i've used cinema 4d for several years ... a great piece of 3d software. i think it comes from a unix background, so when they implemented distributed platform-agnostic rendering a couple of years ago I was impressed but not overly surprised. I tried this once under OS 9 with two other macs. Even under OS9 it was all configured and ran over TCP/IP. C4D is not particularly well-known to many Mac users, but I always loved the idea of buying a copy of NET, sticking the free client on any macs and pcs in the studio, and having your own pet render farm
I have not seen a similar approach to any compositing/editing/post-production software for the mac, and i wouldn't be surprised to see FCP, AE, combustion .... or shake adopt this.
Under X this thing should fly with a render server and client ... any chance we will see apple offer an FCP server component or similar? would be sweet with cinema tools or other hi-end, high definition requirements. Hmm, HDTV distributed rendering anybody?
As for the original rumour posting, i'm not too sure about its validity. I too think it's fishy anyone should be able to to reel off 4 projects like that without fear of getting gangbanged by apple's legal team within hours. however, the idea of clustering or network rendering has legs, and i hope to see more of it soon
<strong>As for the original rumour posting, i'm not too sure about its validity. I too think it's fishy anyone should be able to to reel off 4 projects like that without fear of getting gangbanged by apple's legal team within hours. however, the idea of clustering or network rendering has legs, and i hope to see more of it soon </strong><hr></blockquote>
At least somebody agrees with me. This IS fake. And based upon his fake application of the software system wide tasks would go toward the 'mega computer processor orgy'. I'm thankful for the Admin's post which says this is can work, but NOT for desktops. People have to seperate the difference between personal and professional use. And NO, and ad agency with 20 employees is NOT professional. Using a 'render farm' collection of Xserves is surely probably using the software described. But can you honestly tell me on a campus with about 400 or so macs and four times as many PCs that all these macs, (once upgraded) would function as 'one' computer? I don't buy this. My reasons were clear above and no followup as contradicted me. The original post which said...
"-ALL applications will have access to ?Wolf?
- The network is similar to Peer to Peer
- Metro cities will see higher responses than rural areas based upon number of Macs in regions.
"
HOW can ALL applications do this? How is this at all similar to peer to peer? So wait, now your telling me regional setting effect speed? Can you reexplain the optic backbone to this poor guy? How the *uck can you say that?
All I'm doing is crying him out for the phoney he is. Mis-informed. This isn't witch hunt! Can nobody back me up and say flat out that "Wolf" rumors about making every mac connect to run faster is plain BS. For an elite few who run rendering server setups using Xserves this is possible/practicle but NOT for anyone else.
First of all, I don't necessarily believe a word of the original post.
That said, however, a bunch of the grounds for rejection seem (to me) to be a little off-base. Internet-wide distributed computing has been happening for a couple of years now with the SETI@Home and RC5 cracking projects. Each of these teams had to hand-code their software to do all of that work. All the distributed rendering software out there had to code-their-own too. It would be a really good thing if Apple developed a system-level API that provided a well designed and well implemented mechanism that applications could use to much more easily achieve distributed computing.
[quote]ALL applications will have access to "Wolf"<hr></blockquote> would simply mean that its an API available to an MacOSX app. It does not imply that every application can suddenly use it without being modified. Nor does it mean that it is appropriate for every application.
[quote]The network is similar to Peer to Peer<hr></blockquote> simply means the machine doesn't have to be a dedicated compute server.
There is plenty of need for this kind of technology -- you don't have applications that need this kind of power because you don't have this kind of power. Its like the chicken and the egg... which comes first? Well this sort of technology would enable more distributed compute applications to be built, and built far more easily than is currently possible.
Yes there are tons of issues as listed above (lost work, security, performance impact on shared compute engines, etc etc etc). That doesn't change the fact that it is a very desirable technology and Apple should be working on ways to bring it to the Mac before the competition has something better.
So believe this guy or not, that's up to you... but there is a place in the world for an Apple-provided technology of this nature.
<strong>People have to seperate the difference between personal and professional use. And NO, and ad agency with 20 employees is NOT professional.</strong><hr></blockquote>
No, people need to separate "professional" from "computationally intensive." After all, a professional writer doesn't even need the bottom-of-the-line PowerMac. A technology like Wolf could easily find itself in K-12 education, if the school (or school district) was big enough and using multimedia heavily. Servers and server apps such as databases (Cumulus, anyone?) can benefit from clustering as well.
Distributed computing is also very cool, but it's aimed at a complimentary problem set. You'd want your 3D app to be able to use one or more machines other than your workstation to handle renders, but you'd also want one or more of those "machines" to actually be a cluster. Obviously, if Apple made it easy to do both, that would be the best of all possible worlds.
Elvis Presley's infamy was secured by the way he moved his hips.
[disclaimer: I am not saying anything either way about the validity of these rumors, just speculating]
[quote] 3) Clustering, farming, parallelism. In the ACG, and in close connection with other Apple groups, research is underway on the important ? and formidable ? problems of clustering/farming under Mac OS X, with a focus on those scenarios that beg for exploitation of G4 vectorization. Although clustering is in principle orthogonal to vectorization, one should seek a symbiosis of the two. So, while we have no specific product plans, we are looking vigorously at the general clustering and compute farm problem for Mac OS X users. The ACG is working with various outside agencies, along with the carrying out of its own internal research program. Which brings us to...
4) The legend of ?Zilla?. The public websites and list servers are replete with legends and rumors of the old ?Zilla,? which was a premier clustering application at NeXT, Inc. a decade ago and is now owned by Apple as both a concept and as legacy code. The public dialogue is welcome, for Zilla remains a source of pride, as she was indeed one of the very first screen saver-type distributed computation systems, winning the ComputerWorld-Smithsonian National Science Prize in 1991 as a ?community supercomputer.?
Just to clear up the various legends: In the late 1980s Zilla was created on the idea that machines should do what they were originally intended to do which is to work how, when, and where people are not and cannot. Zilla was based, if you will, on the politic of noninterference. (Indeed, the very name Zilla is reminiscent of a certain monster of legend, whom many of us believe, had a noninterventive soul ? until provoked! In fact this was the original motive for her very name.) The concept of taming a Zilla task as a background Unix process turned out to be way too harsh on the swapping mechanism; hence the adoption of a screen-saver motif which is, by now, a commonplace notion.
Zilla was not used to find record-setting prime numbers, as is often supposed; instead, it was used to develop, through factoring and other number-theoretical calculations, certain cryptographic systems, tests, and algorithms such as Fast Elliptic Encryption (FEE), described below. Zilla was also used at one point to render thousands of color-graphical frames from a sophisticated animation project. It is noteworthy that Zilla was not precisely equivalent to some modern-day models such as that of SETI@home, whereby volunteer users ship back processed data. Instead, while Zilla did involve volunteering of machines, the application Zilla would handle all the volunteers in a moderately coupled fashion.
Part of the ACG?s ongoing research is an in-depth analysis of the best of the old Zilla, in regard to Mac OS X, or more precisely, how the best of Zilla can be fused with the best of what is now out there ? commercially and experimentally ? in the clustering field.
?Wolf? has been a project for over 3 years and will be shipped later this year. The summer of 99, Apple acquired elite kernel and assembler programmers from some friends in Mountain View. This team is less than 6 and has been working on the ?Wolf? project and my understanding is they are flipping brilliant.
<hr></blockquote>
Heh, MountainView is an awesome geekatorium. They come to the Princeton Engineering job fair every year with some really cool handouts. (Well, cool if you're an EE). They have done microcode work with PowerPC chips and have lots of contracts with black-project military stuff. (Like development work on the B-2's comptuer systems way back when). These guys are the best of the best.
At least I appreciate the clever project names that AllenMcJones is using. His project "Wolf" seems to be a nod to the term "Beowulf Cluster", a clustering setup.
I get the feeling that some of our readers aren't aware that there are existing Mac apps that already use network rendering, like Electric Image for instance. EI is a rendering app that can mete out individual frame renders to other slave units on the local network. Each machine chews on the calculations for one frame then hands the finished rendered frame back to the master render engine which assembles them into a final animation.
Kind of like the SETI or FOLDING @ HOME stuff, but all in-shop.
I have no clue if Mr. AllenMCJones is for real or not, but I have to say I like his "rumors". Let's do a quick re-cap, shall we?
Glove - Mac OS X unlimited license agreement
Lucida - underwater digital Firewire camera
Wolf - clustering software at the kernel level
Presley - two "button" mouse
First, I like the code names. It reminds me a little of the urban myth how the Carl Sagan (can't remember which computer that was-- the 8500?) became the BHA (butt-head astronomer?) after Sagan sued Apple.
Second, the concepts seem plausible and even desirable. But they're plausible with a twist. They're not the: "iMac will get an LCD screen"-type rumor. Those are too obvious. This has at least some originality to them.
We'll probably have to wait a few months to see if these are real or not, but at least these are more interesting than the: "is there a G5 or not" threads.
(I'm holding off for project Autobahn, where the Mac finally gets the enough motor to sit in the left lane, forcing all others out of its way as it barrels down the computing roadway.)
Whether or not there is any truth to this rumor, clustering software would be, IMHO, a Good Idea?.
Why? Well currently the Power Mac is pretty crap as a workstation. As a desktop computer it's debatable, but it is not intended as a graphics workstation.
Clustering software would enable the Xserve to be a Mac solution for movie making (not that any of those hollywood bastards deserve Macs), and other high-intensity tasks.
I just had an idea! Apple writes an Open Firmware TCP/IP extension to use the components of other computer over a network as the components of a master computer! How cool would that be! Target CPU mode!
The arguments about "Who'd ever use that kind of pseudo clustering? REAL PROfessionals just buy REAL clusters if they need them" are totally invalid.
One, anyone with a smidgeon of contact with the academic world knows that now, more than ever, the budgetary restraints mean cuts in both research and office hardware. A solution that enables cross-over utility will push the doors to academic purchases wide open (especially as an Apple Only? technology).
Two, the arguments are akin to the ones put up when desktop publishing was first introduced. This will open up for a LOT of new areas - and be used by a lot of non-ivory tower professionals.
As for the argument about endangering Apple hardware sales, I believe the system-level "clustering" technologies discussed here will demand purchase of next-generation Apples (disabled from use on present G4s for financial reasons mainly). Furthermore, the appeal of such a feature will increase the market (not enormously; we ARE talking about niche customers, after all) and pay for itself.
Now, the inherent problems in this "leak" has already undergone a sufficiently competent rumortopsy above, so I will just point to the problems of bandwidth, data integrity, and security...
Still - communitizing processing cycles sure sounds hip, Stevie, my man!
PS - Does anyone have any comments on the goulash of rumors presented by several French Mac-sites and quoted by MacRumors (among others) ?
Its relevance for this thread lies in the following piece of carrot:
- the core following of jaguar will be clusterisable on the system level... one then supposes a very evolutionary machine out of turn of xServe. (On a standard 42u rack, 39 Xserve dual 1Ghz machines and an xserve Raid with 1,68 To, using only one screen)
Presumably this was put forward by loose-lipped (now deceased) French Apple employees at a series of Apple Road Shows (or their French equivalents)...
then say you rip lots of qt movies. wolf would let you steal cycles from the accounting and secretary machines and decrease the rip time big time.</strong><hr></blockquote>
So? Just about everyone doesn't care about that level of performance. If you have a great need to rip movies at lightening speed as part of your job then you just buy your own cluster. Get it? 99.9% of Mac users don't need that level of power and the other 0.1% just buy what they need.
<strong>The arguments about "Who'd ever use that kind of pseudo clustering? REAL PROfessionals just buy REAL clusters if they need them" are totally invalid.
One, anyone with a smidgeon of contact with the academic world knows that now, more than ever, the budgetary restraints mean cuts in both research and office hardware. A solution that enables cross-over utility will push the doors to academic purchases wide open (especially as an Apple Only? technology).
Two, the arguments are akin to the ones put up when desktop publishing was first introduced. This will open up for a LOT of new areas - and be used by a lot of non-ivory tower professionals.
As for the argument about endangering Apple hardware sales, I believe the system-level "clustering" technologies discussed here will demand purchase of next-generation Apples (disabled from use on present G4s for financial reasons mainly). Furthermore, the appeal of such a feature will increase the market (not enormously; we ARE talking about niche customers, after all) and pay for itself.
Now, the inherent problems in this "leak" has already undergone a sufficiently competent rumortopsy above, so I will just point to the problems of bandwidth, data integrity, and security...
Still - communitizing processing cycles sure sounds hip, Stevie, my man!</strong><hr></blockquote>
You're wrong. Anyone in science that needs a fast cluster doesn't buy macs. Overpriced and underperforming. Reality sucks huh?
Trust me. I work in academia and people just buy what they need. There's no great cry in the night for a super software solution to steel cycles from the secretaries? over powered G4.
The 7100 was originally code named "Carl Sagan". Sagan heard about it, and brought it to court for Apple to change it. Sagan won, Apple changed the code name to BHA (butt-head astronomer). Sagan heard about this, had his lawyers call Apple. Apple changed the code name to LAW (Lawyers are wimps).
No-one has pointed out that when someone buys a cluser it usually isn't Apple branded. This means that the cluster market currently has 0 revenue for Apple. A product like wolf can only increase this market for Apple.
Also, look at companies like Genentech . They have said they are buying racks of XServes to process gene sequencing, which is much faster on a G4 due to altivec. Right now, they have to program the cluster support into the app, and administrate the connection of new computers, etc. Wolf is the solution to this. They decide that they need some more speed. Plug in another rack.
Comments
Check <a href="http://www.maxon.de/pages/products/c4d/net/c4d_net_e.html"; target="_blank">this</a> out for a product overview.
I have not seen a similar approach to any compositing/editing/post-production software for the mac, and i wouldn't be surprised to see FCP, AE, combustion .... or shake adopt this.
Under X this thing should fly with a render server and client ... any chance we will see apple offer an FCP server component or similar? would be sweet with cinema tools or other hi-end, high definition requirements. Hmm, HDTV distributed rendering anybody?
As for the original rumour posting, i'm not too sure about its validity. I too think it's fishy anyone should be able to to reel off 4 projects like that without fear of getting gangbanged by apple's legal team within hours. however, the idea of clustering or network rendering has legs, and i hope to see more of it soon
<strong>As for the original rumour posting, i'm not too sure about its validity. I too think it's fishy anyone should be able to to reel off 4 projects like that without fear of getting gangbanged by apple's legal team within hours. however, the idea of clustering or network rendering has legs, and i hope to see more of it soon
At least somebody agrees with me. This IS fake. And based upon his fake application of the software system wide tasks would go toward the 'mega computer processor orgy'. I'm thankful for the Admin's post which says this is can work, but NOT for desktops. People have to seperate the difference between personal and professional use. And NO, and ad agency with 20 employees is NOT professional. Using a 'render farm' collection of Xserves is surely probably using the software described. But can you honestly tell me on a campus with about 400 or so macs and four times as many PCs that all these macs, (once upgraded) would function as 'one' computer? I don't buy this. My reasons were clear above and no followup as contradicted me. The original post which said...
"-ALL applications will have access to ?Wolf?
- The network is similar to Peer to Peer
- Metro cities will see higher responses than rural areas based upon number of Macs in regions.
"
HOW can ALL applications do this? How is this at all similar to peer to peer? So wait, now your telling me regional setting effect speed? Can you reexplain the optic backbone to this poor guy? How the *uck can you say that?
All I'm doing is crying him out for the phoney he is. Mis-informed. This isn't witch hunt! Can nobody back me up and say flat out that "Wolf" rumors about making every mac connect to run faster is plain BS. For an elite few who run rendering server setups using Xserves this is possible/practicle but NOT for anyone else.
That said, however, a bunch of the grounds for rejection seem (to me) to be a little off-base. Internet-wide distributed computing has been happening for a couple of years now with the SETI@Home and RC5 cracking projects. Each of these teams had to hand-code their software to do all of that work. All the distributed rendering software out there had to code-their-own too. It would be a really good thing if Apple developed a system-level API that provided a well designed and well implemented mechanism that applications could use to much more easily achieve distributed computing.
[quote]ALL applications will have access to "Wolf"<hr></blockquote> would simply mean that its an API available to an MacOSX app. It does not imply that every application can suddenly use it without being modified. Nor does it mean that it is appropriate for every application.
[quote]The network is similar to Peer to Peer<hr></blockquote> simply means the machine doesn't have to be a dedicated compute server.
There is plenty of need for this kind of technology -- you don't have applications that need this kind of power because you don't have this kind of power. Its like the chicken and the egg... which comes first? Well this sort of technology would enable more distributed compute applications to be built, and built far more easily than is currently possible.
Yes there are tons of issues as listed above (lost work, security, performance impact on shared compute engines, etc etc etc). That doesn't change the fact that it is a very desirable technology and Apple should be working on ways to bring it to the Mac before the competition has something better.
So believe this guy or not, that's up to you... but there is a place in the world for an Apple-provided technology of this nature.
The initial post made quite a bit of sense, given the background of Zilla.
Here is some more information:
<a href="http://www.apple.com/scitech/research/papers/acg/"; target="_blank">Zilla</a>
<strong>People have to seperate the difference between personal and professional use. And NO, and ad agency with 20 employees is NOT professional.</strong><hr></blockquote>
No, people need to separate "professional" from "computationally intensive." After all, a professional writer doesn't even need the bottom-of-the-line PowerMac. A technology like Wolf could easily find itself in K-12 education, if the school (or school district) was big enough and using multimedia heavily. Servers and server apps such as databases (Cumulus, anyone?) can benefit from clustering as well.
Distributed computing is also very cool, but it's aimed at a complimentary problem set. You'd want your 3D app to be able to use one or more machines other than your workstation to handle renders, but you'd also want one or more of those "machines" to actually be a cluster. Obviously, if Apple made it easy to do both, that would be the best of all possible worlds.
Elvis Presley's infamy was secured by the way he moved his hips.
[disclaimer: I am not saying anything either way about the validity of these rumors, just speculating]
[ 06-22-2002: Message edited by: Amorph ]</p>
[quote] 3) Clustering, farming, parallelism. In the ACG, and in close connection with other Apple groups, research is underway on the important ? and formidable ? problems of clustering/farming under Mac OS X, with a focus on those scenarios that beg for exploitation of G4 vectorization. Although clustering is in principle orthogonal to vectorization, one should seek a symbiosis of the two. So, while we have no specific product plans, we are looking vigorously at the general clustering and compute farm problem for Mac OS X users. The ACG is working with various outside agencies, along with the carrying out of its own internal research program. Which brings us to...
4) The legend of ?Zilla?. The public websites and list servers are replete with legends and rumors of the old ?Zilla,? which was a premier clustering application at NeXT, Inc. a decade ago and is now owned by Apple as both a concept and as legacy code. The public dialogue is welcome, for Zilla remains a source of pride, as she was indeed one of the very first screen saver-type distributed computation systems, winning the ComputerWorld-Smithsonian National Science Prize in 1991 as a ?community supercomputer.?
Just to clear up the various legends: In the late 1980s Zilla was created on the idea that machines should do what they were originally intended to do which is to work how, when, and where people are not and cannot. Zilla was based, if you will, on the politic of noninterference. (Indeed, the very name Zilla is reminiscent of a certain monster of legend, whom many of us believe, had a noninterventive soul ? until provoked! In fact this was the original motive for her very name.) The concept of taming a Zilla task as a background Unix process turned out to be way too harsh on the swapping mechanism; hence the adoption of a screen-saver motif which is, by now, a commonplace notion.
Zilla was not used to find record-setting prime numbers, as is often supposed; instead, it was used to develop, through factoring and other number-theoretical calculations, certain cryptographic systems, tests, and algorithms such as Fast Elliptic Encryption (FEE), described below. Zilla was also used at one point to render thousands of color-graphical frames from a sophisticated animation project. It is noteworthy that Zilla was not precisely equivalent to some modern-day models such as that of SETI@home, whereby volunteer users ship back processed data. Instead, while Zilla did involve volunteering of machines, the application Zilla would handle all the volunteers in a moderately coupled fashion.
Part of the ACG?s ongoing research is an in-depth analysis of the best of the old Zilla, in regard to Mac OS X, or more precisely, how the best of Zilla can be fused with the best of what is now out there ? commercially and experimentally ? in the clustering field.
<hr></blockquote>
?Wolf? has been a project for over 3 years and will be shipped later this year. The summer of 99, Apple acquired elite kernel and assembler programmers from some friends in Mountain View. This team is less than 6 and has been working on the ?Wolf? project and my understanding is they are flipping brilliant.
<hr></blockquote>
Heh, MountainView is an awesome geekatorium. They come to the Princeton Engineering job fair every year with some really cool handouts. (Well, cool if you're an EE). They have done microcode work with PowerPC chips and have lots of contracts with black-project military stuff. (Like development work on the B-2's comptuer systems way back when). These guys are the best of the best.
I get the feeling that some of our readers aren't aware that there are existing Mac apps that already use network rendering, like Electric Image for instance. EI is a rendering app that can mete out individual frame renders to other slave units on the local network. Each machine chews on the calculations for one frame then hands the finished rendered frame back to the master render engine which assembles them into a final animation.
Kind of like the SETI or FOLDING @ HOME stuff, but all in-shop.
Drew
And sometimes not quite reliable...
That's why those in the know call it Rendertrauma...
Unless you got the secret Lewno Renderpig Enabler algorithms...!
;^p
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> Maya for Mac OS X <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
- Glove - Mac OS X unlimited license agreement
- Lucida - underwater digital Firewire camera
- Wolf - clustering software at the kernel level
- Presley - two "button" mouse
First, I like the code names. It reminds me a little of the urban myth how the Carl Sagan (can't remember which computer that was-- the 8500?) became the BHA (butt-head astronomer?) after Sagan sued Apple.Second, the concepts seem plausible and even desirable. But they're plausible with a twist. They're not the: "iMac will get an LCD screen"-type rumor. Those are too obvious. This has at least some originality to them.
We'll probably have to wait a few months to see if these are real or not, but at least these are more interesting than the: "is there a G5 or not" threads.
(I'm holding off for project Autobahn, where the Mac finally gets the enough motor to sit in the left lane, forcing all others out of its way as it barrels down the computing roadway.)
Why? Well currently the Power Mac is pretty crap as a workstation. As a desktop computer it's debatable, but it is not intended as a graphics workstation.
Clustering software would enable the Xserve to be a Mac solution for movie making (not that any of those hollywood bastards deserve Macs), and other high-intensity tasks.
Barto
[ 06-23-2002: Message edited by: Barto ]</p>
Another OMFG moment,
Barto
One, anyone with a smidgeon of contact with the academic world knows that now, more than ever, the budgetary restraints mean cuts in both research and office hardware. A solution that enables cross-over utility will push the doors to academic purchases wide open (especially as an Apple Only? technology).
Two, the arguments are akin to the ones put up when desktop publishing was first introduced. This will open up for a LOT of new areas - and be used by a lot of non-ivory tower professionals.
As for the argument about endangering Apple hardware sales, I believe the system-level "clustering" technologies discussed here will demand purchase of next-generation Apples (disabled from use on present G4s for financial reasons mainly). Furthermore, the appeal of such a feature will increase the market (not enormously; we ARE talking about niche customers, after all) and pay for itself.
Now, the inherent problems in this "leak" has already undergone a sufficiently competent rumortopsy above, so I will just point to the problems of bandwidth, data integrity, and security...
Still - communitizing processing cycles sure sounds hip, Stevie, my man!
Its relevance for this thread lies in the following piece of carrot:
- the core following of jaguar will be clusterisable on the system level... one then supposes a very evolutionary machine out of turn of xServe. (On a standard 42u rack, 39 Xserve dual 1Ghz machines and an xserve Raid with 1,68 To, using only one screen)
Presumably this was put forward by loose-lipped (now deceased) French Apple employees at a series of Apple Road Shows (or their French equivalents)...
<strong>here's another reason why you are wrong:
Let's say you have a office of 75 macs. (10.2)
then say you rip lots of qt movies. wolf would let you steal cycles from the accounting and secretary machines and decrease the rip time big time.</strong><hr></blockquote>
So? Just about everyone doesn't care about that level of performance. If you have a great need to rip movies at lightening speed as part of your job then you just buy your own cluster. Get it? 99.9% of Mac users don't need that level of power and the other 0.1% just buy what they need.
<strong>The arguments about "Who'd ever use that kind of pseudo clustering? REAL PROfessionals just buy REAL clusters if they need them" are totally invalid.
One, anyone with a smidgeon of contact with the academic world knows that now, more than ever, the budgetary restraints mean cuts in both research and office hardware. A solution that enables cross-over utility will push the doors to academic purchases wide open (especially as an Apple Only? technology).
Two, the arguments are akin to the ones put up when desktop publishing was first introduced. This will open up for a LOT of new areas - and be used by a lot of non-ivory tower professionals.
As for the argument about endangering Apple hardware sales, I believe the system-level "clustering" technologies discussed here will demand purchase of next-generation Apples (disabled from use on present G4s for financial reasons mainly). Furthermore, the appeal of such a feature will increase the market (not enormously; we ARE talking about niche customers, after all) and pay for itself.
Now, the inherent problems in this "leak" has already undergone a sufficiently competent rumortopsy above, so I will just point to the problems of bandwidth, data integrity, and security...
Still - communitizing processing cycles sure sounds hip, Stevie, my man!</strong><hr></blockquote>
You're wrong. Anyone in science that needs a fast cluster doesn't buy macs. Overpriced and underperforming. Reality sucks huh?
Trust me. I work in academia and people just buy what they need. There's no great cry in the night for a super software solution to steel cycles from the secretaries? over powered G4.
This is not urban myth. This is fact.
Also, look at companies like Genentech . They have said they are buying racks of XServes to process gene sequencing, which is much faster on a G4 due to altivec. Right now, they have to program the cluster support into the app, and administrate the connection of new computers, etc. Wolf is the solution to this. They decide that they need some more speed. Plug in another rack.