I don't believe high performance computing clusters typically use a PCI bridge as an interconnect.
It depends upon at which level you are interconnecting. PCI may be used to access a computation accelerator such as a GPU.
This probably sounded like a good idea to you and others, but I have yet to see to see it suggested by anyone with even remote knowledge of engineering. The other issue is that no one has really pointed out how such a thing could be set up.
Well there is a question of scale here, of you are thinking HPC center type clusters I see no future for TB there. However if a cluster to you is two to four Mac Pros tied together there is potential. It all depends upon software support as far as I can see. It also depends uponf how you define a cluster, there are many ways to group machines together to solve a problem, Apple could simply use TB to send well defined work units to another box.
I am guessing this is a joke? It's really silly how no one considers logistics such as the case for using a cluster vs a single workstation.
I do tend to agree that people are getting excited over things they don't understand. Realistically you could cluster Macs together now to solve problems.
Does thunderbolt support zero copy protocols? Will we see thunderbolt switches?
Probably not. I really beleive that Apple and Intel had well defined goals for TB when they started development, combine this with the fact that Intel has a superchip near release with a cluster interface built in and I have to say no to TB "switches". I had to quote that because TB is built around crossbar switch technology.
Does GCD support a PCI bridge interconnect? Does it have some method of failover support? Have they started on thunderbolt switches? You'd need some kind of switch for issues like traffic shaping to deal with multiple IO requests. Details seem to be ignored.
While those are all good questions you seem to mis the point here, if Apple is about to market a cluster solution it will be dead simple and defined for a specific group of uses. It won't be a solution where huge racks of Mac Pros are tied together. At best it will be a desktop "clustering" solution. At best it would tie a few machines together, we're few means less than seven. The reality is there are far better solutions in heavy deployment for larger clusters.
Infiniband is that solution. As can be seen here http://www.computerworld.com/s/article/9225703/Intel_superchip_will_be_aimed_at_high_performance_computing Intel is on the road to integrating the tech into its processors directly. It would be most interesting if Apples long delay is not for Sandy Bridge E but rather for this mystery chip with InfiniBand integrated on board. Such a chip would make for a very interesting approach for the next Mac Pro, a machine where clustering isn't an after thought at all. By the way Apple would still need TB on such a machine.
All of the above sounds good, but I'm not so certain that it is a chip that will be ready this year. Hard to tell though, you wouldn't think Intel would announce anything at all unless e hardware was close.
How is that a personal attack? Your comment clearly indicates that you don't understand what would be involved in having every single aspect of Apple's products designed and made in the US. In fact, it's impossible without having to reinvent many aspects and using inferior variants at much higher costs due to components and licenses from companies that reside outside the US that would need to be circumvented.
Then you have the issue of even having a factory whose building and equipment will be built with products from all around the world. But you're not caring about all that, are you? You just want the actual assembly to be in the US without regard for anything else to satisfy your myopic and cliched commentary.
So now you're argument is that no one else but you and Cook want jobs to come to the US? Did you actually read the article or the transcript of his interview. He very clearly stated. ""We will do as many of these things [in America] as we can do," What you propose is myopic and foolish, what Cook, myself and many others have expressed is a desire but a realism that all things simply aren't possible.
And you still haven't addressed how components owned and made by other companies could get remade by Apple without infringing on their IP. Cook addressed it alright.
So now you're argument is that no one else but you and Cook want jobs to come to the US? Did you actually read the article or the transcript of his interview. He very clearly stated. ""We will do as many of these things [in America] as we can do," What you propose is myopic and foolish, what Cook, myself and many others have expressed is a desire but a realism that all things simply aren't possible.
And you still haven't addressed how components owned and made by other companies could get remade by Apple without infringing on their IP. Cook addressed it alright.
Clearly you need to read my original post. Both Cook and I want jobs back in the US. Period. You are imagining a lot of stuff.
The fact that Cook mentioned that a lot of parts for Apple products such as the iPhone and iPad are made in America makes me feel good. As someone who is now trying to slowly buy all or mostly American, I am just getting sick of things made in China not built to last.
The difference being that Cook understands it's impossible, while you don't seem to get it.
Some things are made here. Samsung was building in Texas. Micron has some facilities within the US. The only computer assembly I know of in terms of desktop computing would be Boxx in Texas. Lower volume/higher margin businesses like that can pull off such a thing, and it allows for a certain measure of control. Much like Apple they don't compete predominantly on price.
Edit: That doesn't mean it's practical for a company like Apple. I was only saying some manufacturing does occur here. I think Intel was making ssds here too a couple years ago.
Quote:
Originally Posted by wizard69
While those are all good questions you seem to mis the point here, if Apple is about to market a cluster solution it will be dead simple and defined for a specific group of uses. It won't be a solution where huge racks of Mac Pros are tied together. At best it will be a desktop "clustering" solution. At best it would tie a few machines together, we're few means less than seven. The reality is there are far better solutions in heavy deployment for larger clusters.
Infiniband is that solution. As can be seen here http://www.computerworld.com/s/article/9225703/Intel_superchip_will_be_aimed_at_high_performance_computing Intel is on the road to integrating the tech into its processors directly. It would be most interesting if Apples long delay is not for Sandy Bridge E but rather for this mystery chip with InfiniBand integrated on board. Such a chip would make for a very interesting approach for the next Mac Pro, a machine where clustering isn't an after thought at all. By the way Apple would still need TB on such a machine.
All of the above sounds good, but I'm not so certain that it is a chip that will be ready this year. Hard to tell though, you wouldn't think Intel would announce anything at all unless e hardware was close.
The new quoting system is pretty bad, but yeah... desktop clustering already works, and Infiniband looks impressive. The post I was responding to there involved using thunderbolt to create a 128 mini supercomputer via thunderbolt. You couldn't do that with daisy chaining. Even if it was possible, you'd lack reasonable failover measures there and run into many other issues with stability, latency, and traffic shaping given the pipeline like nature of daisy chained connections. Traffic shaping isn't really the correct term there as we're not really talking about multiple IO connections from the host computer. I just couldn't think of a better description for those IO management problems, so I slightly bastardized the term. You'd be running on a serialized connection with a lot of potential bottlenecks on upstream requests and multiple points of latency between the latter part of the chain and the host computer. I don't know that thunderbolt is designed to be robust in such a situation, and I wouldn't personally trust it without real testing. Anyway I do think thunderbolt has a reasonably good future if it sees wider adoption rates and efficient development. The performance to cost ratio doesn't seem bad at all for higher bandwidth devices, but Intel was slow on releasing any kind of SDK last year. Hopefully that will improve, or maybe it has already been addressed.
Some things are made here. Samsung was building in Texas. Micron has some facilities within the US.
There are many manufacturing plants still in the US. Of course not all of these are consumer electronics suppliers. Ownership of the plants is often not obvious. For example I just found out today that Westinghouse is owned by Toshiba. At least the nuclear plant division is.
The only computer assembly I know of in terms of desktop computing would be Boxx in Texas. Lower volume/higher margin businesses like that can pull off such a thing, and it allows for a certain measure of control. Much like Apple they don't compete predominantly on price.
I don't know about your area but there are a number of regional assemblers of Computers. They just don't have the profile of Dell, HP or Apple.
Edit: That doesn't mean it's practical for a company like Apple. I was only saying some manufacturing does occur here. I think Intel was making ssds here too a couple years ago.
The new quoting system is pretty bad, but yeah... desktop clustering already works, and Infiniband looks impressive. The post I was responding to there involved using thunderbolt to create a 128 mini supercomputer via thunderbolt. You couldn't do that with daisy chaining.
I don't ever expect to see TB used for such large clusters. I just don't think that is where Apple is headed with the port, though I could be wrong. I could see them using TB for small clusters of machines, say two to four boxes.
Even if it was possible, you'd lack reasonable failover measures there and run into many other issues with stability, latency, and traffic shaping given the pipeline like nature of daisy chained connections. Traffic shaping isn't really the correct term there as we're not really talking about multiple IO connections from the host computer. I just couldn't think of a better description for those IO management problems, so I slightly bastardized the term.
The problem I see here is that you are thinking big systems, I don't see Apple ever playing in that field. Rather I see them supporting small clusters of machines to support specific uses. All of that fancy management stuff wouldn't be needed.
You'd be running on a serialized connection with a lot of potential bottlenecks on upstream requests and multiple points of latency between the latter part of the chain and the host computer. I don't know that thunderbolt is designed to be robust in such a situation, and I wouldn't personally trust it without real testing. Anyway I do think thunderbolt has a reasonably good future if it sees wider adoption rates and efficient development. The performance to cost ratio doesn't seem bad at all for higher bandwidth devices, but Intel was slow on releasing any kind of SDK last year. Hopefully that will improve, or maybe it has already been addressed.
TB will be successful when Intel/Apple license it too the general market. When you can pick up an ARM or pic processor or a gate array with TB support built in then TB usage will explode. At this point I see a very very slow acceptance rate for the port. From Apples stand point they probably don't care, they now have a really neat way to dock their laptops which probably was why TB was developed in the first place.
TB will be successful when Intel/Apple license it too the general market. When you can pick up an ARM or pic processor or a gate array with TB support built in then TB usage will explode. At this point I see a very very slow acceptance rate for the port. From Apples stand point they probably don't care, they now have a really neat way to dock their laptops which probably was why TB was developed in the first place.
The forum is still chopping quotes for me. Anyway I only mentioned one assembler, because I don't many of them. It doesn't surprise me much on a smaller scale, and it allows for good quality control when testing and assembly are done in house. I expected to see people people who have been turning increasingly toward laptops to eventually embrace some kind of docking solution for a larger display. The single cable docking solution makes this far less irritating, which probably has a lot to do with its success. I would like to see thunderbolt take off. It provides excellent options for some things. I was only stuck on the large system thing because the topic came up previously.
"China is so vast in natural resources, labor and land..."
You don't know anything about China do you? The only thing China has lots of is labor. Land? Ever hear of the Gobi desert, the largest in Asia? Only 15 % of land is useful for agriculture (compared to 21% in the U.S., which has a much smaller population). Most of that is on the central eastern coast and around the Yangtze and Yellow River valleys, which, thanks to industry, are rapidly becoming toxic. Natural resources? See the previous.
TB isn't fast enough and instead of requiring active copper cables should have been implemented as OPTICAL as originally intended. When it is at 16x PCIe equivalent and optical then it will be useful. As it stands now, having seen how mediocrely it performs when more than one high bandwidth needing device attached (i.e. high speed raid and deep bit 2k monitoring), meh!
TB isn't fast enough and instead of requiring active copper cables should have been implemented as OPTICAL as originally intended. When it is at 16x PCIe equivalent and optical then it will be useful. As it stands now, having seen how mediocrely it performs when more than one high bandwidth needing device attached (i.e. high speed raid and deep bit 2k monitoring), meh!
Optical has a problem of the last inch (or centimeters, if you will). Optical is still an option, the plan is that the transceiver is in the cable, not the device, not unlike how a GBIC connects into a network switch. Optical has to be converted somewhere, Intel (and maybe Apple) chose to do the conversion in the cable and not in the device, and this gives a fair amount of flexibility. Then there is the matter of cost.
The only consumer optical I'm aware of is for digital audio, and that is a one-way link. Thunderbolt is about half the size of a fiber ethernet connector. There probably are other issues that I hadn't heard of.
Comments
While those are all good questions you seem to mis the point here, if Apple is about to market a cluster solution it will be dead simple and defined for a specific group of uses. It won't be a solution where huge racks of Mac Pros are tied together. At best it will be a desktop "clustering" solution. At best it would tie a few machines together, we're few means less than seven. The reality is there are far better solutions in heavy deployment for larger clusters.
Infiniband is that solution. As can be seen here http://www.computerworld.com/s/article/9225703/Intel_superchip_will_be_aimed_at_high_performance_computing Intel is on the road to integrating the tech into its processors directly. It would be most interesting if Apples long delay is not for Sandy Bridge E but rather for this mystery chip with InfiniBand integrated on board. Such a chip would make for a very interesting approach for the next Mac Pro, a machine where clustering isn't an after thought at all. By the way Apple would still need TB on such a machine.
All of the above sounds good, but I'm not so certain that it is a chip that will be ready this year. Hard to tell though, you wouldn't think Intel would announce anything at all unless e hardware was close.
Guess I'm not the only one who things US jobs are a good thing to have...
How myopic was that?
So now you're argument is that no one else but you and Cook want jobs to come to the US? Did you actually read the article or the transcript of his interview. He very clearly stated. ""We will do as many of these things [in America] as we can do," What you propose is myopic and foolish, what Cook, myself and many others have expressed is a desire but a realism that all things simply aren't possible.
And you still haven't addressed how components owned and made by other companies could get remade by Apple without infringing on their IP. Cook addressed it alright.
Clearly you need to read my original post. Both Cook and I want jobs back in the US. Period. You are imagining a lot of stuff.
Quote:
Originally Posted by realwarder
Clearly you need to read my original post. Both Cook and I want jobs back in the US. Period.
The difference being that Cook understands it's impossible, while you don't seem to get it.
The fact that Cook mentioned that a lot of parts for Apple products such as the iPhone and iPad are made in America makes me feel good. As someone who is now trying to slowly buy all or mostly American, I am just getting sick of things made in China not built to last.
Quote:
Originally Posted by Tallest Skil
The difference being that Cook understands it's impossible, while you don't seem to get it.
Some things are made here. Samsung was building in Texas. Micron has some facilities within the US. The only computer assembly I know of in terms of desktop computing would be Boxx in Texas. Lower volume/higher margin businesses like that can pull off such a thing, and it allows for a certain measure of control. Much like Apple they don't compete predominantly on price.
Edit: That doesn't mean it's practical for a company like Apple. I was only saying some manufacturing does occur here. I think Intel was making ssds here too a couple years ago.
Quote:
Originally Posted by wizard69
While those are all good questions you seem to mis the point here, if Apple is about to market a cluster solution it will be dead simple and defined for a specific group of uses. It won't be a solution where huge racks of Mac Pros are tied together. At best it will be a desktop "clustering" solution. At best it would tie a few machines together, we're few means less than seven. The reality is there are far better solutions in heavy deployment for larger clusters.
Infiniband is that solution. As can be seen here http://www.computerworld.com/s/article/9225703/Intel_superchip_will_be_aimed_at_high_performance_computing Intel is on the road to integrating the tech into its processors directly. It would be most interesting if Apples long delay is not for Sandy Bridge E but rather for this mystery chip with InfiniBand integrated on board. Such a chip would make for a very interesting approach for the next Mac Pro, a machine where clustering isn't an after thought at all. By the way Apple would still need TB on such a machine.
All of the above sounds good, but I'm not so certain that it is a chip that will be ready this year. Hard to tell though, you wouldn't think Intel would announce anything at all unless e hardware was close.
The new quoting system is pretty bad, but yeah... desktop clustering already works, and Infiniband looks impressive. The post I was responding to there involved using thunderbolt to create a 128 mini supercomputer via thunderbolt. You couldn't do that with daisy chaining. Even if it was possible, you'd lack reasonable failover measures there and run into many other issues with stability, latency, and traffic shaping given the pipeline like nature of daisy chained connections. Traffic shaping isn't really the correct term there as we're not really talking about multiple IO connections from the host computer. I just couldn't think of a better description for those IO management problems, so I slightly bastardized the term. You'd be running on a serialized connection with a lot of potential bottlenecks on upstream requests and multiple points of latency between the latter part of the chain and the host computer. I don't know that thunderbolt is designed to be robust in such a situation, and I wouldn't personally trust it without real testing. Anyway I do think thunderbolt has a reasonably good future if it sees wider adoption rates and efficient development. The performance to cost ratio doesn't seem bad at all for higher bandwidth devices, but Intel was slow on releasing any kind of SDK last year. Hopefully that will improve, or maybe it has already been addressed.
What is the basis for that argument?
Quote:
Originally Posted by realwarder
What is the basis for that argument?
Your stated belief that it's possible… Some jobs can be done in the US, but they can't do anywhere near the entirety of it here.
TB will be successful when Intel/Apple license it too the general market. When you can pick up an ARM or pic processor or a gate array with TB support built in then TB usage will explode. At this point I see a very very slow acceptance rate for the port. From Apples stand point they probably don't care, they now have a really neat way to dock their laptops which probably was why TB was developed in the first place.
Quote:
Originally Posted by wizard69
TB will be successful when Intel/Apple license it too the general market. When you can pick up an ARM or pic processor or a gate array with TB support built in then TB usage will explode. At this point I see a very very slow acceptance rate for the port. From Apples stand point they probably don't care, they now have a really neat way to dock their laptops which probably was why TB was developed in the first place.
The forum is still chopping quotes for me. Anyway I only mentioned one assembler, because I don't many of them. It doesn't surprise me much on a smaller scale, and it allows for good quality control when testing and assembly are done in house. I expected to see people people who have been turning increasingly toward laptops to eventually embrace some kind of docking solution for a larger display. The single cable docking solution makes this far less irritating, which probably has a lot to do with its success. I would like to see thunderbolt take off. It provides excellent options for some things. I was only stuck on the large system thing because the topic came up previously.
"China is so vast in natural resources, labor and land..."
You don't know anything about China do you? The only thing China has lots of is labor. Land? Ever hear of the Gobi desert, the largest in Asia? Only 15 % of land is useful for agriculture (compared to 21% in the U.S., which has a much smaller population). Most of that is on the central eastern coast and around the Yangtze and Yellow River valleys, which, thanks to industry, are rapidly becoming toxic. Natural resources? See the previous.
TB isn't fast enough and instead of requiring active copper cables should have been implemented as OPTICAL as originally intended. When it is at 16x PCIe equivalent and optical then it will be useful. As it stands now, having seen how mediocrely it performs when more than one high bandwidth needing device attached (i.e. high speed raid and deep bit 2k monitoring), meh!
Macs running Windows?
Optical has a problem of the last inch (or centimeters, if you will). Optical is still an option, the plan is that the transceiver is in the cable, not the device, not unlike how a GBIC connects into a network switch. Optical has to be converted somewhere, Intel (and maybe Apple) chose to do the conversion in the cable and not in the device, and this gives a fair amount of flexibility. Then there is the matter of cost.
The only consumer optical I'm aware of is for digital audio, and that is a one-way link. Thunderbolt is about half the size of a fiber ethernet connector. There probably are other issues that I hadn't heard of.