Apple and PC users alike criticised Motorola when the G4 CPUs failed to keep pace with the aggressive clock speed ramping of processors from AMD and Intel. The newly announced 8641D with dual cores, dual DDR-II memory controllers and more is set to change all that.
The latest 7447A G4 runs at up to 1.5GHz and can be found in Apple’s 15″ and 17″ PowerBooks. Freescale (previously Motorola Semiconductor) have concentrated on the embedded market where low power consumption rules rather than absolute maximum performance, thus the clock speed has not soared like the Pentium 4 but then neither has pipeline length or the power consumption. The highest performing Pentium 4s already use in excess of 100 Watts, the next generation G4 7448 is expected to require less than 10 Watts at 1.4 GHz. Of course when performance is needed there’s always the Altivec unit which was never been lacking, to encourage it’s use an increasing number of routines are being made available by Freescale for use by their customers.
The G4 however has been at a disadvantage as regards it’s memory bandwidth. While it’s bus has increased to 166MHz (200MHz on the 7448), it does not use the DDR or QDR signalling technology used in PCs thus limiting it’s memory bandwidth. This impacts the performance of some applications but in general usage this doesn’t seem to be much of a problem as applications are more often latency bound and PCs generally have no advantage over the G4 in this respect (apart from AMD64 systems).
All the CPU manufacturers have hit a wall though, clock speeds have rising rapidly but power consumption has been rising faster, the most visible sign of this was the cancellation of Intel’s Tejas processor and their new found focus on dual core processors.
Intel however have also introduced the low power Pentium-M processors which give the majority (if not more) of the performance of a Pentium 4 at a fraction of the power consumption and a considerably lower clock rate. The Pentium-M CPU has been praised for this achievement, a rather odd situation considering the criticism the G4 got; with the Pentium-M Intel is following the exact same strategy the G4 designers have been implementing all along – with much the same results.
New Chips
It’s been known for some time that Freescale were working on new PowerPC processors. Now they’ve been announced and they are substantially different from the previous G4 designs, they still follow the same philosophy of low power consumption but this time doing so at the same time as increasing overall system performance.
Freescale could have increased performance by upping the clock speed but in doing so it would would have sent the power consumption rocketing to PC CPU levels, something unacceptable in their primary market. Instead, they took another path and added a second core, the move to a 90nm SOI (Silicon on Insulator) process cuts power consumption considerably so the second core doubles performance without doubling power consumption. A larger cache (1MB cache per core) and some architectural changes will boost performance further.
They not only added a second CPU core and increased the CPU performance but integrated dual 667MHz DDR-II memory controllers sending aggregate memory bandwidth up by a factor of 8 to 10 GBytes / Second in a single leap, of course placing the memory controllers on die also reduces memory access latency.
As with the Pentium-M the G4 delivers much of the power of a Pentium 4 even given it’s low clock frequency, consequently I fully expect a pair of these cores will outgun any current single core PC CPU though obviously not on single threaded operations. That alone would have brought Freescale back into the reckoning but they didn’t stop there, they have added a slew of on board peripherals not only reducing costs but increasing overall system performance whilst simultaneously reducing power consumption.
AMD’s Opteron moved a Hypertransport connection and a memory controller on board the CPU die, Transmeta’s Efficeon went further adding an AGP X4 controller into the mix. The 8641 family goes much further than either of these by also adding dual PCI Express interfaces, a RapidIO bus (similar idea to HyperTransport) and 4 Gigabit Ethernet MACs with hardware TCP/IP acceleration and other interfaces all onto the same die as the CPU cores.
Despite all the interfaces, controllers and dual CPU cores power consumption is currently estimated at 15 – 25 Watts in typical usage.
The 8641D and 8641 (single core version) are not expected to sample until 2H 2005 so they won’t be turning up in machines in the immediate future but they should prove very interesting when they do. The CPU performance will increase significantly due to the inbuilt memory controllers and clock speed boost (no speeds have been quoted yet other than “at least 1.5GHz” but 1.7GHz seems to be expected) but more than this the overall system performance should increase considerably. This is in contrast to PC designs which go all out to boost CPU performance and leave the rest of the system lagging behind. The new PCI Express will give additional bandwidth but the current 11 year old PCI bus will just about handle a single Gigabit Ethernet port, never mind four!
For years embedded version of the PowerPC CPUs have been available which include all manner of connectors and functionality but the highest performance PowerPC CPUs are typically stand alone components requiring north and south bridges for interfaces. It is an interesting development that Freescale have now chosen to introduce similar integrated functionality on their highest performance CPU. However, this device is not aimed primarily at PCs so it cannot be taken as a sign that PC CPU vendors are planning on going to this level of integration in the immediate future. While users measure systems by pure CPU benchmarks they’ll get systems which are designed to perform well in pure CPU benchmarks, even if you actually use the complete system.
Servers may not seem an obvious target for a G4 based CPU but these are the perfect target for a CPU which optimises system performance over raw CPU performance, servers after all, are doing many things at once so a 8641D could perform surprisingly well. High network throughput requires a lot of computing power, the TCP/IP hardware assist will help with this leaving the CPU to work on the actual content to be served. Low latency connections mean that data can be moved around and accessed for processing quickly.
The low power consumption is also a factor to be considered if a large number of servers are in use, high power consumption CPUs produce heat which has to be removed, usually by cooling systems which cost considerably more than the computer hardware. CPUs with low power consumption can be cooled easier and cheaper and having the north-bridge integrated on the CPU saves even more power.
Operating Systems
It’s a pity something like this wasn’t around when BeOS was still about, it was designed for multiprocessor systems from the get go so would have ran like a dream on such a CPU, hopefully Haiku and Zeta will be able to pick up the benefits though. Of course there are many multiprocessor capable operating Systems available nowadays including of course, Linux and the BSDs so there’ll be no shortage of options for desktop or server users. There will also be, as you would expect, several embedded OSs to choose from.
One feature the 8641D has is the ability to not only run multiprocessor operating systems but also to run an OS per core, bringing new meaning to “dual boot” systems in the process! Having two operating systems running independently has some very interesting possibilities for OS or application development though modified drivers will be required for any operating systems which utilise this feature. This also gives developers the ability to create an uber stable OS environment. If all the I/O goes via a primary OS on core 1 a secondary OS could be set up to run completely independently on core 2 with only minimal communication between them, even if the first OS completely fails and needs to reboot, the second OS is never affected. You could conceivably leave a task running on one core and reboot multiple times into different systems without ever stopping the task. Useful if you want to boost your RC5 score.
Who’ll use it?
The 8641 / 8641D is expected to sample in the second half of next year so I don’t expect to see products turning up for at least another year.
Apple’s iBook would be an obvious target for this CPU as I think they’ll look at the G5 for the PowerBook, however Apple could then find themselves in a situation where the iBook outperforms the PowerBook. They could however use the single core version of the chip in the iBook leaving the option of either the dual core 8641D or a G5 in the PowerBook. Of course as usual Apple will probably use higher clocked “P” rated parts which aren’t generally advertised on the web site.
The 8641D and G5 will be strong in different areas so will not compete directly for the most part, the G5 is strong on floating point and like PC CPUs goes for high single threaded performance. The 8641D will give better overall system and multi-threaded performance with lower power consumption. The G5 is also 64 bit whereas the 8641 is 32 bit, Freescale do have a have a higher clocked 64 bit CPU in development but no details have been announced as yet.
Macs aside this CPU is mainly targeted at embedded applications so it’s presence for the most part will not be visible. It will turn up in some desktop systems though, for general consumer computing the existing G4 provides more than enough computing power so a pair of them at a higher speed with much faster memory should provide quite a potent system for that market.
There has been no official announcement but Genesi have said their future products will track Freescale CPU development and my information is they do indeed plan to use this processor at some stage in the future.
Conclusion
The 8641D may not get the single threaded performance of high end PC chips but it’s not designed for that, having said that this CPU will probably deliver the majority of that level of performance even if it doesn’t quite reach 100%. The clock rate is highly deceptive as CPUs spend much of their time sitting around waiting for data so raising the clock rate doesn’t necessarily translate to an equivalent level of higher performance. The Itanium II CPU (also at 1.5GHz) with 6.4GBytes / second memory bandwidth and a huge 6MB cache has been studied and even when running a highly optimised benchmark is doing absolutely nothing for 50% of the time – and that’s one of the world’s fastest CPUs (i.e. it’ll happily eat any x86).
On the desktop bragging rights matter, they are completely irrelevant in the most of the embedded world if you can’t perform in a low power budget. An extra 10-20% performance isn’t so as important as to be worth the additional effort and cost. A dual core PC CPU may be faster than this on raw computing speed but don’t expect to see them in many embedded systems any time soon.
The already high power consumption of single core PC processors means manufacturers will have to lower the clock of the CPU cores otherwise a pair will use far too much power and end up approaching 200 Watts, AMD are doing exactly this with the dual core Opteron, they will run 3 or 5 speed grades below the top single core Opteron. Freescale have a lot more headroom so don’t need to do this, they can run both cores at full speed and still come in at less than half the power consumption of other PC CPUs. This problem will also affect the G5 unless IBM can lower it’s power consumption, currently a dual 1.8GHz G5 should be possible at under 100 Watts (Maximum).
It’s clear to me in some respects that the embedded and desktop markets are converging in this CPU, Freescale have been shipping highly integrated CPUs for years now but they’ve never appeared anywhere near desktop systems. PC CPUs wont be able to race away from embedded CPUs as they have over the last few years, if anything I think the gap will tend to close as everyone starts using multiple cores.
The 8641D and other family members are targeted at a different market from PC CPUs so direct comparisons are not terribly meaningful unless you’re interested in using one as a desktop. PCs are upgraded at a rapid pace, the current top end is often a few notches down within a few months if not sooner, within a couple of years they’re not even being made any more. In the embedded world you expect these parts to be on sale and in use for many years. I bet there’s not many people still using 10 year old CPUs in their main system, would many of those systems even work now these days? The 8641D is expected to be around for a long time – it’s got a reliability rating of 10 years at 105¡C.
What if?
Freescale didn’t need to drop the speed of their cores to put two on a single die but if they did they could build some very potent processors. Dropping the cores to 1.4GHz should allow 4 cores to be used and still remain under 50 Watts. They could even build an 8 core device if they wanted and still stay under 100 Watts, it would outgun every PC CPU on the market several times over and still not use as much power as a P4 Prescott!
Perhaps the criticism levelled at the G4 should have been sent in another direction…
© Nicholas Blachford, October 2004
About the Author
Nicholas Blachford lives in Paris. He is currently helping out on the Yoper Linux disto, learning French, Python and dreaming up a GUI for advanced consumer entertainment systems, but not necessarily all at the same time.
If you would like to see your thoughts or experiences with technology published, please consider writing an article for OSNews.
Thank you for giving us this info about Freescale CPUs.
But I think your comment below is wrong. The part about “in contrast”, you make it sound as if PC are crap and are in the dark ages.
Please explain what you mean.
The CPU performance will increase significantly due to the inbuilt memory controllers and clock speed boost (no speeds have been quoted yet other than “at least 1.5GHz” but 1.7GHz seems to be expected) but more than this the overall system performance should increase considerably. This is in contrast to PC designs which go all out to boost CPU performance and leave the rest of the system lagging behind. The new PCI Express will give additional bandwidth but the current 11 year old PCI bus will just about handle a single Gigabit Ethernet port, never mind four!
with Intel and AMDs recent focus on hih clock rates, it has been impossible for them to put anything ese on the die except certain normal components which ends up adding to latency of device access.
this new proc is state of the art. is is 90nm and basically integrates all the system controllers that you would see on the motherboard into the chip which makes device access much much faster.
Although, considering where I am, I’m not suprised. Let’s get a fifteen year-old single-user OS and put it on new hardware! With that line of reasoning, he should be nostalgic for OS9, too. If he didn’t notice, OSX has good SMP support, too.
I’m excited about what Freescale is doing with their PPC’s. IBM is also rumored to be working on a low-power PPC based off the old 750 (G3) core. While these aren’t 64-bit chips, they should be very nice in Apple notebooks, as well as in embedded applications. If we see a 64-bit PPC notebook, I’m more inclined to think it’ll be a step up from one of these processors, instead of a stripped-down Power4. They’re just totally different beasts.
I have always loved my g4. It is just a well designed chip that works perfectly for everyday tasks. For an idealist, it perfect. For more pragmatic types, just get the most power you can (x86). I think the author of this article is an idealist like me.
It is good to see it getting a new lease on life with the upcoming dual core chips.
A good article.
Raises the bar for other OSNews writers.
Thanks
Nicolas in his article said:
> There has been no official announcement but Genesi have said
> their future products will track Freescale CPU development
> and my information is they do indeed plan to use this
> processor at some stage in the future.
There has been a semi-official “announcement” at the last SNDF in Frankfurt. THe PPCNUX report in fact says:
“Pegasos III is not cancelled but will come with a different CPU. The close cooperation with Freescale made bplan choose the new SoC G4 with DDR2-667 support and PCI-Express to base their next-gen system on.”
http://www.ppcnux.de/modules.php?name=News&file=article&sid=4208
Kind regards,
Andrea
Although, considering where I am, I’m not suprised. Let’s get a fifteen year-old single-user OS and put it on new hardware! With that line of reasoning, he should be nostalgic for OS9, too. If he didn’t notice, OSX has good SMP support, too.
Huh? First of all BeOS isn’t 15 years old. R5 is in fact about 4 years old. And it is in fact in many ways a modern OS. OS9 doesn’t even compare to it.
If BeOS is 15 years old, then how old is WinXP or even OSX? Why would you put that old crap on new hardware?
It’s obvious that you’ve never tried BeOS or understood what it’s about. You just want to feel important.
Well, modern boy, read up or shut up.
so… they’ll have dual-core chips in 2H of 05 ? But AMD&Intel will have such in 05 too …
Elswere (I don’t remember where) I read that Apple’s adoption of the dual core chips for notebook was improbable due to the fact that the pin incompability with present motherboards would require a new motherboard design (mainly a new chipset).
This seems to be in contrast with what the author says.
Does anybody know something more precise??
Thanks
A bit negative at Intel I reckon:
“The Pentium-M CPU has been praised for this achievement, a rather odd situation considering the criticism the G4 got; with the Pentium-M Intel is following the exact same strategy the G4 designers have been implementing all along – with much the same results”
Well that’s hardly fair – the Pentium-M’s are well faster than the G4’s – there’s no 1.7GHz G4 part so you can’t really call it the same result.
And the big difference is a Pentium-M is a dedicated notebook chip; it’s meant to be power efficient but not the fastest possible. Whereas the G4 was powering Apple’s desktops as well, which is where it really had problems; they weren’t so far behind in the notebook line.
The article says:
“The Itanium II CPU (also at 1.5GHz) with 6.4GBytes / second memory bandwidth and a huge 6MB cache has been studied and even when running a highly optimised benchmark is doing absolutely nothing for 50% of the time – and that’s one of the world’s fastest CPUs (i.e. it’ll happily eat any x86).”
Please state the source of the benchmark and the TYPE of the benchmark. Was this highly optimized benchmark CPU bound? I very much doubt that. You have written an incorrect generalization that a 750 Mhz Itanium 2 would do just as well as a 1.5 Ghz one. I don’t think Intel engineers are that completely dumb to produce the higher model.
Well, I don’t know what I’m talking about and have no sources but based on the article I would think this new CPU’s design is so radically different it would require all new designs for all surrounding components, begining with of course the motherboard.. the fact of it supporting ddr2 vs. ddr1 alone would require a new motherboard as it’s pin incompatible. Also, since this thing is near a system on a chip (SoC ™ ) half of the components that were on the support chipset (north/ south bridge ) such as the mentioned memory controller (on-die ala amd64) are now on the cpu constituting a full system redesign anyhow. Either way this looks like it’s a bit down the road and wouldn’t apply to anything but a total redesign.
Hope that was at least a bit helpful.
I agree – what benchmark is this exactly? It’s trivial to produce a benchmark that will utilise the CPU to practically 100% because it requires very little I/O. This could be considered “highly optimised” for that purpose – even if it’s probably pretty useless in real life.
If you’re going to complain about I/O times and do benchmarks on “real life” situations, like a loaded database server or mail server etc, it’ll be massively *disk* (not memory) I/O bound. That’ll throttle any reasonably quick chip, hdd speeds being what they are – and this chip won’t do anything to change that. I’m not sure memory bandwidth is *that* much of a factor – not as much as they’re trying to make out.
Also, since this thing is near a system on a chip (SoC ™ ) half of the components that were on the support chipset (north/ south bridge ) such as the mentioned memory controller (on-die ala amd64) are now on the cpu constituting a full system redesign anyhow.
This has made Apple’s job easier and cheaper
It would be safe to say that with a large cache such as the cited Itanium, memory latency is much less of a system bottleneck compared with this new G4.
But then, the Itanium comes with a hefty price tag. I don’t know the price of those new G4 but I guess it would be unfair to compair the two, and the article should never have mentioned Itanium 2 in this way.
It could be very powerfull although I doubt that AltiVec or Floating-Point performance is going to be impressive: quite often CPU for embedded market reduce the number of gates for these function as they are not critical..
And the FPU and vector-units are areas where the gate count matter..
Huh? First of all BeOS isn’t 15 years old. R5 is in fact about 4 years old. And it is in fact in many ways a modern OS. OS9 doesn’t even compare to it.
R5 would be about four years old, yes, but so is OS9. And, yes, OS9 does compare to BeOS — an old school single-user OS with good multimedia performance.
If BeOS is 15 years old, then how old is WinXP or even OSX? Why would you put that old crap on new hardware?
WinXP and OSX are under active development. XPSp2 is only about six weeks old, and OSX is about a year old. A new version of OSX will be out in a few months, and a new version of Windows will be out in 05 or 06.
It’s obvious that you’ve never tried BeOS or understood what it’s about.
I do understand what it was, and why it was important. But it’s not relevant today. A single-user OS just won’t cut it today. The multimedia performance back in the day was impressive, no doubt. But it fell short in other areas (network stack, internet applications, security), and a nice, robust multiuser OS is what most people want and need today. I loved OS/2 — OS/2 doesn’t really cut it these days, either.
I fail to see how a G4 is going to outperform, or even be in the same performance class as available P4 processors when they are launched. Adding a bunch of bandwidth isn’t going to make a processor itself magically fast as the author seems to claim. There are other bottlenecks at work here. After all, the chip is still a G4.
Read all about it here:
http://www.genesi.lu/olp.php
A few milestones along the way…
http://www.genesi.lu/press.php?date=20040930
http://www.genesi.lu/press.php?date=20041004
http://www.genesi.lu/press.php?date=20041009
http://www.genesi.lu/press.php?date=20041011
And, how about that quote: “The PowerPC™ architecture offers many benefits when combined with open source software. As an Avalanche member we look forward to the progressive development of this resource in our quest to reduce our total IT cost of ownership,” said Andrew Black, CIO of Jostens.
There is more to come.
That was a great article Nicholas. By design Freescale is insuring pin to pin compatibility increasing the speed at which each generation of CPU can find purpose in hardware designed to do something. Speed to market is the key and we are finally getting in a position to leverage that. The G4 (or G3) and the PegasosPPC is 90% a terrific story and 10% not G5, not 64 bit, not XXX certified yet, not available in polka dots. There are many customers where the 90% will provide 100% of their solution. We think there is a market for the low power, low cost, small footprint this package can offer and we are about to really attack it.
R&B
A few more tidbits:
http://www.freescale.com/webapp/sps/site/overview.jsp?nodeId=018rH3…
http://www.terrasoftsolutions.com/realworld/showcase/oem/freescale/
http://www.gentoo.org/news/en/gwn/20041011-newsletter.xml
http://solutions.mysql.com/other/hardware/?item=6
Genesi is a Certifed MySQL Partner. There is information on SONY (Europe) using the Pegasos Server here: http://www.genesi.lu/press.php?date=20040922
And, plenty of good info and discussion here:
http://www.ppczone.org
and here: http://www.morphzone.org
Thanks Eugenia…;-)
R&B
The 7448 is said to be pin compatible with the 7447(A) currently powering the iBook and Powerbook ranges. Hence Apple will release updates to these product ranges as soon as there are supplies – expect them by Christmas!
The last (IBM) G3 in an iBook was low on power consumption, with larger cache and higher clock speed than the subsequent iBook G4 (Motorola). However, it lacked altivec and was phased out as Apple moved its range to G4 and above.
Freescale have raised the bar with the 7448 – greater cache still, smaller die size etc. Now if only Apple would release a 12″ widescreen notebook – It’d blow the Dell 600m out of the water.
The newer chips will require a motherboard redesign but will extend the life of 32-bit PPC for a couple of years. After all, isn’t double the battery life more important than raw performance for most? I know which I’d prefer…
How Apple handles the transition to an entirely 64-bit platform is another matter. They’ve managed to pack a G5 into an iMac but no signs G5 ‘books yet due to battery life/heat/noise etc. I’m sceptical an underclocked G5 in a notebook would give much of a performance boost over a tried and true 7448 evolutionary chip. Particularly if Apple managed to use a 200Hmz bus…
WinXP and OSX are under active development. XPSp2 is only about six weeks old, and OSX is about a year old. A new version of OSX will be out in a few months, and a new version of Windows will be out in 05 or 06.
True. But WinXP SP1, which in your perspective would be considered an old OS is being installed and used on brand new hardware every day.
Heck, a lot of people even install Win98 and Win2k on thier new machines.
I just don’t get why new hardware would require new software if there’s still drivers available.
If I upgrade my hardware I do so because I want better performance. If I upgrade the software I do that because I need those features or fixes.
a nice, robust multiuser OS is what most people want and need today.
Really? Most people I know doesn’t care about multi-user features. Most poeple I know have all their users set on admin level. People want a multi-profile OS. They couldn’t care less if it’s multi-user.
Besides, we are talking about a niche OS here. How much need for a multi-user environment is there on media workstations?
WinXP still doesn’t cut it for me when it comes to audio production. And I would need new (expensive) hardware to be able to use OSX. And linux is just a joke when it comes to media.
BeOS, or rather the idea of BeOS is still relevant today.
And as you probably know, I’m not the only one with that opinion.
go motorola!
/me loves motorola
BeOS is plenty relevant. I still use it as my main OS today. Most people couldn’t care less about having a multiuser OS. I know I don’t. If I am the only one using the computer why would I need a multiuser OS?
Anyways, what you posted is your opinion.. And I won’t hold it against you
Can I just mention that the use of the Itanium as a high end example of workloads is exceptionally poor, since the Merced design intentionally passes intricate processing onto the compiler. This is intended to lighten pre-processing on the processor itself, but it does cause issues with compiler development.
In short, Intel have taken almost 15 years producing the current compiler, and it is a long way from being sufficient. It is, however, relatively good for certain tasks, especially floating point maths. If you compare a G4’s FP performance to an Itanium, the Itanium will smack it about something rotten. But the Itanium is not a good desktop chip, and looking at constant die activity is not a good benchmark of performance in any way.
If you want to compare a G4 with a desktop processor, use a PIII, P4 or Athlon XP. They’re all desktop 32bit chips of a similar age. The G5 should be compared with an Athlon 64, or an Opteron, since they are of similar cost.
“WinXP still doesn’t cut it for me when it comes to audio production. And I would need new (expensive) hardware to be able to use OSX.”
I do audio work on OSX(866P4 powerbook) and XP(workstation). please tell me how audio production on a XP box sucks.
the audio hardware works the same under both platforms, and with exception of apps such as MOTUs Digital Performer, I can’t find any production apps I need/use that i cant use on either platform.
I have no stability issues with either platform.
hell, i usually record multitrack and mixdown on the powerbook, but master on my workstation.
please leave unfounded, baseless claims to the trolls please. . .
Well, I have to reply to the BeOS comments first of course.
Stating that BeOS is irrelevant today is a sign of serious ignorance. First of all, ex-Be engineers currently work at Apple, and one can feel that when using OS X. Secondly, Microsoft, Apple and Linux can only dream of the performance BeFS delivers– Apple and MS are pouring lots of money into WinFS and Spotlight– which both, in the end, are just exttra layers on top of existing filesystems. All the functionality given by WinFS and Spotlight are available already to BeOS users (including me) all over the world for years. And thirdly: no operating system in the world can deliver the speed and responsiveness of BeOS.
BeOS is clean, and designed as a pure, 100% desktop operating system. That’s its strength: it doesn’t pretend to be more than that it really is.
Anyway, back to topic:
Nicholas, this was a great article, well written and very clear. The new FreeScale CPU’s are definitely something to pay attention to.
It would indeed be great if Genesi could use these CPU’s for their new systems. Pegasos/MorphOS is something I’d really like to have, and if they were to use these new CPU’s in the future, the justification of buying a Pegasos-system is a lot closer. For me, that is.
I wasn’t comparing the Itanium II and the G4, I mentioned that an Itanium benchmark (actually one from the SPEC suite IIRC) had been analyzed and it was found the CPU was waiting on RAM something around 50% of the time. This is true for all CPUs but the fact this was measured on an Itanium II indicates that despite huge chunks of Cache and RAM it still has to wait on memory.
The Itanium and G4 are for totally different markets so there’s no point comparing them. I mentioned this to illustrate a point.
“Freescale have raised the bar with the 7448 – greater cache still, smaller die size etc. Now if only Apple would release a 12” widescreen notebook – It’d blow the Dell 600m out of the water.
”
They will launch a widescreen 13,3″ PowerBook..
http://www.macosxrumors.com/index.php#09
It’ll be similar to this vaio:
http://akiba.sorobangeeks.com/revue_23.html
Expect it in late November~December (2004), in january 05 if the WiFi Forum doesn’t finalize the 802.11n standard at time. (Airport Ultra will be present in next iBooks & PowerBooks)
Probably, Dell will launch their own 13,3″ notebook also..
Dual cores were a gimmick from the start. Its cool and whiz-bang and all that but the majority of the newer software i have on my P4 1.4Ghz HT Windows XP machine is single threaded and the the P4 HT processors have been out how long? The major problem with the PPC architecture is the snails pace that they release new processors, by the time this processor comes out the P4 will be up to at least 4Ghz and i am sure the Athlon 64 will be on par also, so who cares about a year old CPU design dual cores or not. Prime example the G5 mac was the fastest, most powerful PC in the world for how long? a month?
Thanks for this excellent review of the Freescale core. I’ve been looking for a good bunch of information on this cpu for a while now and this is exactly what I wanted. Thankyou.
On a completely unrelating note when will we stop spending so much time and effort developing CPU’s that provide only a few seconds of speed faster access times…. when hardrives are the main bottleneck of every system. We have been sitting at ATA100/133 for how long now???? Serial ATA looked promising but hasn’t got much in terms of real world performance difference. Why are we not seeing faster hardrives? Surely if they can cram more speed in something as small as a processor they can make hardrives a lot faster! And it’s not even the hardrives so much as the bus that goes to them. ATA100 was fast 4 years ago… now it’s just LAG.
Concerning harddrives, I think we might see that in the future the OS and often accessed data will be on Flash-like cards, while other data such as photos and documents and whatever will be on conventional harddrives.
This would also push developpers to create more compact and better written code for their OS.
Read up on NCQ.
Next version of sata is on its way soon anyway.
What about a hard drive with double the amount of drive heads? Position the other set 180 degrees around the other side of the disk. If this causes problems then spin the drive half as fast. Then your rotational latency stays the same, but the drive itself will be able to perform two reads/writes at once.
We were promised faster HD access with SATA already… supposedly 155Mb/sec. Real world performance gains? Mimimal.
I haven’t read up on NCQ (thx for the heads up) but with minimal gains in the first round of SATA will this be just more hype? 64bit Processors have been overhyped too. Having 64bits of addressable memory is great, but not many programs utilize this increased address space so the gains again are minimal. It’s all just a big marketing gimic. I suppose a few years down the road we’ll have lots of software that utilizes it, but right now it’s just a reason to spend you hard earned dollars.
Here’s some great info on ncq
http://www.cooltechzone.com/index.php?option=content&task=view&id=4…
You are right, HDs are the real cause of every OSes dreadfull perf when it comes to working with Gigs of files, but the bad news is there is nothing practical/cheap on the horizon, not yet.
You can get HDs about 2x as fast as your typ avg $ specials, ie 15K RPMs but they run very very hot, less Gigs, low life expectancy etc, big $. My brother used to work for 1 of the HD companies, I picked up a ton of useless stories about how they work. The world should be pretty gratefull they even have HDs, but they are mostly only pushing up on the density curve. The HD industry like the cpu industry are all stuck inside their own little rat cages competing on just one pt.
Flash could/should be an interesting option, right now flash drives even with USB2 are stuck at about 5Mbytes/sec because the market they are aimed at doesn’t really need more.
But the eeprom nand device could quite easily be configured for many * bulk speed improvement as it is solid state, but it would make for a pretty expensive drive. Theres also ram drives, also expensive.
Anyway pretty good article Nick, my opinion of Freescale just turned around. I usually forget about companies that change names.
BeOS, whats to be said, single user is all I need when I use it. Its just as modern if not more so than the W2K I usually use since W2K has tons of things I don’t want and not much I do want except the apps. Would be neat to see Haiku on this Freescale dual though.
JJ
I love BeOS but I wouldn’t really claim “And thirdly: no operating system in the world can deliver the speed and responsiveness of BeOS. ”
Risc OS and Amiga OS are horribly fast too
I guess on ppc and x86 BeOS might be the best responsive but AmigaOS 4 will probably take over that position from what I can tell on the PPC side 
oh well, long live beos!
so… they’ll have dual-core chips in 2H of 05 ? But AMD&Intel will have such in 05 too …
The big difference is that the G4-based CPU includes a memory controller, a PCIe controller, a couple GigE controllers, a couple RapidIO connections, and a graphics bus controller. None of the AMD or Intel CPUs include that much. The Athlon64/Opteron includes a memory controller and a couple HyperTransport links, but none of the Intel parts includes anything beyond the CPU.
These are getting close to systems-on-chip. All that’s really missing is a storage device controller. These would make great little laptop / palmtop / notebook CPUs, as well as 1U or blade server systems.
R5 would be about four years old, yes, but so is OS9. And, yes, OS9 does compare to BeOS
…but only if you use disingenuous, arbitrary criteria like:
— an old school single-user OS with good multimedia performance.
That doesn’t really deserve to be dignified with a response, but by most of the criteria that the modernity of OSes is judged by, BeOS and OS9 are vastly different beasts.
One of them has protected memory, pre-emptive multi-tasking, partial POSIX compliance, and doesn’t require apps to be specifically written to support SMP. The other is OS9. BeOS may not have all the bells and whistles of other modern OSes, but it certainly has much more in common with NT, Linux, OS X, et al, than it does with classic MacOS.
its funny how you seem to imply that macs invented on-chip memory controllers and pci-express. Yes most PCs still use the older technology, but i do believe amd has the onchip memory controller for a while now, and pci-express is a new standard created mainly by…….pc companies. And intel is already having pci-express chipsets out.
Nicholas Blachford should have stuck with writing an informative article. Instead his info is so tainted with opinion that it throws doubt on the believeabitlty of this piece.
He keeps comparing the new offering from Freescale to the Pentium 4, the Pentium M, Athlon 64 and Opteron. He even drags the Itanium out of the grave to give it a kick or two. The problem is he forgets to compare this chip to the one CPU it will be competing against. The only other chip Apple uses in its product line, the G5 from IBM. That chip has very similar strengths and weaknesses of an AMD Opteron. The architecture is similar, they both run 32 and 64 bit apps natively, they both use hypertransport and have onboard memory controllers. In unbiased benchmarks they compare very favorably against each other. Using the authors logic (which has no benchmarks to back it up just conjecture) if the Freescale will beat anything AMD and Intel have to offer, then it should completely dominate the G5 as well.
This is important because IBM and Freescale will both be selling CPU’s to Apple. Can you imagine the outrage with Power Mac users if the new notebooks performed much better then their giant boxes. G5 chips would dissappear from the Power Mac line very quickly.
Nicholas also repeats the lie that a dual core CPU is twice as fast as a single core. Even under the best circumstances a dual CPU solution can only give about 1.5 times the performance. With dual core CPU’s it is even less, you might see on average around 30% gains. Nothing to sneeze at, but still a far cry from twice as fast as the author suggests.
Next time Nicholas leave the bias at home.
Yes I agree the stupid spining things slow all my systems down. It is not the interface but the performance of the mech. I mean no SATA drive exceeds the performance of ATA133. Why you would think a slow mech on a faster bus would be faster I don’t know. Many of the these big SATA drives are actually designed to be big and quite not fast at all. And note book drives are horrible.
If you don’t understand why having the system controller running on the same die as the cpu at the same speed will improve performance more the MHZ you should read more about computer systems. Most of what I do is wait for disks to spin. Or wait for the graphics subsystem. Or the memory subsystem. This chip addresses both of the these issues well.
I would love a notebook/tablet/thin client with one of these. You could build a very small motherboard that would handle alot of functionality.
I said that XP doesn’t cut it for me. And that’s not an unfounded baseless claim, because I should obviously know, since I am me(like it or not).
It does cut it for a lot of people, but not for me and some other people. Some of those other people are using macs, some of them (including me) wish for a third alternative.
It’s a personal opinion. And I could tell you the reasons why it doesn’t cut it for me. But I sense that you really wouldn’t care anyway.
getting back on top? upcoming?
are you kidding me?
they’re reasonably saturdated in network processors. thats not a small feat at all; its a very good place to be.
on the other hand, they are developping some HORENDOUSLY badass solutions for other markets. The ColdFire just gained a mmu and would make an astounding mobile processor. Really. The desktop chip in this article is very well integrated and has recoculous memory bandwidth. I remain skeptical about its scaling abilities, i dont think it was designed for massive scaling. Dual core is one thing, to say “just add two more” is naive, especially for a mobile platform where the ability to even utilize two cores is in question.
the g-enesis project is interesting. it seems like they’re trying to play two different games though. they’re trying to argue taht a single platform will simplify network setup. However, a single software platform does not mandate a single hardware platform. Particularly where linux is concerned. A deployable pure linux platform would be equally effective, whether or not you have ppc, x86 or a mixed environment. asuming everything cross compiles…
The real killer is that hte US distributor for this platform costs the mobo + cpu at $775. I could get a dual channel A64 of some blitzingly fast variety for that sort of damage. The processors themselves are cheap, but the development costs v. the market space are not in the platforms favor.
This post wasnt supposed to be so negative, I like PPC, and genesis is kind of cool.
I really do need to get a good readup of PPC as an instruction architecture. Altivec does worry me. IBM is clearly very ho-hum about the whole thing as illustrated by the large latencies and lack-of-inclusion on Power.
ps: in my last post, “they” meant freescale. sorry.
Everyone is biased to one degree or another.
I was comparing the different approaches of the x86 companies to that of Freescale – If I didn’t I may of well just written a press release.
I did mention the G5 as I said the strengths are going to be in different areas, some existing benchmarks show the G4 as performing much the same as the G5, at the same clock speed others show a wild difference, the G5 is much stronger on (64 bit) Floating point work.
For going dual CPU (or dual core) the scaling depends on the applications and especially the OS. This has been shown in DP Mac benchmarks where the second processor can get very close to doubling performance in some instances (see http://www.barefeats.com).
As for the Itanium, see my comment earlier, it was used to illustrate a point.
You say you were comparing the different approaches of the x86 manufacturers to Freescale, I saw that. I also saw that you were cherry picking, indivudal weaknesses from the full line of x86 processors, and glaringly omitting the same or similar weaknesses of the original G4s and IBM’s new G5s from your comparison. If you want to compare the fact that the latest P4 uses 100 watts and the Freescale single core only uses 10, then let us continue that comparison and see how the rest of the abilities of the two chips match up. Since the intended market is probably notebooks, let us compare the chips destined for that market. Instead of a notebook and a desktop or even a server CPU.
As for your claims that you’ve seen benchmarks in which a dual processor Mac “almost double” (what happened to double?) the performance of a single processor Mac, I hope you can hear me scoff from where ever you are. The overhead of communication between processors erode a good chunk of the potential advantage of DP systems. I guess that a synthetic benchmark which could run within each individual CPU without having to communicate might get a better than 50 percent increase. In real world applications using current hardware this increase isn’t likely. Still I wanted to make sure I wasn’t completely mistaken,. so I went to barefeats.com (your suggested website) and I could not find any benchmarks to support your claim. I did find a page where he confirms what I have been saying however.
http://www.barefeats.com/fcp4.html
Basically he states that DP G4 Macs are about 45 percent faster than SP.
There are no benchmarks to compare single or dual processors G4s to dual core Freescale “G4s” at this time. When they finally do launch then they can be tested and the performance evaluated against their PPC and x86 counterparts.
I have been awaiting more information about these chips ever since I heard they were announced. Your piece was the most comprehensive overview I have seen so far. Still the anti-anything-but-mac tone was nauseating. In the future I hope you realize your audience at OSnews isn’t limited to those who buy the same computer as you.
Could you imagine Amiga OS on one of these processors, holy cow. OS 4.0 is beyond responsive now, I saw it on a G4 machine.
You haven´t looked enough, see http://www.barefeats.com/g5.html
I wonder, Apple handheld, running a graphicly-scalled-back OSX.
but, I feel like I should say SoC is not necessarily a good thing. SoC is really good for some reasons: It reduces BOM cost; It reduces board size; It reduces board complexity, but it’s really bad for somethings: Larger chips = lower yield; Larger chips = more spins of silicon; More integrated = harder to upgrade (This one is particularly important because, one of the issues facing AMD right now is that DDR2 is faster, and northbridge manufacturers cannot simply create a new northbridge that supports DDR2 because it’s integrated into the CPU die. AMD had to respin their whole CPU and that’s a really big deal);
I can’t imagine that this new G4 chip could possibly have parallel interfaces to all of the peripherals dispite their being on die, so I recon it’ll suffer from most of the bottlenecks of the non-integrated CPUs.
On another note, for mobile purposes, pushing the technology isn’t necessarily a good idea. I can tell you right now that 90nm is BAD for mobile. It’s really really really leaky and suffers from all kinds of power consumption problems. Notebook chips need to keep a low transistor cound and a small die size in order to reduce leakage. The problem with leakage is that you can’t gate it off. If you’re using 0% of your CPU cycles and you’ve turned off everything (gated clocks and whatnot) leakage is still there. For laptops where the computer might be sitting idle for long periods of time it’s super important to get rid of leakage. This means larger processes (higher dynamic power, but lower leakage) and fewer transistors (cutting down cache and FPUs and such).
In a typical desktop you usually have single-threaded programs, which means that you will almost NEVER be able to drive your System into 100% processor utilisation.
If you have a single core system you will have 100% CPU utilisation when you start up a program (for a very short time). approximately 20-30% is used for system tasks, the rest is for the application at that point in time. So effectively you could speed up your “program start” by 20 – 30% by using a dual core (where each core is clocked at the same speed as the single- core). If you have to clock down by 20-30% you are on par with the single – core system.
However if this new G4, run as single core has maybe 70% of the raw computing power of a P4 it might end up being as fast as the P4, when run in a system maybe even faster due to better memory bandwidth and latency.
However scaling this scenario up to an 8-core chip is not realistic (at least not on the desktop) because you cannot get by the single-threadedness of the application programs.
You would need to run multi-threaded software to get the advantage a multcore system offers.
Additionally not all application software although multi-threaded does not scale up to 8 processors. At the company I work for we run structural stress analysis programs which typically scale up to 4 processors, so the performance question is: What is the fastest machine with maximum 4 processors? Is it a 4-way Itanium or a single core P4 or a 4-way AMD64 or a 4-way G4 or whatever. We would not buy a 128 – CPU Altix machine because we would not be able to USE this computing power.
Nevertheless, 20 W power consumption will be excellent for Notebooks, routers, etc.
This is the most informative and best written article I’ve seen on this site in a long time… well done, Nicholas. Your personal life seems to mirror mine somewhat too, born in the Republic of Ireland, lived in Haarlem just outside Amsterdam, Paris too… good luck with the job hunt.
“Huh? First of all BeOS isn’t 15 years old. R5 is in fact about 4 years old. And it is in fact in many ways a modern OS. OS9 doesn’t even compare to it.
If BeOS is 15 years old, then how old is WinXP or even OSX? Why would you put that old crap on new hardware?”
Not to mention BSD, Solaris, QNX, more or less Linux, HPUX, AIX and more… 30 year old OSes and they are still at the top.
Note that AMD64 has other ISA improvements(i.e. GPR/XMM increase) other than the memory addressing extensions.
Jul 27, 2004:
“IBM’s Microprocessor division is readying a dual-core version of its PowerPC 970 processor, called the PowerPC G5 by Apple Computer Inc. According to sources, the new chip, code-named Antares, will make its way into workstations and servers sometime next year.”
Motorola was either incapable or unwilling to up the G4’s clock speed, as the article mentions, their PRIMARY focus is embedded applications and NOT performance desktops/workstations/servers. Apple was perfectly correct in re-aligning with a powerpc developer(IBM) who DOES design their processors specifically for such applications, or at least workstations and servers. Also, as we see by the above quote we can expect at least dual cores from IBM soon, sayonara Freescale and don’t let the door hit you… [fill in the rest as they exit stage left]
Scaling isn’t something that we can predict across the board. If you go here:
http://www.netlib.org/benchmark/performance.pdf
and go to page 48 and down a ways, you’ll find some interesting things about scaling performance. Some chips, os’s, and compilers are just terrible at scaling. But others are great. Excluding clusters, the worst performer was an Suprenum SICI (2 cpu) at 1.5 times to a Sun Sparc2000 at 2.11 (2 cpu) which did even better with four cpu’s at 4.28.
While those were specialized tests, it shows that there is no real way that we can predict scaling performance until the systems come out.