Two New AMD Mobile Chips Launched

to_kallon writes "Advanced Micro Devices (AMD) has unleashed two new mobile microprocessors today. One processor belongs to the company's 64-bit Mobile Athlon64 line while the other one comes from the 32-bit Athlon XP-M product line. According to CNET News: 'Like other Athlon 64 chips, and Intel's Prescott, the new Athlon 64 3400+ will block many security threats automatically in conjunction with Windows XP Service Pack 2. The delayed SP2 is slated to come out in August. The Athlon 64 3400+ will also run a 64-bit version of Windows, due now at the end of the year.It runs at 2.2GHz and comes with 1MB of cache. Gamer-PC maker Alienware will insert the chip in a notebook later in the month. Meanwhile, the Athlon XP-M 2200+ comes from the company's older line of chips. It runs at 1.6GHz and is built around an older processor core and comes with a 512KB cache. Averatec, a small computer manufacturer, has put the chip into a notebook that can convert into a tablet PC, marking the first time AMD's chips have been used in a tablet'."

Comments

Old Joke

[Apocalypse111]

AMD has also recently released a new multiprocessor motherboard configuration for its low-end processors. Machines based on this technology will specialize in playing 80's MP3's.

They're calling it the Duron-Duron.

Re:Old Joke

[Ari_Haviv]

I met him on a Monday and my heart stood still
Da du ron ron ron, da du ron ron
Somebody told me that his name was Bill
Da du ron ron ron, da du ron ron
Yeah, my heart stood still, yeah, his name was Bill
And when he walked me home, da du ron ron ron, da du ron ron

And don't forget Celine Duron

Re:Old Joke

[dj245]

Soon they will release a new processecor to help compete in the low-end adult computing market. They call it the Sempr0n.

"unleashed two new mobile microprocessors"

[Anonymous Coward]

What a vivid image! I can imagine to "mobile" chips, both in leashes, one pulling, the other restlessly stalking back and forth on silvery leg-pins until... Bam! the leashes are cut, and they bound off into the distance...

Cooling

[growlydog]

With all these new advances in processor and video-card technology, when are we going to see some advances towards cheaper, and quieter cooling solutions? These devices keep getting hotter and hotter!
In order to keep my gaming computer cool I have something like 7 or 8 fans in there, and the box sounds like a jet-engine taking off... I've looked into water-cooling but virtually every water-cooling setup costs upwards of $200. Is it really *that* expensive for the equipment? What other alternatives are out there?
And with a notebook... isn't heat going to soon be a real serious issue with laptops?

Re:Cooling

[Anonymous Coward]

I just bought one of those mini fridges. I drilled a hole in the top, popped in a grommet to help keep air in, and ran the cords through. I don't pay for power so I don't care...

Re:Cooling

[irokitt]

BTX may take care of that on the desktop end, but only if Intel can convince case and motherboard manufacturers to pick it up and run with it-so far none of them are happy with the idea. AMD has not espressed any interest in BTX, but I wonder if they would pick it up anytime in the future. AMD Athlon 64 chips run much cooler now that they have a heat spreader on the chip.

As for laptops, I guess we may have to start looking at liquid nitrogen;)

Re:Cooling

[the unbeliever]

get larger fans that move more air at lower rpm.

reduce cable clutter with rounded cables and zip ties/wire mesh.

Re:Cooling

[alwaystheretrading]

These devices keep getting hotter and hotter!

The Athlon 64 runs a lot cooler than the Athlon XP processor. I'm running a Athlon 64 2800+ overclocked to 3200+ and the chip sits at 38 Celsius (100 Fahrenheit) with the stock heat sync. My current computer is cooler than I've had in years and it's quiet!

Re:Cooling

[StateOfTheUnion]

Do you find that 7-8 fans are really necessary?

I will admit that I don't run the latest and greatest machine (AMD Barton core 2200 and a midrange graphics card), but I slowly started shutting down all those extra case fans and slowed down my CPU fan. I'm now running the cpu fan, the power supply fan, GPU fan, and a motherboard fan that sits on either the southbridge or the northbridge (don't remember). I run this system in an unairconditioned house near the Mediteranean (read:hot and humid) and I've never experienced overheating issues.

Not to say that there aren't systems/users out there that need better cooling than what I have, but I've found that sticking with the basics seems to work pretty well . . . I know a lot of people that drill big holes in their cases, mount gargantu-fans in the case, buy gazillion dollar heat transfer compund, etc. . . but most of the people that I know that do this stuff do it before doing any system characterization.

The fact that its hard to find a handheld device with a remote temperature probe (thermocouple) at a normal computer shop is indicative to me that most of the folks that go overboard on the cooling have not characterized their system before going out and buying the "superduper cooling kit." Does it really make sense to install XXX number of fans in a case before doing any serious temperature monitoring and characterization to identify whether they really need to go to these sorts of extreme measures. (I know, I know, there's usually a couple of thermocouples on the MB . . . but do the third and fourth case fan really have a significant effect on the MB thermocouples, or would it make more sense to also measure temperatures near other heat sensitive components in nooks and crannies of the case that one would expect to have poor ventilation (e.g. a graphics board installed next to another PCI board) . . .for that a probe would be useful to determine which fans where have the most significant effect)

And for that matter what about characterizing the temperature profile after installing the "supercooling solution" to determine whether it made a significant difference.

And for that matter, how much of a difference is really significant anyway?

I almost think that people like to brag about how much cooling their case has whether the system needs it or not "I have 6 cooling fans" . . . "Oh, yeah well I have 7" . . . "oh yeah, well I have a liquid circulating cooling kit on my overclocked system that cost me more than just buying a faster processor would have." . . . do you see the logic here? I don't.

I don't claim to understand it, but I guess its just cool to be cool . . .

Re:Cooling

[Gldm]

Perhaps you've missed AMD's introduction of the HE and EE series opterons, which have rated outputs of 55W and 30W respectively. It seems like low heat output is quickly becoming the new Mhz. When your transistor budget is so much more than what your logic requires, you can always throw another pipeline or even an entire extra cpu core on the die for more performance, but if it's too hot you're screwed.

Re:Cooling

[vluther]

I just bought a desktop replacement AMD 64 3200+ Laptop. It's got top of the line hardware, and has never gotten warmer than say me rubbing my hands together for a little while. It's not the blazing hot like my old thinkpad (p4 2.0 ghz), and it's not loud like my old desktop an amd 2200. It's amazing this much power and so cool.

Battery life may be suckier than a centrino, but this is a desktop replacement, or a portable desktop for me. It rocks.

Re:Cooling

[MindStalker]

I know this is a joke, but seriously I'm pretty sure most computers would product more heat that a fridge could pump out, and as its insulated your really screwed.

Re:Cooling

[bubkus_jones]

One question. Why did you overclock your processor? With such a high power processor to begin with, is it really necessary?

Re:Cooling

[Anonymous Coward]

Asbestos Condoms!

Re:Cooling

[JensLH]

The thing is... Laptops with large fans and less cable clutter just don't sell that well. I dunno why, tho'. :-)

Re:Cooling

[Anonymous Coward]

Heat is becoming recognized as a big issue in server environments, too. You can stuff a rack full of 1U boxes for cheap, but if you have to buy a big HVAC plant to cool it all you just drove up your total cost of ownership.

I'd love to see a bunch of low-power rackmount server boxes, down around laptop heat output.

Re:Cooling

[Epistax]

The solution to cooling is to use less power in the first place. As technology increases were using less power to do the same amount-- however we're doing more with every processor so power consumption is going up. My personal feeling is that I'd be willing for my laptop to lag an entire 1 ghz or more behind the crowd so that it may utilize the latest technology for power benefits over performance.

While perhaps not directly related there must be a strong correlation between energy used and heat dissipated (for a given technology). One thing I'm a fan of is on the fly clock adjustment. I may want my laptop to run without a fan (that is, say 15 dB or less if on) for one to several hours. I don't expect a lot of performance but then again I wouldn't need it.

On a side note I think it'd be really nifty if someone put the original pentium 1 design into the latest technology generation just to see how fast it could be clocked. Maybe also see how many you can fit on one die. I'd say to make it work too but that'd take a few months...

Re:Cooling

[Aphrika]

"...and the box sounds like a jet-engine taking off..."

Have a look at Quiet PC - they have a huge range of components and fans specifically aimed at reducing noise. I have just invested in some case fans myself after (not?) hearing them in action in a friends recording studio.

Re:Cooling

[MarcQuadra]

OOH! It's already here! It's called PowerPC. IBM makes these CPU chips that are zippy, but also generate about half the heat of similar-performing x86 chips.

The PowerPC 750GX is a -GREAT- CPU, it's got a modernized G3 core and 1MB full-speed on-die L2 cache. You can get one soon on an ATX board from PegasosPPC. They run at about 1GHz, but they are pretty quick. Heat dissipation stands at about 8 watts.

The Motorola 7447a and 7457 are also good CPUs, but they've got memory-starvation issues because they use the same old bus as the 750-series. I wouldn't bother unless you have altivec-dependent stuff to do (compression).

The IBM 970 is a lightning chip with 64-bitness and HyperTransport. It puts out about 40 watts at 2GHz, compared to over twice that for a similarly-beefy Pentium4.

The whole PowerPC line has some level of compatability with each other, and Linux runs on them all.

Happy Camping!

Re:Cooling

[Anonymous Coward]

I have a Barton 2500 and a Radeon 9600 pro in my desktop machine, with 3 hard drives. Its an Antec case (don't remember which one but it was really cheap) with a power supply fan, and a big 120mm fan below the PSU in back. Its quiet (not whisper quiet but in a quiet room it doesn't bother me) and the CPU temp hovers in the high 30s Celsius. This is with the stock AMD processor fan.

Just ancedotal evidence I know, but I've never had a problem with overheating using stock cooling components, not on the Barton, or on a 3GHz Pentium 4.

Re:Cooling

[Anonymous Coward]

Does it really make sense to install XXX number of fans in a case before doing any serious temperature monitoring and characterization to identify whether they really need to go to these sorts of extreme measures.

It's hard to characterize the effects of humidity variation, a plugged air filter, fan failure, and air conditioning failure. For an expensive machine that runs 24x7, it's better safe than sorry.

Re:Cooling

[seitentaisei]

You are absolutely correct. I have a total of four fans in my case, and those are only in the places they either came pre-installed or are basically required to go: --One in the back, next to the processor --One in the front, near the floor and the PCI slots --One on my graphics card (preinstalled) --One on my processor Mind you, that one on my processor is really freaking loud. But it's a nice Dr. Thermal that has kept me at an almost completely constant 45 degrees celcius since I've had the computer, and I bought it for 20 bucks. All that cooling total cost me 30, including the thermal grease on the processor die. So far, my secret has been to make sure the hot air is drawn out by one case fan and cold air drawn in by the other. Also, founded cables have improved the airflow from one fan to the next, leaving my computer nice and cozy.

Re:Cooling

[networkBoy]

for the record the humidity helps you with heat dissapation. Water has a higher heat density than air, thus humid (water bearing) air will absorb and carry away more heat. This does not address the fact that water and electronics are not the best of friends ;)
-nB

Re:Cooling

[Vancorps]

Will stick with the Athlon64 running about 30 watts, but that's good info.

Re:Cooling

[ncc74656]

On a side note I think it'd be really nifty if someone put the original pentium 1 design into the latest technology generation just to see how fast it could be clocked.

There are enough architectural enhancements in later processors that even if you got a Pentium core into the 3-5 GHz range, even a Celeron would probably walk all over it. It'd be like cranking a 6502 up to 10 GHz...it could be neat for bragging purposes, but it wouldn't give you all of the performance boost you'd expect.

Re:Cooling

[radish]

Do you really need all those fans? I have a Athlon 3000+ on an nForce2 ultra, with a 2x HDD and 2x optical, 1gb ram, sound, video etc. My fan count is 3: CPU, PSU, Exhaust. No fan on the northbridge (just replace it with a heatsink), no fan on the GPU (slow card, of course not suitable for a gaming rig). Only need one exhaust fan due to being careful with wiring. Not only that, but not one of my fans is over 20dbA. The PSU is around 17, the others about 19 I think. The disks are much louder than the fans (I have a near silent Seagate and a much louder Maxtor which I will replace soon).

Making a quiet PC isn't really hard if you're careful with components!

Re:Cooling

[Anonymous Coward]

If he didn't have to up the voltage, then I'd just ask, why not? Then (if it doesn't significantly affect the longevity of the chip) it's just simply using the full potential of the chip. They are sometimes marked down, you know: market demand differs from what the production line happens to spit out...

Many multimedia related tasks (video editing, music encoding, etc.) and some games eat up anything you can throw at them -- and this all is still within the domain of "home computing"!

Re:Cooling

[bubkus_jones]

Is there a way of telling if the overclocking is going to FUBAR the chip before doing it, or is it just a "do-it-and-see-if-it-explodes" kind of trial and error?

I'm just wondering if trying to get that "full potential" is worth the risk of frying the processor, if you only get a small (if noticable) increase in speed (especially with the power of modern processors). I overclocked my PII-350 to 466, and even running my most intensive games, I wasn't able to see much of a difference, aside from it being less stable at times.

Re:Cooling

[Afrosheen]

Well, the AMD 64 chips also have a smaller die size and use less voltage than the older chips. That .13 micron process really does help in more ways than one.

AMD still has a bad reputation for the original Athlon series that tended to space-heat rooms. Great for winter since you can turn off the heater. However, the 64bit chips are a big step in the right direction.

Now if you could only do something about your new, hot, power-hungry Nvidia FX6800.

Re:Cooling

[alwaystheretrading]

Why did you overclock your processor?

As a hobby I do 3D rendering and video production. I'm always on a tight budget so I can't afford the hottest processor on the market but I need every bit of performance I can get. It is amazing how little extra heat the Athlon 64 produces when overclocked.

Re:Cooling

[Anonymous Coward]

Agreed.

I'd say the best way of cooling is just to have a good airflow through the box.

I have the PSU fan and a single other fan on the back of the machine blowing air out, plus one on the CPU. My machine is always cold inside. The CPU never gets anywhere near overheating. If I take the side of the case off and kill the airflow the system overheats very quickly.

Cutting holes in the case would seem a bad idea as you're screwing up the airflow.

The one thing I would say in favour of many fans, is being able to run them all quite slowly ala Mac G5 and have a very quiet PC.

Re:Cooling

[Anonymous Coward]

your really screwed.

"you're".

Re:Cooling

[bhtooefr]

So put some alcohol in it. Just remember NOT to run the thing at 10x it's rated FSB while drunk on the coolant, as those poor souls found out the hard way ;-)

Re:Cooling

[bhtooefr]

BTW, I don't think the motherboard they show in the parts list is the motherboard they used - it's a "very old" Super 7 ATX board with AGP, but they used a 486SX-25.

As you can see in the third page, they're not using that mobo - the one they're REALLY using has a green PCB, whereas the one in the parts list had a brown PCB.

Anyway, a 1000% overclock... It was indirectly alcohol-cooled.

Re:Cooling

[bhtooefr]

Which is why we need Whitefield (multicore Dothan Xeon), or even a single-core Xeon M based on Dothan.

Blade servers are already using the Pentium M to cut down on heat output and power consumption (remember, those Xeons have to get power to throw the heat).

The Opteron HE/EE is probably the best choice if you need low power multiprocessing, and it's close between the Oppie and the P-M for low-power uniprocessing (the Pentium M only puts out 24.5W in it's highest power model, the 1.7GHz Banias, and the new Dothan chips only put out 21W, even the 2GHz chip, thanks to 90nm).

Re:Cooling

[bhtooefr]

You might be wrong there, based on the fact that a Pentium M DOES walk all over a Celeron. The Pentium M is a revised P6 core, and the Pentium 3 (the previous P6 chip), while it can't come close to the P-M, would probably give a Celeron a run for it's money if it would ramp up to that clock speed (that would be one hell of a stretch, though - it appears to only be capable of 1.4GHz - if Intel would put it on 90nm, and we could liquid nitrogen cool it, though). Seeing as there wasn't much difference in the pre-P-M P6 chips (PPro, P2, P3) in performance (there was some, mostly because of cache arrangements), and the PPro, while it was better in 32-bit that the Pentium, wasn't much better, and it sucked balls in 16-bit. I think a theoretical 3GHz Pentium I might just have a chance against a Northwood Celeron (although the Northwood wouldn't need LN2 or 90nm to get the job done). Now, a Preshott Celery? I don't know about that.

A lot of the architectural "enhancements" in later processors, especially NetBurst chips, and to a lesser extent Pentium Ms, are to either increase the max clock speed or counteract the disadvantages of making the changes to increase the clock speed. So, a REVISED Pentium I with those second changes (like an improved branch predictor, etc., etc.) could definitely give a Celery a run for it's money, even at a lower clock speed.

Re:Cooling

[bhtooefr]

Heh, my computer is MUCH noisier, and it's an old Pentium I system. A large 120mm exhaust fan, a crappy 80mm fan in the PSU, and a 40mm fan on the CPU. The 120 is probably the noisiest, but I don't think they're undervolted at all...

Of course, my Quantum Bigfoot CY probably dwarfs any of the fans in noise - ditto on the seek noise of my Seagate 8.4GB (however, I've NEVER had a problem with a Seagate - I've seen Deathstars, Western Digitals (anyone got a derogatory name for WD?), and a Maxtor (it was Sudden Maxtor Death Syndrome, too) die, but never a Seagate).

Re:Cooling

[bubkus_jones]

So, your overclocking experience has yet to produce any noticable negative impacts? How long have you had it OC'd? Yeah, I've heard the latest generation AMD chips have better heat control.

Re:Cooling

[alwaystheretrading]

The greater the temperature difference between overclocked and regular, the more likely you will have a reduced life. My new Athlon 64 is only 4 degrees hotter overclocked. It should last 99% as long as if it were not overclocked. Since it will be obsolete years before it reaches that point the reduced life won't affect me.

Overclocking always depends on which processor you get and which specific processor out of each batch you get. I also have several 2500+ Barton cores that overclocks nicely to between 2700 + and 3000+ they run hot but with good cooling they should have a good life yet. On the other hand I just replaced a 2100+ that never would overclock. With that chip, just a few more mhz and it would freak out. I had an Athlon 600 once that would overclock to 1000 but it took the best heatsync I could get at the time to do it. With a modern heatsync I might be able to get another couple hundred mhz out of it, but with new cheap processors what's the point? My guess is I reduced the life of that processor by as much as a year but I'll never know because I upgraded enough systems that it has finally dropped out of my render farm alltogether.

Re:Cooling

[Jeff DeMaagd]

I have seen two dual processor capable systems operate with only two fans. I kid you not.

One is a Compaq Evo W6000, capable of up to two 2.8GHz Xeon chips. There is one fan in the power supply, another on the exhaust side of a fan duct that joins like a "Y". The fan duct also has gaps around the chipset heatsink, so it too gets forced air.

The other is a Compaq SP700, capable of up to two 550MHz Xeon CPUs. There is one 12cm fan on the intake of the power supply, and another 12cm fan that forces air inbetween two processors with their heat sinks back-to back.

Both systems are pretty quiet. People rave about Shuttle's quiet little cubes, but I don't know why. They use a nice heat pipe system but then attach a racket of a fan to the other end of the heat pipe.

Haha!! I'm one of the first!!!

[protect_the_code]

Yay! I've always been a fan of AMD's architecture. I definitely to see AMD take more of the laptop market.

Re:Haha!! I'm one of the first!!!

[protect_the_code]

This sucks! Fine, I'll say something more worthwhile. Notebook makers should really drop both AMD and Intel processors and pick up Via's stuff. They put out miniscule amounts of heat(~20 watts, I think) and while not as fast can still run at 1.8Ghz. This is what notebooks really need.

Re:Haha!! I'm one of the first!!!

[PeteyG]

Yay! I've always been a fan of AMD's architecture. I definitely to see AMD take more of the laptop market.

That's probably a good thing! AMD architecture NEEDS big fans for how hot they run.

Re:Haha!! I'm one of the first!!!

[Anonymous Coward]

Except the Via is horrible for supporting hardware that exists, and isn't really usefull except in situations where full computers are needed (desktop solutions).

Re:Haha!! I'm one of the first!!!

[kannibal_klown]

While I agree with your feelings, I do not agree with actually persuing such a thing.

I want laptops that run COOL and have MASSIVE battery life. In a laptop, performance comes third or later (in my opinion).

That said, I also need it to do work on when on a business trip. The VIA chips are lacking some serious components, so they run REALLY slow compared to Intel and AMD running at similar clock speeds. I think they're missing float-point arithmetic, or something else that makes anything somewhat complex take FOREVER to perform.

I don't just use my laptop for browsing the 'net or checking my email, I use it to do all of my work when I'm at home or on a trip. It's easier for me to make my laptop replicate the business environment as much as possible (and throw in some development tools: eclipse, JBuilder, VS, etc). Then, use my desktop for my gaming, browsing, emailing, etc. This way, I have a delopment machine for work, and an "everything else" machine for play (or personal development).

I think the Centrino is a step in the right direction. But I think companies need to just start making a model that runs slow, but long and cool.

After all, not everyone is alike. Some people want pure battery life and a cool unit, while others want more of a balance between SOME performance and good batery life.

Re:Haha!! I'm one of the first!!!

[Anonymous Coward]

That's probably a good thing! AMD architecture NEEDS big fans for how hot they run.

Maybe this is why Intel's stuff has become hotter than AMDs as of late. Because they are losing alot of their fanbase? Har har!

Re:Haha!! I'm one of the first!!!

[Vlad_the_Inhaler]

Via processors top out at around 1.2Ghz, which is normally fast enough for me - this PC I am using is a Via 866.

What they do not have and AMD (+Intel?) do have is a mechanism which slows the processor down in milliseconds if there is nothing for it to do. That is how they save power. I got the chance to speak to an AMD VP about 20 months ago and he indicated that they were very aware of the need to reduce power consumption.

I don't know how Transmeta compare.

Re:Haha!! I'm one of the first!!!

[Vlad_the_Inhaler]

The VIA chips are lacking some serious components, so they run REALLY slow compared to Intel and AMD running at similar clock speeds. I think they're missing float-point arithmetic, or something else that makes anything somewhat complex take FOREVER to perform.

You exaggerate. They certainly have a Floating Point Unit. They are a bit slower than Intel/AMD machines running at the same frequency.

Where I notice problems is when looking at large PDFs, or running .mpeg or .mp3 files whilst doing something else. PDFs are painfully slow to load and the other two need the machine to themselves. As I said in another posting, this is on a Via 866.

Re:Haha!! I'm one of the first!!!

[Anonymous Coward]

As opposed to the Intel Prescotts which run as cool as a jet engine?

Re:Haha!! I'm one of the first!!!

[homer_ca]

The Via C3 is seriously underpowered (har!) for desktop use. Really, it would get spanked by a P-III Coppermine which are pretty energy efficient already, 20-30W. It does have a FPU, but the earlier C3's ran the FPU at half the clock speed. The newer ones run the FPU at full clock speed, but it's still not very powerful. It's fine for embedded applications where it has just enough power like small servers and home theater PCs that play videos. It also helps that they built co-processors for MPEG decoding and AES encryption into the motherboard chipsets. Via know who the target market is, and full powered desktop CPUs it is not.

Re:Haha!! I'm one of the first!!!

[Rob Simpson]

Yup. I had a 1 GHz C3 in my quiet system, but it was annoyingly slow (slower than a 500mhz Pentium III, and though it was one of the Nehemiah ones with a "full speed" fpu, it was appallingly slow with 296 Whetstone mflops in SiSoft Sandra - slower than a PII), so I replaced it with a 1.3Ghz Celeron (a newer one with 256k cache) which has six times the FPU speed and still stays at less than 40C with the fan speed set to minimum.

While the C3 cpus are nice for ultra-low power situations, a Centrino is probably a better (if more expensive) choice for most uses. I'm pretty sure the clock speed can be set low enough to use a comparable wattage - I just wish they were available in desktop socket 478. While there are desktop motherboards that take the notebook chips, they are ridiculously expensive. The mobile Athlons can be used in desktop motherboards, though...

Uggghhh...

[k4_pacific]

"marking the first time AMD's chips have been used in a tablet"

Prior to this, they were either injected or used in suppository form.

Re:Uggghhh...

[Psiren]

... or used in suppository form.

Ouch! The sooner they move to pinless chips the better!

NX capability

[110010001000]

Interesting. The 64 bit version will enforce the NX data segment capabilities that is the cause of so many buffer overflows. This can only enhance the security of our servers, no matter what OS they are running. Although SP2 has added this, Linux and the BSD's also can leverage this feature.

Work with XP SP2

[wolenczak]

I don't want a freaking CPU that knows it's running WindowsXP SP2, or linux, or anything else, I only want a processor that does processing. Let the software do the security stuff

Re:Work with XP SP2

[Anonymous Coward]

You're right it should be the other way around...
... like GENTOO! *Bling KARMA Bling*
The OS should be built around the proc, not the other way around!

Re:Work with XP SP2

[oostevo]

If I recall correctly, it isn't an issue of the chip "knowing" anything. I'm fairly sure this technology just enables the chip to respond to security events passed along to it by Windows. If other operating systems emulate this feature, there'd be no way for the chip to "know" that it is using anything other than an operating system with such enhancements. Why do you care if your chip knows what operating system you're running anyway? It's not like it's going to email Microsoft all by itself and tell them you're running a pirated version of Windows XP.

Re:Work with XP SP2

[strictnein]

but how are we supposed to enforce the DRM? We have to make sure the artists rights are protected!

Re:Work with XP SP2

[mapmaker]

I don't want a freaking CPU that knows it's running WindowsXP SP2

It doesn't, it's the other way around - this CPU has a feature (NX bit) that WinXP will be aware of as of SP2.

Re:Work with XP SP2

[Anonymous Coward]

OR IS IT!?!!?!?!

Re:Work with XP SP2

[kpansky]

Umm... you do realize this is referring to the NX (no execute) extension of the CPU right? It merely means that if you are running SP2 you will be able to utilize this hardware feature of the processor. Linux also has support of this in all recent 2.6 kernels.

Re:Work with XP SP2

[Wordsmith]

Don't get all paranoid. The processor supports the NX flag, which SP2 will use to help prevent buffer overflow attacks. Any other operating system can impliment this as well (and if i'm not mistaken, Linux either has this or its in test kernels already - but I could be wrong there).

It's not OS-specific, but the majority of users aren't likely to see a benefit until SP2 arrives, because that's the first time Windows will impliment the function. NX support is coming eventually from Intel as well.

Re:Work with XP SP2

[Anonymous Coward]

Don't worry the proc has no idea what OS is running. Its just that it supports a noexec instruction which lets you mark areas of memory as data vs. code, this makes the buffer overflow attact a little harder because the chip won't do instructions in an area that is flagged data(where the array you are exploiting would be). SP2 will give the windows OS the ability to support performing that instruction in some instances and therfore enchance security. I have already heard noises of implementing this in Linux and *BSD. Its complex from a software stand point to make really effective use of as you have to be able to predict the future to some degree but its probably a good security tool noone the less and could be extreemly benificial in many cases.

Re:Work with XP SP2

[Anonymous Coward]

no, but it could cripple itself in the name of "trusted computing"

Re:Work with XP SP2

[Anonymous Coward]

Do you even know what the hell you are talking about? This will allow operating systems to use the NO EXECUTE instructions that are built into the chip. This is a well overdue feature that will cause less headaches for administrators as well as end users. And this feature will not be required or anything, so don't get your panties in a ruffle.

Re:Work with XP SP2

[Lord Omlette]

Are you sure? In Linux, when you want to watch a video, tell teh CPU that you're running WinXP Pro and you're good to go!

How did you think the CPU was going to identify the operating system?

Re:Work with XP SP2

[Lord Kano]

I don't want a freaking CPU that knows it's running WindowsXP SP2, or linux, or anything else, I only want a processor that does processing. Let the software do the security stuff

You dumb ass, the processors have features that HELP the software "do the decurity stuff".

A padlock doesn't care if it's keeping out a robber or you when you've forgotten your key. It's the same thing here.

NX will help to prevent buffer overflow exploits. These processors neither know nor care what OS they're running. NX will help provide security for Windows, Linux, *BSD or whatever OS is developed next.

LK

Re:Work with XP SP2

[steveb964]

This is a well overdue feature that will cause less headaches for administrators as well as end users.

I've never done any research on the 'nx' feature, but is this just a feature so that developers can remain lax at actually writing non-vulnerable code?

Re:Work with XP SP2

[Eric604]

It also means faster code, no need for overflow checking code anymore

Re:Work with XP SP2

[Chirs]

The only reason they're making a big deal about this is that traditionally x86 has not had separate write/execute flags for memory regions. If you could write to it, you could run code from it.

Other architectures (sparc, for instance) have had separate read/write/execute flags for ages, and Linux has support for them. Now x86 gets that support as well.

Since POSIX has separate flags for each of the three attributes, properly-written programs will *just work*. Stuff that relies on the old x86 behaviour may need to be fixed.

Re:Work with XP SP2

[Anonymous Coward]

Not really - your code will still crash if you're not doing checking. It just may not do so much damage on the way down.

Huh?

[Illissius]

This is news? This isn't some sort of new processor design, or a new socket, it's not even a new speed grade, it's just two minor additions into a (relatively) less significant product line. Name a processor launch in the past few months, *any* processor launch, and chances are it's more significant than this one. Not that any of those would (should) be news, either...

Re:Huh?

[callipygian-showsyst]

C'mon now! By this logic, /. shouldn't have covered any of this past Mac World announcements!

And it's exciting to think that we may get 64-bit laptops soon!

Re:Huh?

[bdaehlie]

HP and Compaq 64-bit laptops already exist (both found on hp.com). The guy in the office next to me has the Compaq one. I haven't been much of a fan of Compaq but that laptop is awesome (and $2000 with most of the options). Runs Linux like a champ, though if you get it with the built-in wireless card it'll only work in Windows.

Re:Huh?

[Anonymous Coward]

Just a note, the latest Linux kernels seem to have a problem seeing the keyboard with APIC enabled (Compaq R3000Z).

Also, PCMCIA does not work on most default configurations, you have to specify an alternate range for memory/IO (you can usually use the values in lspci for the PCI-PCI bridge in the laptop and it will work).

I ended up getting rid of Linux because I dont use it much anyway and I can get most of the tools I need in Windows. (My server still runs Linux, so its not an ideology issue with me).

The built in Broadcom wireless can be used in Linux (I havent tried though0, only with a 32 bit kernel though, I beleive using one of the avaliable Windows driver loaders. Havent tried this yet.

Re:Huh?

[Illissius]

We already *have* 64-bit laptops, plenty of them. There's already ones with Mobile 2800+ up through 3200+, and DTR (DeskTop Replacement) 2800+ through 3400+. And the only difference between the Mobile and DTR is that the former is 62W, while the latter is 80-something... not an insigificant difference, to be sure, but the only difference it'll actually make is that your desktop replacement notebook will have slightly less horrible battery life and be a bit less scalding. Not anything you could actually call mobile ("portable" is the correct term). (For reference, Intel's fastest Pentium M (Centrino) processor is 21W, and there's ultra-low voltage versions under 10W which are used in ultra-mobile thingies. AMD also has a "line" of 35W Mobile A64s (2700-2800+) which could also be used in something other than a desktop masquerading as a notebook.)

it would be nice to see

[asv108]

IBM putting one of these 64 bit chips in a highend thinkpad. While I really like the Athlon64 processor, I would not even consider buying anything but a thinkpad for a PC laptop.

Re:it would be nice to see

[fantom2000]

amen

Re:it would be nice to see

[Anonymous Coward]

However, having been strung along for months by Thinkpad support, and having had the story I was told change multiple times from the same person, I will not buy a Thinkpad ever again.

Re:it would be nice to see

[jeffphil]

Would you settle for an emachine:

http://www.emachines.com/products/products.html?pr od=eMachines_M6811

Emachines has it listed as 2.0 GHz, but several people on forums called Emachines and AMD and confirmed this is the 2.2 GHz 3400+.

Supposedly as well Best Buy already has them, and J&R Electronics as well with a $100 rebate.

Re:it would be nice to see

[Anonymous Coward]

Any particular reason you would only buy a thinkpad? I haven't used them myself, but my Acer Ferrari 3200, despite the color (red), seems very nice to use (it's based on the older mobile Athlon 64 2800+). Ok, the finish quality isn't quite up to Apple's level, but I haven't seen any PC where it was...

Re:it would be nice to see

[salimma]

IBM does not have much an incentive to put an Athlon 64 on a Thinkpad then, if you won't even consider buying another brand that does use that CPU.

The first of IBM, Sony, Fujitsu or Toshiba .. perhaps even Dell, but that seems a very remote possibility - to ship with Mobile Athlon 64 will probably get my business. Bought a Centrino-based Acer laptop for my sister and she was not too impressed by it, so I think I'll skip Acer for now.

Re:it would be nice to see

[WhiteDragon]

I have one of the eMachines amd64 laptops, and overall it is quite nice. The only thing is that the ACPI is somewhat strange, so whenever I shut or open the lid, I get screens full of "unknown acpi event" messages.

fixes for windows BUT

[Man in Spandex]

as long as it doesn't affect linux I guess I don't care much about the security fixes although it's nice to see these tweaks being made.

Just wondering, IF linux was more used than windows, would AMD and Intel make those fixes of MS's windows xp even though it's less popular than another O/S?

Re:fixes for windows BUT

[vuvewux]

It's not a Windows specific fix. It's NX support (so that data marked as such won't be executed) and Linux can benefit from it as well.

Re:fixes for windows BUT

[Anonymous Coward]

You don't know what you are talking about. 2.6 kernels also make use of this feature, it just does not say so in the article.

Re:fixes for windows BUT

[DaHat]

You make it sound as if AMD one day went to Microsoft and said "Hey, we've got this really cool feature we are adding to our chips, it'll make your operating system much more secure."

NX is not a 'security fix' it is an added feature, a feature not unlike MMX or SSE. An ability that sits on the chip waiting to be used and is generally only used if it is called (ie software support).

This is nothing new! It amazes me that you think it's so radical to hardware and software supporting the same feature.

BTW: NX support now exists for linux and has for some time. By your logic... does it's presence mean that AMD is fixing linux?

Re:fixes for windows BUT

[Anonymous Coward]

Even calling NX a feature is stretching it. It's a minor fix that finally addresses an omission in the paged memory model introduced in the 80386. Many other processors have separated read and execute permissions of virtual memory pages for a long time...

Re:AMD

[KillaKen187]

What are you talking about. Intel at one point hired AMD to make processors for them. So you want to call AMD poor quality when they are the 2nd largest supplier of x86 compatible processors, and a leading supplier of non-volatile flash memory?

Get the byte outta here.

Kudos to Hardware Engineers

[MirgNave]

Not meant as a troll:

Kudos go out to hardware engineers for stepping up and cleaning up the dangerous wake of bad programming practices. I'm not familiar with exactly how these security features would work, but I assume they will be automatic and thus will help clean up problems in older software which carry little hope for patches.
It does bring questions to mind thought.Will software developers get lazy and no longer even attempt to adhere to good programming practices? Will it matter if they do or not?

Re:Kudos to Hardware Engineers

[FooAtWFU]

The trick is basically a permissions system for memory. If the memory isn't in a certain range, you can't (write to|execute) it. This keeps you from executing your data section, or writing over your code. This prevents buffer overflows from being exploted with the "arbitraty code injection" that you hear so much about.

But it doesn't prevent the overflow, just the injection. This leaves your program free to (behave poorly|crash).

Re:Kudos to Hardware Engineers

[yohan1701]

Am I the only one who thinks this is a bad idea ? Why is my processor doing work tha should have been done by the programmer ?

I want my one thing and that is blindly doing whats it told as fast as it can

Re:Kudos to Hardware Engineers

[MindStalker]

Sorry, it doesn't for for old program. A program specisfically has to say, "here is non executable data, don't let anyone execute it." Thats basically all it does.

Re:Kudos to Hardware Engineers

[Anonymous Coward]

Actually you're right to think the processor shouldn't need to do this, but you're wrong to think the programmer should be the one to do this. The job of memory allocation/deallocation belongs to the programming language, libraries, and compiler. Programmers shouldn't allocate buffers at all. Runtime libraries should allow you to index beyond the bounds of an array, etc. The problem isn't (for once) a problem with Windows or with processors. It's a problem with C becoming the de facto business programming language.

There are plenty of ways to screw up security that SHOULD BE the programmer's job. Just not this one.

BTW, Linux has had a fix for this issue that did not require a hardware change--it changed the location of memory allocations so that you could never overflow a buffer and get a reliable code injection working. And because Linux works on more platforms, the injected code may not match the platform the OS is running on and may not work. So Linux didn't have this problem as much, and managed to fix it. Windows got chip manufacturers to fix it for them.

Re:Kudos to Hardware Engineers

[Anonymous Coward]

This tweak, already present on some less common hardware, will be used by the OS to mark pages which have no reasonable to be executed, the stack being an example virtually everyone is agreed on. Most application software won't be changed in any way, and most programmers need never know this change was introduced.

This raises the barrier for those trying to insert their code into your machine through a bug, because one of the easiest places to put such code is now useless to them.

It could potentially make software engineers lazier, in that the consequence of certain mistakes will now just be crash reports rather than the bad press of security break-ins through your software. It will certainly, as one person has mentioned, reduce the pressure to switch from unsafe-but-fast C or C++ to safer-but-slower languages for work that's not performance critical. Depending on your perspective that may or may not be a good thing.

Overall once this technology is popularised it will be impossible to go back, and the ongoing overhead for the extra transistors to implement it is tiny compared to saving even a small proportion of users from the consequences of a security hole.

Re:Kudos to Hardware Engineers

[Anonymous Coward]

Uh, where did you get your information? Its horribly inaccurate. Its WRONG.

All that has to be done is the OS has to turn the feature on and handle the exception properly (usually by terminating the program).

Re:Kudos to Hardware Engineers

[Anonymous Coward]

Not quite. First of all, range restrictions were possible earlier (using segments), just not used very much because every sane operating system ditched the segmented memory model of x86 and went to flat, paged virtual memory.

The thing is, on the 386, virtual memory page mappings didn't separate read permissions and execute permissions, so any page that was readable was executable. Note that writing over your code hasn't been a problem, since write permissions have always been separate.

Some operating systems (OpenBSD, at least) have used segments to separate the stack into an address range that isn't executable, disabling many buffer overflow exploits on any x86, but that's a bit of a kludge.

Also note that a no-execute bit (or a separate "allow execute" bit, which is how the Unix mmap(2) /mprotect(2)API exposes the protection - as the PROT_EXEC flag) is by no means a perfect solution.

A buffer overflow exploit doesn't necessarily need to execute any code written by the exploit. Especially a stack overflow can just replace the return address with the address of a library function it wants to call, with the rest of the stack set up as suitable arguments for that call...no code needed, only data.

The best protections against exploits are - reduce the privileges of programs accepting external, potentially dangerous, inputs, and, to the extent possible, fix the bugs that allow the exploits in the first place. Changing programming languages, as advocated by some, isn't sufficient, since there are plenty of bugs that allow exploits that aren't buffer overflows.

Heat and AMD

[charnov]

The Hammer based chips run cooler than their older AthlonXP brothers. They also add cool-n-quiet for power management. My Athlon 64 laptop (with a DTR chip) rarely gets very warm and the lower power portable Athlon 64 chips are extremely cool (nearly in the G3 - G4 power range which is really impressive for an x86 chip). The P4 based laptops, however, can literally burn you. If you go the Intel route, the Centrino platform is excellent and one of the best chips ever by Intel.

Re:Heat and AMD

[mog007]

I'm sure that heatspreader helps a lot in that reguard. I'm suprised AMD didn't start adding them in the barton core XP line.

Re:Heat and AMD

[tier 3 geek]

talk about great quotes

"It compiles... ship it!"

"[RIAA] says its concern is artists. That's true, in just the sense that a cattle rancher is concerned about its cattle."

First day on \. and I'm already loving it.

Well to look at things, I have an AMD XP 2400+ with a Thermaltake Volcano 7.
You would think my heating issues would be no problem, especially since its kept open, with the AC at 70...

But that's not the case. It runs at roughly 50+ Celcius, and the problem resides in case design -> The power supply is situated right over the fan and heatsink, which leads me to believe its sucking all the cool air for itself... Go figure.

Re:Heat and AMD

[Ruediger]

I think the reason that the A64 runs cooler than the XP line is the lower default voltage on the A64.

Besides running at a lower voltage the A64 are not clocked a lot higher then the XP family (sometimes even lower than a comparable XP).

From what I have seen the heatspreader doesn't help since there is a layer of cheap thermal goo between it and the chip core. Without the heatspreader you could use AS5 or Ceramiquè and the core would be making contact with the cooper part of your heatsink.

On the other hand, the heatspreader protects the core, which probably worth temperature increase.

Re:Heat and AMD

[Anonymous Coward]

And hows the battery life?

Re:Heat and AMD

[mog007]

You'd better make sure your Volcano is spinning at full speed, or use some artic silver. Those are very high temperatures... Unless you're OCing the processor.

Re:Heat and AMD

[Mesaeus]

You might also want to try undervolting your cpu. At it's rated speed, it probably does not need the default voltage specified, and each 0.025 Volt you can lower it (my Abit KR7A mainboard handles core voltage in steps of 0.025), you gain 1 or two degrees celsius. I can undervolt my old Athlon XP 1700+ to 1.60 volt from the standard 1.75 and it makes my processor about 7 degrees cooler than usual under load, and it's free. You do have to make sure your processor is still stable at the lower voltage, I use Prime95 to make sure my cpu does not make a single error even under stress.

Re:Heat and AMD

[Anonymous Coward]

it's rated speed

"its".

Re:Heat and AMD

[tier 3 geek]

Nope, not OCing, sadly...

And I've sort-of-kind-of-not-really checked the fan speed, but everything seems to be in line.

About the artic silver, I also heard grease is also a very good heat dispenser, and I could probably use bacon grease at that!

Explanation please!

[jasoncc]

Someone please explain the statement about blocking security threats in conjunction with the operating system. My imagination is failing me. New instructions that do....what?

Re:Explanation please!

[jasoncc]

http://www.amd.com/us-en/Weblets/0,,7832_11104_111 05,00.html?redir=CPVP01/

Here's what AMD says about the new security features. Lots of hype on this page but not much info.

"Protects against certain classes of viruses" ... any guesses as to which classes these might be?

Re:Explanation please!

[Anonymous Coward]

it disables IE and Outlook.

Re:Explanation please!

[Holi]

Buffer overruns

Re:Explanation please!

[Anonymous Coward]

NX. Sections of memory that are marked as containing data ONLY, which cannot be made to execute. If you try to execute this memory, an exception is raised.

Helps you with things like buffer overflows that are such a PITA today because of sloppy code.

Re:AMD

[Anonymous Coward]

WalMart is the largest retailer and McDonalds is largest restaurant chain.

Volume != Quality.

And intel's reply:

[silverfuck]

Intel has 'unleashed' new low voltage and 'ultra low voltage' Centrinos and Celerons.

Re:And intel's reply:

[drinkypoo]

WTF does low voltage or ultra low voltage mean? 1V? 500mV? Typically, the register eschews the technical details.

Prescott NX? Not yet!

[vuvewux]

Notably, Prescott chips with NX support aren't yet shipping.

*shakes head*

[Flower]

The processor comes with a "no exec" op code that an operating system can use to enhance security. OpenBSD or Linux could use this just as readily as XP. As a matter of fact OBSD already does iirc and MS isn't doing anything truely original on this platform (yes, I'm aware other processors and OSes have offered this for years.)

Virus Protection

[l8f57]

You can get Virus Protection without an AMD processor here.

Speaking of Averatec...

[Anonymous Coward]

I just got to say how much I love their notebooks. Easily some of the best made on the market. I purchased one last year, based off the Athlon 1600+ Mobile chip. It ran pretty decently, but like most notebooks... the graphics chip was a little underpowered. They came out with a new version of their 12" notebooks (size and weight was another factor) that contained the 2000+ Mobile CPU and a mobile/integrated version of the S3G DeltaChrome chipset (Unichrome, if I remember correctly)... it may not be the greatest for games, but it plays HalfLife/CS at 30FPS+ at 640x480 and that is good enough for me. UT2004, with everything set to minimum, can still play at around 20fps. So it does have a little bit of muscle in it.

The new Averatec 12" notebooks are the bomb... Pretty decently loaded and skips out on most of the legacy stuff, like parallel/serial/ps2... Gets a pretty respectable 3.5hours average battery life. Even when doing something intensive or playing DVDs (moving parts will help suck up the battery life).

The best thing about their notebooks is that they are sturdy as hell. Very, very well built. the hinges never feel like they're going to come apart, the screen holds taut and stays there. Some of the larger 15" (and especially widescreen models) almost feel like they'll bend and wobble at the corners.

I was impressed with the performance of my 2000+ Mobile (Barton Chipset) based laptop that I've given up on using my desktop, permanently. And it's a watercooled 2600+ with more drives than I care to count. It was very easy to get used to using the Averatec and I'm in love with them.

I'd highly recommend getting them. I'm on my second (gave the first one to me mum) and if they keep it up, I can't see how they'll have any trouble breaking into the tight notebook market even further.

(Sorry for the slightly off topic post, but Averatec was mentioned and I just felt like posting my raves for this company.)

Oh, hey, not that anyone here really needs it... But free 24/7 tech support (toll free #) that actually uses people in the US... not india. That's major kudos in my book. (*Spits on HP*)

The Averatec notebooks are also very reasonably priced. You pay more for quality, of course... but the deal is still quite outstanding. Currently you can get the model I have, 2000+ AMD Athlon Mobile CPU, Delta/UniChrome Video chipset, 12" Screen, built in broadcomm wifi 54g adaptor, touchpad with scrolling areas, 40GB HD, 256MB Ram, CDRW/DVD Drive, 3xUSB2 (each port gets full bandwidth, not three plugs shared across one bus), 1 PCMCIA, 10/100, 56k, VGA Out, sound, etc... For about $850 now.

My only gripe is the sound. It's AC97 Audio, so it's not that great. Just noisy. In my case, since I do professional mixing, I would rather have an mAudio USB 24-Bit sound card anyway. (I mixed two CDs so far on this laptop without a hitch. It's been absolutely wonderful.)

My longest uptime (damn you, Windows) has been 3 weeks, 6 1/2 days. Not bad, considering previous experience with laptops they tend to crash easily due to the heat or whatnot. No problems, heat related here... Just had to friggin reboot for an IE patch. Oh well.

Re:Speaking of Averatec...

[Anonymous Coward]

I bought one of their 12" laptops last year and returned it within a few days. Performance wise it was alright but the case was engineered poorly. Cheezy plastic case that scratched all to hell and the keyboard rubs up against the screen when closed.
I quickly ditched this thing for a much better cousin put out under the Sharp brand. Biggest gripe about either is the fan can get annoying.

Re:Speaking of Averatec...

[Anonymous Coward]

Both issues fixed with the 3200 series. The case is actually a magnesium alloy. Not plastic. The keyboard only rubs the screen if pressure was put on top of the case when closed. Again, not a problem with the 3200 series, but if you have the older models... get a thin sheet of foam plastic. Despite this design flaw, they're still some of the best made laptops you can generally find in stores now.

Re:AMD

[tier 3 geek]

Perhaps you would care to enlighten the more elite vendors out there then. And by your analogy, Intel is the largest retailer, making it worse than AMD.

Any low noise AMD 64 Notebooks out there?

[Rauchbier]

I'm looking for a low noise (quiet) AMD 64 Notebook. Do you know any? Thanks a lot for your reply!

Re:Any low noise AMD 64 Notebooks out there?

[Brian Stretch]

Yeah, this one. Pity it uses an ATI video chip (which lacks the OUTSTANDING 64-bit Linux drivers nVidia has), but at least they used the best one (128MB Radeon 9700). And pity about the Broadcom wireless and wired network chips, but at least it's 802.11g and gigabit, respectively. Other than those two complaints, if you want to make TiBook owners drool, this is the notebook to get. (No, I don't have one. Having played 64-bit UT2004 on my desktop I'm holding out for a proper AMD64 nVidia notebook.)

Re:Any low noise AMD 64 Notebooks out there?

[Sayan]

Another good option is http://www.emachines.com/. I plan to buy one for myself soon.

Re:Any low noise AMD 64 Notebooks out there?

[Anonymous Coward]

I own both a TiBook and an Acer Ferrari 3200...I like both - the TiBook has an overall better feel of quality, is lighter and more compact, but it's also older and slower...

I run FreeBSD on the Ferrari 3200, and it works very nicely, supporting all the hardware in i386 mode and most of it in amd64 mode... Sadly, the wireless requires i386 mode, for now, since it has to use the NDIS compatibility layer and the 32-bit Windows driver.

And, of course, I don't get 3D acceleration under FreeBSD (but the X.org 2D support works fine), but that's not something I care a great deal about.

I'm not sure what you have against the Broadcom ethernet, it works very well under FreeBSD with a native driver (supported under both -current and -stable), AFAIK Linux 2.6 should have a driver as well, although 2.4 doesn't.

Re:AMD

[boarder8925]

AMD = poor quality.

I think you misspelled "Intel."

Using mobile chips in a desktop

[kilonad]

I'm getting ready to build a new PC and I've decided on the Athlon64 (it was such a difficult decision... not.). It looks like the mobile chips are basically identical to their desktop brethren except that they don't put out as much heat nor do they have integrated heat spreaders. I really like the fact that they put out less power, perform identically (I think), and only cost a few dollars more.

Does anyone know if you can take a mobile A64 and just plop it in a desktop motherboard (for regular A64s) and have it just work? Or does the BIOS have to be aware of the fact that it's a mobile proc? Will the heatsinks designed for the desktop versions work with a mobile version? Does an A64 really require a 400W+ power supply as many sites suggest?

Re:Using mobile chips in a desktop

[MindStalker]

Well honestly your gonna get more speed out of the non mobile. And the desktop processors 64's run pretty cool as it is. Me thinks its a bit silly, but go ahead, and try I guess, its a neat idea.

Re:Using mobile chips in a desktop

[theMerovingian]

The regular chip runs relatively cool, just as-is. I have one at home, and have never had any problems with it.

Conversely, I have dual xeons at work. I have to run a fan under my desk to keep my feet from sweating.

Re:Using mobile chips in a desktop

[SnakeJG]

Why do you think the non-mobile version will run faster? Do you have access to information that tells you that the cores are actually different (and that the mobile performs worse)?

Do you have anything to back up your statement? Perhaps a site benchmarking, or stating that some power saving optimization in the mobile processors hinders performance?

Also, at least for the Mobile Athlon XPs, there was no difference between the cores, they were just hand picked to be able to run at a lower core voltage...any reason why you think the case is different with the mobile Athlon 64s?

Re:Using mobile chips in a desktop

[SnakeJG]

I would suggest checking out this thread

From what I can tell, most heatsinks will not work without some modification (sometimes as simple as adding washers), but read the thread, it answers a lot of questions.

Re:Using mobile chips in a desktop

[steveha]

Does an A64 really require a 400W+ power supply as many sites suggest?

If you want a quiet and reliable system, it's a good idea to buy a power supply that is rated for more Watts than you actually need. Near the top of the possible power output, power supplies are less efficient and emit more heat. If they aren't working very hard, they run cooler, and they should have a long life and not require as much noisy cooling.

I'm not certain how much of a difference the above really makes, but that's my rule of thumb.

Anyway, what you will be doing also matters. If you plan to put in a ferocious nVidia graphics card, and play CPU-intensive games for hours, you will need more power than if you just use an ordinary graphics card and surf the web. If you plan on having a whole bunch of hard drives, you will need more power than if you just have one.

steveha

Re:Using mobile chips in a desktop

[MindStalker]

What I meant was that you could buy faster non mobile chips than you can buy mobile chips. Fastest mobile athlong 64 out right now seems to be the 3200, though quite surprising is the fact that it only cost 2 dollars more that the non mobile 3200. So yea if your going 3200, mobile all the way.

Next month CPU releases!

[DungeonCoder]

- AMD is lauching the new AMD Athlon Greece 2004, because the real Athlete use Athlon!
- Intel is lauching the new Pentium 4 Olympic Games Edition, with 20 new "Olympic Instructions" to make the Olympic Games coverage 50% faster!

Oops...

[kilonad]

I meant to say the Desktop Replacement (DTR) version of the Athlon 64, not the true Mobile A64.

Re:Oops...

[dago]

For motherboard, check silentpcreview.com (esp. the forums).

For the PSU, there you'll have also a impressive list of "what can run on a 300W PSU". FYI, my dual Athlon 2600 is between 100 and 170W (not overclocked). So far for the 500+W PSU myth.

Re:Oops...

[Illissius]

They are theoretically compatible, but it depends on whether the motherboard in question supports them. The Asus K8N-E Deluxe looks like it does a good job of that, though information is scarce as it's pretty new. As for heatsinks, they aren't inherently compatible, but the Thermalright SLK-948U is one that is.
And a true mobile will work just as well as a DTR, and have even better thermals (in the case of the 35W mobile 2700+/2800+, magnitudes better).

Re:Oops...

[nelsonal]

I think the reason for a 500W PS is that companies rate PS differently. One company's 500W supply might be able to peak at 500W, but only output 250-300 cleanly. I'd guess it's a lot like stereo amps where boom boxes list peak wattage (on a sunny day in ND with no wind) while a company like NAD rates its applifier's wattage based on how much power it can pump without exceeding a certain deviation from the input signal. So a 65W NAD amp might be considerably louder than a 300W Walmart special.

Re:Oops...

[Lord Kano]

About a decade ago I learned that lesson. I bought a couple of Amps from Radio Shack because their peak wattage was so much higher than similarly priced amps from a car audio shop.

After about two monthe of fuxoring around with them, having them overheat and shutdown I finally broke down and bought a Pyramid Gold amp for about $300. I am still using this amp today.

Always buy with quality in mind. In the long run you'll save money because 1 good product will last longer than 4 cheap ones.

LK

Re:Oops...

[dago]

Yes, but this kind of argument comes only after. And in rare case does people mention that a (put-your-favorite-brand) 300-350W PSU is more than enough for a normal* PC.

As I said, my dual AMD which consume max 170W from the wall !. I guess an athlon64 won't go over 120W ... if I have time, I'll check at work tomorrow and I'll tell you back.

Note : there are some reasons to buy oversized PSU. But just saying that a 500W PSU is needed for an Athlon64 is pure BS.

* = not overclocked, not intel P4, no top graphic card

Re:Oops...

[dago]

Ok, a athlon64 3000+, a radeon 9200, 2 HDD, 1 DVD are consuming from the wall, 80W idle and 130W max (3d+cpu). With a cheap 300W PSU, not stability problems.

OT, Your Sig

[Lord Kano]

You know you've been IMing too long when you almost say 'lol' out loud to a non-geeky friend...

You n00b!

IMs indeed. lol

LK

64 ~= 32?

[Zx-man]

Excuse me, but did they mention that any AMD64 processor comes out of the power-off state as well as every interrupt (even the _timer_ one) in the 32-bit mode? It, _architecturally_, isnt even close to the level of the IBM G5 or SUN Spark IIIi!

Re:64 ~= 32?

[Blaskowicz]

I thought PCs "comes out of the power-off state" in 16-bit mode :D

AMD seems ready to compete

[foidulus]

against Intel now in the laptop market as well. AMD has steadily been gaining against Intel is the desktop market, but I really think that the laptop market is the place to be(in terms of cash of course) Also nice that they are attacking at both ends, the performance end and the budget end.
With the advent of wi-fi, I see a lot more people ditching desktops all together and using a laptop as their only machine. Why not? Laptops easily have enough power to check email, browse, use an office suit all at the same time. And you can take them anywhere. And with a low end laptop only a few hundred more than a lowend desktop, there really isn't the financial motivation to get a desktop anymore.
Desktops aren't dying, but the real growth(at least in terms of the mature world economies) will probably be in laptops in the coming years.

Re:AMD seems ready to compete

[drinkypoo]

Also nice that they are attacking at both ends, the performance end and the budget end.

Computers are like cars in that you sell a lot more budget models, but people judge you based on your top of the line. Dodge was getting creamed in the sixties until they brought out a whole line of muscle cars that walked all over GM and Ford. Nissan had no sports cars and all but died. Turns out it's hard to spin a really sexy marketing campaign when your flagship's a minivan. The same is true of computer parts. Try getting someone [that is not a slashdot regular or similar] excited about a transmeta chip. "Transwhata?" But show them a 3GHz celeron from intel and they'll think it's fast just because of the name.

AMD has no choice but to make both high and low end chips. Low end is the proverbial meat and potatoes, you can sell absolute loads of them and the more you sell of something that has a high research cost, the cheaper each one becomes, right? Well, from a certain standpoint. But, you can't sell cheap chips in volume unless you have a good name, and you can't have a good name without fast chips.

Re:AMD

[Luke-Jr]

That used the be the case many years ago, but these days, AMD = superior quality

Re:AMD

[KillaKen187]

Well... since you ignored my wikipedia link, let me show you some examples of AMD's quality [from http://en.wikipedia.org/wiki/AMD]:

In 2001, Intel released the Pentium 4 architecture (code-named Willamette) which had a radically different microarchitecture than the Athlon or the P6 cores. While sporting a dramatically higher clock rate, the per-clock architectural performance of the Pentium 4 appears to be much slower than the Athlon or even Intel's own P6 core based processors. This lead some to believe that the Pentium 4 had higher performance because of its higher clock rate, despite benchmark performance.

AMD responded with a new K7 core (code-named Palomino) which had superior memory pre-fetching mechanisms, SSE (a set of floating-point extensions first featured on the Pentium III), an on-chip L2 cache and also re-branded them based on Model numbers which would approximately project the clock rate relative performance of these new Athlons versus the earliest versions of the Athlon. The net effect of this was for the Model numbers to be more comparable to the Pentium 4's actual clock rate. For AMD processors of a given Model number, the comparable Pentium 4 by corresponding clock rate showed rough parity on performance in a wide variety of benchmarks.

Hmmm... looks like AMD only continues to make Intel look stupid in proformance, year after year. So with that in mind does profomance = quality, I would think in the processor world it would be. You are right being #1 in a market doesn't make you the best, but having better benchmarks does. If AMD were poor quality they would have died off long ago, but yet here they stand. Ain't competetion great?

Proud AMD User

Mobile Athlons the Celeron 300A of 2004!

[mr_rizla]

I picked up a Mobile Athlon XP (unlocked) 1600+ last week here in the UK. Its now running happily at 2.3GHz, equivalent of a Athlon 3400+, and the best bit? Only £28+vat (about 50$ total)! Now that is some cheap processing power... definitely recommend anybody on a budget to pick up one of these babies...

Averatec

[Dr.Dubious DDQ]

The Averatec "3150H" that I'm posting from here has been a pretty nice little machine. Small and light, and (most importantly for me) every single component has Linux drivers available - including the software-driven modem (drivers developed by the modem company themselves, no less).

The downside is that their support blows. I reported an annoying BIOS problem (if the "Auto-dim" feature [which automatically dims the screen when you disconnect from AC and switch to battery] is turned on, the touchpad stops working when you disconnect (or, if you started off of battery, when you connect) the AC adapter. Yes, I checked it in Windows, too - it happens there as well. No response from them at all, and the BIOS download has never been updated.

Their newer systems, sadly, also use "Broadcomm" (or was that "Broadcom"?) chipsets for the built-in wireless. Naturally - the company with the most contemptuous attitude (among wifi-chipset manufacturers) towards linux users is the one they pick...

That said, though, The "Athlon XP-M 1600+" in this machine chugs along just fine. I can't exactly play Unreal Tournament 2004 on it (While I have the DRI CVS drivers for hardware accelerated 3D installed, the "S3 ProSavage/DDR" is not known for its high performance...) but for just about everything else it's been great.

Impossible specs? AMD64/nVidia/17" TFT/laptop

[Wills]

Has anyone got a suggestion for a laptop with

• AMD64 CPU at or above 3000+ AMD rating
  --- must have
  
  
• nVidia graphics
  --- not ATI or anyone else
  
  
• 17" TFT display
  --- not 15"

It seems every laptop fails to meet one or more of my above criteria.

Re:Impossible specs? AMD64/nVidia/17" TFT/laptop

[Anonymous Coward]

Has anyone got a suggestion for a laptop with

I doubt you're going to find one.

16" or larger screens

Not to mention the fact that 17" LCD makes for a huge laptop that won't fit in most bags. (I'd rather have a hi-res 14-15" LCD with 1400x1050 or 1600x1200 for portability, using an external monitor for the size.)

Heatspreaders Don't Help

[Anonymous Coward]

Sorry, but the heatspreader doesn't help anything. It doesn't increase the core area any (for better heat dissipation) but it adds an extra discontinuity to the metal-to-metal contact from core to heatsink. Furthermore, every serious overclocker knows that the P4 heatspreaders are far from flat surfaces, and the bolder pry them away before installing the CPU (and the heatsink) -- dunno if AMD makes them better though.

Ditto with heatspreaders in RAM sticks: they are just cosmetics that serious overclockers get rid of for better results.

Overclockers.com et al. have lots of good articles on the subject.

Yes, I too wonder why AMD and Intel bother with heatspreaders. They do protect the fragile core from they heavy heatsinks of today...

Re:But..

[Anonymous Coward]

But.. (Score:-1, Troll)

To quote Nelson, "Ha Ha!"

Re:Using mobile chips in a desktop--Done it

[Brian Stretch]

I put a Mobile Athlon 64 3200+ (62W) in my ASUS K8V Deluxe motherboard, replacing a standard 3200+ desktop chip. It works exactly as you'd expect. Getting a heatsink to fit was tricky since the notebook chips are "lidless" (no aluminum lid protecting the core), but Zalman's CNPS7000A-AlCu (don't use the all-copper version, it's twice the weight) fit. Alpha's didn't. Just be real careful not to overtighten the screws. Supposedly the lidless CPUs can be cooled better but that lid was put there for a reason.

That said, you can undervolt most of the Athlon 64's quite a bit. I've used ClockGen under WinXP to undervolt the DTR 3200+ in my notebook to 1.3V @ 2GHz, down from the standard 1.5V. Perfectly stable. See here, and check silentpcreview.com (which got me started on all this).

Anyone know how to manually set core voltage levels under 64-bit Linux? Then verify that I didn't push things too far?

Re:AMD

[sigaar]

"AMD = poor quality."

Yeah, quality does tend to drop dramatically if you're stupid enough to remove the heatsink from a running CPU.

Re:i'm sorry, but you have no idea what quality is

[sigaar]

Can't say what the figures are, but I've replaced more Intel CPUs than AMDs. Not that that means anything. CPUs fail every now and then for various reasons, regardless of their perceived "quality." If you put an AMD chip on a quality motherboard you won't have problems with it. If you put an Intel chip on a crap motherboard your chances of having trouble increases dramatically.

"quality is measured in #/% defects"

While we're on the subject of "quality" let's not forget the pentium bug, shall we?

http://support.intel.com/support/processors/pent iu m/ppiie/
http://kuhttp.cc.ukans.edu/cwis/units/IP PBR/pentiu m_fdiv/pentgrph.html

Intel have quality issues from time to time too.

uhhh

[nilbog]

I have an old tablet PC that has an AMD 100Mhz processor in it. The story is incorrect.

Re:Will they support Ninnle?

[Anonymous Coward]

This is a troll, why?