I just got confirmation from my buddy in San Fran, the upcoming Mac Pro will be a modular design. The base module will have no SATA, only PCIe blade SSD storage (four slots on the prototype he used). It retains PCIe expansion slots, and yes, a 7970 was installed. He was cagey about the CPUs but said just look at Intel's roadmap for 3rd quarter 2013. That would make them E7 Xeons, with possibly up to 12 cores each. Apple will undoubtedly gimp it with 8-10 cores/cpu.
He says the base module was disguised in a plastic case, but that it's considerably thinner than the current Mac Pro, lol. Shorter too. Most impressively, even quieter.
The second module stacks on the base module, and houses up to 8 HDDs. It connects to the base module and is listed in System Information as being on the PCIe bus. Thunderbolt? The connector is different from the current implementation, so maybe it's a new revision with more lanes? Speeds are far in excess of x4 lanes.
It appears Apple are hedging against HDDs with this design. The base computer is a long term design, while the HDD module can be phased out once NAND prices make four 1-2 TB SSDs possible at a sane price.
Most intriguing is the possibility of Apple pricing at least one Mac Pro model at sub-$2000 prices. With no ODD or HDD, and thus a small PSU and case, Apple should be able to hit a fairly low price point on a single socket version. Will their rectal-cranial syndrome interfere with what is a no-brainer decision?
Junkyard...if that is the future, there'll be drooling Pro heads turning their heads. Sounds very exciting. I'll take the info' on face value. I always do. And thanks for it.
A new Mac Pro HAS to come. It HAS to change shape now. It's a dinosaur. It's price is outrageous. It's design O.T.T.
A cheaper base model would make a lot of sense. Times have moved on.
Nobody is buying shed loads of towers at 2k for a crappy quad core. Get with the program, Apple.
A modular SSD base, slimmer unit would be a kick ass Pro 'tower' (Mini tower?) with a kick ass gpu.
The modular unit dies as platter drives die?
Makes sense.
Get any more info? Pass it on.
I'm listening.
...is this like the previous 'Cluster' system Cube/Blade/Galaxy thing?
Just asking.
And yes, if Apple price the new Pro with their anal pricing structure? They want reaming.
Most intriguing is the possibility of Apple pricing at least one Mac Pro model at sub-$2000 prices. With no ODD or HDD, and thus a small PSU and case, Apple should be able to hit a fairly low price point on a single socket version. Will their rectal-cranial syndrome interfere with what is a no-brainer decision?
This is more junk. They could hit it with what they have today if that was their desire. If you look at the current design choices, they are about keeping costs down. They recycled the external case many times over, used the backplane + daughterboard design to re-use as many parts as possible without paying for a dual package chipset on the single models, and used budget parts for cpu and gpu at the $2500 mark. The mental gymnastics as to how they could cut price show very little in the way of insight. You don't see a new machine there today, so it leaves you with a blank canvas to project your ideas, even if they aren't very logical.
Just passing along what my buddy told me. I had to garble it a bit so he can't be indentified, otherwise, it's what he's been using at his studio.
3rd quarter 2013 is a long time though. Apple could easily kill the Mac Pro before then, especially if their computer sales keep tanking. I don't trust them to keep producing Macs more than 10 years or so.
I get that you don't wish to implicate him even if it is just a rumor. I wouldn't call that confirmed, and the E7 Xeons portion is just absolute nonsense. Apple has never used an EX variant. I can't think of any workstation vendor who would use them. They aren't designed for anything outside of server use. The only part that is correct is that EX type sockets are only refreshed once every tick tock cycle. They release EX with a die shrink, so currently it's still using Westmere.
It's the E5 v2 Xeon that's due for volume production in the 3rd quarter, I'm sure that's what he meant.
He's a video guy, he doesn't really know hardware tech any more than necessary to get his work done.
I know plenty of them. None of them are going to use something built with engineering samples to complete their work. Beyond that it's not like a next generation Mac Pro exists in the wild. The only reason I didn't assume this to be trolling is that you aren't known for troll posts.
Ever hear of a test mule? You think Apple just tests a totally new hardware design at Apple HQ for a few months, then says "That's good enough!" and puts it into production?
Yeah, they're called 'a select few Apple employees'.
You think Apple just tests a totally new hardware design at Apple HQ for a few months, then says "That's good enough!" and puts it into production?
Yes. Otherwise we would have known everything about every hardware release they've ever done up until now.
And yet this one of yours is the only one we've ever heard being used in this manner. At least, the only one that isn't directly from a website that lies about absolutely everything they've ever done, without pity or remorse, for the sole purpose of, well, lying (and nothing else).
So please don't forgive us for being skeptical; it's just, well, you hate most, if not all, of what Apple does right now, so making something up off the top of your head to suit your needs (real, imagined, or otherwise) makes a little more sense than the classic "heard it from a guy who told me what he has except he changed some of it so that when I told you he wouldn't be known". Never mind that Apple would know exactly who had what in the first place, since even if they gave unreleased, industry-reinventing hardware to third-parties (they don't), they would give DIFFERENT hardware to different places so that they could know exactly where any one leak came from.
You can ask one Olympian to lift 500 pounds or you can ask 10 normal people to help lift 50 pounds. Which is reality based? Both are. Is one solution better than the other? It depends. For many applications, many hands make light work. An ARM processor may be like the normal people and the latest x86 like the Olympic weight-lifter, but do I really care what is inside the machine if they can both accomplish the same work? No I don't.
One Olympic weight lifter cannot scale above its roughly 500 pound limit. But average people are cheap to come by. 100 average people can lift way more than one Olympic athlete. If each average person can lift 50 pounds, than 100 people can lift 5,000 pounds - way more than the weight lifter. Sometimes quantity has a quality all its own.
There is no question audio, video, and photo work are massively parallel problems. Are these not the main source of taxing the CPUs? Then you would be better served by a parallel solution not a serial solution. Therefore ARM chips are not a stupid solution.
Of course, one could come back and say what about having 10 Olympic weight lifters, not just one? This too is a valid solution, and in fact, the one we currently have. Which is better? I really do not care which solution I have if the work outcome is the same. Therefore, it all comes down to cost. How much does cost to hire 10 Olympic weight lifters verses how much does it cost to hire 100 average people? Realistically, how much do x86 chips cost verses a boatload of equivalent ARM chips? If they are the same cost than I really do not see why Apple should do anything different. But if one solution is much cheaper, then that is what Apple should do.
Actually, cost is not the only factor. Heat is another big deal. I am not a CPU guru so I do not know the answers to these questions. Maybe somebody else here does. I am sure Apple has this data. The fact that Apple seems to be dragging their feet on the Pro side indicates to me that a significant shift is coming soon.
This theory fits the data better than proposing Apple just drop all its pro level machines and software. What is pro level today becomes the consumer machine of tomorrow. That is why Apple needs to stay in the Pro market and why they will. Therefore I see the Xeon Phi or a massively parallel ARM new MacPro being the way forward in the near future.
Apple isn't going to migrate to a new architecture again, not with Intel at the top of their game. And if Apple were to decide to migrate, the Mac Pro would be the LAST model they would migrate.
Beyond that, Apple obviously doesn't care about pushing computer tech limits with the Mac Pro. They aren't even interested in taking what Intel can give them at this point.
I would say, in the past 5 years even when Steve was around, Apple HAS changed, from your classic Apple. Now that there is Tim Cook now too, Anything is possible, seriously...
I just got confirmation from my buddy in San Fran, the upcoming Mac Pro will be a modular design. The base module will have no SATA, only PCIe blade SSD storage (four slots on the prototype he used). It retains PCIe expansion slots, and yes, a 7970 was installed. He was cagey about the CPUs but said just look at Intel's roadmap for 3rd quarter 2013. That would make them E7 Xeons, with possibly up to 12 cores each. Apple will undoubtedly gimp it with 8-10 cores/cpu.
Well this sort of mirrors what I've been suggesting. To build a Pro suitable for the next few years Apple needs to overhaul the architecture massively.
As for the CPUs, Intel is working on many options that could end up in a new Mac Pro. However a 12 core model would be a nice high end model and Intel does have the intention to introduce such a chip.
He says the base module was disguised in a plastic case, but that it's considerably thinner than the current Mac Pro, lol. Shorter too. Most impressively, even quieter.
Sounds like previous Apple practice.
The second module stacks on the base module, and houses up to 8 HDDs. It connects to the base module and is listed in System Information as being on the PCIe bus. Thunderbolt? The connector is different from the current implementation, so maybe it's a new revision with more lanes? Speeds are far in excess of x4 lanes.
Faster maybe. The problem with the current TB connector is that it isn't exactly a high reliability connection for a workstation. As far as speeds I'm not sure how an array of hard disks can be that much faster than the current TB connector.
It appears Apple are hedging against HDDs with this design. The base computer is a long term design, while the HDD module can be phased out once NAND prices make four 1-2 TB SSDs possible at a sane price.
That would make sense. Though mass storage needs continual to grow at a very fast rate so I don't see hard drive arrays going away anytime soon.
Most intriguing is the possibility of Apple pricing at least one Mac Pro model at sub-$2000 prices. With no ODD or HDD, and thus a small PSU and case, Apple should be able to hit a fairly low price point on a single socket version. Will their rectal-cranial syndrome interfere with what is a no-brainer decision?
A base model in that price range would be ideal. It would have to be suitably powerful, but they should be able to cut costs significantly with a new architecture. There is nothing more important than getting the base model back to a reasonable price point.
and obviously Thunderbolt but if they don't have an exclusive, it's more likely to be June than April and even then the shipping date is likely to be further out.
They could be at NAB to dispel some of the remaining negativity around FCPX - it would be good if they had a keynote about real-world cases where it's being used and the workflows as well as the features they added. That would give people assurance that it's capable of handling heavy duty workloads. A new 64-bit QT X Pro with feature parity to QT 7 Pro wouldn't go amiss either (that includes the UI not covering the frame when you hover over it).
Where it really shines is content creation - Audio, Video, Photo. I am sure there are a few other things too, but the point is that these apps can all benefit from massively parallel processing. Even the current Mac Pros are not nearly powerful enough for what we wish to do.
There's a good article about the use of cloud computing here, which is another factor to think about:
"There is a small client-based application and plugins that allow the jobs to be run directly from NUKE and Maya. Its also a web based management so it doesn't matter if the studio is PC, Linux or Mac-based.
When launched straight from Autodesk Maya, ZYNC Render system allows the user to launch as many instances as they require, then drop them when they don’t need them, using Amazon EC2. It’s up to the user how many machines they want and how fast they want the job done.
With this arrangement, we’re seeing that people will be able to render more. They’ll be able to pick off hundreds, if not thousands of nodes if they need it, and then only pay for rendering as a utility without having this upfront cost of having to buy this whole farm.
Todd talks about a situation he heard about where a director needed a change and was coming in, in two hours. The outfit needed the power of 500 nodes. With ZYNC, the solution was possible with the flick of a switch.
There are also up to five nodes in NUKE 7 that are GPU-accelerated and over time, we’ll see more and more existing products move from CPU to GPU power and we’ll be able to provide that on the Cloud.
ZYNC has a list of individuals who have created amazing work with really heavy lifting graphics, working from home as a freelancer. Alex Brubaker, does work for ESPN on their opener graphics, can start off with a 15-node render job without any of the worry of getting render hardware of his own."
Cloud rendering isn't always ideal. You'd need a fast network to be able to upload multiple GBs of source content so for the likes of compositing where you might have 100GB of footage, that's better done locally. This will ultimately be done in real-time though. For CGI and other compute work, the source files can be a lot smaller but a fast network will still help. It would be difficult at any point in time for an individual to buy 15 top of the line render nodes outright and impossible to be able to scale it up to 500 or more instantly.
If the main part of the work is done real-time, it can be done on almost any machine and one MP on its own wouldn't be enough in the high-end cases so cloud rendering makes it less important to have one. For the cases where it's not feasible to upload a lot of data, having a powerful local machine has its benefits but all the quad-i7 machines are powerful machines.
The video above shows that it's been used on hundreds of commercials and 7 feature films and delivered all of the ~400 visual effects shots for the movie Flight.
[VIDEO]
There are few higher-end usage scenarios than the visual effects shots for a major movie production.
"There is a small client-based application and plugins that allow the jobs to be run directly from NUKE and Maya. Its also a web based management so it doesn't matter if the studio is PC, Linux or Mac-based.
When launched straight from Autodesk Maya, ZYNC Render system allows the user to launch as many instances as they require, then drop them when they don’t need them, using Amazon EC2. It’s up to the user how many machines they want and how fast they want the job done.
With this arrangement, we’re seeing that people will be able to render more. They’ll be able to pick off hundreds, if not thousands of nodes if they need it, and then only pay for rendering as a utility without having this upfront cost of having to buy this whole farm.
Todd talks about a situation he heard about where a director needed a change and was coming in, in two hours. The outfit needed the power of 500 nodes. With ZYNC, the solution was possible with the flick of a switch.
There are also up to five nodes in NUKE 7 that are GPU-accelerated and over time, we’ll see more and more existing products move from CPU to GPU power and we’ll be able to provide that on the Cloud.
ZYNC has a list of individuals who have created amazing work with really heavy lifting graphics, working from home as a freelancer. Alex Brubaker, does work for ESPN on their opener graphics, can start off with a 15-node render job without any of the worry of getting render hardware of his own."
I like how they used vray. That is an amazing piece of software. I don't think cloud computing will completely remove the need for optimization to control costs and cpu time requirements. You still pay for the time there. It provides smaller shops with some amount of scalability, which is very cool. It doesn't actually displace the need for workstation hardware anywhere things must be addressed in real time, especially in terms of gpu hardware. GPUs get stressed quite a bit. It's still common to see low rez proxies used to set up a scene or animate even with the use of powerful gpus.
My point was it provides a solution where routine use isn't enough to justify a step up in hardware purchases. It still costs money, and there are some things that it won't replace, but it provides a huge resource for 1-10 man shops.
Comments
Confirmed: Mac Pro to be a modular design.
I just got confirmation from my buddy in San Fran, the upcoming Mac Pro will be a modular design. The base module will have no SATA, only PCIe blade SSD storage (four slots on the prototype he used). It retains PCIe expansion slots, and yes, a 7970 was installed. He was cagey about the CPUs but said just look at Intel's roadmap for 3rd quarter 2013. That would make them E7 Xeons, with possibly up to 12 cores each. Apple will undoubtedly gimp it with 8-10 cores/cpu.
He says the base module was disguised in a plastic case, but that it's considerably thinner than the current Mac Pro, lol. Shorter too. Most impressively, even quieter.
The second module stacks on the base module, and houses up to 8 HDDs. It connects to the base module and is listed in System Information as being on the PCIe bus. Thunderbolt? The connector is different from the current implementation, so maybe it's a new revision with more lanes? Speeds are far in excess of x4 lanes.
It appears Apple are hedging against HDDs with this design. The base computer is a long term design, while the HDD module can be phased out once NAND prices make four 1-2 TB SSDs possible at a sane price.
Most intriguing is the possibility of Apple pricing at least one Mac Pro model at sub-$2000 prices. With no ODD or HDD, and thus a small PSU and case, Apple should be able to hit a fairly low price point on a single socket version. Will their rectal-cranial syndrome interfere with what is a no-brainer decision?
Junkyard...if that is the future, there'll be drooling Pro heads turning their heads. Sounds very exciting. I'll take the info' on face value. I always do. And thanks for it.
A new Mac Pro HAS to come. It HAS to change shape now. It's a dinosaur. It's price is outrageous. It's design O.T.T.
A cheaper base model would make a lot of sense. Times have moved on.
Nobody is buying shed loads of towers at 2k for a crappy quad core. Get with the program, Apple.
A modular SSD base, slimmer unit would be a kick ass Pro 'tower' (Mini tower?) with a kick ass gpu.
The modular unit dies as platter drives die?
Makes sense.
Get any more info? Pass it on.
I'm listening.
...is this like the previous 'Cluster' system Cube/Blade/Galaxy thing?
Just asking.
And yes, if Apple price the new Pro with their anal pricing structure? They want reaming.
Lemon Bon Bon.
Yes, in that it's also completely made up.
Quote:
Originally Posted by Junkyard Dawg
Most intriguing is the possibility of Apple pricing at least one Mac Pro model at sub-$2000 prices. With no ODD or HDD, and thus a small PSU and case, Apple should be able to hit a fairly low price point on a single socket version. Will their rectal-cranial syndrome interfere with what is a no-brainer decision?
This is more junk. They could hit it with what they have today if that was their desire. If you look at the current design choices, they are about keeping costs down. They recycled the external case many times over, used the backplane + daughterboard design to re-use as many parts as possible without paying for a dual package chipset on the single models, and used budget parts for cpu and gpu at the $2500 mark. The mental gymnastics as to how they could cut price show very little in the way of insight. You don't see a new machine there today, so it leaves you with a blank canvas to project your ideas, even if they aren't very logical.
Just passing along what my buddy told me. I had to garble it a bit so he can't be indentified, otherwise, it's what he's been using at his studio.
3rd quarter 2013 is a long time though. Apple could easily kill the Mac Pro before then, especially if their computer sales keep tanking. I don't trust them to keep producing Macs more than 10 years or so.
Quote:
Originally Posted by Junkyard Dawg
Just passing along what my buddy told me.
I get that you don't wish to implicate him even if it is just a rumor. I wouldn't call that confirmed, and the E7 Xeons portion is just absolute nonsense. Apple has never used an EX variant. I can't think of any workstation vendor who would use them. They aren't designed for anything outside of server use. The only part that is correct is that EX type sockets are only refreshed once every tick tock cycle. They release EX with a die shrink, so currently it's still using Westmere.
It's the E5 v2 Xeon that's due for volume production in the 3rd quarter, I'm sure that's what he meant.
He's a video guy, he doesn't really know hardware tech any more than necessary to get his work done.
Originally Posted by Junkyard Dawg
…it's what he's been using at his studio.
Yeah, I'm sure Apple lent out an industry-redefining product to a third party¡
Quote:
Originally Posted by Junkyard Dawg
It's the E5 v2 Xeon that's due for volume production in the 3rd quarter, I'm sure that's what he meant.
He's a video guy, he doesn't really know hardware tech any more than necessary to get his work done.
I know plenty of them. None of them are going to use something built with engineering samples to complete their work. Beyond that it's not like a next generation Mac Pro exists in the wild. The only reason I didn't assume this to be trolling is that you aren't known for troll posts.
Ever hear of a test mule? You think Apple just tests a totally new hardware design at Apple HQ for a few months, then says "That's good enough!" and puts it into production?
Originally Posted by Junkyard Dawg
Ever hear of a test mule?
Yeah, they're called 'a select few Apple employees'.
You think Apple just tests a totally new hardware design at Apple HQ for a few months, then says "That's good enough!" and puts it into production?
Yes. Otherwise we would have known everything about every hardware release they've ever done up until now.
And yet this one of yours is the only one we've ever heard being used in this manner. At least, the only one that isn't directly from a website that lies about absolutely everything they've ever done, without pity or remorse, for the sole purpose of, well, lying (and nothing else).
So please don't forgive us for being skeptical; it's just, well, you hate most, if not all, of what Apple does right now, so making something up off the top of your head to suit your needs (real, imagined, or otherwise) makes a little more sense than the classic "heard it from a guy who told me what he has except he changed some of it so that when I told you he wouldn't be known". Never mind that Apple would know exactly who had what in the first place, since even if they gave unreleased, industry-reinventing hardware to third-parties (they don't), they would give DIFFERENT hardware to different places so that they could know exactly where any one leak came from.
Quote:
Originally Posted by visionary
You can ask one Olympian to lift 500 pounds or you can ask 10 normal people to help lift 50 pounds. Which is reality based? Both are. Is one solution better than the other? It depends. For many applications, many hands make light work. An ARM processor may be like the normal people and the latest x86 like the Olympic weight-lifter, but do I really care what is inside the machine if they can both accomplish the same work? No I don't.
One Olympic weight lifter cannot scale above its roughly 500 pound limit. But average people are cheap to come by. 100 average people can lift way more than one Olympic athlete. If each average person can lift 50 pounds, than 100 people can lift 5,000 pounds - way more than the weight lifter. Sometimes quantity has a quality all its own.
There is no question audio, video, and photo work are massively parallel problems. Are these not the main source of taxing the CPUs? Then you would be better served by a parallel solution not a serial solution. Therefore ARM chips are not a stupid solution.
Of course, one could come back and say what about having 10 Olympic weight lifters, not just one? This too is a valid solution, and in fact, the one we currently have. Which is better? I really do not care which solution I have if the work outcome is the same. Therefore, it all comes down to cost. How much does cost to hire 10 Olympic weight lifters verses how much does it cost to hire 100 average people? Realistically, how much do x86 chips cost verses a boatload of equivalent ARM chips? If they are the same cost than I really do not see why Apple should do anything different. But if one solution is much cheaper, then that is what Apple should do.
Actually, cost is not the only factor. Heat is another big deal. I am not a CPU guru so I do not know the answers to these questions. Maybe somebody else here does. I am sure Apple has this data. The fact that Apple seems to be dragging their feet on the Pro side indicates to me that a significant shift is coming soon.
This theory fits the data better than proposing Apple just drop all its pro level machines and software. What is pro level today becomes the consumer machine of tomorrow. That is why Apple needs to stay in the Pro market and why they will. Therefore I see the Xeon Phi or a massively parallel ARM new MacPro being the way forward in the near future.
Apple isn't going to migrate to a new architecture again, not with Intel at the top of their game. And if Apple were to decide to migrate, the Mac Pro would be the LAST model they would migrate.
Beyond that, Apple obviously doesn't care about pushing computer tech limits with the Mac Pro. They aren't even interested in taking what Intel can give them at this point.
I would say, in the past 5 years even when Steve was around, Apple HAS changed, from your classic Apple. Now that there is Tim Cook now too, Anything is possible, seriously...
EOL
As for the CPUs, Intel is working on many options that could end up in a new Mac Pro. However a 12 core model would be a nice high end model and Intel does have the intention to introduce such a chip. Sounds like previous Apple practice. Faster maybe. The problem with the current TB connector is that it isn't exactly a high reliability connection for a workstation. As far as speeds I'm not sure how an array of hard disks can be that much faster than the current TB connector. That would make sense. Though mass storage needs continual to grow at a very fast rate so I don't see hard drive arrays going away anytime soon. A base model in that price range would be ideal. It would have to be suitably powerful, but they should be able to cut costs significantly with a new architecture. There is nothing more important than getting the base model back to a reasonable price point.
I'm not saying this rumor has any truth to it just that Apple has sent out test chassis in the past.
Originally Posted by wizard69
Why would they stop? Apple has done so many times in the past with Macs.
I'm not saying this rumor has any truth to it just that Apple has sent out test chassis in the past.
Wait, really? When? For what?
What are the chances we'll see an announcement of the new Mac Pro at NAB, like they did with FCPX in 2010?
Surely the largest remaining target audience of a true pro machine are filmmakers, videographers, compositers and animators.
Ivy Bridge EP doesn't come out until at least June / July:
http://www.tweaktown.com/news/23076/we_shouldn_t_see_ivy_bridge_e_processors_until_mid_2013/index.html
They've had early exclusives from Intel before:
http://www.bit-tech.net/news/hardware/2009/10/16/rumour-32nm-xeon-a-timed-apple-exclusive/1
and obviously Thunderbolt but if they don't have an exclusive, it's more likely to be June than April and even then the shipping date is likely to be further out.
They could be at NAB to dispel some of the remaining negativity around FCPX - it would be good if they had a keynote about real-world cases where it's being used and the workflows as well as the features they added. That would give people assurance that it's capable of handling heavy duty workloads. A new 64-bit QT X Pro with feature parity to QT 7 Pro wouldn't go amiss either (that includes the UI not covering the frame when you hover over it).
There's a good article about the use of cloud computing here, which is another factor to think about:
http://www.cgsociety.org/index.php/CGSFeatures/CGSFeatureSpecial/zync
"There is a small client-based application and plugins that allow the jobs to be run directly from NUKE and Maya. Its also a web based management so it doesn't matter if the studio is PC, Linux or Mac-based.
When launched straight from Autodesk Maya, ZYNC Render system allows the user to launch as many instances as they require, then drop them when they don’t need them, using Amazon EC2. It’s up to the user how many machines they want and how fast they want the job done.
With this arrangement, we’re seeing that people will be able to render more. They’ll be able to pick off hundreds, if not thousands of nodes if they need it, and then only pay for rendering as a utility without having this upfront cost of having to buy this whole farm.
Todd talks about a situation he heard about where a director needed a change and was coming in, in two hours. The outfit needed the power of 500 nodes. With ZYNC, the solution was possible with the flick of a switch.
There are also up to five nodes in NUKE 7 that are GPU-accelerated and over time, we’ll see more and more existing products move from CPU to GPU power and we’ll be able to provide that on the Cloud.
ZYNC has a list of individuals who have created amazing work with really heavy lifting graphics, working from home as a freelancer. Alex Brubaker, does work for ESPN on their opener graphics, can start off with a 15-node render job without any of the worry of getting render hardware of his own."
Cloud rendering isn't always ideal. You'd need a fast network to be able to upload multiple GBs of source content so for the likes of compositing where you might have 100GB of footage, that's better done locally. This will ultimately be done in real-time though. For CGI and other compute work, the source files can be a lot smaller but a fast network will still help. It would be difficult at any point in time for an individual to buy 15 top of the line render nodes outright and impossible to be able to scale it up to 500 or more instantly.
If the main part of the work is done real-time, it can be done on almost any machine and one MP on its own wouldn't be enough in the high-end cases so cloud rendering makes it less important to have one. For the cases where it's not feasible to upload a lot of data, having a powerful local machine has its benefits but all the quad-i7 machines are powerful machines.
The video above shows that it's been used on hundreds of commercials and 7 feature films and delivered all of the ~400 visual effects shots for the movie Flight.
[VIDEO]
There are few higher-end usage scenarios than the visual effects shots for a major movie production.
Quote:
Originally Posted by Marvin
There's a good article about the use of cloud computing here, which is another factor to think about:
http://www.cgsociety.org/index.php/CGSFeatures/CGSFeatureSpecial/zync
"There is a small client-based application and plugins that allow the jobs to be run directly from NUKE and Maya. Its also a web based management so it doesn't matter if the studio is PC, Linux or Mac-based.
When launched straight from Autodesk Maya, ZYNC Render system allows the user to launch as many instances as they require, then drop them when they don’t need them, using Amazon EC2. It’s up to the user how many machines they want and how fast they want the job done.
With this arrangement, we’re seeing that people will be able to render more. They’ll be able to pick off hundreds, if not thousands of nodes if they need it, and then only pay for rendering as a utility without having this upfront cost of having to buy this whole farm.
Todd talks about a situation he heard about where a director needed a change and was coming in, in two hours. The outfit needed the power of 500 nodes. With ZYNC, the solution was possible with the flick of a switch.
There are also up to five nodes in NUKE 7 that are GPU-accelerated and over time, we’ll see more and more existing products move from CPU to GPU power and we’ll be able to provide that on the Cloud.
ZYNC has a list of individuals who have created amazing work with really heavy lifting graphics, working from home as a freelancer. Alex Brubaker, does work for ESPN on their opener graphics, can start off with a 15-node render job without any of the worry of getting render hardware of his own."
I like how they used vray. That is an amazing piece of software. I don't think cloud computing will completely remove the need for optimization to control costs and cpu time requirements. You still pay for the time there. It provides smaller shops with some amount of scalability, which is very cool. It doesn't actually displace the need for workstation hardware anywhere things must be addressed in real time, especially in terms of gpu hardware. GPUs get stressed quite a bit. It's still common to see low rez proxies used to set up a scene or animate even with the use of powerful gpus.
My point was it provides a solution where routine use isn't enough to justify a step up in hardware purchases. It still costs money, and there are some things that it won't replace, but it provides a huge resource for 1-10 man shops.