power mac won't get any faster
The latest news.
Power Mac will be stuck at rougly 2.5GHz for the next 3 years.
This is a known fact.
Every processor company is hitting a power wall and its not likely to change.
The next power mac are 2.5GHz multicore designs.
powerbook will use motorola embedded devices because they have a better future (all chip designers now are shifting from raw speed to power consumption as their main focus)
Power Mac will be stuck at rougly 2.5GHz for the next 3 years.
This is a known fact.
Every processor company is hitting a power wall and its not likely to change.
The next power mac are 2.5GHz multicore designs.
powerbook will use motorola embedded devices because they have a better future (all chip designers now are shifting from raw speed to power consumption as their main focus)
Comments
A single dual 9XX core from IBM may be a good powerbook chip even at reduced voltage...Delivery time for IBM dual-core?
Originally posted by Nr9
This is a known fact.
Link?
Oh wait, aren't you the guy who last year said:
The PowerBook G5 has 2 MCMs with 2 processor each
4 processor PowerBook G5 with modified 440 core with altivec
Link
Right. You speak many facts....
Originally posted by kupan787
Link?
Oh wait, aren't you the guy who last year said:
Link
Right. You speak many facts....
Well, in all fairness, he does have a good point. Clock speed is likely to move sideways or even downwards with dual-core designs, even though performance will improve dramatically.
We've been hitting a bit of a wall on the PC side of the fence for the past couple of years here. For example, I'm using a 2.1GHz P4 that I built a little over 2 years ago, which for most purposes is still pretty competitive with the latest and greatest chips that are currently shipping. 2005 however is going to be mighty interesting.
Originally posted by Chagi
even though performance will improve dramatically.
To be fair, that depends on the aplication. If you are using a single threaded aplication, then preformance will go down if you are using a multi CPU computer. This preformance hit diminishes if you are using many aplications at the same time though.
Nr9.
Rumour has it that Apple has samples of dual core chips in August.
Getting on for two months now.
Rumour says IBM will begin production in January.
So, what kind of schedule are we on, Nr9?
A Jan' ramp for a March announcement?
So, 2.5 gig and no further? I guess that means no 'bump' to 2.8 in Jan'?
Why bother when dual core designs are right around the corner? Who would buy a 2.8 rev' when it's common knowledge Wintel are going dual core and that Apple is going to have to go with that in order to protect what PowerMac marketshare they do have.
Timeline, Nr9?
Mid' 2005 is my guess at latest. Otherwise it would be a major shock for AMD to beat IBM to dual core designs.
Apple must strike out the stadium this time after the lacklustre rev B PowerMac update.
Lemon Bon Bon
I thought this chip may take us to 3 gig? Supposedly has deeper pipelines with variable depth according to integer, floating point with more efficient 'instruction cracking' to offset length of pipelines.
Deeper pipelines = more ghz.
Go over to Thinksecret's message boards, Nr9 and check out Morpheus' posts on the 'next gen' IBM cpu.
So, you have my attention.
Lemon Bon Bon 8)
I thought this chip may take us to 3 gig? Supposedly has deeper pipelines with variable depth according to integer, floating point with more efficient 'instruction cracking' to offset length of pipelines.
Deeper pipelines = more ghz.
Go over to Thinksecret's message boards, Nr9 and check out Morpheus' posts on the 'next gen' IBM cpu.
So, you have my attention.
Lemon Bon Bon 8)
Originally posted by Nr9
The latest news.
Power Mac will be stuck at rougly 2.5GHz for the next 3 years.
This is a known fact.
Every processor company is hitting a power wall and its not likely to change.
The next power mac are 2.5GHz multicore designs.
powerbook will use motorola embedded devices because they have a better future (all chip designers now are shifting from raw speed to power consumption as their main focus)
I am more confident that your intelligence will not improve in the next 3 coming years.
The PPC 970 was supposed to reach 3 ghz. I am confident that we will se faster G5.
However you are right that manufacturers are becoming more focused on raw power rather than clockspeed marketing.
Originally posted by Lemon Bon Bon
Can you confirm the 'Antares' chip aka 970fx dual core based on Power5 architecture?
I thought this chip may take us to 3 gig? Supposedly has deeper pipelines with variable depth according to integer, floating point with more efficient 'instruction cracking' to offset length of pipelines.
Deeper pipelines = more ghz.
Go over to Thinksecret's message boards, Nr9 and check out Morpheus' posts on the 'next gen' IBM cpu.
So, you have my attention.
Lemon Bon Bon 8)
Antares is being worked on - this much I do know. From what I know, the 970 series will not go much beyond 3.25 GHZ, but we should see a clockrate increase to go along with architectural refinement, L2 increase, and the addition of more cores.
The Power5 derivative should be around at some point next year, with dual cores and multi-threading still in the pipeline from big blue.
As for the PowerBooks, the low-voltage 970 is progressing and there is a specific dual-core 970 designed for laptops that is being worked on - similar to the Pentium M in increasing the cache. The PB chip will be based off the current 970 architecture and not yet incorporate the Power5 derivative, though that is in the planning for the future.
I agree with Nr9 that the clock rate race is, by and large, over for now. I've been waiting for this to happen for some time, and it looks like it has finally come to pass. IBM might be able to push the 970 family's clock rate a bit higher (i.e. the low 3's) with process and pipeline refinements, but it won't be a dramatic leap and most of the performance increases will come from other directions (e.g. multiple cores, execution unit level improvements, improved memory architectures). Same for Intel/AMD.
It'll be interesting to see if the rather high promised clock rates for the POWER6 actually come to fruition, and how they accomplish that. But that is 2-3 years out.
Note: the title of this thread is very deceptive. The PowerMac's performance will improve substantially, but its clock rate won't get significantly higher. More effort will have to go into writing high performance multi-threaded software, but software developers have had it too easy for too long and most software is horribly inefficient on modern hardware.
Originally posted by Programmer
Hannibal's latest post on Ars has me wondering if there will be no POWER5-lite, simply enhancements to the 970's existing design. This fits well with the 970MP rumors, and ... other things.
From what I've been told, all dual-core designs for desktops are POWER5 based. The only dual-cores based off the current design (with modifications) are the "mobile solution" versions that we'd see in PowerBooks.
The POWER5 derivatives were meant for the end of this year, but with all the delays IBM has seen the latest estimates would put their availability sometime near summer.
the future of software development is toward heavily threaded apps
lemon bon bon:
deeper pipelines is cancelled.
the higher frequency proved to be too power hungry and the chips are melting down. you dont know how many next generation processor projects were cancelled due to power problems.
powerdoc:
antares won't reach 3GHz
programmer:
the high clock rate design of POWER6 is officially cancelled. IBM will switch into aggressively multi-core designs, and try to design low power cores.
this is what happened:
they were designing the POWER6, everything looked fine, until they looked at the power consumption, which was over 200 watts per core. then, they cancelled the design because there was no way to fix it.
Originally posted by Nr9
the future of software development is toward heavily threaded apps
Multicore, Multi-processor machines / clustured computing... Apple is ahead of the herd on these with Xgrid and Tiger... We will see better and faster Power Macs.
The title of this thread is wrong.
Originally posted by Nr9
this is what happened:
they were designing the POWER6, everything looked fine, until they looked at the power consumption, which was over 200 watts per core. then, they cancelled the design because there was no way to fix it.
That is a bit of a stretch, IBM could always buy the technology from somebody else to fix the problem. Just because IBM is unwilling or unable to fix the power disapation issue with it's 90nm process doesn't imply that others won't overcome the issues.
What is undeniable is that IBM has some of the hottest chips out there on 130 & 90nm processes. Especially considering performance of the 970 series.
Personally I do believe that IBM is stuck at 2.5GHz and maybe isn't even truely there yet. It would be interesting to learn just how well the 2.5GHz chips are yielding. My suspicion is the yield truely sucks based on what appears to be random thermal performance of the chips. It appears that this is the result of a couple of things: IBM's sucky 90nm process and their design tools that apparently could use some sort of diet.
As to multiple cores I think this is exactly what a portable needs especially if power management can be done core by core. What is interesting is how far Freescale is along with these chips. Having to wait what is now only a couple of months for the new years production would not be that bad. Waiting until late into 2005 would not be so good. The reality is that Apple needs portable upgrades real soon, so I wouldn't be surprised if at least one of Freescales processors is already for production.
Dave
EVERYONE has power problems. its the physical limitation of CPUs
Nr9. Likewise.
I suppose even at 2.5 dual core. You're going to get a very impressive chip. Especially as you dual the dual core.
X sees four cpus. That is going to kick-ass 3D rendering along some. Especially with increase programability of GPUs as co-processors.
I am happy with this scenario. I would still like IBM to touch down on 3 gig, though.
but software developers have had it too easy for too long and most software is horribly inefficient on modern hardware
I don't think I've seen a computer stretch since the days of the c64. The programmers of that machine pushed it way beyond its limits. Big sprites. More than 8 appearing on screen at a given time. Fitting games into a tiny 64k!!! The most diverse array of games ever.
It seems that code is middle of the road...for the average machine. Knowing as they do, the next fastest machine will pick up the bloatware tab.
Just my perception. It took Amiga programmers ages just to get smooth hardware scrolling! Something c64 programmers cracked early-ish on in its life.
That aside,
I find it hard to believe we won't squeak over the 3 gig milestone. Even if it halts right there.
It will be great, after all these years, Intel's Prescott cannot go dual core under its current power outtage..? So they downclock...or go for the Pentium-M which is really lower clocked.
Which means we could see an Antares dual core at 3 gig! And a Pentium dual core at...2-2.5 gig?
And the 970fx is ahead of AMD64 in mhz?
What won't be happening is dual core Pentiums at 4 gig? Or the 5 gig Intel once boasted about?
After all that mhz talk...PPC might eventually come out on top next year?
It seems more and more the refinement of 'X' via 'Spotlight' features that make your existing machine faster in the sense of 'more productive' is the way to go. And with Apple's 'X-Grid' and emphasis on multi-threading...they're ahead of the game. Time to make programming an art? Like it was in the days of the old Apple II and the c64?
Seems to me...are we really unlocking the power we already have? Or are we about to..? eg with things like meta-data and more refined and optimised code with newer versions of compilers?
Lemon Bon Bon
IBM Going to 10GHz