I agree. It's not like Apple's gonna say "Oh, wow, this new chip just landed. I guess we better hack it into Jaguar!"
The more likely scenario is that when news reached Apple that the 970 was coming in ahead of schedule (which it appears to be doing), Apple realized that they could release the machine sooner, except that Panther wouldn't be ready. So they started rolling the necessary subset of functionality into Jaguar so that they'd have something to ship the machines with.
This would probably have happened several months ago, at least. Alternately, they were careful and launched a Jaguar-on-970 project even before it was clear that Panther wouldn't ship with the 970s, just because they knew that projects had a way of slipping behind schedule. After all, I'm sure they're more eager than we are to ship a 970-based Mac as soon as is practicable.
Now for a bit of a tangent: we have a contribution from AppleLust's Joe Carson. It's something of an intemperate rant, he stubbornly refuses to link to or even credit the sites he gleans his information from, and he offers the MacBidouille benchmarks as given. He also says something about the Opteron that I'm not sure about:
Quote:
The pseudo-64 bit Opteron can run 32 bit x86 code directly, and it an run it's own specially cranked up extensions for 64 bit... but not simultaneously. You an run your old software and Windows as is, but if you need to run a 64 bit x86 app, you have to shut own the machine and reboot into a 64 bit operating system, While in 64 bit mode you cannot run your old 32 bit x86 apps or operating systems. Oops!
As I understood it, the Opteron has to switch between "64 bit mode" and "32 bit mode." I was unaware that this required a system reboot? and frankly I'd be surprised if it did. Anyone more familiar with the processor care to comment? Or is this article as slapdash as it reads?
The pseudo-64 bit Opteron can run 32 bit x86 code directly, and it an run it's own specially cranked up extensions for 64 bit... but not simultaneously. You an run your old software and Windows as is, but if you need to run a 64 bit x86 app, you have to shut own the machine and reboot into a 64 bit operating system, While in 64 bit mode you cannot run your old 32 bit x86 apps or operating systems. Oops!
This is bunk. Check out Ars Technica's artile on the x86-64 ISA, specifically the following (emphasis mine):
Quote:
These modes are set for each segment of code on a per-segment basis by means of two bits in the segment's code segment descriptor. The chip examines these two bits so that it knows whether to treat a particular chunk of code as 32-bit or 64-bit.
In addition, there are other Opteron tests around the net (http://aceshardware.com/read.jsp?id=55000251 for one) that benchmarked the Opteron running both 32 and 64 bit code, and nary a reboot was mentioned. I don't know where Joe Carson got this reboot idea from, but he's probably on crack.
Well, I wasn't carrying my Xserve around much - though I am a sissy-girl.
Perchance we'll finally be seeing the much rumored iTMS Wurlitzer. I could see Steve keeping that on the side of the stage for each keynote, cranking out Beach Boys tunes...
...[*] Availability? 1 or 2 single processor 970 based Macs in 30 to 60 days.[*] Dual processors to follow with the release of Panther, sometime in the fall.
...
I have a gut feeling that what was wrong with the original list was the whole "first lets release a wimpy single processor version on a hacked version of Jaguar" business. How about having an all dual lineup and waiting for Panther? I think that would make a bit more of a splash: Apple would have machines which really would trash Pentiums. The demos would show them trashing Pentiums and as they became available it would be pretty clear that they really do trash Pentiums. If you have a bunch of singles coming out on a hacked version of Jaguar, sure they will trash the current G4s, but the people at PC magazine are going to run a bunch of benchmarks showing that these new machines are still slower than the P4 and by the time the duals come out the news value of the new chip will have evaporated.
Steve Jobs is going to introduce a new chip tower. In fact it won't be a tower, nor a cube, but a brand new mac . The Apple. This computer will definitively mix art with value. Imagine a computer with the shape of an apple. The cut size of the apple will have the DVD r slot in.
Just 10 days more to wait . Cannot wait any longer to see this mervellous mastercraft piece of design.
Yes , you have understand it well : SJ will introduce the Apple
Steve Jobs is going to introduce a new chip tower. In fact it won't be a tower, nor a cube, but a brand new mac . The Apple. This computer will definitively mix art with value. Imagine a computer with the shape of an apple. The cut size of the apple will have the DVD r slot in.
Just 10 days more to wait . Cannot wait any longer to see this mervellous mastercraft piece of design.
Yes , you have understand it well : SJ will introduce the Apple
Hmmmm. There's a post in the WWDC thread that made me think of an Apple branded blade server running the rumored xGrid clustering software. Would a full rack of blades be heavy enough?
I have a gut feeling that what was wrong with the original list was the whole "first lets release a wimpy single processor version on a hacked version of Jaguar" business. How about having an all dual lineup and waiting for Panther? . . .
The changes to Jaguar are minor according to IBM, and will not slow down the 970. It just will not run Mac 64-bit applications, of which there are none and won't be for a little while. There is no reason a Smeagol version of Jaguar would not run on both dual and single processor G5 Power Macs.
... think of an Apple branded blade server running the rumored xGrid clustering software. Would a full rack of blades be heavy enough?
My thread from a year ago "Blade Runner ~ Modular Mac " Sounds better now that IBM has announced their intentions to release a 970 powered blade for their blade chassis. LINK
What about Apple just licensing OSX Server to IBM for these 2.5 GHz blades?
The changes to Jaguar are minor according to IBM, and will not slow down the 970. It just will not run Mac 64-bit applications, of which there are none and won't be for a little while. There is no reason a Smeagol version of Jaguar would not run on both dual and single processor G5 Power Macs.
The point was not that it wouldn't run but that it wouldn't run as fast.
After have compelling various threads and rumor my feeling is that Apple will only release the single PPC 970 running on Smeagol. How did i manage to go to this conclusion ? few points :
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
2) The only reason of Smeagol is to support the PPC 970 before the release of Panther (due in september). If there is not PPC 970, there is no need of smeagol. Apple do not work for nothing, especially when a real 64 bit os in on the way : panther.
After have compelling various threads and rumor my feeling is that Apple will only release the single PPC 970 running on Smeagol. How did i manage to go to this conclusion ? few points :
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
Where did you get this bit about NUMA? NUMA is a performance hit on any system where it's not strictly necessary, there's no real reason to make the 970 systems NUMA at all. Apple has to design a new memory controller for the 970, and either they did so in a brain-damaged fashion tha requires two memory controllers (which I find incredibly hard to believe) or it's a straight and normal memory architecture.
There's no significant difference between dual and single 970 systems from the perspective of memory architecture unless Apple is doing something incredibly stupid (or supporting dual processor systems using an architecture designed for 64-processor systems!).
After have compelling various threads and rumor my feeling is that Apple will only release the single PPC 970 running on Smeagol. How did i manage to go to this conclusion ? few points :
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
2) The only reason of Smeagol is to support the PPC 970 before the release of Panther (due in september). If there is not PPC 970, there is no need of smeagol. Apple do not work for nothing, especially when a real 64 bit os in on the way : panther.
Comments
Originally posted by haunebu
I agree. It's not like Apple's gonna say "Oh, wow, this new chip just landed. I guess we better hack it into Jaguar!"
The more likely scenario is that when news reached Apple that the 970 was coming in ahead of schedule (which it appears to be doing), Apple realized that they could release the machine sooner, except that Panther wouldn't be ready. So they started rolling the necessary subset of functionality into Jaguar so that they'd have something to ship the machines with.
This would probably have happened several months ago, at least. Alternately, they were careful and launched a Jaguar-on-970 project even before it was clear that Panther wouldn't ship with the 970s, just because they knew that projects had a way of slipping behind schedule. After all, I'm sure they're more eager than we are to ship a 970-based Mac as soon as is practicable.
Now for a bit of a tangent: we have a contribution from AppleLust's Joe Carson. It's something of an intemperate rant, he stubbornly refuses to link to or even credit the sites he gleans his information from, and he offers the MacBidouille benchmarks as given. He also says something about the Opteron that I'm not sure about:
The pseudo-64 bit Opteron can run 32 bit x86 code directly, and it an run it's own specially cranked up extensions for 64 bit... but not simultaneously. You an run your old software and Windows as is, but if you need to run a 64 bit x86 app, you have to shut own the machine and reboot into a 64 bit operating system, While in 64 bit mode you cannot run your old 32 bit x86 apps or operating systems. Oops!
As I understood it, the Opteron has to switch between "64 bit mode" and "32 bit mode." I was unaware that this required a system reboot? and frankly I'd be surprised if it did. Anyone more familiar with the processor care to comment? Or is this article as slapdash as it reads?
Renouned Mac online journalist Joe C. Carson was found dead in his home last night, apparently after overdosing on crack
The pseudo-64 bit Opteron can run 32 bit x86 code directly, and it an run it's own specially cranked up extensions for 64 bit... but not simultaneously. You an run your old software and Windows as is, but if you need to run a 64 bit x86 app, you have to shut own the machine and reboot into a 64 bit operating system, While in 64 bit mode you cannot run your old 32 bit x86 apps or operating systems. Oops!
This is bunk. Check out Ars Technica's artile on the x86-64 ISA, specifically the following (emphasis mine):
These modes are set for each segment of code on a per-segment basis by means of two bits in the segment's code segment descriptor. The chip examines these two bits so that it knows whether to treat a particular chunk of code as 32-bit or 64-bit.
In addition, there are other Opteron tests around the net (http://aceshardware.com/read.jsp?id=55000251 for one) that benchmarked the Opteron running both 32 and 64 bit code, and nary a reboot was mentioned. I don't know where Joe Carson got this reboot idea from, but he's probably on crack.
[Edit: Sheesh, he OD'd!!]
That way AMD has side-stepped the obtuse refusal of developers to learn a new processor instruction set.
Dude, it's not the developers, its all the old apps! Maybe he should look up 'compiler' in the dictionary!! Besides, we're not all obtuse...
despite the fact that every 64 bit RISC processor in existence...will stomp "faster" 32 bit Intel processors
500 MHz UltraSPARC, anyone? Truly bizarre.
Originally posted by moki
There will be some heavy things about at WWDC just don't expect to be able to carry 'em home.
so.... it's bigger than a breadbox?
Originally posted by BuonRotto
so.... it's bigger than a breadbox?
Yep. Time to start hittin' the weight room.
Originally posted by moki
Yep. Time to start hittin' the weight room.
Well, I wasn't carrying my Xserve around much - though I am a sissy-girl.
Perchance we'll finally be seeing the much rumored iTMS Wurlitzer. I could see Steve keeping that on the side of the stage for each keynote, cranking out Beach Boys tunes...
Originally posted by Shaktai
So be it! Here is a (cough) revised list.
...[*] Availability? 1 or 2 single processor 970 based Macs in 30 to 60 days.[*] Dual processors to follow with the release of Panther, sometime in the fall.
...
I have a gut feeling that what was wrong with the original list was the whole "first lets release a wimpy single processor version on a hacked version of Jaguar" business. How about having an all dual lineup and waiting for Panther? I think that would make a bit more of a splash: Apple would have machines which really would trash Pentiums. The demos would show them trashing Pentiums and as they became available it would be pretty clear that they really do trash Pentiums. If you have a bunch of singles coming out on a hacked version of Jaguar, sure they will trash the current G4s, but the people at PC magazine are going to run a bunch of benchmarks showing that these new machines are still slower than the P4 and by the time the duals come out the news value of the new chip will have evaporated.
Just 10 days more to wait . Cannot wait any longer to see this mervellous mastercraft piece of design.
Yes , you have understand it well : SJ will introduce the Apple
Originally posted by BuonRotto
so.... it's bigger than a breadbox?
It's this big:
Introducing the G1945.
Originally posted by Powerdoc
Steve Jobs is going to introduce a new chip tower. In fact it won't be a tower, nor a cube, but a brand new mac . The Apple. This computer will definitively mix art with value. Imagine a computer with the shape of an apple. The cut size of the apple will have the DVD r slot in.
Just 10 days more to wait . Cannot wait any longer to see this mervellous mastercraft piece of design.
Yes , you have understand it well : SJ will introduce the Apple
Is this a parody of Kormac?
Originally posted by moki
Yep. Time to start hittin' the weight room.
Hmmmm. There's a post in the WWDC thread that made me think of an Apple branded blade server running the rumored xGrid clustering software. Would a full rack of blades be heavy enough?
Originally posted by JBL
I have a gut feeling that what was wrong with the original list was the whole "first lets release a wimpy single processor version on a hacked version of Jaguar" business. How about having an all dual lineup and waiting for Panther? . . .
The changes to Jaguar are minor according to IBM, and will not slow down the 970. It just will not run Mac 64-bit applications, of which there are none and won't be for a little while. There is no reason a Smeagol version of Jaguar would not run on both dual and single processor G5 Power Macs.
Originally posted by Tomb of the Unknown
... think of an Apple branded blade server running the rumored xGrid clustering software. Would a full rack of blades be heavy enough?
My thread from a year ago "Blade Runner ~ Modular Mac " Sounds better now that IBM has announced their intentions to release a 970 powered blade for their blade chassis. LINK
What about Apple just licensing OSX Server to IBM for these 2.5 GHz blades?
Aphelion ...
Originally posted by snoopy
The changes to Jaguar are minor according to IBM, and will not slow down the 970. It just will not run Mac 64-bit applications, of which there are none and won't be for a little while. There is no reason a Smeagol version of Jaguar would not run on both dual and single processor G5 Power Macs.
The point was not that it wouldn't run but that it wouldn't run as fast.
Originally posted by Barto
Is this a parody of Kormac?
Kormac ? : i don't know, but a parody : yes definitively
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
2) The only reason of Smeagol is to support the PPC 970 before the release of Panther (due in september). If there is not PPC 970, there is no need of smeagol. Apple do not work for nothing, especially when a real 64 bit os in on the way : panther.
Originally posted by Powerdoc
After have compelling various threads and rumor my feeling is that Apple will only release the single PPC 970 running on Smeagol. How did i manage to go to this conclusion ? few points :
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
Where did you get this bit about NUMA? NUMA is a performance hit on any system where it's not strictly necessary, there's no real reason to make the 970 systems NUMA at all. Apple has to design a new memory controller for the 970, and either they did so in a brain-damaged fashion tha requires two memory controllers (which I find incredibly hard to believe) or it's a straight and normal memory architecture.
There's no significant difference between dual and single 970 systems from the perspective of memory architecture unless Apple is doing something incredibly stupid (or supporting dual processor systems using an architecture designed for 64-processor systems!).
Originally posted by Powerdoc
After have compelling various threads and rumor my feeling is that Apple will only release the single PPC 970 running on Smeagol. How did i manage to go to this conclusion ? few points :
1) The dual PPC 970 will be based upon NUMA architecture and more precisely the chip will communicate via Hypertransport. This is a new technology, more difficult to do than a single mobo without it. So my guess is (and some rumors said so) that the single mobo are ready but not the DP. I'll add that it's possible that Steve show a demo of this beast, but will say that it won't ship until september.
2) The only reason of Smeagol is to support the PPC 970 before the release of Panther (due in september). If there is not PPC 970, there is no need of smeagol. Apple do not work for nothing, especially when a real 64 bit os in on the way : panther.
Actually I am thinking of the same thing....
Single CPU PM comes out first running Smeagol
then....later in September
Dual CPUs PM come but only running 10.3