in anyone's studio yet, available though they may be.
Here's what I don't get: if a laptop HAS USB3, what's the big deal about having a Thunderbolt box to USB3 when you can just get a USB3 hub from the native port? There's only real value in a several hundred dollar Thunderbolt box if it provides ports that the laptop itself doesn't give you. I personally have zero reason to pay for USB 2 or 3 to be coming out of such a box as long as I can still use a native port for it.
Here's what I don't get: if a laptop HAS USB3, what's the big deal about having a Thunderbolt box to USB3 when you can just get a USB3 hub from the native port? There's only real value in a several hundred dollar Thunderbolt box if it provides ports that the laptop itself doesn't give you. I personally have zero reason to pay for USB 2 or 3 to be coming out of such a box as long as I can still use a native port for it.
There are a couple reasons to include USB on this device. Surely it adds more USB ports without adding too much cost to the device itself but I think the most likely reason is that it allows you to keep this device on your desk with peripherals plugged in so when you disconnect your laptop you are only unplugging a power cable and a Thunderbolt.
The Sonnet box is better, as it costs "only" $400-$800 depending on the model, but it only works with half-length cards.
Does anyone know why these things cost so much? If Thunderbolt is really just external PCI-Express, it seems that a TB->PCIe adapter should be a very simple piece of equipment, shouldn't it?
It's not just video. There is a major part of the audio recording interface market that's between USB2 and PCI cards who live and die by FW. USB3 maybe be great but it's generally been ignored by recording gear manufacturers, and FW is certainly gone from new computers but is still what these interfaces are being made with, even new ones showing at NAMM. It's not a matter of getting new drives. I'm not about to sell my two year old interfaces which work great and haven't even been replaced withThunderbolt or USB3, except for a few pieces from companies, which I personally have yet to see in anyone's studio yet, available though they may be.
Here's what I don't get: if a laptop HAS USB3, what's the big deal about having a Thunderbolt box to USB3 when you can just get a USB3 hub from the native port? There's only real value in a several hundred dollar Thunderbolt box if it provides ports that the laptop itself doesn't give you. I personally have zero reason to pay for USB 2 or 3 to be coming out of such a box as long as I can still use a native port for it.
Like SolipsismX said, it's an all-in-one dock, single cable connects everything.
Also, something like this (or the Belkin) can add USB 3 to Thunderbolt Macs that are pre-USB 3. Which would be 12-18 month's worth of Macs production.
So the Magma does support GPUs but you still need to find a PCIe GPU that supports it with Thunderbolt -and- Mac OS X drivers.
Hmmm... interesting. I always assumed that the only external GPU PCI cards that would be possible (for Mac) would be the ones Apple officially has drivers for—meaning, the cards Apple sells for Mac Pros. But it sounds like you also need special TB drivers as well? In that case, a PCI card via TB must not appear to the system in the same way a PCI card normally appears. I wonder if that's a potential compatibility problem for all manner of PCI cards—sounds that way.
I don't ever see GPUs over TB as being cost effective. For. The money being talked about you might as well upgrade to a Haswell based AIR. The number of GPUs Apple supports itself is minimal and you still take a 30% hit in performance.
Their compatibility list seems to contain no graphics cards, just a note: “External graphics support is a feature many users desire and we’ll keep you informed."
Orto put it another way, nobody is working on support for their GPUs. The potential sales volume doesn't justify the development effort.
I hold out hope for the future... I’d spent $100-$150 for a chassis that let me connect a decent dedicated GPU to a MacBook Air. (Hot plugging not expected... I can accept a reboot if I have to. But I do want it to be OS X! OS X is where I game, while Windows remains safely virtualized and strictly used for occasional testing of web stuff in IE.)
You would be far better off trying to get Apple to build an XMac or bring back a Mini with discrete GPU. That of course is a second computer, wait a bit longer and even an AIR might have respectable GPU performance. You really have to ask what incentive is there for a GPU solution over TB, I'd have to say there is very little.
One thing people seem to have problems grasping here is that TB is a serial port. As such the bandwidth is fixed. Thus all of these demands for more ports and capability in the docking boxes doesn't make a lot of sense. You can only push so many bits per second through today's TB cables. TB isn't magic and has to deal with the same physical realities any other serial port has to deal with.
One thing people seem to have problems grasping here is that TB is a serial port. As such the bandwidth is fixed. Thus all of these demands for more ports and capability in the docking boxes doesn't make a lot of sense. You can only push so many bits per second through today's TB cables. TB isn't magic and has to deal with the same physical realities any other serial port has to deal with.
20Gpbs full duplex serial port at that. The only thinkg I'd consider unrealistic in this thread is the graphics. You could max out four independent USB3 ports and that might be at your limit. FW800 is a drop in the bucket.
20Gpbs full duplex serial port at that. The only thinkg I'd consider unrealistic in this thread is the graphics. You could max out four independent USB3 ports and that might be at your limit. FW800 is a drop in the bucket.
To be fair, in current TB implementation the bandwidth is split in half between DisplayPort video signal and DATA. But I agree with you, beside graphics TB is more than enough for any other use.
Re market size... is it worth writing the drivers? Some people assume no, but I think they are guessing. Some questions to consider:
• Are TB drivers hard to write? Will that get easier? If the GPU driver already exists, how complicated is the TB driver?
• Are ultrathin laptops going to sell better or worse over time?
• Are Macs going to sell better or worse over time?
• Are Mac games and 3D pro apps going to sell better or worse over time?
• Are ultrathin GPUs going to improve, but slotted GPUs are going to stand still, eroding the advantage?
• Are TB ports going to become more common or less over time?
• Are TB accessories going to get more expensive or less over time?
Any way I look at it, the viability of making TB drivers seems likely to improve rather than worsen. GPU makers want more customers for higher-end products, and laptops are where sales growth is in the PC industry. I’m just sad to see it go so slowly.
As for the hardware... just a box with a slot or two, serving many different niches—plus, for my niche, the cards you can already get for a Mac Pro. I’m not expecting anyone to broaden that selection, but it would still be a welcome choice to have. It looks like the hardware side is already solved, it just needs to drop in price. Next up: software?
I agree that eventually a future Air’s internal GPU may be “good enough”... except... the CURRENT Air has GPU power that would have blown a gamer away a few years ago, and now it’s not quite “good enough” for everyone. I like my detail turned up
Quote:
Originally Posted by wizard69
No
Nobody promised GPU support over TB except for a few poorly informed pundits on the web. The idea has no merit at this time.
We’re using “promise” in two different ways. It holds promise—not that someone promised to deliver it. The idea has huge merit if people want ultrathin laptops AND want more GPU power than they can contain. Such a market isn’t mainstream (neither is video capture or RAIDs) but it exists.
To be fair, in current TB implementation the bandwidth is split in half between DisplayPort video signal and DATA. But I agree with you, beside graphics TB is more than enough for any other use.
My best understanding is that you can get it all as TB data if there isn't a DP device in the chain. But you do lose half once you have a DP device.
Lack of thunderbolt out is a deal breaker for me. No biggie, I can wait and find something else, but my MBP only has one TB out and I want a dock that would allow running two external displays. Also, I'd rather have those displays connected via displayport then hdmi/dvi, but can live with it.
The lack of universal USB 3 seems ridiculously cheap as well. I get some people will use this for an external keyboard and mouse, but many will want a couple hard drives attached and USB 2 is a terrible bottleneck for hard drives at this point.
• Are TB drivers hard to write? Will that get easier? If the GPU driver already exists, how complicated is the TB driver?
Normally TB is hardware and drivers transparent. But there is one catch, being external it doesn't allow main RAM aperture required for high performance 3D video card and forced them to rewrite drivers to not use AGP features.
My best understanding is that you can get it all as TB data if there isn't a DP device in the chain. But you do lose half once you have a DP device.
In my understanding, in current implementation on Apple laptops the TB is a dual 10Gbps per channel port, with one channel dedicated for video and the other for data.
So even without any DP devices connected on the TB bus, it can't use the video channel for data.
Update:
My infos seams to be incorrect, according to Apple devs notes, uses of channels is dynamically assigned, but it can't bond both channels.
Comments
Quote:
Originally Posted by digitalclips
Nice! Any idea on price of this device!
As stated upthread, it is planned to be $400 and not out till 1Q2013
It is worth to take a look at this almost year old piece:
http://www.sonnettech.com/product/echoexpresscard34thunderbolt.html
I'm waiting for the Belkin version. It has what I need. The Matrox DS1 is simply missing too much.
At $400 the Belkin is worth it. Completely worth it.
Quote:
Originally Posted by nagromme
Glad to see more TB products out there, but how about a TB external GPU—or slot in which to install one? Whatever card the drivers can handle! (Meaning, I suppose, whatever OS X expects a Mac Pro to handle.)
I can’t give up Air portability, but TB has held out the promise of greater GPU power when I’m at home at my desk. All promise... no reality, so far. A GPU made for Mac Pro would be FAR better than you’d expect an ultrathin laptop to be able to offer. Come on, TB, let me have my cake and eat it too! Plus then I could choose to keep the same (home-base-only) GPU when getting a new Air.
Realize tha the Mac Book Pro Retina can already drive multiple Monitors.
There will NEVER be a TB External GPU. There isn't much of a market for it. It already costs a lot of money to develop a TB product. So to waste it because there are few people that will buy it.
The best chance of having an external GPU is to have an expansion chassis with full-size PCI slots. The Magma Thunderbolt Expansion box has THREE full-size PCI Slots. BUT, the PCI GPU card manufacturer needs to write Thunderbolt drivers for the cards. Tough luck with that.
Anything else (MacBook Pro or desktop) doesn't need this device in the same way. There's also not going to be a lot of demand on the PC side either, I wouldn't think.
But I agree that the peripheral selection today is, at best, disappointing. But I'm not surprised. Firewire, which in the USB 1.0 days, had real advantages, never caught on much outside the digital video world (It was tough, for example, to find firewire card readers). Unless Thunderbolt becomes a consumer-grade technology present on basic PCs, then this will remain a niche technolgy with high prices, much like firewire and SCSI before that.
I think it has merit. Eventually I expect desktop displays to Retina, but I don't think a MBA or an older MBP could easily run them. Having a display with a built-in video card (or external) would help make that an option. Perhaps it's not a big market but it might be big enough that it's worth vendors looking into. Maybe CES will be favourable to Thunderbolt.
Quote:
Originally Posted by jameskatt2
Realize tha the Mac Book Pro Retina can already drive multiple Monitors.
There will NEVER be a TB External GPU. There isn't much of a market for it. It already costs a lot of money to develop a TB product. So to waste it because there are few people that will buy it.
The best chance of having an external GPU is to have an expansion chassis with full-size PCI slots. The Magma Thunderbolt Expansion box has THREE full-size PCI Slots. BUT, the PCI GPU card manufacturer needs to write Thunderbolt drivers for the cards. Tough luck with that.
The beauty about Thunderbolt comes from its hardware nature of being drivers independents. TB really is an external PCIe 4X BUS where almost any PCI cards in a TB chassis will work without any changes in it's drivers. This is also why Thunderbolt has to be built-in on the controller's motherboard and can't be add to non-thunderbolt ready PC via a PCI card.
This being said, a PCIe 4X is not enough bandwidth for high performance video card, you won't see external video solution suitable for gaming anytime soon.
Quote:
Originally Posted by digitalclips
Well equipped for the video semi-pro market. I don't understand Matrox thinking FireWire isn't necessary. They are missing they very market they specialize in, video editing. Who doesn't have some FireWire gear in this field?
It's not just video. There is a major part of the audio recording interface market that's between USB2 and PCI cards who live and die by FW. USB3 maybe be great but it's generally been ignored by recording gear manufacturers, and FW is certainly gone from new computers but is still what these interfaces are being made with, even new ones showing at NAMM. It's not a matter of getting new drives. I'm not about to sell my two year old interfaces which work great and haven't even been replaced withThunderbolt or USB3, except for a few pieces from companies, which I personally have yet to see in anyone's studio yet, available though they may be.
Here's what I don't get: if a laptop HAS USB3, what's the big deal about having a Thunderbolt box to USB3 when you can just get a USB3 hub from the native port? There's only real value in a several hundred dollar Thunderbolt box if it provides ports that the laptop itself doesn't give you. I personally have zero reason to pay for USB 2 or 3 to be coming out of such a box as long as I can still use a native port for it.
There are a couple reasons to include USB on this device. Surely it adds more USB ports without adding too much cost to the device itself but I think the most likely reason is that it allows you to keep this device on your desk with peripherals plugged in so when you disconnect your laptop you are only unplugging a power cable and a Thunderbolt.
Low volume = high prices
Like SolipsismX said, it's an all-in-one dock, single cable connects everything.
Also, something like this (or the Belkin) can add USB 3 to Thunderbolt Macs that are pre-USB 3. Which would be 12-18 month's worth of Macs production.
20Gpbs full duplex serial port at that. The only thinkg I'd consider unrealistic in this thread is the graphics. You could max out four independent USB3 ports and that might be at your limit. FW800 is a drop in the bucket.
Quote:
Originally Posted by JeffDM
20Gpbs full duplex serial port at that. The only thinkg I'd consider unrealistic in this thread is the graphics. You could max out four independent USB3 ports and that might be at your limit. FW800 is a drop in the bucket.
To be fair, in current TB implementation the bandwidth is split in half between DisplayPort video signal and DATA. But I agree with you, beside graphics TB is more than enough for any other use.
Re market size... is it worth writing the drivers? Some people assume no, but I think they are guessing. Some questions to consider:
• Are TB drivers hard to write? Will that get easier? If the GPU driver already exists, how complicated is the TB driver?
• Are ultrathin laptops going to sell better or worse over time?
• Are Macs going to sell better or worse over time?
• Are Mac games and 3D pro apps going to sell better or worse over time?
• Are ultrathin GPUs going to improve, but slotted GPUs are going to stand still, eroding the advantage?
• Are TB ports going to become more common or less over time?
• Are TB accessories going to get more expensive or less over time?
Any way I look at it, the viability of making TB drivers seems likely to improve rather than worsen. GPU makers want more customers for higher-end products, and laptops are where sales growth is in the PC industry. I’m just sad to see it go so slowly.
As for the hardware... just a box with a slot or two, serving many different niches—plus, for my niche, the cards you can already get for a Mac Pro. I’m not expecting anyone to broaden that selection, but it would still be a welcome choice to have. It looks like the hardware side is already solved, it just needs to drop in price. Next up: software?
I agree that eventually a future Air’s internal GPU may be “good enough”... except... the CURRENT Air has GPU power that would have blown a gamer away a few years ago, and now it’s not quite “good enough” for everyone. I like my detail turned up
Quote:
Originally Posted by wizard69
No
Nobody promised GPU support over TB except for a few poorly informed pundits on the web. The idea has no merit at this time.
We’re using “promise” in two different ways. It holds promise—not that someone promised to deliver it. The idea has huge merit if people want ultrathin laptops AND want more GPU power than they can contain. Such a market isn’t mainstream (neither is video capture or RAIDs) but it exists.
My best understanding is that you can get it all as TB data if there isn't a DP device in the chain. But you do lose half once you have a DP device.
The lack of universal USB 3 seems ridiculously cheap as well. I get some people will use this for an external keyboard and mouse, but many will want a couple hard drives attached and USB 2 is a terrible bottleneck for hard drives at this point.
Quote:
Originally Posted by nagromme
• Are TB drivers hard to write? Will that get easier? If the GPU driver already exists, how complicated is the TB driver?
Normally TB is hardware and drivers transparent. But there is one catch, being external it doesn't allow main RAM aperture required for high performance 3D video card and forced them to rewrite drivers to not use AGP features.
Quote:
Originally Posted by JeffDM
My best understanding is that you can get it all as TB data if there isn't a DP device in the chain. But you do lose half once you have a DP device.
In my understanding, in current implementation on Apple laptops the TB is a dual 10Gbps per channel port, with one channel dedicated for video and the other for data.
So even without any DP devices connected on the TB bus, it can't use the video channel for data.
Update:
My infos seams to be incorrect, according to Apple devs notes, uses of channels is dynamically assigned, but it can't bond both channels.
https://developer.apple.com/library/mac/#documentation/HardwareDrivers/Conceptual/ThunderboltDevGuide/Introduction/Introduction.html