Is Intel's Sossaman powerful enough for PowerMac's?
We all know/suspect that Intel's Yonah will be at the heart of the next-gen iBooks, PowerBooks, Mac minis...
I was wondering, in the case that IBM "can't" provide its new 970 chips to Apple soon... if using Sossaman chips in PowerMacs or iMacs is an option for Apple?
Why?
- the Sossaman is a server/workstation dual core chip.
- it can work in dual processor configurations (dual dualcore)
- it is 64-bits
- it is low enough power (it is a server version of the yonah)
- it will be available in Q1 '06
Here's a link to a german site with some photos, information and benchmarks
What do you (who know about Intel stuff) think?
I was wondering, in the case that IBM "can't" provide its new 970 chips to Apple soon... if using Sossaman chips in PowerMacs or iMacs is an option for Apple?
Why?
- the Sossaman is a server/workstation dual core chip.
- it can work in dual processor configurations (dual dualcore)
- it is 64-bits
- it is low enough power (it is a server version of the yonah)
- it will be available in Q1 '06
Here's a link to a german site with some photos, information and benchmarks
What do you (who know about Intel stuff) think?
Comments
Update: Cinebench scores: G5 dual 2.7GHz: 701, Sossaman quad 2.0GHz: 884 (Of course, a quad G5 may score even higher.)
http://www.barefeats.com/dualcore.html
It probably has PAE allowing it to have 64 GB of memory, but that requires some operating system and developer support. It would simply be more cost effective to wait for Conroe before doing something to Mac OS X to support PAE. Mac OS X is already has some pretty ugly support for 64 bit, not need to muck it up some more.
The iMac however, sweeet!
I bet Intel won't even dare to compare it's 2007 chips to AMD's 2005 Opterons.
Originally posted by ngmapple
Wow it takes 2 Sossamans to beat a single Athlon X2 4800+ chip.
It takes two 1.5 GHz Sossamans to beat a single 2.4 GHz Athlon X2 chip in Cinebench. Sossamans are likely to be in the 2.5 GHz range when they ship, and are likely to have the same performance per clock as the Athlon X2. However, Sossamans will be intended for blade servers and will not compete with Opterons or even Athlon X2s.
To make things worse for Intel, the X2 4800+ is already availible!!! Intel's slower Sossaman won't even be availible till '06. By then AMD will be even further ahead perhaps an X2 5200+??. AMD is soo far ahead of Intel right now.
If you just looked, you can see that Intel has a dual core Pentium XE that is competitive with the Athlon X2. Intel plans on having dual core Netburst chips throughout 2006 at higher clock rates. They've be competitive. They won't outclass Athlon/Opterons, but will be competitive.
We should also note that Apple is likely not going to use Sossaman or any Netburst chip. It's going to be Yonah and Merom-based chips.
If you just looked, you can see that Intel has a dual core Pentium XE that is competitive with the Athlon X2. Intel plans on having dual core Netburst chips throughout 2006 at higher clock rates. They've be competitive. They won't outclass Athlon/Opterons, but will be competitive.
Competitive but still not as fast (this is just a 32-bit benchmark, wait till you see that Xeon compared to the X2 in a 64-bit test), Apple is a brand associated with being the very best. Check out this bench mark showing AMD's 2005 chips beating Intels 2006 chips by leaps and bounds:
For someone working with digital video editing, especially HD content, that extra speed is worth it. Esspecially since you're getting a faster cpu for roughly the same price.
We should also note that Apple is likely not going to use Sossaman or any Netburst chip. It's going to be Yonah and Merom-based chips.
Yonah may end up in iBooks, but deffinately not in PowerBooks, not as a 32-bit only cpu or a bus speed capped at 667Mhz. Merom looks a little more worthy given the fact it will support EMD64.
Originally posted by THT
We should also note that Apple is likely not going to use Sossaman or any Netburst chip. It's going to be Yonah and Merom-based chips.
But Sossaman is Yonah.
Originally posted by ngmapple
Competitive but still not as fast (this is just a 32-bit benchmark, wait till you see that Xeon compared to the X2 in a 64-bit test), Apple is a brand associated with being the very best. Check out this bench mark showing AMD's 2005 chips beating Intels 2006 chips by leaps and bounds:
For someone working with digital video editing, especially HD content, that extra speed is worth it. Esspecially since you're getting a faster cpu for roughly the same price.
Apple has rarely been a brand associated with the fastest CPU performance. The best hardware/software integration and industrial design, yes. CPU performance, no. The price performance advantage with Yonah and Merom-based chips is the optimal choice for the Apple brand: near the absolute best performance in an elegantly designed package.
And again, your referenced chart and the conclusion you make is fanboyish. 2 Sossamans are needed to beat one Athlon X2 4800+? Yes, sure, but those are 1.5 GHz Sossamans, and we all know that Sossamans will be clocked up to 2.5 GHz, and that dual core Pentium XEs will be in the 3.6 GHz range in Q1 2006.
Intel isn't the fastest now, they won't be the fastest in 2006, but they will be competitive. And where they will be head of the class in, performance/watt, is what Apple is most interested in because they like to make pretty boxes.
Yonah may end up in iBooks, but deffinately not in PowerBooks, not as a 32-bit only cpu or a bus speed capped at 667Mhz. Merom looks a little more worthy given the fact it will support EMD64.
Yonah will end up in Powerbooks. It is the best option overall including the AMD and PPC options. What we should be crossing our fingers on is a 1.6 GHz dual-core Yonah appearing in an iBook. I wouldn't be surprised at all if iiBooks use single-core Yonahs.
Originally posted by wmf
But Sossaman is Yonah.
Yes, but it is Yonah + 20 to 40 more Watts and slightly higher clock rate and dual independent bus support from the core logic. Whether Apple will use them in Power Mac/Intel machines is up for debate. I think most everyone is saying no even though they would be nice machines. A 2.6 GHz dual-Sossaman Power Mac would be a very nice system, but 4 GB memory support might be limiting to customers used to being able to get 8 GB.
Originally posted by THT
Apple has rarely been a brand associated with the fastest CPU performance.
This is not at all true, when they first unvieled the PowerMac G4/G5 as well as the PowerBookG4 they touted them as the "World's first Personal SuperComputer" and went on and on about how PowerPC was faster than x86 by leaps and bounds. Not to mention the countless performance comparisons (often photoshop and video tests) demonstrated at MacWorld Expos durring the keynotes.
Now I think we all agree that the Intel's Pentium M offerings on the market now are deffinately ahead of the G4 used in current PowerBooks, but I don't think they can touch the G5 yet (but I agree they are likely two within the next 2 years given the ability for Intel to manufacturer at smaller process sizes than IBM). Intel vs. AMD in performance per watt can be argued, but in terms of performance and features Intel's future chips are behind AMD and the G5. I'm talking about 64-bit architecture standard in all cpu's.
Anyway, the point I'm trying to make is that if Apple can move the PowerBook to 64-bit architecture with current cpu offerings (either the new lower power 970FX or something from AMD) why would they wait for an Intel chip that isn't even released yet only to be stuck with another 32-bit machine? That just doesn't make sense to me. OS X / Darwin is a high performance operating system designed to take advantage of 64 bit architecture, why constrain it to 32-bit??? Unless Yonah can deliver 12+ hours of battery life, I don't see the benifit in sacrificing that much performance.
Originally posted by ngmapple
OS X / Darwin is a high performance operating system designed to take advantage of 64 bit architecture, why constrain it to 32-bit???
OS X is based on Mach. Mach is a microkernel. Microkernel-based operating systems are not designed for speed : they come with a very clean architecture that separates the core kernel functions (memory management, multitasking, etc...) and the low-level services (file system, network, etc...). This is a lot cleaner, but it slows down disk access, for example. Monolithic kernels such as Linux include file system routines right inside the kernel.
Furthermore, OS X uses files to swap memory, while high performance operating systems use dedicated partitions. This really hits performance for obvious reasons.
And OS X is not designed to take advantage of 64 bit architecture. It mostly uses dirty tweaks to do so.
Apple uses the "high performance" and "64-bit" arguments to market its OS, but it clearly is not the leader on those fronts...
Originally posted by ngmapple
This is not at all true, when they first unvieled the PowerMac G4/G5 as well as the PowerBookG4 they touted them as the "World's first Personal SuperComputer" and went on and on about how PowerPC was faster than x86 by leaps and bounds. Not to mention the countless performance comparisons (often photoshop and video tests) demonstrated at MacWorld Expos durring the keynotes.
Not my problem if you believed in marketing. PowerPC had a brief moment in the Sun when the 604 was introduced. After the Pentium Pro was introduced, PowerPC has never achieved performance superiority to this day. A few moments of parity, but was never faster again.
Now I think we all agree that the Intel's Pentium M offerings on the market now are deffinately ahead of the G4 used in current PowerBooks, but I don't think they can touch the G5 yet (but I agree they are likely two within the next 2 years given the ability for Intel to manufacturer at smaller process sizes than IBM).
Why bother with the comparison? Pentium M CPUs go into laptops. IBM 970 CPUs do not go into laptops. I'm sure IBM and Apple really really wanted the 970fx to go into laptops, but they never were able to get the 970fx down to laptop power levels over the last 18 months, so there really is no point in making the comparison.
I was a big proponent of a desktop replacement Powerbook G5 for the longest time. Believe me, I want a Powerbook G5 to happen, but Apple doesn't want to.
Intel vs. AMD in performance per watt can be argued, but in terms of performance and features Intel's future chips are behind AMD and the G5. I'm talking about 64-bit architecture standard in all cpu's.
For now, 64-bit doesn't matter for laptops: virtually all laptops only have 2 memory slots thus limiting laptops to less than 4 GB, there is no performance advantage between 64 bit and 32 bit for PPC except for long integer math never mind that laptop G5s aren't available, and Athlons burn more power than Pentium Ms for about the same performance. Pentium Ms are the best choice.
For high-end desktops, it is Apple's choice whether to use 64-bit Netburst CPUs or to wait for Conroe and Woodcrest.
Anyway, the point I'm trying to make is that if Apple can move the PowerBook to 64-bit architecture with current cpu offerings (either the new lower power 970FX or something from AMD) why would they wait for an Intel chip that isn't even released yet only to be stuck with another 32-bit machine?
Because Pentium M CPUs offer the highest performance. Actually, because Yonah CPUs will be the hands down the highest performance laptop CPU.
Because low power 970fx CPUs still consume too much power for laptops.
Because AMD laptop solutions aren't better than Intel laptop solutions.
OS X / Darwin is a high performance operating system designed to take advantage of 64 bit architecture, why constrain it to 32-bit??? Unless Yonah can deliver 12+ hours of battery life, I don't see the benifit in sacrificing that much performance.
For the million-gazillionth time, 64-bit software and hardware isn't any faster than 32-bit software and hardware.
In PowerPC-land, there is about zero performance gain with two exceptions: greater than 4 GB memory support, and long integer math support. There are very few applications that require 64-bit integer math. Laptops, especially Apple laptops, only have enough room for 2 memory slots.
For x86 CPUs, there is a real-world 64-bit performance advantage due to 8 additional registers which equate to some 5% improvement on some codes. That's pretty much the only thing outside of the usual 4+ GB memory support and long integer math.
And the Mac OS X being a 64 bit operating was answered by the Rescuer. Mac OS X currently provides very little 64 bit support. RIght now, it's only been changed to support 4+ GB memory and 64-bit command line code. No GUI code is 64-bit.
Originally posted by THT
Not my problem if you believed in marketing. PowerPC had a brief moment in the Sun when the 604 was introduced. After the Pentium Pro was introduced, PowerPC has never achieved performance superiority to this day. A few moments of parity, but was never faster again.
What saved Apple in the field of performance were its dual-CPU configurations. They were pioneers in the field of consumer-level dual offerings. Although the gain was only limited to a few Photoshop filters before Mac OS X...
Quote:
And the Mac OS X being a 64 bit operating was answered by the Rescuer. Mac OS X currently provides very little 64 bit support. RIght now, it's only been changed to support 4+ GB memory and 64-bit command line code. No GUI code is 64-bit.
I'm not a rescuer, I'm someone to rescue, TNT... Oh sorry, THT
Originally posted by THT
Because low power 970fx CPUs still consume too much power for laptops.
I have an AMD64 laptop that peaks around 90Watts with about 3hrs of battery life and you're telling me a 20Watt 970FX consumes too much power for a laptop???
Originally posted by ngmapple
I have an AMD64 laptop that peaks around 90Watts with about 3hrs of battery life and you're telling me a 20Watt 970FX consumes too much power for a laptop???
Hope that you are not a male, or that you already have children : to me, 3 hours with some stuff running a 90W CPU on your lap equals to testicle suicide!
"New Apple Powerbook Yonah : waste your time, save your balls"
Originally posted by The One to Rescue
"New Apple Powerbook Yonah : waste your time, save your balls"
Originally posted by ngmapple
I have an AMD64 laptop that peaks around 90Watts with about 3hrs of battery life and you're telling me a 20Watt 970FX consumes too much power for a laptop???
What kind of battery is this?
Originally posted by ngmapple
I have an AMD64 laptop that peaks around 90Watts with about 3hrs of battery life and you're telling me a 20Watt 970FX consumes too much power for a laptop???
Apple isn't interested in thick laptop designs. They like a 1 inch thick form factor. So yes, anything over 25 to 30 Watts is too much for them.
There is a 1.8 GHz 37 Watt part, but that apparently is too much for the form factors they like. The problem with the 20 Watt 970fx is that it runs at <1.6 GHz. Doesn't provide much benifit over the existing 1.67 GHz G4s.
Originally posted by THT
There is a 1.8 GHz 37 Watt part, but that apparently is too much for the form factors they like. The problem with the 20 Watt 970fx is that it runs at <1.6 GHz. Doesn't provide much benifit over the existing 1.67 GHz G4s.
What if Apple used the 970MP and underclocked it to run @ about 30 watts. A dual core 970MP at ~1Ghz per core would still be a pretty strong performer. Where would this fit in on a time line?
PowerBook G4 (now) -> PowerBook 970MP 2x1Ghz (soon) -> PowerBook "Merom" (late 06 / early 07)