Slashdot Mirror

On that first option. Has anyone compared the available development tools for linux?

http://www.intel.com/cd/software/products/asmo-na/eng/compilers/clin/277618.htm

http://developer.amd.com/cpu/Pages/default.aspx

and nvidia vs ati

http://www.nvidia.com/object/cuda_home.html#

http://developer.amd.com/gpu/Pages/default.aspx

http://www.nvidia.com/page/drivers.html

http://ati.amd.com/support/driver.HTML

http://support.intel.com/support/graphics/sb/CS-028594.htm
(^Ewww)

Do you want to use this to handle AI?

It depends on what kind of AI you are talking about. Path finding actually maps nicely to the GPU. AMD released a demo that showcases this by running a path finding simulation on the GPU for several tens of thousands of agents. Read all about it in Chapter 3 the Advanced Real-Time Rendering course notes from SIGGRAPH 2008. Demo and screen shots here: Froblins Demo

Do you want to use this to handle AI?

It depends on what kind of AI you are talking about. Path finding actually maps nicely to the GPU. AMD released a demo that showcases this by running a path finding simulation on the GPU for several tens of thousands of agents. Read all about it in Chapter 3 the Advanced Real-Time Rendering course notes from SIGGRAPH 2008. Demo and screen shots here: Froblins Demo

the relationship to other cores and to memory aren't fully 'cross-bar'

In AMD multicore processors and the latest i7 Intel chips with integrated memory (rather than a shared FSB), the system memory is "stacked to the processor" already (one bank of memory per physical chip). You just need to make sure that your working set per core is allocated from the memory that is directly connected to the CPU the you are working on.

FWIW, Windows already has a solution for this. You run each thread on a core with a set processor affinity so it doesn't get swapped around. Then you perform your memory management with NUMA allocations so your working set comes from the memory directly connected to your core.

You will have some memory that is "shared" for communication and other uses but as long as your primary working set is directly attached to your CPU, all should be roughly equivalent to the "stacked memory" custom supercomputer architectures mentioned in the article.

I'm sure that other OS's support NUMA allocation as well.

FWIW, you can further increase performance by "chunking" data into segments that will fit into L1 for inner loops if possible so that multiple cores in a single die don't compete for L2/L3 and Memory Bus bandwidth as often.

Finally, avoid having shared/overlapped data between multiple cores. Especially for "false sharing" which is two separate data elements which are not actually shared but are co-located in the same cache line such that independent accesses to them cause cache updates/invalidations.

Mana,

I agree.

Then eleventypie will need to get another...better laptop, Damn the delay... the Dell is now old shit... paid for new... get new.

Start, What is new/available:
AMD: http://wheretobuy.amd.com/notebooks.html
Intel: http://www.intel.com/Consumer/Shop/core2x_62.htm

Then use PriceGrabber, Yahoo/Google shopping, TigerDirect, NewEgg, Amazon, to find the best deal that will deliver to APO.

For the latest MB&Processors scale to affordable cost with Brand OEM-URLinks.

Yes, AMD's Stream technology. I don't think it is used as much as CUDA in practice.

Soon to be supplanted by OpenCL as being OpenCL compliant.

Yes, AMD's Stream technology. I don't think it is used as much as CUDA in practice.

Intel has been hit with antitrust charges in several countries. I assumed a Slashdot reader would be familiar.

http://arstechnica.com/news.ars/post/20070726-eu-slaps-intel-with-formal-antitrust-charges.html

http://www.amd.com/us-en/Weblets/0,,7832_12670_13242,00.html

http://www.pcworld.com/article/142443/intel_and_antitrust_a_brief_history.html

http://en.wikipedia.org/wiki/AMD_v._Intel

"This is because the Korea Fair Trade Commission has issued a fine of US$25.4 million against Intel."

Several vendors have come forward to corroborate AMD's story, and issue statements against Intel for these cases.

I could keep going, but that should suffice.

Well, games that are writing to your user folder are doing the "correct" thing, according to Microsoft.

http://ati.amd.com/developer/SwedenTechDay/02_Preparing_Games_for_Windows_Vista.pdf

It's not just about sharing - it's about security practices as well. If you are not running with Administrator privileges (which normal users are *supposed* to be running at), then these user directories are the only legitimate places an application can freely write to. In Vista, writing to the program files directory will be virtualized and shunted to a user-specific location anyhow. Games today still can't ignore running well on XP, but neither can they ignore the requirements of Vista.

Unfortunately, Microsoft has been pretty inconsistent about the recommended location of these types of files in the past - it's a bit of a moving target - as the inconsistency of your game saves shows. And unfortunately, Windows users (your post illustrates this exactly) have been wrongly trained to expect user-specific data to simply be written to a subfolder under the program's install directory or something like that (because this mechanism just hasn't been used or enforced on the Windows platform previously).

I empathize with the "messiness" of it, though. Incidentally, are you aware that you can point your "My Documents" folder anywhere, even a separate partition, or a common folder on your C drive? That might help you to keep things organized a bit better.

Yer Intel Captains can't do that anyway matey. The kind of bosuns Intel hires are the finest on the seven seas! The finest sailors won't sit on their arses and grind their swords, them kinds like to be up and doing! They like the smell of fresh booty in the morning! If Intel let those people sit, they'd keel-haul the bosses and set sail for new horizons! YARRRRRRRRR

The first time I'm aware of that an FPU was integrated into a common CPU was the 486 DX. In 1989. Hypertransport was introduced in 2001. Incidentally, 486 DX parts that had defective FPUs were sold as 486 SX processors to recover some of the cost that would otherwise have been lost. Later 486 SX processors were made without the FPU, to save die area (and therefore cost).

AMD claims that integrating the memory controller into the CPU greatly reduces latency (improving performance). Everybody seems to agree with them.

So no, HyperTransport is not designed to interconnect things on the motherboard with the speeds and latencies that are available on the CPU die itself. No external bus can do that, now that CPUs are running at speeds where clock pulses can barely traveling across the die fast enough.

Yes, the article advocates moving away from GPUs to something much more general purpose and much more programmable. That will not be a graphics processing unit. Integrating such a complementary processing unit with the CPU might make sense, though it will be expensive. Integrating a high performance GPU does not.

"AMD would lose every single Govt. and Big American company contract the day they do such deal with Chinese govt. Don't forget "AMD gets support from human rights abusers!" trolls too, millions of them, amateur or professional."

Let's then remember that China significantly floats FANNIE MAE and FREDDIE MAC (or, let's make them Frannie Mae/Feddie Mac or Manny Fae, and Meddie Frack...). Now, it seems the South Korea Government will be buying (or is trying to buy/floating the idea that they will buy) Lehman Brothers:

http://www.iht.com/articles/2008/09/02/business/lehman.php

If a European bank were floating this idea, there'd be not much of any uproar. If China were trying to buy Lehman, there'd be an uproar. With Korea, it's something in-between. That, well, is paraphrasing what I heard last night on BBC/NPR/PRI about the situation. So, SK is taking things slowly.

Now, if China bought AMD, would China really HAVE to care/worry that the US government would nix projects? Even if the SEC allowed it (after ordering the purging and dual-use/DOD-related files and dies/machinery from the company's assets/possession) China would STILL have a significant windfall. Actually, they may already have that windfall without having to buy the company...

Anyway, given AMD's financial position (depending on what you read into things), it might make an easier buy for China. Even if AMD ends up in China's sights, India may well step into the fray.

http://www.amd.com/us-en/Corporate/AboutAMD/0,,51_52_484,00.html

Do those laws apply if say the microprocessors are fabbed in Germany/Israel/Ireland, assembled/packaged in Malaysia, and then exported to China?

How about if the microprocessors are assembled/packaged in China itself?

http://www.intel.com/pressroom/kits/manufacturing/manufacturing_qa.htm

http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51__104_543~117787,00.html

Do you consider your CPU good enough? Or if not, which of the zero high performance CPUs with open source microcode do you run?

My CPU conforms to a documented and published interface, which is all that is necessary for it to support free software.

And that's also exactly what's necessary for any system to "work completely" - can't check that something works if its behavior isn't precisely specified.

The conditions necessary to be able to state that a bit of hardware "works completely" are the same as those necessary to support free software - conformance to a published interface specification.

Is it good if, in addition, the hardware is free (as in freedom)? Sure. But that's not necessary for it to support free software.

Because modern video cards have more and more support for hardware encoding and decoding of video. While the support for encoding is only starting to show up, the hardware decoding makes a big difference in previewing HD video in real time with little CPU usage.

Perhaps if the XBOX gets its blu-ray finally I can not only rent Blu-ray films, play games on my HD tv, but I could also stream it, all for a lower monthly than any Cable/Satellite company and only have my equipment in the loop

I've been doing my research for a couple of weeks now. While I've been using SageTV for the past 4 or 5 years after dumping Tivo, it is looking more and more like Windows Media Center might be the best option for alot of folks. Not because it is the best at anything that it does, but simple because of the leverage power Microsoft has to tying other option into their system. I was reading today about thisATI TV Wonder product. It is just looking more and more like you either chose a million devices that hang off your TV, or something with the might behind it to tie in all the various components so you can make a simple, nice media server and extend that to the various displays around the house with an Xbox360 or similar other device.

But after just getting off the phone with my local cable company, they want $4.50/mo per CableCard, and that I just feel is ridiculously high. Why does the FCC always seem to be half-assed with everything they do, always leaving some loophole for the consumer to get screwed over.

The paper is called March of the Froblins

Unfortunately that won't work with envy:

https://answers.launchpad.net/envy/+question/23594

The only way to get a working binary for the older cards is to install an ATI blob from version 8.28.8 http://ati.amd.com/support/drivers/linux/radeonprevious-linux.html or before. A lot of laptops use 9100s, they were dropped along with a heap of other models after this release.

You just have to give up in the end and lose most of your acceleration, or install a old distro from cd and not update it.

They are a hugely popular supplier of mobo components like usb or network ports.

At least their I/O ports are just metal and plastic (with an occasional LED or two). They either work or they don't. Anything with active components, on the other hand, needs to be designed and built by people with a clue.

BTW, google searching for "foxconn" yields the text "Motherboard manufacturer, certified by ATi." - which to ATi, means: Purchasing products with the "Graphics By ATI - Certified logo" means that ATI has certified and tested for product quality, reliability and stability, tested to ensure a good customer experience. . With ATi owned by AMD, and AMD firmly committed to Linux, Foxconn may find themselves in a bit of a bind here.

They are a hugely popular supplier of mobo components like usb or network ports.

At least their I/O ports are just metal and plastic (with an occasional LED or two). They either work or they don't. Anything with active components, on the other hand, needs to be designed and built by people with a clue.

BTW, google searching for "foxconn" yields the text "Motherboard manufacturer, certified by ATi." - which to ATi, means: Purchasing products with the "Graphics By ATI - Certified logo" means that ATI has certified and tested for product quality, reliability and stability, tested to ensure a good customer experience. . With ATi owned by AMD, and AMD firmly committed to Linux, Foxconn may find themselves in a bit of a bind here.

You make a good point to the extent that the TDP is an oversimplification of power consumption. But I think it has value as a proxy metric for this purpose. Intel does in fact, as you point out, use this specification for determining system design for thermal issues. However, AMD uses (or used to) TDP to refer to energy consumption (I realize that the Atom is an Intel, not an AMD product). Now AMD looks at ACP (average CPU power) because it thinks that that metric more accurately reflect real world experience rather than the max figure provided by TDP. However,I think it's also important to note that in addition to giving a max TDP, even Intel provides a minimum or idle mode TDP figure, which I think, supports the contention that TDP is a reasonable proxy figure. If you're interested here is a link to a white paper from AMD further discussing these issues. http://www.amd.com/us-en/assets/content_type/DownloadableAssets/43761C_ACP_WP.pdf

Yep, same here. My upstair room can go up over 85 degrees(F) with a fan and central unit AC running! I have a bunch of electronics in my tiny room. I try to unplug stuff I don't use like TVs and printer. I also try to use AMD Cool'n'Quiet on my AMD systems. I also turn off machines that I don't use, but Linux/Debian box and wired network devices stay 24/7. Same for my old VCR (yes, still use one for recordings). :(

• The OpenSSL MD5 implementation is 14% faster in 64-bit mode than the 32-bit version.
• Ditto for RC4 which is about 50-60% faster.
• I have seen the sequential disk read throughput of an old SATA box jump by +30-40% with a 64-bit kernel, because of the paging overhead of a 32-bit kernel required to access high-memory (ie. memory between 1GB and 4GB).

May I suggest Myths and facts about 64-bit Linux for your reading pleasure ?

I agree there is a subset of game programmers who can take on the roll of scripting. Lua itself and Lua-like clones are facilitating this. But the industry really wants lower level programmers. Guys who can think not in terms of object-think wait()-do()-wait(), but guys who can implement a smartptr system to create a better garbage collection system or guys who plan thread synchronisation rather than just wait and see.

Matt

AMD also gave a lot of help to initially port the Linux kernel to AMD64 processors (they kind of coordinated the whole thing via www.x86-64.org), and they continue to be an important contributor to the Linux kernel (I see many patches related to chipset drivers, SATA, etc). The also contribute to OpenSolaris, OpenJDK, KVM (NPT support, Vista / XP 64-bit bugfixes, etc). AFAIK most of these open source contributions are made by AMD employees from the OSRC (Operating System Research Center) group.

I would like to see video cards slow down clock cycles like CPUs (e.g., AMD Cool'n'Quiet). I don't always play games, 3D stuff, etc. Most of the time, it is surfing the Web, e-mails, newsgroups, watch videos and Flash (might need video card's accerelation for this so speed it up), etc.

Or...you could try actually performing a product search:

http://www.nextag.com/mac-tv-tuner/search-html

There is also this:

http://ati.amd.com/products/tvwonder650/usbmac/index.html

Quite a few products take up only a USB slot, or alternatively, you can find MacOSX drivers for several Hauppauge cards and the like if you want an internal card. I've even found Linux drivers and cards for crying out loud :)

The AMD export compliance specifications are a little easier to find on AMD's site than Intel's. Search AMD.com for "CTP calculations". CTPs (Composite Theoretical Performance) are a synthetic benchmark, but do show the relative performance of various CPUs.

• 1st generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800%5E14415~72730,00.html
• 2nd generation Opteron: http://www.amd.com/us-en/Processors/SellAMDProducts/0,,30_177_4458_3505%5E8796%5E8800%5E15227~118076,00.html
• 3rd generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800~124990,00.html
• and others

The AMD export compliance specifications are a little easier to find on AMD's site than Intel's. Search AMD.com for "CTP calculations". CTPs (Composite Theoretical Performance) are a synthetic benchmark, but do show the relative performance of various CPUs.

• 1st generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800%5E14415~72730,00.html
• 2nd generation Opteron: http://www.amd.com/us-en/Processors/SellAMDProducts/0,,30_177_4458_3505%5E8796%5E8800%5E15227~118076,00.html
• 3rd generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800~124990,00.html
• and others

The AMD export compliance specifications are a little easier to find on AMD's site than Intel's. Search AMD.com for "CTP calculations". CTPs (Composite Theoretical Performance) are a synthetic benchmark, but do show the relative performance of various CPUs.

• 1st generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800%5E14415~72730,00.html
• 2nd generation Opteron: http://www.amd.com/us-en/Processors/SellAMDProducts/0,,30_177_4458_3505%5E8796%5E8800%5E15227~118076,00.html
• 3rd generation Opteron: http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_572_863%5E8800~124990,00.html
• and others

That's a pretty serious mischaracterization. Though Dirk Meyer worked at Intel for 3 years early in his career, he spent the next 9 years at DEC working on the early Alpha CPUs (see here). Many of AMD's other current & former top engineers on the CPU side are ex-DEC people as well (Mike Uhler, Rich Witek, Jim Keller in the past, etc.). IBM is the other main source of top AMD engineers (Rich Oehler, Chuck Moore, until recently Phil Hester, etc.).

There are some top folks from Intel too but they're definitely in the minority.

Notice that there are no Intel people mentioned here.

AMD needs to start developing they're own chipset.

In case you've not been following the news lately, AMD has been producing it's own chipsets for quite some time.
They started back in the K-6 era, and then after acquiring ATI, they built quite a few more using additional technology from ATI.
Their current 790 line of chipset has received good critics.

With CPU's sitting on third party crappy chipsets, the true performance of the CPU will never be revealed.

AMD processors are much less sensitive to crappy chipsets, simpy because the high speed sensitive parts are INSIDE the processor. The CPU has direct access to memory and handles everything thanks to its on-board memory controller.

The northbridge on AMD platform is mostly a glorified bridge between the HyperTransport and the PCIe lanes.
(Not exactly. It is also in charge of providing all the "on-board" features available on the mother board. But what I mean is that it doesn't have a major impact on processing speed and, as long it is an efficient HT-PCIe bridge and provide enough PCIe lanes, it doesn't impact gaming & graphics speed).

The best part is, as the HyperTransport is controlled by a consortium, it is less tighly controlled by AMD and much more open to concurrence and 3rd party builders.

Thus you can find origial ATI/AMD chipsets, as well as chipsets from nVidia (specially for SLI enthusiasts) or VIA (for the low-segment)

Lets take for example Intel, I have not had any luck with any third party chipset manufacturer for Intel CPU's.

Intel is a bad example.
With their processors the memory is entirely handled by the northbridge. The bus between the northbridge and the CPU package is of critical importance. Specially in recent multicore systems, because they are technically separate die slapped together in the same package : They don't talk to each other (much), instead they all communicate (mostly) with the northbridge, as if they were discrete packages on a multi CPU system (with multiple sockets).

Put a crappy chipset (anything that doesn't supports the latest bus speed "du jour" from Intel and/or is bad at memory handling) and you basically have a CPU sitting idle, waiting for data to arrive. This is even more difficult, because Intel often upgrades its bus, but doesn't give licenses to all player in the field (the wikipedia article mentions that ATI/AMD doesn't have a license for the 1333Mhz bus, thus the latest series of chipsets are AMD-only).

That's also one of the minor reason why Intel engineers would probably like to move to QuickPath : they will be putting the CPU and the critical parts of the northbridge in the same package, and leave the chipset only to its "glorified bridge + features" job, thus opening the possibility of lots of Intel motherboard with 3rd party chipsets, without much impact on the performance.

This is specially more important recently. Up until recently, because of the tight relationship between CPU and the northbridge, Intel's chipset have been the single best soltuion (except during the whole Rambus fiasco back in the Pentium III era).

Now, users *have* some time to choose other chipsets. Nvidia is a rather an import player in the market for hardcore gamers. But for example doesn't license their SLI-over-PCIe technology to 3rd party chipsets manufacturer. So either gamers go for the original Intel but lose SLI OR choose to have SLI and thus must use nVidia chipset but maybe won't have the latest Intel bus OR must buy expensive monstruosities featuring both chipsets as the case with latest Intel top-level-hardcore motherboards as recently mentionned on /. .
Also, embed graphics is starting to evolve into hybrid solutions, whereas an onboard grap

Same thing with a monitor. If you are using an old CRT, ok sure the Integrated video is probably fine. However if you have a new professional LCD, maybe it is worth the money to buy a graphics card that properly supports it (for example has enough RAM to run at native rez and has a DVI port).

I get your audio argument, but that doesn't really fly with graphics. Integrated graphics don't have any problems driving large LCDs, and some even have HDMI on top of DVI outputs. That particular chipset easily beats a bunch of discrete video cards on the market, and you won't notice any difference between it and a high-end video card in most cases, no matter what monitor you use. In the audio world, you will always notice the difference between on-board and discrete audio, if you use a good pair of headphones.

> Intel can do it. ATI has promised to do it and now so does VIA. Why is NVidia different?

ATI hasn't just promised, they did:

http://ati.amd.com/developer/open_gpu_documentation.html
http://www.phoronix.com/scan.php?page=article&item=842&num=1

Didn't we already have a 9800? http://ati.amd.com/products/radeon9800/radeon9800pro/index.html

...I've just bought an AMD Phenom 9600 (and an Asus M3A32-MVP Deluxe) and I'm fucking impressed... and I'm not usually easy to impress.
The dead one should be Intel.

--

I'm neither associated with AMD nor Asus.

You also forgot this venture, the Xilleon(TM) Panel Processors. With the advent of digital only TV brodcasts (US) in 2009 there is a huge market for these kinds of processors to make TV's display HD content perfectly on flat panel displays. I would estimate that this market is even bigger than the PC market.

Your funny math makes my brain hurt.

The number of vendors is a horrible measurement. Try variants on for size:

How many CPU variants does one of the two manufacturers currently support? Try over 125. http://products.amd.com/en-us/DesktopCPUResult.aspx?f1=&f2=&f3=&f4=&f5=&f6=&f7=&f8=&f9=& Oh, and that's just for the desktop.

GPU? NVidia has 38 families of chipsets. At ~5 products for each chipset, you're over 190.

What about all of the other hardware on a motherboard? Bluetooth, USB, Firewire, network drivers and modems are some of the largest contributers to OS development overhead/headache; tell me, have you ever tried to load up Feisty Ubuntu using a Broadcomm wireless device?!?

By the way, this doesn't include all of the half-assed components people drop onto their computers like humping dog memory sticks or coffee warmers let alone all of the out of date drivers people have installed on their systems (have you checked your BIOS rev lately?).

This isn't "Microsoft's way of saying 'we don't know what's wrong with Vista,'" it's Microsoft's way of saying, "we're trying our damnedest to clean this up, but you idiots keep pissing in the pool."

I don't pay close attention to the GPU market in general, though lately I've been interested in a few numerical modeling projects that could benefit from high-performance computing. The AMD Firestream 9170 is supposed to be released in the first quarter of this year, with a peak speed of 500 GFLOPS, most likely single-precision, but the beauty part is that it should also support double-precision, the numeric standard for most computational modeling. NVidia's option in this space is the Tesla C870; I wonder whether this move to purchase another GPU line will divert resources away from their number-crunching-first GPUs.

Anyone remember the ATI Rage Fury MAXX? I've still got one in use. It was a monster in its day. Dual Rage 128 Pro GPU's and 64MB of RAM. But for some reason the way that they jury rigged them on one board didn't work properly in XP, so it only uses one. Oh well, still s nifty conversation piece.

At the risk if being off-topic, can anyone explain how Intel is culpable for any of the numerous missteps AMD has made dating back to the acquisition of ATI in 2006?

IMO, Intel still is the M$ of the processor market, at least in home PCs.

Still, if it hadn't been for IBM deciding in the early 80s that having all its CPUs come from one manufacturer was stupid, AMD would never have had access to the x86 chip specifications, and thus would most likely be a footnote in the history of CPUs.

You don't have to take my word for it either: AMD themselves say this:
"1982
At IBM's request, AMD signs an agreement to serve as a second source to Intel for IBM PC microprocessors"

http://ati.amd.com/products/radeon9550/index.html

All current socket AM2/AM2+ AMD processors (Opteron 1000 series, Phenom, Athlon X2, etc) support a maximum of four unbuffered DDR2 memory sticks. All current socket F AMD processors (Opteron 8000 and 2000 series) support a maximum of eight registered DDR2 memory sticks. (You can find this info in AMD's public datasheets).

As of today, unbuffered and registered DDR2 memory sticks of 4 GB or more are extremely expensive because the technology cannot be inexpensively mass-produced (yet). Only 2-GB DDR2 sticks can be found at reasonable prices.

For these financial and technical reasons, your are restricted to a total of 8 GB per socket AM2/AM2+ processor, or 16 GB per socket F processor. Therefore the cheapest option for an AMD mobo supporting more than 8 GB of memory is to buy a single socket F model. Newegg sells one for $136 (open box, though). Add a $180 Opteron 2212 processor, $240 for eight 2-GB sticks of registered DDR2-667, and you end up spending only $556 for a dual-core 2.0 GHz 16 GB barebone server assuming you have a chassis and a PSU lying around.

I'll leave other people comment on your Intel options. I am not very familiar with Intel server motherboards.

Intel's desktop processor page
The Dells have either a E4500 or E6550
2.2 GHz clock / 2 MB cache / 800 MHz FSB
2.33 GHz / 4 MB / 1333 MHz
All ship with only 667 MHz DIMMs

Intel's mobile processor page
The iMacs have either a T7300, T7700, or X7900.
2.0 GHz / 4 MB / 800 MHz
2.4 GHz / 4 MB / 800 MHz
2.8 GHz / 4 MB / 800 MHz
All ship with only 667MHz SO-DIMMS

BTW, if a 128MB 2400XT is a joke, then WTF do you call integrated video?
The highest specced XPS ONEs have Mobility Radeon HD 2400's, memory is unlisted
while BASE iMacs have Mobility? Radeon HD 2400 XT's with 128MB GDDR3
The highest end iMacs have Mobility? Radeon HD 2600 PROs with 256MB GDDR3.

The iMac has better specs, flat out. It most likely uses a lot less power and weighs less also. The XPS ONE is a very well integrated _PC_ for sure, and has other nice features the iMac doesn't. I think Dell did a great job with the hardware integration and bundled features, and it is somewhat on par with the iMacs, IF you leave Leopard and iLife out of the picture anyway.

Personally, those speakers have got to go!

Intel's desktop processor page
The Dells have either a E4500 or E6550
2.2 GHz clock / 2 MB cache / 800 MHz FSB
2.33 GHz / 4 MB / 1333 MHz
All ship with only 667 MHz DIMMs

Intel's mobile processor page
The iMacs have either a T7300, T7700, or X7900.
2.0 GHz / 4 MB / 800 MHz
2.4 GHz / 4 MB / 800 MHz
2.8 GHz / 4 MB / 800 MHz
All ship with only 667MHz SO-DIMMS

BTW, if a 128MB 2400XT is a joke, then WTF do you call integrated video?
The highest specced XPS ONEs have Mobility Radeon HD 2400's, memory is unlisted
while BASE iMacs have Mobility? Radeon HD 2400 XT's with 128MB GDDR3
The highest end iMacs have Mobility? Radeon HD 2600 PROs with 256MB GDDR3.

The iMac has better specs, flat out. It most likely uses a lot less power and weighs less also. The XPS ONE is a very well integrated _PC_ for sure, and has other nice features the iMac doesn't. I think Dell did a great job with the hardware integration and bundled features, and it is somewhat on par with the iMacs, IF you leave Leopard and iLife out of the picture anyway.

Personally, those speakers have got to go!

The key point here is that you can get the best performance/watt around from ARM chips. AMD's Geode series has a 1.5-watt Geode LX900 (600MHz) and a 0.9-watt Geode LX800 (500MHz) (link). Note: AMD's site rates these at higher power (2.6W and 1.8W respectively) here.

ARM chips have always been more efficient than X86 chips and always will be due to CPU architecture and the way that every instruction is encoded. Each ARM opcode has got a 4-bit conditional field that governs whether that opcode is executed or not. In an IC, you've got quiescent power (always there from the moment you switch on) and dynamic power. Dynamic power comes from switching transistors on and off. If an instruction isn't executed, there is less switching and less power consumption.

With a "save the planet through electronic design" attitude, I'd love to see a large proportion of X86 desktops replaced with ARM-based machines. Especially when you consider that saving even 1 Watt per PC scales to many thousands of megawatts , especially when you see how many PCs are in use now.

As ARM CPU speeds increase beyond the point where you can have a modern, complex OS and good office software running at a comfortable speed to the user, isn't that a goal worth aiming for? The practical sides of that dream are daunting. I'd be naive to think that the world will port its software just because it's a good idea to save electricity where possible. A fresh start would be bigger than Haiku in it's ambition. Is it worth it? I'd like to think so. What's 10 years of OS and application development that could make a good dent in global power consumption that would last forever?

AMD makes a whole line of embedded x86 processors. http://www.amd.com/us-en/ConnectivitySolutions/ProductInformation/0,,50_2330_9863,00.html