Activity › Forums › Creative Community Conversations › Pretty amazing Thunderbolt demo.
-
Pretty amazing Thunderbolt demo.
Walter Soyka replied 14 years, 3 months ago 26 Members · 119 Replies
-
Frank Gothmann
January 20, 2012 at 12:12 pmI agree that it’s an impressive demo. And I agree that it’s great for anybody who needs fast io on the road with mobile devices. Never disputed that.
All I am saying is that there’s a lot more to the whole story than the one-example-fits-all approach.
There’s the driver issue, energy is a problem since most GPUs draw much more power than the external box can provide.
PCIe 3.0, on a dual cpu tower, will pack 80 lanes vs TB’s 4x so even a future revision of TB will have to go a long way to match that. Power is not an issue and Sandy Bridge Xeons will outperform regular i7s.
And PCIe 4.0 is in development, as is external PCIe.
So.. why limit the possibilities, why build even more walls? My clients don’t walk in with TB or FW800 drives. Most come with USB 2.0, some USB 3.0 or eSata. I doubt this will change anytime soon.Wouldn’t it be convenient to have these built in, no adapters, especially since it’s only politics why Apple chooses not too, no technical reason whatsoever.Do I need that kind of performance provided by PCIe 3.0? Maybe, maybe not, but it’s there, today, so why should I settle for less if the mobility advantage of TB has zero value for me as I simply don’t work on the road and the cost of deploying TB is more expensive, less compatible and has limited cross platform capabilities.
It has zero advantage for me so I refuse to cheer at the prospect of this being the only choice for high bandwidth io in Apple’s future. I’ll happily cheer at the prospect of it being an addition, not a replacement.Yes, Apple probably won’t have to build MacPros anymore to further their business. To me, that’s bad.
-
Jeremy Garchow
January 20, 2012 at 3:02 pm[Frank Gothmann] “All I am saying is that there’s a lot more to the whole story than the one-example-fits-all approach”
Of course there is. But we also know Apple’s history. If this is the way it’s going, and I want to keep using OSX, then I can only use the tools that I have available. This demo proves to me, that there’s a lot of potential here. Not to replace desktops, but to have all the power and connectivity of a desktop in a much more flexible form factor (for my needs, perhaps not yours).
[Frank Gothmann] “There’s the driver issue, energy is a problem since most GPUs draw much more power than the external box can provide.”
Yeah, manufacturers will need to retool their drivers. It’s already happening. As far as power, not sure what you mean. Why will it need more than what you can get from the outlet on the wall?
[Frank Gothmann] “And PCIe 4.0 is in development, as is external PCIe.
So.. why limit the possibilities, why build even more walls? My clients don’t walk in with TB or FW800 drives. Most come with USB 2.0, some USB 3.0 or eSata. I doubt this will change anytime soon.Wouldn’t it be convenient to have these built in, no adapters, especially since it’s only politics why Apple chooses not too, no technical reason whatsoever. “Macs don’t have USB ports? I’m not following you here. Are there other computers that have built in sata? Clients always send me USB drives too as they are the cheapest/most ubiquitous.
It is true, 4x can’t cut it for everyone, but it’s just the beginning. That demo is a beginning. There’s no reason for me, personally, for PCIe 4.0 at the moment. Thunderbolt presents more opportunity, for me and my needs, than a faster pcie bus. I like fast CPUs as well, but this demo was an eye opener.
[Frank Gothmann] “It has zero advantage for me so I refuse to cheer at the prospect of this being the only choice for high bandwidth io in Apple’s future. I’ll happily cheer at the prospect of it being an addition, not a replacement.
Yes, Apple probably won’t have to build MacPros anymore to further their business. To me, that’s bad.”
We see it differently and that’s fine. Thunderbolt is more flexible, which is good for me personally. I’ll take that over more lanes.
I do think we will see another MacPro or two, but for apple, it doesn’t make sense to keep producing them if they don’t sell. I do believe they will offer an alternative that won’t be an iMac or a MacPro. If not, there’s plenty of PCs that will fit the bill.
Jeremy
-
Frank Gothmann
January 20, 2012 at 3:53 pm[Jeremy Garchow] “As far as power, not sure what you mean. Why will it need more than what you can get from the outlet on the wall?”
The box can only acomodate cards that draw a max of 150 Watts (the smaller enclosure only 80, same for the MSI). Most GPUs draw more than that. A Quadro 4000, according to spec, draws 142, that’s too close for compfort if it spikes. A current Radeon HD 7970 peaks at over 270 w.
[Jeremy Garchow] “Macs don’t have USB ports? I’m not following you here. Are there other computers that have built in sata? Clients always send me USB drives too as they are the cheapest/most ubiquitous.”
What I am saying is that I’d like to see a broad range of IO options based on real world usage. There is USB2, FW800 and TB in Apple’s future. Why can’t we have USB3, eSata, FW800, TB, hdmi and whatever is actually out there. Apple’s concept of pushing a certain technology by eliminating or keeping out others has not worked in the past, why is it always the same story all over again where people end up on an island having to buy adapters that introduce more issues (eg display port) and cost.
-
Jeremy Garchow
January 20, 2012 at 4:32 pm[Frank Gothmann] “The box can only acomodate cards that draw a max of 150 Watts (the smaller enclosure only 80, same for the MSI). Most GPUs draw more than that. A Quadro 4000, according to spec, draws 142, that’s too close for compfort if it spikes. A current Radeon HD 7970 peaks at over 270 w. “
Ah. That can be fixed with more powerful electronics. Not too worried about that. This is just a start.
[Frank Gothmann] “What I am saying is that I’d like to see a broad range of IO options based on real world usage. There is USB2, FW800 and TB in Apple’s future. Why can’t we have USB3, eSata, FW800, TB, hdmi and whatever is actually out there. Apple’s concept of pushing a certain technology by eliminating or keeping out others has not worked in the past, why is it always the same story all over again where people end up on an island having to buy adapters that introduce more issues (eg display port) and cost.”
it’s the way Apple has worked for a long time, so I guess it’s not surprising to me. Yes, thunderbolt might require an adapter, but you buy what you need and don’t buy what you don’t. It’s very flexible.
-
Walter Soyka
January 20, 2012 at 4:48 pmVery cool demo!
Once again, this makes it clear that many of us have different needs and thus different definitions of “power.”
Some need massive throughput — doable with TB on an Air.
Others need massive compute power — doable with a “24 sizzle core beast.” (Jeremy, this is my new favorite expression!)
Some niche within a niche needs both.
I’m glad we all have so many options to meet our needs, and I hope that continues to be the case for a long time to come.
All that said, let’s put Thunderbolt in perspective: Thunderbolt’s big advantage is not its absolute speed. There are faster interconnects. The thing that makes Thunderbolt very special is that it’s external (so you can expand outside the case and make it work on itty bitty laptops), while being relatively cheap and relatively fast.
[Jeremy Garchow] “That’s the thing. you wouldn’t need a whole new computer with it’s own OS and infrastructure, you’d just plug in and double your CPU.”
This is really, really speculative. I’d argue that suggesting that Thunderbolt will be capable of this kind of expansion is very premature.
Thunderbolt is DisplayPort plus PCIe. It is a not a CPU interconnect. There are no PCIe cards that allow you to just plug a card in to add a CPU. This simply doesn’t reflect current computer architecture.
Intel’s multiprocessing uses a separate high-speed CPU bus, unrelated to the expansion bus. The current interconnect is called QuickPath, and it runs at 25.6 GB/s (that’s a capital B for bytes) on 3.2 GHz processors. This is way faster than Thunderbolt at 10 Gb/s. (Incidentally, this is why the Core i7 cannot be configured in a multi-processor configuration. It doesn’t implement QPI.)
Looking at the end of Thunderbolt’s roadmap and suggesting it’d be fast enough to be a good CPU interconnect presupposes that there will be no further advancement in CPUs or dedicated CPU interconnects like QuickPath in the interim. I don’t think that’s realistic. The rest of the computing industry is not standing still while Thunderbolt develops.
That said, Thunderbolt could make a nice cluster interconnect, but clustering is not the same as just plugging in and doubling your CPU.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
Craig Seeman
January 20, 2012 at 5:10 pmLaCie eSata to Thunderbolt
https://modmyi.com/content/6535-ces-2012-lacie-hub-allows-esata-drives-connect-thunderbolt-macs.htmlBelkin USB2, Firewire 800, HDMI to Thunderbolt.
https://modmyi.com/content/6548-ces-2012-belkin-adds-new-dock-list-thunderbolt-accessories.htmlSonnet Express Cart to Thunderbolt
https://www.sonnettech.com/product/echoexpresscard34thunderbolt.htmlThe only thing I’m not seeing on the way is USB3 to Thunderbolt. One can speculate why not.
Although it’s been alluded to, the advantage of Thunderbolt is that, unlike PCIe cards, such devices will be able to move from desktop (once implemented) to portable device right on down to the MacBook Air and, soon, Acer and Lenovo laptops as well. Unless you’re a “Pro” that has no use for portable computing, this is a major breakthrough in convenience. When you consider Magma Expansion chassis, even certain PCIe cards become portable as well as cross platform. Granted there are PCIe needs such as 16x GPUs which really are dedicated to workstations given Thunderbolt limitations but just because there’s some exceptions, doesn’t make this a major PROFESSIONAL improvement.
One might edit on a MacBook Air, they are now viable for use in the field if you need to cut together a few shots to show a client in the field during a shoot.
-
Jeremy Garchow
January 20, 2012 at 5:25 pm[Walter Soyka] “This is really, really speculative. I’d argue that suggesting that Thunderbolt will be capable of this kind of expansion is very premature.”
But of course. I mentioned that.
Here’s to the future.
-
Richard Herd
January 20, 2012 at 5:26 pm[Bill Davis] “I now find that scrolling through pages and pages of clip icons is something I only want to do during times I select for creative exploration. For most edit purposes, where I already know what clip I need, keyword search was is so very much faster and more powerful that “clip scrolling” becomes a chore rather than something enjoyable.”
Seconded.
In legacy FCP I used a lot of markers. The keywords are better markers, IMO. Instead of marking a single frame, I mark (keyword reallY) a range.
The real problem I’m having is trying to keep the keywords regularized. I still don’t have a nomenclature system yet.
Any good advice would be welcome, of course.
For example, I’m cutting a film, the writer/director’s first attempt. My first pass at keywords was with the script and I’d label them as Scene 1. Then selected scene 1 and all the scene 1 coverage shows up. Keyword again for WS, MS, CU etc. Then I keyworded the coverage for favorites.
It seemed to work okay. Generally speaking I screened each scene twice and had a decent assemble of each scene based on the following hierarchy search: Scene 1/Favorites. My favorite WS, MS, CU etc were the only pieces of footage taking up screen real estate when it was time to blade it.
One more cool thing! You can double click each edit point and dial in the precision.
The actual cutting process of X is pretty cool.
-
Frank Gothmann
January 20, 2012 at 6:00 pmWhich is what I have said: lots of adapters, at a cost: your product links incl. the Magma box will set you back 2.000 bucks. Pile them up and you have a “tower”, both in size and weight. And most of that cost will be on the Mac user bill because hdmi, esata, usb3 etc. are standard ports on most Lenovo and Acer portables already so prices are not likely to go down too much for the Mac market alone.
Raid connectivity etc. to an Air or Macbook Pro – yeah, that is great. No doubt. Still no reason not to have standard ports built in. -
Richard Herd
January 20, 2012 at 6:07 pm[Frank Gothmann] “Why is it that people here tend to take a very specific example and try to generalise it.”
All learning environments are inductive. So is science. Only mathematics and formal logic proceed deductively.
Reply to this Discussion! Login or Sign Up