<strong>Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when.</strong><hr></blockquote>
As head of Pixar, Jobs surely has a precise vision of what it will take to someday say, "We aint gonna work on Intel's, Sun's, or anyone else's render farm no more."
</strong><hr></blockquote>
Pixar currently uses a large renderfarm of unix based machines. If Stevie is truely trying to push macs as renderfarm computers, wouldn't pixar be one of the earliest adopters?
<strong>Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when.</strong><hr></blockquote>
If I'm not mistaken, rack mountable, server and cluster capable hardware based on Apple PPC/OSX is already available. Take a look at <a href="http://www.terrasoftsolutions.com."; target="_blank">www.terrasoftsolutions.com.</a> The following specs were taken off the GVS 9000 2U rack system product info:
T H E D U A L G H z G V S 9 0 0 0
There is not a great deal we can say ... when a product speaks for itself.
Apple technology. A GVS 9000 2U Rack System. And it ships with both Yellow Dog Linux, Mac OS 9, & Mac OS X pre-installed ... ok, we'll say that one more time --YDL, Mac OS 9, & Mac OS X!
The ultimate workstation, server, or cluster node, the GVS 9000 offers the same technology as the latest dual 1 GHz Apple G4 --in half the space. Rack, stack, and compact your cluster. Add a second drive, more RAM, and a PCI expansion card, or RAID storeage solution.
Boot, format, install, and reboot a cluster node in less than 5 minutes with Black Lab automation for Yellow Dog Linux.
CPUs: dual 1 GHz PowerPC G4 w/AltiVec.
Memory: 1 GB PC-133 SDRAM - 2 DIMMs.
Storage: 80 GB IDE drive.
Removable: DVD-RW Super Drive.
Communications: Gigabit, FireWire, USB.
Video Card: ATI Radeon 7500 dual (supported with a kernel patch)
Expansion: single 64/66 PCI slot.
Chassis: GVS 9000 2U Rack System.
Keyboard & Mouse: Apple USB.
Operating System: YDL & Mac OS X pre-installed; complete box sets.
Warranty Service Plan: GVS One year - Hardware.
Weight: 50 lbs.
The GVS 9000 is pre-configured with Yellow Dog Linux 2.2, Mac OS 9, Mac-On-Linux (allows you to run Mac OS 9 on YDL), and Mac OS X.
Perhaps Apple is hoping to sell cluster nodes to the high end market which generally prefers rack mounted systems to save space. I'm not a techie but wouldn't a cluster node of DP G1's or higher be able to run high end apps like Shake/Tremor faster than competing systems. Apple did announce recently that Shake/Tremor would be supported on Irix/Linux/OSX but not Windows after ver 2.5. That fits nicely with the GVS 9000 which runs on Linux and OSX. Just a thought.
<strong>POWER4 based high-end machines in the $12K+ range.
That's my guess. flame away.
TING5</strong><hr></blockquote>
Sorry to dissappoint you, but I'm afraid I agree ...
I think Apple's got some sort of eye on some sort of late-in-life plan IBM has for bringing some sort of scaled down version of the POWER4 to the masses - if not directly ( it would probably still be too pricey) then by way of racks in their IT departments.
But I must admit, this is total shot-In-The-Dark territory.
[quote]"IF" Apple releases servers they MUST have the following options:
Redundant power supplies, Redundant cooling, hot swap HD's, the support for Intel NICs with port trunking, etc.
<hr></blockquote>
Apple doesn't have the balls to make such a product. I think they would sell quite well, but this product doesn't fit with Apple's "phat margin" philosophy. But I hope I'm wrong.
Generally it loads a movie from startup in about 1 bounce (mov: divx[720x480] mp3,vbr) on slower machines. If this is a system wide Quicktime improvement, things are going to go from sticky to very smooth real quick.
I'd say it's a whole lot beta' than anything OS9+Quicktime has ever brought us.
[quote]I'm to the point where I don't care what Apple does as long as they manage to get revenues back up to the 1.8-2.2 billion mark and get quarterly PowerMac sales back up to 1999 levels.<hr></blockquote>
Yup, especially considering that Apple has a very, very slim performance advantage between the high-end iMac and the low-end Quicksilver. This is ludicrous, but I refuse to believe Apple has serious hopes of maintaining this for long.
I share your pain in its entirety. However, by common-sensical reasoning one cannot expect Apple to return to such Pro sales levels without some drastic change of the status quo.
Nonsuch:
[quote] Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when. <hr></blockquote>
Yes, the discussion feels more engaging (at least to me) than in the other threads, although I must admit that they are a fun and entertaining read.
The rack-mountable macs...
I fully agree with your implict point that this approach to dealing with its new position as a player in the high-end non-linear hardware/software industry is a good one. But I almost take this for granted, as this is indeed the natural form-factor of hardware in this industry. It seems silly for Apple not to do this, as even SciTech customers (such as my school) complain about Apple's intransigence regarding this issue.
However, the "when" question's appeal fades if we are to understand this as given (Note: if Apple doesn't do this [make r-m macs], Rubenstein should be fired. It's that simple to me.) As you and Lemon Bon Bon insightfully suggest, there are several logical queue point for this hardware to surface but then this also depends on what Apple choses to enclose in a rack-mountable case. It wouldn't make much sense for the rack-mountable mac to have a significantly similar spec (feature-set, not benchmark) to a regular Pro machine. These r-m macs would be performing in very different, specialized environments. BTO is not flexible enough. Apple can cut costs by removing features which are not necessary in these environments. This has obvious advantages for Apple and the customer. Mike provided some great insight into this by going into detailed "wants" in a r-m server.
[quote] This having been said, "IF" Apple releases servers they MUST have the following options:
Redundant power supplies, Redundant cooling, hot swap HD's, the support for Intel NICs with port trunking, etc.<hr></blockquote>
It would also be silly for them not to do so because the customers would at the very least expect Apple to demonstrate some serious commitment to their specialized needs. From my experience in speaking with people employing clusters (whatever kind) serious capital deepining in these markets axiomatically follows a favorable assesment of investment viability / longevity. This is most often a highly subjective measurement but Apple can do some simple things in order for it to be well-received in such communities.
Moreover, Apple would ensure that it really is maximizing performance in these machines (both unitary, but more importantly clustered performance). Apple has significant leeway for this given the strong price inelasticity of this market.
So yes, I agree with your comments fully, but the points that I feel importantly missing from the r-m mac consideration are the aforementioned ones. We must understand that the performance these machines should exhibit (hopefully, that should will translate into do when the machines arrive) will not necessarily mean that it will be performance that us, the average end user, will enjoy (conceptually, it would be performance that we would never enjoy).
Lemon Bon Bon:
I also agree with you in that the processor situation is a mess.
The one intriguing facet of considering the existence (near market presence?) of a "G5" is that its development has been so well kept secret. One thing is for Apple to lay down the iron curtain on its product line, but making Moto do this so efficently quite surprising (even if you point out to me the 8450...)
Also, other Apple competitors must believe that Apple must be up to something if Apple wants to keep even a single customer in the long run. And the time elapsed since the G4 came into being really is the "long run" in the computer industry (remeber that the G4 is fundamentally a 604...) so, as we all probably feel, Apple's processor revamp is overdue.
So, I really don't care if we get a G5, or if Apple's "clever" PR decides to call it GX (read G Ten) to match "OS X", or if we get an IBM Power1000, or anything along those lines (AMD/Intel is a different story...) as long as we get some serious power.
It's important to not get fixated in the "Apple didn't deliver the G5 once again..." We don't need a G5 specifically, we need performance.
Matsu:
[quote] I don't think a proc card is a band-aid at all, just a recognition of how people tend to work in computationally intensive fields.
Whatever you give them, they'll put it in front of a render-farm. Right now the render farm comes from a slew of cheap boxes or rack-mounts. Little farms (1-6 machines) can add quite a boost. I know of people using one and two DP Athlon boxes as render farms at home! Nutters, die hards really, but you can imagine that businesses with a real interest would certainly cough up the cash.
<hr></blockquote>
Ok, some misunderstanding between us. Entirely my fault.
I fully comprehend the point you make regarding the "render farm card approach." In fact I have upgraded several cards in a Chem lab here at my school. I had understood you as advocating for Apple to address this new market (Film, but really any computing-intensive field, such as science in which Apple seems to be currently making significant inroads) by accepting the current unacceptable performance of the G4 (Altivec is an entirely seperate issue) and just addressing this problem by applying "best practices" mentality from cluster markets and selling such flexibility as a solution.
In such a case, as I had understood, introducing this card flexibility as a problem-solver would be a band aid.
However, I full agree with you in that in the long-run (hopefully short-run, accompanied by a major processor revision) Apple should pursue such flexibility.
Whis makes the current scenario of Apple purchasing the PowerPC rights for $500 million all the more interesting. Apple, in fact, should offer proc card upgradability if it wants to attract serious cluster customers. But it also would want to hedge its bet that it could remain the sole supplier of these cards. Imagine the losses Apple would suffer unless it remained cut-throught competitive in a proc card market. And even then so, profit margins would be really slim considering the effect-compounding fact that this design would also offer a dis-incentive to purchase new hardware in the short to middle-run.
However, even for $4 billion, $500mil is still a hefty amount. An Apple purchase would unequivocably signal a strong belief in the remaining potential of the PowerPC architecture, and would even require a certain degree of hubris as Apple would be implicitly saying that it thinks it can do a better job with PPC than Moto has (meaning a strong investment in added Chip Design employees that could create overhead...remeber the IBM do-it-all approach was easily shot down by Compaq back in the eighties)
But, it isn't hard to recognize that Moto has not delivered as expected (at least for Apple) with the PPC...
Eupfhoria:
[quote] Pixar currently uses a large renderfarm of unix based machines. If Stevie is truely trying to push macs as renderfarm computers, wouldn't pixar be one of the earliest adopters?<hr></blockquote>
Apple's not there yet in terms of performance. Not there by a long shot. Firms in the movie industry have firm deadlines they must meet. They will not mess around until the are assure that they are adopting proven technology. Apple has yer to prove anything in this regard (other than a big wallet).
I wouldn't believe anybody at Pixar would accept a move by Jobs to move Pixar to the Mac anytime soon.
And from media reports (and understandably) Jobs apparently isn't too heavily involved in Pixar operations as (they seem a highly autonomous firm, which is a good thing for both Apple and Pixar, as we do get more of Jobs' great work devoted to our own cause...)
AirSluf:
[quote] Matsu's idea is a fairly decent one and in no way a mere band-aid. As long as the PCI-farm cards had their own large memory you gain scale quite nicely within an individual box. Essentially trading PCI for Gig-E networking internally. Some master-slave work would need to be done ensuring the PCI bus stays optimized for HD traffic and results reporting/submission, making these pure computation engines will help reduce a lot of that bus traffic though. With how much in the way of main-MB extras you can leave off, it looks to be a high margin product area that could add significant performance at less cost than a full machine. And be faster than a fully external cluster <hr></blockquote>
I agree, see my post regarding Matsu's arguements.
Junkyard Dawg:
[quote] Apple doesn't have the balls to make such a product. I think they would sell quite well, but this product doesn't fit with Apple's "phat margin" philosophy. But I hope I'm wrong. <hr></blockquote>
Yup, if Apple pursues this its up for a really bitter reality check. For one, Apple will have to emerge from its high-profit bubble...
But, some of the points made throughout the discussion (with respect to the server and other enterprise markets) ignore that a promising business model is necessary for Apple to succed. Making the hardware / software good enough doesn't cut it. As an extreme example, look at JLG's BeOS. Great technology, absolute failure. We have to demand from Apple that it follow a sensible strategy in attempting to enlist enterprise customers as its clients.
As a generally comment, I wouldn't mind Apple pursuing a "OS X" hardware strategy of intoducing a impressively good new product and then improving it along the way. I don't expect Apple to fix all of the shortcommings of its hardware in one revision, but we should atleast be looking at new hardware that will blow our minds away (ala Mac OS X). From then on, it's not a catch-up game anymore (which is harder to play) but rather a keep-ahead-of-the-curve game.
Well with today's announcement of new neverbefore seen rackmount hardware at Apple, the reality of Apple trying to make inroads towards the corporate network enviroment is becoming realised. Since Steve has returned (I am not going to give him all the credit but at least part) the Apple hardware spectrum has begun a whole new revival from a modest company to cutting edge. It takes time to produce new products, especially if they are constantly working outside of the 'box' unlike many such as Dell and Compaq which use the same form factor much longer and don't go out on a limb to try new things. Since it seems that they are trying to woo the corporate market especially those dealing in digital content companies as it may seem, the need for performance to increase may hopefully bring a new set of hardware changes faster, especially to the upper echelon of products changing their business model slightly but to stay more competetive.
There has been much confusion lately to the CPU front is it going to be a G5, or maybe just an upgraded G4? It doesn't really matter what they do, but it better outperform the current line of CPU's and by a larger margin than ever before. I am sorry but if AMD/Intel boxes are beating the 'Ultimate' mac setup by a large margin at it's own bread and butter, digital content, something drastic has to happen. If Apple can present a new machine that can match the performance (reguardless of MHz) of the top Intel/AMD boxes or perhaps a little better, you ultimately have the advantage with OS X. With the release of Jaguar this summer, we should have an even faster OS and better CPU's as well by MWNY.
I think and sure hope that Apple has something real special up their sleeves that will just make Mac fans salivate this summer. We are getting the best of software with the Apple developed apps and new hardware just down the pipeline, there has never been a better time to be a Mac user in my opinion. I hope Apple brings the necessary and requested capabilites to the network enviroment with the new Rackmount Servers. To me that would show they really are competetive in more than just consumer and education and are cutting edge with the brawn to compliment those good looks.
Comments
<strong>Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when.</strong><hr></blockquote>
<strong>
As head of Pixar, Jobs surely has a precise vision of what it will take to someday say, "We aint gonna work on Intel's, Sun's, or anyone else's render farm no more."
</strong><hr></blockquote>
Pixar currently uses a large renderfarm of unix based machines. If Stevie is truely trying to push macs as renderfarm computers, wouldn't pixar be one of the earliest adopters?
<strong>Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when.</strong><hr></blockquote>
If I'm not mistaken, rack mountable, server and cluster capable hardware based on Apple PPC/OSX is already available. Take a look at <a href="http://www.terrasoftsolutions.com."; target="_blank">www.terrasoftsolutions.com.</a> The following specs were taken off the GVS 9000 2U rack system product info:
T H E D U A L G H z G V S 9 0 0 0
There is not a great deal we can say ... when a product speaks for itself.
Apple technology. A GVS 9000 2U Rack System. And it ships with both Yellow Dog Linux, Mac OS 9, & Mac OS X pre-installed ... ok, we'll say that one more time --YDL, Mac OS 9, & Mac OS X!
The ultimate workstation, server, or cluster node, the GVS 9000 offers the same technology as the latest dual 1 GHz Apple G4 --in half the space. Rack, stack, and compact your cluster. Add a second drive, more RAM, and a PCI expansion card, or RAID storeage solution.
Boot, format, install, and reboot a cluster node in less than 5 minutes with Black Lab automation for Yellow Dog Linux.
CPUs: dual 1 GHz PowerPC G4 w/AltiVec.
Memory: 1 GB PC-133 SDRAM - 2 DIMMs.
Storage: 80 GB IDE drive.
Removable: DVD-RW Super Drive.
Communications: Gigabit, FireWire, USB.
Video Card: ATI Radeon 7500 dual (supported with a kernel patch)
Expansion: single 64/66 PCI slot.
Chassis: GVS 9000 2U Rack System.
Keyboard & Mouse: Apple USB.
Operating System: YDL & Mac OS X pre-installed; complete box sets.
Warranty Service Plan: GVS One year - Hardware.
Weight: 50 lbs.
The GVS 9000 is pre-configured with Yellow Dog Linux 2.2, Mac OS 9, Mac-On-Linux (allows you to run Mac OS 9 on YDL), and Mac OS X.
Perhaps Apple is hoping to sell cluster nodes to the high end market which generally prefers rack mounted systems to save space. I'm not a techie but wouldn't a cluster node of DP G1's or higher be able to run high end apps like Shake/Tremor faster than competing systems. Apple did announce recently that Shake/Tremor would be supported on Irix/Linux/OSX but not Windows after ver 2.5. That fits nicely with the GVS 9000 which runs on Linux and OSX. Just a thought.
That's my guess. flame away.
TING5
<strong>POWER4 based high-end machines in the $12K+ range.
That's my guess. flame away.
TING5</strong><hr></blockquote>
Sorry to dissappoint you, but I'm afraid I agree ...
I think Apple's got some sort of eye on some sort of late-in-life plan IBM has for bringing some sort of scaled down version of the POWER4 to the masses - if not directly ( it would probably still be too pricey) then by way of racks in their IT departments.
But I must admit, this is total shot-In-The-Dark territory.
bam!
Redundant power supplies, Redundant cooling, hot swap HD's, the support for Intel NICs with port trunking, etc.
<hr></blockquote>
Apple doesn't have the balls to make such a product. I think they would sell quite well, but this product doesn't fit with Apple's "phat margin" philosophy. But I hope I'm wrong.
Quicktime 6 is VERY fast.
Generally it loads a movie from startup in about 1 bounce (mov: divx[720x480] mp3,vbr) on slower machines. If this is a system wide Quicktime improvement, things are going to go from sticky to very smooth real quick.
I'd say it's a whole lot beta' than anything OS9+Quicktime has ever brought us.
-walloo
[ 05-05-2002: Message edited by: willywalloo ]</p>
Sell What they can then to the mass markets.
Keep the consumer line.
LowLevel: iMac, iBooks
Mid-Level: G4/5/6 Towers
High-Lev: Editing Workstations running a unix flavor, sleep with SGi, while there at it. (did I say co-produce a product?)
-walloo
Nostradamus
[quote]I'm to the point where I don't care what Apple does as long as they manage to get revenues back up to the 1.8-2.2 billion mark and get quarterly PowerMac sales back up to 1999 levels.<hr></blockquote>
Yup, especially considering that Apple has a very, very slim performance advantage between the high-end iMac and the low-end Quicksilver. This is ludicrous, but I refuse to believe Apple has serious hopes of maintaining this for long.
I share your pain in its entirety. However, by common-sensical reasoning one cannot expect Apple to return to such Pro sales levels without some drastic change of the status quo.
Nonsuch:
[quote] Some great discussion going on in this thread. I don't have the technical knowledge to add much, but I notice no one has made mention of the recently revived rumors concerning rack-mountable Apple servers. Machead's post convincingly argues that Apple wouldn't be embarking on this new market without the necessary hardware in the pipeline, and rackmounted 7500s released to coincide with the first Apple-branded release of Shake, or maybe FCP 4, would send a resounding shot across the Wintel bow. That really would make for the paradigm-shift in performance JYD alluded to, and I don't see it as something very farfetched at all; the only question is when. <hr></blockquote>
Yes, the discussion feels more engaging (at least to me) than in the other threads, although I must admit that they are a fun and entertaining read.
The rack-mountable macs...
I fully agree with your implict point that this approach to dealing with its new position as a player in the high-end non-linear hardware/software industry is a good one. But I almost take this for granted, as this is indeed the natural form-factor of hardware in this industry. It seems silly for Apple not to do this, as even SciTech customers (such as my school) complain about Apple's intransigence regarding this issue.
However, the "when" question's appeal fades if we are to understand this as given (Note: if Apple doesn't do this [make r-m macs], Rubenstein should be fired. It's that simple to me.) As you and Lemon Bon Bon insightfully suggest, there are several logical queue point for this hardware to surface but then this also depends on what Apple choses to enclose in a rack-mountable case. It wouldn't make much sense for the rack-mountable mac to have a significantly similar spec (feature-set, not benchmark) to a regular Pro machine. These r-m macs would be performing in very different, specialized environments. BTO is not flexible enough. Apple can cut costs by removing features which are not necessary in these environments. This has obvious advantages for Apple and the customer. Mike provided some great insight into this by going into detailed "wants" in a r-m server.
[quote] This having been said, "IF" Apple releases servers they MUST have the following options:
Redundant power supplies, Redundant cooling, hot swap HD's, the support for Intel NICs with port trunking, etc.<hr></blockquote>
It would also be silly for them not to do so because the customers would at the very least expect Apple to demonstrate some serious commitment to their specialized needs. From my experience in speaking with people employing clusters (whatever kind) serious capital deepining in these markets axiomatically follows a favorable assesment of investment viability / longevity. This is most often a highly subjective measurement but Apple can do some simple things in order for it to be well-received in such communities.
Moreover, Apple would ensure that it really is maximizing performance in these machines (both unitary, but more importantly clustered performance). Apple has significant leeway for this given the strong price inelasticity of this market.
So yes, I agree with your comments fully, but the points that I feel importantly missing from the r-m mac consideration are the aforementioned ones. We must understand that the performance these machines should exhibit (hopefully, that should will translate into do when the machines arrive) will not necessarily mean that it will be performance that us, the average end user, will enjoy (conceptually, it would be performance that we would never enjoy).
Lemon Bon Bon:
I also agree with you in that the processor situation is a mess.
The one intriguing facet of considering the existence (near market presence?) of a "G5" is that its development has been so well kept secret. One thing is for Apple to lay down the iron curtain on its product line, but making Moto do this so efficently quite surprising (even if you point out to me the 8450...)
Also, other Apple competitors must believe that Apple must be up to something if Apple wants to keep even a single customer in the long run. And the time elapsed since the G4 came into being really is the "long run" in the computer industry (remeber that the G4 is fundamentally a 604...) so, as we all probably feel, Apple's processor revamp is overdue.
So, I really don't care if we get a G5, or if Apple's "clever" PR decides to call it GX (read G Ten) to match "OS X", or if we get an IBM Power1000, or anything along those lines (AMD/Intel is a different story...) as long as we get some serious power.
It's important to not get fixated in the "Apple didn't deliver the G5 once again..." We don't need a G5 specifically, we need performance.
Matsu:
[quote] I don't think a proc card is a band-aid at all, just a recognition of how people tend to work in computationally intensive fields.
Whatever you give them, they'll put it in front of a render-farm. Right now the render farm comes from a slew of cheap boxes or rack-mounts. Little farms (1-6 machines) can add quite a boost. I know of people using one and two DP Athlon boxes as render farms at home! Nutters, die hards really, but you can imagine that businesses with a real interest would certainly cough up the cash.
<hr></blockquote>
Ok, some misunderstanding between us. Entirely my fault.
I fully comprehend the point you make regarding the "render farm card approach." In fact I have upgraded several cards in a Chem lab here at my school. I had understood you as advocating for Apple to address this new market (Film, but really any computing-intensive field, such as science in which Apple seems to be currently making significant inroads) by accepting the current unacceptable performance of the G4 (Altivec is an entirely seperate issue) and just addressing this problem by applying "best practices" mentality from cluster markets and selling such flexibility as a solution.
In such a case, as I had understood, introducing this card flexibility as a problem-solver would be a band aid.
However, I full agree with you in that in the long-run (hopefully short-run, accompanied by a major processor revision) Apple should pursue such flexibility.
Whis makes the current scenario of Apple purchasing the PowerPC rights for $500 million all the more interesting. Apple, in fact, should offer proc card upgradability if it wants to attract serious cluster customers. But it also would want to hedge its bet that it could remain the sole supplier of these cards. Imagine the losses Apple would suffer unless it remained cut-throught competitive in a proc card market. And even then so, profit margins would be really slim considering the effect-compounding fact that this design would also offer a dis-incentive to purchase new hardware in the short to middle-run.
However, even for $4 billion, $500mil is still a hefty amount. An Apple purchase would unequivocably signal a strong belief in the remaining potential of the PowerPC architecture, and would even require a certain degree of hubris as Apple would be implicitly saying that it thinks it can do a better job with PPC than Moto has (meaning a strong investment in added Chip Design employees that could create overhead...remeber the IBM do-it-all approach was easily shot down by Compaq back in the eighties)
But, it isn't hard to recognize that Moto has not delivered as expected (at least for Apple) with the PPC...
Eupfhoria:
[quote] Pixar currently uses a large renderfarm of unix based machines. If Stevie is truely trying to push macs as renderfarm computers, wouldn't pixar be one of the earliest adopters?<hr></blockquote>
Apple's not there yet in terms of performance. Not there by a long shot. Firms in the movie industry have firm deadlines they must meet. They will not mess around until the are assure that they are adopting proven technology. Apple has yer to prove anything in this regard (other than a big wallet).
I wouldn't believe anybody at Pixar would accept a move by Jobs to move Pixar to the Mac anytime soon.
And from media reports (and understandably) Jobs apparently isn't too heavily involved in Pixar operations as (they seem a highly autonomous firm, which is a good thing for both Apple and Pixar, as we do get more of Jobs' great work devoted to our own cause...)
AirSluf:
[quote] Matsu's idea is a fairly decent one and in no way a mere band-aid. As long as the PCI-farm cards had their own large memory you gain scale quite nicely within an individual box. Essentially trading PCI for Gig-E networking internally. Some master-slave work would need to be done ensuring the PCI bus stays optimized for HD traffic and results reporting/submission, making these pure computation engines will help reduce a lot of that bus traffic though. With how much in the way of main-MB extras you can leave off, it looks to be a high margin product area that could add significant performance at less cost than a full machine. And be faster than a fully external cluster <hr></blockquote>
I agree, see my post regarding Matsu's arguements.
Junkyard Dawg:
[quote] Apple doesn't have the balls to make such a product. I think they would sell quite well, but this product doesn't fit with Apple's "phat margin" philosophy. But I hope I'm wrong. <hr></blockquote>
Yup, if Apple pursues this its up for a really bitter reality check. For one, Apple will have to emerge from its high-profit bubble...
But, some of the points made throughout the discussion (with respect to the server and other enterprise markets) ignore that a promising business model is necessary for Apple to succed. Making the hardware / software good enough doesn't cut it. As an extreme example, look at JLG's BeOS. Great technology, absolute failure. We have to demand from Apple that it follow a sensible strategy in attempting to enlist enterprise customers as its clients.
As a generally comment, I wouldn't mind Apple pursuing a "OS X" hardware strategy of intoducing a impressively good new product and then improving it along the way. I don't expect Apple to fix all of the shortcommings of its hardware in one revision, but we should atleast be looking at new hardware that will blow our minds away (ala Mac OS X). From then on, it's not a catch-up game anymore (which is harder to play) but rather a keep-ahead-of-the-curve game.
As always, let me know what you think.
There has been much confusion lately to the CPU front is it going to be a G5, or maybe just an upgraded G4? It doesn't really matter what they do, but it better outperform the current line of CPU's and by a larger margin than ever before. I am sorry but if AMD/Intel boxes are beating the 'Ultimate' mac setup by a large margin at it's own bread and butter, digital content, something drastic has to happen. If Apple can present a new machine that can match the performance (reguardless of MHz) of the top Intel/AMD boxes or perhaps a little better, you ultimately have the advantage with OS X. With the release of Jaguar this summer, we should have an even faster OS and better CPU's as well by MWNY.
I think and sure hope that Apple has something real special up their sleeves that will just make Mac fans salivate this summer. We are getting the best of software with the Apple developed apps and new hardware just down the pipeline, there has never been a better time to be a Mac user in my opinion. I hope Apple brings the necessary and requested capabilites to the network enviroment with the new Rackmount Servers. To me that would show they really are competetive in more than just consumer and education and are cutting edge with the brawn to compliment those good looks.
Just my 3 cents.