Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
IBM on PowerPC 970MP Power Savings
- Thread starter MacRumors
- Start date
- Sort by reaction score
emotion said:
Funny, I hear the opposite from my peers. I have found AIX to be the best platform for enterprise computing that I have ever had the pleasure of using. Either way I still like IBM. I don't have anything against Intel. I am sure the new Intel Apples will be awesome.
stockscalper said:
What? Those Dothans are out, on time and they're every bit as fast as they said they would be.
Don't confuse The Pentium M line with all this Pentium D/4/Netburst crap they've had out...that stuff ain't going anywhere near a mac.
In any case IBM/Moto have touted their fair share of vapourous hardware over the years and I'm willing to give another set of jokers a chance
(as are Apple). At least this vapour has a chance at being cheaper because of the volumes made.To give you an idea of how fast a Dothan is in comparison to G5 and G4 here's a performance list for my main application (Ableton Live). Admittedly the app needs some optimisation for Apple hardware but the results speak for themselves.
http://www.ableton.com/forum/viewtopic.php?t=24715
Should cheer you up a bit.
Steamboatwillie said:
Well, it's what you're used to I guess. I'm not as experienced with AIX (for HPC, which is what I do) as I am with Irix, Unicos and Linux.
gekko513 said:
Opteron's are available in 55w as well.
The AMD TDP rating is absolute maximum under worst possible conditions (i.e. at 96w the chip or mobo will probably go up in smoke). The G5 rating is expected wattage at full-load and the Intel rating is 75% of maximum.
Seriously, even if you try to massively overclock the AMD proc you will never get it to 95w.
avus said:
I thinkApple may have a deal with Intel that bars them from working with AMD for at least a few years. I don't think Intel would spend all this money on Apple's transition to X86 if Apple could easily just jump to AMD once the changeover is complete.
I bet the contract is for a decade or more.
This is HILARIOUS.
But,but,but Uncle Steve told us that PPC sucks now and this fantastic Intel ROADMAP.
The ROADMAP!!!
Love my iPod Steve, but ya really shouldn't have tried to nickel and dime IBM over the mobile G5 chip...
Have fun with the Intel processor trainwreck! You earned it Stevie Boy! Too bad I won't be along for the ride. I'll be off in the computational promised land called Cell.
But,but,but Uncle Steve told us that PPC sucks now and this fantastic Intel ROADMAP.
The ROADMAP!!!
Love my iPod Steve, but ya really shouldn't have tried to nickel and dime IBM over the mobile G5 chip...
Have fun with the Intel processor trainwreck! You earned it Stevie Boy! Too bad I won't be along for the ride. I'll be off in the computational promised land called Cell.
Yes, but the Intel chip is about as likely to hit 100% as an Apple system is to hit the GFLOPS ratings that Apple quotes - in other words 0%.BGil said:
The absolute maximum assumes all data in L1 and L2 caches, all pipelines and execution units always busy, .... One cache miss or bubble and you're below the max. (Sounds like Apple and Intel are using roughly the same formula.)
In addition, the Intel chips automatically reduce the heat generation if the chip gets too hot - so in effect the Intel chip cannot exceed the capability of whatever heat sink system you put on it.
AidenShaw said:
Okay but that still doesn't really tell you how much power or heat is being pushed through those processors. The best way to measure the differences would be to use similar configs and test the power draw at the wall. that's what a few PC sites do.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2353&p=4
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2410&p=2
Intel's stated TDP of the P4 630 (3ghz) is 84w and AMD's stated TDP of a 2.4ghz dual core Athlon X2 4800+ is 110w but the AMD proc uses 52 less watts at full-load. That just shows you how much those TDP numbers aren't directly comparable. Considering AMD sells dual core opterons at 55w max I'd have to say AMD is far ahead of Intel right now and probably ahead of whereever Intel is going to be in the next few years. The Yonah has about a 3 month window then it'll be surpassed by the dual core Turion as well.
Steamboatwillie said:
I hadn't used AIX in years and had become a hardened Solaris guy until I just took a new job. I am blown away by the pSeries machine where we're running 20+ LPARS of AIX and Linux. AIX has come a long way and while I still prefer the user friendly Solaris administration I'm slowly becoming a believer.
Well said...
This is a really great point, and I'm glad you brought it up. I'm running a dual 450 MHz G4 with 1.5 GB of RAM, with ~480 GB of storage. To be completely honest, I'm still very happy with it. I'm going to get a new Dual core G5 now; seeing as my computer turns 6 come January. However, this doesn't mean I'm going to get rid of it; it still has a lot of life left in it, and I have grown rather attached to it. Mind you, it doesn't encode video so fast, doesn't go through my lab data very quickly (faster than the machines in the lab however! : ), but it's a solid, reliable machine; I don't think I've turned it off for more than 2 weeks total since I've had it! Anyhow, I plan the same sort of life for my Quad, and can't wait!
longofest said:
This is a really great point, and I'm glad you brought it up. I'm running a dual 450 MHz G4 with 1.5 GB of RAM, with ~480 GB of storage. To be completely honest, I'm still very happy with it. I'm going to get a new Dual core G5 now; seeing as my computer turns 6 come January. However, this doesn't mean I'm going to get rid of it; it still has a lot of life left in it, and I have grown rather attached to it. Mind you, it doesn't encode video so fast, doesn't go through my lab data very quickly (faster than the machines in the lab however! : ), but it's a solid, reliable machine; I don't think I've turned it off for more than 2 weeks total since I've had it! Anyhow, I plan the same sort of life for my Quad, and can't wait!
So for an upcoming film/editing student...
So now with the new Powermacs released, I'm wondering whether or not to wait until the Intel/x86 updates roll around next year, or just stop hesitating and get a Quad (either way, it'll be the best offered at the time cause it's gotta last me awhilie). I won't officially be enrolled and going to school until next Fall (which is when the Intel's might come out), but there are soooo many projects I would love to work on and figure out right now. So basically: should the Intel architecture be that much faster that it would just be worth waiting?
Also, regarding Intel Powerbooks, how would those possibly handle programs like Final Cut Pro and (more importantly) Motion? Thanks everyone.
So now with the new Powermacs released, I'm wondering whether or not to wait until the Intel/x86 updates roll around next year, or just stop hesitating and get a Quad (either way, it'll be the best offered at the time cause it's gotta last me awhilie). I won't officially be enrolled and going to school until next Fall (which is when the Intel's might come out), but there are soooo many projects I would love to work on and figure out right now. So basically: should the Intel architecture be that much faster that it would just be worth waiting?
Also, regarding Intel Powerbooks, how would those possibly handle programs like Final Cut Pro and (more importantly) Motion? Thanks everyone.
In terms of Intel on desktop Macs it isn't looking as good as it once was. I was looking forward to the convergence of IA64 and IA32 (Itanium and Xeon) but it seems that now this is delayed. I think Apple would do well to keep IBM sweet for a little while yet.
http://www.theregister.co.uk/2005/10/29/intel_xeon_2009/
http://www.theregister.co.uk/2005/10/29/intel_xeon_2009/
To answer the last posters question. If I was gonna go for a desktop mac that quad would be awesome if your apps are multiprocessor aware.
I believe that the "convergence" merely meant that Xeon and Itanium processors could share a common chipset and maybe motherboard.emotion said:
It would not make x86/x64 run IA64 software natively, or vice versa.
Note that "IA32" is the 32-bit architecture - virtually all of the current CPUs except the laptop line are x64 (AKA EM64T) 64-bit architecture.
AidenShaw said:
It was not specifically the convergence of ABIs more the fact that it'll all end up as one unified server/high end workstation architecture.
I had hoped that that was waht Apple would pick up after the G5s.
It's not a difference in ABIs, it's completely different ISAs.emotion said:
There's no reason that IA64 and x64 can't have the same ABI.
AidenShaw said:
Yeah true, must reply in less of a hurry
When Intel lose netburst and move more toward IA64/EPIC i'll find that more interesting.
For those of us that do scientific computing, changing to the Xeon or P4 will mean Apple has no intrinsic advantage over any white box PC.
With the Altivec capability, for example, a 10 hour work unit went to 7 hours (Einstein@Home) ... I run a dual G5 2.0 GHz and it beats my 3.4 GHz dual Xeons for most things still ...
Of course I run 24/7 and have no interest in turning down the processor, but want the fastest throughput ... I will love to see what a 1M cache does compared to the 256M cache I currently have ... alas, no cash at the moment...
With the Altivec capability, for example, a 10 hour work unit went to 7 hours (Einstein@Home) ... I run a dual G5 2.0 GHz and it beats my 3.4 GHz dual Xeons for most things still ...
Of course I run 24/7 and have no interest in turning down the processor, but want the fastest throughput ... I will love to see what a 1M cache does compared to the 256M cache I currently have ... alas, no cash at the moment...
Paul D. Buck said:
there is no way that apple will go with netburst architecture (well i'd be very surprised).
for scientific computing the itanium architecture is very very nice when your code is suitably dealt with (either by hand or failing that a good compiler).