[quote]Personally, knowing his bias, I think he is afraid of showing his findings in that it totally obliterates anything on the x86 side or processor design. <hr></blockquote>
Agreed.
Posted by Bigc:
[quote]Who initiated Altivec anyway, Apple, IBM, MOT? IBM didn't want it but who thought up the idea? <hr></blockquote>
Well, it sorta goes like this... Apparently, it was Keith Diefendorff of Apple (previously of Moto) who pushed for adding vector instructions to PPC. All 3 companies worked on joint development of this SIMD unit, as all 3 were still part of Somerset at this point, although a significant chunk of people involved on all sides were not actually at Somerset itself. The internal development name that the 3 companies decided on was VMX. Moto later trademarked the name AltiVec for it's own usage. Apple got Velocity Engine.
I'll offer a few admittedly less concrete counterpoints, just as food for thought:
NeXTStep had clustering support, and Apple is rumored to be working on kernel-level clustering support for OS X.
The above + Rendezvous + OS X server admin tools = the easiest clustering solution to set up and maintain.
Xserve is, according to Apple, an entry level solution and the first of a line. So Apple might well have a plan here.
Xserve is not as dense in terms of processing power per rack as dual-CPU IBM blades would be, but on the other hand it has gobs of storage per rack unit, where the blades require a SAN or other storage solution. In other words, it's tailored to serve a different market than is served by the processing-power-at-all-costs blades.
Apple had something very much like a blade system designed for them years ago, by frogdesign - the same design firm responsible for the original Macintosh. It was a "bookcase," into which "books" representing storage or CPUs or RAM could be placed. It was way ahead of its time, and well outside any market that Apple had any hope of penetrating then. But I wouldn't be too surprised if someone in Cupertino has dusted some of that research off. If anyone can do a fully modular, plug-and-play blade system it's Apple.
Everything that I've read says that the FSB will be 'up to 900Mhz, which is always tagged on to the whole 1.8Ghz introduction thing, which is now an understatement. Anyway, that suggests that anything with a lower clock speed might not have a 900Mhz FSB, but I never read anything about clear 2:1 ratio (due to DDR) though I suppose that one could assume that. I assume that I'm missing something if that's the case for sure, so can someone put up a link to where IBM states that it IS a 2:1 ration between clock speed and FSB?
<strong>Right, how would you start the PowerMac 970 sales intro?</strong><hr></blockquote>
"This here's a dual 1.4 GHz G4+ maxxed out. Corner two has a Pentium whatsit-max." Blah blah. Start demo. Keep talking. (demo still running) (5 minutes pass). Oh, this is taking too long. Pulls sheet off a new case. "I think I'll do it on my new computer." Hits space bar. 970 finishes first. Rubs 970 affectionately.
Turns to crowd "$3799 for this one, singles start at 1400, duals, and quads available, and IBM is now an authorized reseller of Mac OS X if you'd like more than 4 processors." (Big rack with big _I_B_M_ on it rolls from darkness) Hits spacebar. _Still_ finishes before the G4 finishes (which is still chugging along).
"I told you we'd have options. I'd like to say 2004 is officially the year of the PowerMac"
Imagine the look on the face of the poor unsuspecting not-in-tune-with-rumors soul that purchases a Dual 1.45 GHz G4s and discovers a couple weeks later that a 2.5 GHz 970 is now selling for the same price and offers 3+ times the performance.
I think the look would probably be equivalent to the those of the people that bought a 23" cinema display right before the huge price cuts.
[ 02-27-2003: Message edited by: kim kap sol ]</p>
Good news on Hannibal's part 2 of his PPC 970 article!
From Ars Technica's discussion thread on the IBM PPC 970 blade server news:
[quote]By Hannibal:
And regarding the PPC 970 article, I've yet to get a response from IBM either via phone or email. I'm going to go ahead without the info for now, because I'm determined to get it out there. It's half written already.<hr></blockquote>
I would do a lot more than rub it affectionately, but then I'd probably get arrested.
Seriously, I was excited enough by 1.8GHz - but 2.5 GHz with the potential for more when it goes to 0.09 has me planning a celebratory bottle of wine whilst watching the keynote. And then, I'll just wait for the 970-powered Powerbook.
I want to see Michael Dell's face when this stuff gets released, especially when we see how fast Virtual PC runs on Mac.
[quote] · Runs at frequences ranging from 1.8 GHz - 2.5 Ghz
Therefore the IBMPowerPC 970 is the fastest PowerPC so far.
Further technical highlights of the PowerPC 970:
· Onchip 512 KB L2 Cache
· Altivec ? Vector/SIMD unit
· 6,4 GB/s I/O system bus throughput<hr></blockquote>
Doesn't that imply that it'll only be a 900MHz bus? It lists the 2.5Ghz speed on that page so, if it was a 2:1 ratio regardless of frequency, wouldn't they have rated the bus speed as "up to 8.8GB/s" (going off Amorph's numbers)?
Comments
...oh and please put your goggles on.
<hr></blockquote>
...yeah, 2.5 ghz, you could take somebody's eye out with that...
Lemon Bon Bon <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/cancer.gif" border="0" alt="[cancer]" />
[ 02-27-2003: Message edited by: Lemon Bon Bon ]</p>
Me too. <begins wild specualtion on why IBM aren't spilling the beans...>
Right, how would you start the PowerMac 970 sales intro?
Also. What's stopping IBM from producing say a 1.2Ghz or 1.3Ghz 970. Surely that would run cool enough for a powerbook.
[quote]Personally, knowing his bias, I think he is afraid of showing his findings in that it totally obliterates anything on the x86 side or processor design. <hr></blockquote>
Agreed.
Posted by Bigc:
[quote]Who initiated Altivec anyway, Apple, IBM, MOT? IBM didn't want it but who thought up the idea? <hr></blockquote>
Well, it sorta goes like this... Apparently, it was Keith Diefendorff of Apple (previously of Moto) who pushed for adding vector instructions to PPC. All 3 companies worked on joint development of this SIMD unit, as all 3 were still part of Somerset at this point, although a significant chunk of people involved on all sides were not actually at Somerset itself. The internal development name that the 3 companies decided on was VMX. Moto later trademarked the name AltiVec for it's own usage. Apple got Velocity Engine.
--
Ed M.
I'll offer a few admittedly less concrete counterpoints, just as food for thought:
NeXTStep had clustering support, and Apple is rumored to be working on kernel-level clustering support for OS X.
The above + Rendezvous + OS X server admin tools = the easiest clustering solution to set up and maintain.
Xserve is, according to Apple, an entry level solution and the first of a line. So Apple might well have a plan here.
Xserve is not as dense in terms of processing power per rack as dual-CPU IBM blades would be, but on the other hand it has gobs of storage per rack unit, where the blades require a SAN or other storage solution. In other words, it's tailored to serve a different market than is served by the processing-power-at-all-costs blades.
Apple had something very much like a blade system designed for them years ago, by frogdesign - the same design firm responsible for the original Macintosh. It was a "bookcase," into which "books" representing storage or CPUs or RAM could be placed. It was way ahead of its time, and well outside any market that Apple had any hope of penetrating then. But I wouldn't be too surprised if someone in Cupertino has dusted some of that research off. If anyone can do a fully modular, plug-and-play blade system it's Apple.
[ 02-27-2003: Message edited by: Amorph ]</p>
[ 02-27-2003: Message edited by: FrostyMMB ]</p>
<strong>Right, how would you start the PowerMac 970 sales intro?</strong><hr></blockquote>
"This here's a dual 1.4 GHz G4+ maxxed out. Corner two has a Pentium whatsit-max." Blah blah. Start demo. Keep talking. (demo still running) (5 minutes pass). Oh, this is taking too long. Pulls sheet off a new case. "I think I'll do it on my new computer." Hits space bar. 970 finishes first. Rubs 970 affectionately.
Turns to crowd "$3799 for this one, singles start at 1400, duals, and quads available, and IBM is now an authorized reseller of Mac OS X if you'd like more than 4 processors." (Big rack with big _I_B_M_ on it rolls from darkness) Hits spacebar. _Still_ finishes before the G4 finishes (which is still chugging along).
"I told you we'd have options. I'd like to say 2004 is officially the year of the PowerMac"
Nifty promo' idea.
IBM and Apple use a big 'strap on' (gun) and bend Intel over...
'This won't hurt a bit...'
Lemon Bon Bon
I think the look would probably be equivalent to the those of the people that bought a 23" cinema display right before the huge price cuts.
[ 02-27-2003: Message edited by: kim kap sol ]</p>
From Ars Technica's discussion thread on the IBM PPC 970 blade server news:
[quote]By Hannibal:
And regarding the PPC 970 article, I've yet to get a response from IBM either via phone or email. I'm going to go ahead without the info for now, because I'm determined to get it out there. It's half written already.<hr></blockquote>
<strong>
Rubs 970 affectionately.
</strong><hr></blockquote>
I would do a lot more than rub it affectionately, but then I'd probably get arrested.
Seriously, I was excited enough by 1.8GHz - but 2.5 GHz with the potential for more when it goes to 0.09 has me planning a celebratory bottle of wine whilst watching the keynote. And then, I'll just wait for the 970-powered Powerbook.
I want to see Michael Dell's face when this stuff gets released, especially when we see how fast Virtual PC runs on Mac.
Bring it on!!
<a href="http://www-5.ibm.com/de/pressroom/cebit2003/i/highres/bladeprototype_300_Dpi.jpg"; target="_blank">http://www-5.ibm.com/de/pressroom/cebit2003/i/highres/bladeprototype_300_Dpi.jpg</a>;
--
Ed M.
[quote] · Runs at frequences ranging from 1.8 GHz - 2.5 Ghz
Therefore the IBMPowerPC 970 is the fastest PowerPC so far.
Further technical highlights of the PowerPC 970:
· Onchip 512 KB L2 Cache
· Altivec ? Vector/SIMD unit
· 6,4 GB/s I/O system bus throughput<hr></blockquote>
Doesn't that imply that it'll only be a 900MHz bus? It lists the 2.5Ghz speed on that page so, if it was a 2:1 ratio regardless of frequency, wouldn't they have rated the bus speed as "up to 8.8GB/s" (going off Amorph's numbers)?
<strong>Damn!! Moki !! someonew was sandbanging!!
there may be more than one bag of sand, too... two more that I can think of actually... I posted it here as well:
<a href="http://www.ambrosiasw.com/webboard/Forum64/HTML/001106.html"; target="_blank">http://www.ambrosiasw.com/webboard/Forum64/HTML/001106.html</a>;
[ 02-27-2003: Message edited by: moki ]</p>
<strong>Staggering. That will be the equivalent of a 5 gig G4!?!? Intel won't get to that level of performance until late 2004!!!
I'm in shock. :eek:
Put them in dual formation. That's the equivalent of a 10 gig G4 in terms of performance!
:eek:
Sandbagging. Indeed.</strong><hr></blockquote>
See, there's plenty or reason to not worry and be happy.
<strong>
there may be more than one bag of sand, too... two more that I can think of actually... I posted it here as well:
<a href="http://www.ambrosiasw.com/webboard/Forum64/HTML/001106.html"; target="_blank">http://www.ambrosiasw.com/webboard/Forum64/HTML/001106.html</a>;
[ 02-27-2003: Message edited by: moki ]</strong><hr></blockquote>
two as in dual as in core?
That would be nice... and something that I though was going to happen (and then later quite shocked when it did come to pass)
Dave