What Makes Apple's Power Mac G5 Processor So Hot

An anonymous reader writes "58 million transistors can drive a lot of power. Apparently, Apple appreciated the choices IBM processor architects made when designing the 970 family. This article provides the 64-bit architecture big picture for the 970 family (A.K.A. the Power Mac G5) and the critical issues in IBM's 64-bit POWER designs, covering 32-bit compatibility, power management, and processor bus design."

Comments

anyone else noticed how COOL the AMD-64 chips run?

[linuxbaby]

On a related note, has anyone else noticed how COOL the AMD-64 chips run?

I got my first AMD-64 last month and was assuming it would be burning hot (fast = hot, right?) - but was surprised that even under heavy load, I was able to hold my hand on the heatsink and it was barely warm.

Can anyone explain why this is?

Circular Logic

[WormholeFiend]

If it's so hot, maybe it's not cool enough.

What Makes Apple's Power Mac G5 Processor So Hot?

[Anonymous Coward]

Insufficient cooling?

What makes it so hot (abridged)

[Sensible Clod]

Basically, we took one of our superchips that go into superservers, with a gitastic cache and frontside bus, stripped it down a bit so we don't cut into our own market, and gave it a new name. Isn't that cool?

Re:What makes it so hot (abridged)

[TAGmclaren]

Basically, we took one of our superchips that go into superservers, with a gitastic cache and frontside bus, stripped it down a bit so we don't cut into our own market, and gave it a new name. Isn't that cool?

For us Apple users - you bet!

Well, except for the fact that the processor's so hot, but you know what I mean :)

Re:What makes it so hot (abridged)

[malfunct]

Its not that apples (and most likely some of the x86 based computers too though I haven't really looked into it) aren't supercomputers. Its just that given the current definition of a supercomputer its just not that amazing anymore. Maybe its the definition that needs to change rather than you being upset at people that follow the definition.

Re:What makes it so hot (abridged)

[William Tanksley]

Yes -- because it's part of the ongoing merging of the POWER line with the PowerPC line of chips.

-Billy

Re:What makes it so hot (abridged)

[John Whitley]

Wow. Artful and elegant rebalancing of engineering tradeoffs for very diferent markets reduced to a knee-jerk oversimplification in one fell swoop. And it got a +5 Insightful for that, too boot. Here are some reasons why the "stripped it down a bit so we don't cut into our own market" statement is ridiculous:

1. If selling POWER series chips to Apple was going to undermine IBM's server business, IBM would have a hell of a lot more to worry about from the plain 'ol x86 market.
2. IBM's POWER-series chips are designed to trade away ultra-high-speed clock rates in favor of low failure rates. The design rule (feature size on chip) is pulled back from the bleeding edge and other layout techniques are employed to make these processors rock solid, to avoid costly downtime from hardware failures in business servers.
3. These days Apple is well known for its forays into the cluster computing space -- but that's a far cry from the sort of transactional throughput capacity of IBM's high-end servers. I.e. not the same markets!

Re:What makes it so hot (abridged)

[dthree]

Maybe the reality distortion field isn't calibrated properly.

Re:What makes it so hot (abridged)

[Sentry21]

IBM's POWER-series chips are designed to trade away ultra-high-speed clock rates in favor of low failure rates.

In other words, IBM's customers will pay $2m for a computer that has one processor out of 64 fail once or twice in its operational lifetime thus and has no downtime, instead of $800,000 for a cluster with ten times the processing capacity and an average of two minutes of downtime a year.

These may not be actual numbers (it's a pretty extreme example), but that's the philosophy. Pay more for reliability.

--Dan

Whoah, here's the problem...

[LegendOfLink]

...somebody set this thing to "evil!"

64 bit integers

[Xpilot]

From the article:

...64-bit processors also accelerate complex mathematical calculations through their ability to perform calculations directly on 64-bit numbers...

Don't they mean 64-bit integers? Since floating point registers in most modern CPU's are 64-bit wide already.

Re:64 bit integers

[lisaparratt]

It can perform the 64-bit floating point operations correctly, but it has to load the numbers in a two step process, 32 bits at a time, so it's not direct.

Re:64 bit integers

[francisew]

More importantly, it doesn't have to break the 64 bit operations into many successive 32 bit operations. 64 bit operation are not simply 2x32 bit operations, but can be several dozen operations.

An 8-bit microcontroller can perform 64 bit floating point operations correctly. It just takes a long time.

Re:64 bit integers

[Waffle Iron]

Since floating point registers in most modern CPU's are 64-bit wide already.

Actually, since most modern CPUs are x86 variants, the floating point registers are usually 80 bits wide (and have been since the 1981 introduction of the 8087).

As far as "complex mathematical calculations" go, 64-bit integers aren't really that big a deal. It's pretty rare to need integers bigger than 2^32 but no bigger than 2^64; floating point usually handles big numbers more flexibly.

The big deal with 64-bit CPUs is 64-bit address pointers and operations on them (which usually aren't more complex than adding and shifting).

Re:64 bit integers

[Anonymous Coward]

Unless "complex mathematical calculations" means exact bignum arithmetic where 64x64->128 multiply is likely to provide 3-4 times speedup compared to 32x32->64..

Re:64 bit integers

[cimetmc]

Sorry, but you are completely wrong here.
A lot of 32 bit processors already have a 64 bit data bus right now. This is for example the case for the Intel 32 bit processors. So loading an entity bigger than 32 bit is not an issue. The bus width between the processor core and the L1 cache is even wider allowing even bigger chunks to be loaded in one cycle.
As for the floating point unit, it is also designed to do the operations in double precision (64bit) or even more. Once again, full 64 bit processors have no advantage here.
So for floating point numbers, having a full 64 bit processor compared to current 32 bit processors does not give you any speed advantage. The only advantage that remaines is the bigger address range.

Marcel

Re:64 bit integers

[shotfeel]

IIRC, even the G4 doesn't need to break 64 bit floating point operations into 32 bit operations. IOW, the floating point units in the processor have been 64 bit for some time, its the rest of the architecture that's just "cacthing up".

Re:64 bit integers

[Fulcrum of Evil]

A lot of 32 bit processors already have a 64 bit data bus right now. This is for example the case for the Intel 32 bit processors. So loading an entity bigger than 32 bit is not an issue.

This is not relevant to the instruction stream - you still need two load instructions. Actual bus widths are not visible to executing code - it's simply there to improve bandwidth.

Re:64 bit integers

[fitten]

I don't know of any FPU made in the last 20 years that have to break 64-bit FPU up into 32-bit operations. They make take longer (FPU optimized for 32-bit operations so 64-bit take a few cycles longer) but they are still handled natively.

The previous poster was talking about 32-bit and 64-bit integer values.

Re:64 bit integers

[dgatwood]

Of course, your 64-bit PPC machines have an elastic bus that can do multiple transactions in-flight concurrently. Memory bandwidth will have a much smaller effect on G5s by their very nature.

Also, since the G5 has many, many more registers than your 32-bit x86 machines, it can pre-flight those requests much earlier, thus in many cases, you'll see essentially zero performance hit on the G5 where the x86 CPU would see a significant hit, not because of actual memory bandwidth differences, but because the data was already there before it was needed on the G5.

Re:64 bit integers

[fitten]

This isn't anything new... a good compiler can/will rearrange loops and such so that it initiates fetches for the n+1 iteration during the nth iteration (this is regardless of platform). Other CPUs in the past had special instructions that were specifically designed to initiate memory reads in order to load the caches (L1 and/or L2) without actually doing anything with the data being loaded (not actually put it in any register, for example) simply so it will be in cache when it is needed.

Transactional buses are good for other things, though (more so than preloading data, as mentioned here).

Re:64 bit integers

[dgatwood]

Pentium 4 has eight 32-bit general-purpose registers. The G5 has 32.

You're right that P4 has a lot of rename registers, but those aren't directly accessible. Thus, their usefulness is limited by the CPU's look-ahead. In theory, a compiler can always do better if it has access to the same number of registers because it can look arbitrarily far ahead.

Oh, but you were trolling. You didn't actually expect an answer, did you? Well, you got one.

Nice title...

[gmuslera]

for a moment I thinked that the question was related to temperatura, as in "What makes AMD Processors So Hot?", but in the article temperature issues are almost not touched (just a suggestion of checking liquid cooling, could be a hint?)

Re:Nice title...

[Anonymous Coward]

It's not a matter of total heat dissapation as much as the fact that the head is dissapated in very small areas of an allready tiny chip.

The PPC970FX actually doesn't produce that much heat compared to the current AMD and Intel crop, only dissapating 54W typical at 2.5Ghz (albiet a good deal higher at peak). But due to it's smaller die, it has to dissapate that heat from a smaller area, thus requiring a cooler heat sink to dissapate it into to maintain the same die temperature.

It's basic thermodynamics, but something geeks often overlook.

Re:Nice title...

[shotfeel]

is there another reason to use liquid cooling, other than excessive temperature?

Not sure if this is the reason, but its not just a matter of amount of heat generated. Its the amount of heat generated in a given area. So if you generate the same, or even less, heat in a smaller area, you may need to resort to something more efficient that air cooling to do the job.

Re:Nice title...

[DigitalRaptor]

As I said very clearly in my post, and I'll say it again and try to use smaller words so you can understand, the noise (sound) is only an issue when there is too much heat.

If the processor didn't produce so much heat, the cooling could be achieved with either passive cooling, or slower rotating fans and noise (sound) wouldn't be an issue.

Yes, one of the nicer side effects and possible motivation for using liquid cooling is that it is quieter, but the ENTIRE PURPOSE of liquid cooling is to COOL the processor BECAUSE THERE IS TOO MUCH HEAT.

Join the club. Would you like some milk to help that go down smoother?

Re:anyone else noticed how COOL the AMD-64 chips r

[Noah Adler]

I was able to hold my hand on the heatsink and it was barely warm.

It could be because there's inadequate conduction between the CPU core and the heatsink. Check the temperature monitors to make sure it's actually as cool as you hope it to be. It could be that just most of the heat is staying in on the CPU, which would be a bad thing. Hopefully you've already checked this though.

ob Memory

[GillBates0]

18 exabytes ought to be enough for anybody

-GillBates0, 2004.

Re:ob Memory

[Mark_in_Brazil]

18 exabytes ought to be enough for anybody

-GillBates0, 2004

So where does it end?

This page makes a fairly convincing argument that 256 bit CPUs should be enough (basically, there would be no way to exhaust the amount of memory a 256 bit CPU could access, because the number of memory locations is about the same as the number of atoms in the universe).

--Mark

Re:ob Memory

[gfody]

because the number of memory locations is about the same as the number of atoms in the universe

but each memory location takes more than a few atoms to construct.

Re:ob Memory

[brsmith4]

Dude, that's insane. You'd have to build a ram stick with more mass than the entire universe to exhaust the memory addressing capability of a 256 bit chip.

All i can say is "Whoa".

Re:ob Memory

[PureCreditor]

Umm....what about addressing subatomic particles ? =)

Re:ob Memory

[danalien]

I'd have to agree that it looks like something unthinkable

Buuuttt, if one revises one POV to account for 'Quarks (subatomic particles)' and/or 'String Theory (that there is something beyond atoms and quarks....)' ... the unthinkable just got a nudge toward the thinkable

As Insane as we Humans (the Human Species as a whole) are ... I wouldn't bet against Us from one day doing/achieving the unthinkable :-)

I mean, if an alien race came by 2.5-2.0 Million years ago and saw us then, 'Stupidest Humans that we think has ever lived(Homo habilis)'. Who of them would have thought the unthinkable that we would evolve to 'Insane Rocket scientists(Homo sapiens)' that one day would have the capacity to shoot one of em' down? =)

Re:What Makes Apple's Power Mac G5 Processor So Ho

[sky289hawk1]

The fan belt on its search engine doesn't drive the radiation air emission unit fast enough! SHE NEEDS MORE POWER!

Teflon underside

[suso]

They should make the G5 powerbooks have a teflon underside.

"Turn it over, and you can cook dinner".

Re:Teflon underside

[21chrisp]

My old OC'd PIII doubles as a space heater.

Re:Teflon underside

[suso]

No kidding. Recently we moved two of our computers from our house office into another real office. The average temperature of the upstairs floor dropped 3 degrees F. It was amazing how much heat those were generating from the CPUs and hard drives.

Re:Teflon underside

[Reducer2001]

Ah ha! The REAL cause of global warming. OC'd CPU's.

Re:Teflon underside

[NathanBullock]

Imagine the problems trying to type while your laptop keeps sliding around on your lop.

Re:Teflon underside

[Spydr]

teflon? you mean so you don't stick to it when you lift it up after a long day of work?

i'll be waiting for the thermal tiles from the space shuttle to be added to mine.

Re:Teflon underside

[Fulcrum of Evil]

They should make the G5 powerbooks have a teflon underside. "Turn it over, and you can cook dinner".

That's all well and good until some tool uses a metal spatula on it. At least frying pans are cheap.

Re:Teflon underside

[sp00]

Why turn it over? Just make a heatsink cooker like this guy did.

Re:Teflon underside

[kurt555gs]

I would love to see a G5 powerbook (I have a dual G5 Powermac)

One thing I have noticed is that when I do very processor intensive tasks the fans spool up like a B-52 at take off.

I see the 2.5 Ghz Powermacs have water cooling, which i am sure will be needed in a G5 PowerBook.

Now the only thing left of concern is a battery the size of those used in Cat tractors will be needed to run the thing

Cheers

Re:Teflon underside

[aflat362]

I would also love to see a G5 powerbook.

I have a pretty good G4 powerbook now - it keeps up well with most apps - but anything real processor intensive it kinda bogs down.

BTW I'm jealous of your Dualie G5. Must . . resist . . urge . . . to purchase.

Re:Teflon underside

[Ilgaz]

poor me with 1600mhz G5...

Anyways, the fan system control by kernel does one thing. Processor heat rise is predicted so to keep system low sound, when processor hits 100% by load, for slight time, that B52 full speed starts. Processor gets "advantage" cooling than it gets back to normal. So, you don'T have to hear a fan fighting to get stable cpu speed.

Its called "predictive cooling" or something.

Of course, irrelevant on monsters like 4:2:2 to mpeg2 conversion or protools.

Increased Pointer size

[trigeek]

One side-effect of 64-bit computing that I don't hear a lot of discussion about is the increase in the size of a pointer. A standard implementation of a linked list of integers will now be 50 to 100% larger (depending on if you use 32 or 64 bit integers), simply because the pointers take up more space. If I bought a 64 bit system, simply because it's the "Best", but only got 1GB of RAM, I have less useful memory, because the pointers take up all of my physical RAM. Do the architects of these systems take this into account?

Re:Increased Pointer size

[Eccles]

In most meaningful, sizeable programs, pointers aren't a significant chunk of memory usage. (And for small programs, it doesn't matter.) I would think most modern apps consume most of their memory storing images, which aren't affected by the 32->64 change.

Also, 64-bit pointers allow you to go from a max of 4GB of RAM to 16 billion GB, so the assumption is memory prices will keep dropping and you'll have much more than twice as much RAM on your 64-bit system anyway.

Re:Increased Pointer size

[Myolp]

After reading the privous post, I somehow got to think about viagra spam... "Increase The Size Of Your Pointer Now! Buy A 64-Bit PowerMac G5 And Get That Extra Size Today!"

Re:Increased Pointer size

[zulux]

A linked list is just a container, so yes, if you're using your linked list - pointers will eat a bunch of space if you're storing small bits of data.

But then again, you don't have to use linked lists. The C++ standard template library has all sorts of wonderful containers that may be better suited than simple linked-lists. Java has some neat containers as well.

Re:Increased Pointer size

[HeghmoH]

On Mac OS X, dynamic memory has a granularity of 16 bytes. That means that if your linked list node is only 8 bytes (4 bytes data, 4 bytes pointer) then it will get a 16-byte allocation anyway, and you'll waste the extra space. Using 16 bytes per node won't hurt at all. Allocation overhead makes the standard linked list a fairly wasteful way to store data anyway.

Re:Increased Pointer size

[hattig]

You'd use 32-bit pointers if your dataset would be under 4GB (well, less than that due to other factors, but you get the point) in size, otherwise you'd have more than 4GB of memory installed and it would continue to not be a problem.

Anyway, a linked list of integers of 32-bit size would get 50% larger with 64-bit pointers. A linked list of 64-bit size integers would get 33% larger. Most linked lists don't merely hold a single item however, so the overhead wouldn't be that bad in the common case.

Re:Increased Pointer size

[21chrisp]

Since the AMD64 instruction set can also run code compiled for 32 bit, it can also work with 32 bit pointers. However, any code optimized for AMD64, will probably contain all 64 bit pointers.

I guess if you plan on shelving out for 64 bit, you should plan on getting more ram w/ it.

Re:Increased Pointer size

[tomstdenis]

Use a different data structure?

IIRC also a b-tree like structure could help this. Where you have to have pointers to nodes the actual elements ["leaves"] in the nodes would not need individual pointers.

Also this isn't much of an issue. If you're likely to have a huge database you're already likely to require 64-bit offsets [into memory/file/etc] as 4GB is often not enough.

So not only can you mitigate the "downside" to larger pointers by using more efficient data structures but the issue is largely moot anyways.

And FYI my Gentoo AMD64 box running Xorg6.8.x, GNOME, gaim, xmms, a few xterms, gedit and nedit only is using 200MB of ram [prolly cache too]. Which is comparable to when I was running the same software on my P4.

The bulk of memory used is for code [which could be shorter as the instruction set is more powerful] and it's data [which largely doesn't change with register size].

Tom

Re:Increased Pointer size

[rjshields]

A standard implementation of a linked list of integers will now be 50 to 100% larger (depending on if you use 32 or 64 bit integers)

Use a linked list of arrays of 10 ints instead and use only 5 to 10% more space. They're only ints.. allocating memory for a few more then neceassary won't hurt in most situations :)

Re:Increased Pointer size

[tomstdenis]

It wouldn't be safe to assume a pointer can be stored in 32-bits in 64-bit mode. The OS can map your data anywhere in the 64-bit space [which maps physically to a 40-bit bus, at least on my AMD64]. It's also not portable to do that. Even if your OS maps stuff within 4GB [e.g. top 32-bits are zero] another 64-bit box [sparc, ppc, etc..] may not [and likely would not] do that.

If you build in 32-bit mode [e.g. -m32] you lose the major benefits of the 64 which is namely the extra registers.

Tom

Re:Increased Pointer size

[Jimmy_B]

[blockquote]A standard implementation of a linked list of integers will now be 50 to 100% larger[/blockquote]

That's assuming each node has a 'next' pointer. That's how they teach it in CS classes, but in the real world it makes more sense to use an index into an allocator. It requires cleverness from the language and library designers, of course, but there isn't necessarily a size increase.

Re:Increased Pointer size

[lisaparratt]

On Silicon Graphics 64-bit machines, this was solved by having two ABIs, one 32-bit, one 64-bit. You could still use 64-bit operations in either mode, but pointer size depended on the width of the ABI. This allowed you to optimise memory use on small memory jobs, but still have access to vast amounts of RAM when you had to, just by choosing the right ABI for the job.

Re:Increased Pointer size

[Fulcrum of Evil]

If I bought a 64 bit system, simply because it's the "Best", but only got 1GB of RAM, I have less useful memory, because the pointers take up all of my physical RAM. Do the architects of these systems take this into account?

Umm, duh? You increase your footprint by a small amount (say 10%) to get access to a flat 16EB address space. With memory running $200/G for the good stuff, why not?

Re:Increased Pointer size

[Anonymous Coward]

This is an excellent observation. Please forgive me if this seems offtopic -- my primary development platform is PC, however I've got quite a bit of experience working on 64 bit.

From a developer perspective (at least in software I've worked on) programmers often do not realize that a vast number of projects took dependencies on 4 byte pointers.

Structure alignment, pointer size, etc have plagued all projects I've worked on, even after a cursory perusal at the code indicating "It looks good". Lots of bugs in porting code to 64 bit can be hidden until the right things happen.

Your particular example probably overexaggerates the point (I doubt many modern apps use a majority of memory in integer linked lists). When you think of standard server apps, most data is serialized request data which is a significant constant overhead. For the average desktop user, typing documents & email is their primary utilization of a PC, in which case it shouldn't have a tremendous effect.

That being said, there is no way your memory usage will become more effecient :-).

One additional reason that your memory will drop -- on 64 bit platforms compiled binaries become much larger. This plagues the IA64 chipset more than any other -- due to their "four simultaneous operations" (forgive me if I'm misrepresenting -- this is only through hallway chatter) binaries get an order of 2 to 4 times larger (mostly full of NOPs, no less).

Re:Increased Pointer size

[the_2nd_coming]

umm, if you only got a gig of RAM on one of these systems, then you are not running the software that would require more memory.

Re:Increased Pointer size

[spaceyhackerlady]

On Silicon Graphics 64-bit machines, this was solved by having two ABIs, one 32-bit, one 64-bit.

Sun machines with UltraSPARC processors do this too. They run 64 bit kernels, and applications are 32 bits. Unless you actually need 64 bits, in which case you feed the compiler some differnet options and it makes a 64 bit executable for you.

Both Solaris and Linux do it the same way. When you build a kernel for Linux on an UltraSPARC machine the part about kernel support for different kinds of executables offers you (among other options) 32 bit ELF (which you need), 64 bit ELF (optional), and Solaris emulation (never tried it...).

...laura

Re:Increased Pointer size

[TeknoHog]

umm, if you only got a gig of RAM on one of these systems, then you are not running the software that would require more memory.

Address space is one thing. Register size is another: I could be crunching lots of 64-bit numbers, yet stay within 1GB of memory consumption.

Re:Increased Pointer size

[SnapShot]

The C++ standard template library uses MAGIC PIXIE DUST and MOONBEAMS for the data structures. Java Containers, on the other hand, are held together with PURE THOUGHTS and KITTEN WHISKERS. That's how they avoid using pointers... ;-)

Re:Increased Pointer size

[Midnight Thunder]

One of the better approaches, though a little more complicated is linked arrays. It has the nice advantage of being speedy when iterating through the arrays and can be extended as the room in the current arrays runs out.

Re:Increased Pointer size

[Anonymous Coward]

..and when you have 12 bytes of data and a 4..8 byte pointer?

(Of course efficient linked lists don't malloc every node, nullifying both your point and my counterpoint here)

Re:Increased Pointer size

[Waffle Iron]

With memory running $200/G for the good stuff, why not?

Because bigger pointers eat more cache, and external memory bandwidth is already the biggest performance bottleneck in most systems.

Re:Increased Pointer size

[Jeff DeMaagd]

The instruction density is probably pretty similar. A 64 bit CPU does not automatcially mean a 64 bit instruction word, most are 32 bits for a single instruction.

Re:Increased Pointer size

[Anonymous Coward]

So in your 64bit OS, with your 64 bit application you run into overflow errors in years to come.

We recently ran into a serious undiscovered bug in the original core of our application.
Some wise ass developer thought that a 16bit integer would be enough for storing the index.
All was fine for years and years until the first person had >16383 individual transactions in a job.

Whilst you can't do anything about true future proofing, you can think about handling 64bits now, and making sure your application can expand flawlessly into that realm should be considered.

It does also involve thinking logically, there is no point in using absolute memory references in a small array, but for expandable dictionaries and trees, its better to code in simple chunks.

Re:Increased Pointer size

[fitten]

This probably depends on the compiler (and compiler flags). This is an optimization by the compiler to match memory allocation to cache line length for better performance. If you look, you'll find compiler options to allocate on whatever boundary you want.

Otherwise, it would just be a design of the memory allocation libraries (always allocate on 16-byte boundary) for similar reasons (and to lessen memory fragmentation).

Still good information to know about Mac OSX.

Re:Increased Pointer size

[pclminion]

A standard implementation of a linked list of integers will now be 50 to 100% larger

Yes, but how often do you have a realistic linked list that only has a single data element per node? Even with 32-bit pointers that would be a ridiculous waste. A linked list of integers is a first year programming assignment, not a real application.

The answer is obvious -- assuming the memory allocator has good locality, use 32-bit integer offsets instead of 64-bit pointers. To follow the link, take your own address and add the offset to obtain the 64-bit pointer to the next node.

Re:Increased Pointer size

[budgenator]

modern apps consume most of their memory storing images,
interesting that yu should mention this, Hi-End graphics are moving From 8 bit to 16 bit. This is 16 bits per channel so a RBGA image is 16 X 4 bits /pixel or 64 bits per pixel resulting 2.81475e+14 colors and frames exceeding 70 MB each. Store a hundred frames in a flip-book (about 4 seconds of movie) and you need a shit-pile memory.

Re:Increased Pointer size

[Fulcrum of Evil]

Because bigger pointers eat more cache, and external memory bandwidth is already the biggest performance bottleneck in most systems.

If it's a big deal for you, just run as a 32 bit process (you can do that). You just won't get the 16EB goodness that comes with running 64 bit.

Re:Increased Pointer size

[HeghmoH]

Yes. Go watch Fight Club.

Re:Increased Pointer size

[Atzanteol]

I was *going* to respond to the parent, but I don't think I could top your post. :-)

Re:Increased Pointer size

[TheRaven64]

Actually, I believe that this limitation is imposed for compatibility with the AltiVec unit. The AltiVec unit (or units on a G5) requires input data to be aligned on a 16-byte boundary. To assist with this, OS X automatically 16-byte aligns all allocated memory for you. Of course, this doesn't help when you want to send some data from the middle of a block of memory to the AltiVec unit (as I recently did, and eventually worked out that the byte alignment code would cost me more than doing the same thing on the standard integer unit. I want an Itanium.)

Re:Increased Pointer size

[morcheeba]

That's probably a malloc/gcc limitation. However, if you roll your own memory system, you can pack it any way you want. We did that and it helped a lot (we also got to pack it so our data was more cache-efficient) -- it's not as hard as you'd think if you build a good infrastructure for your programs.

Re:Increased Pointer size

[faragon]

It's up to the compiler and OS used. Back in 1998 I was developing software on an Alpha 21164 (using DEC UNIX and the built-in compiler), building huge tree data structures, and the 'sizeof' of a pointer was 4 bytes -32 bits- (sizeof(int) = 4, sizeof(long) = 8; as the 'long' data type usually represents the uP word length).
(could be nice that someother could report the 'sizeof' pointers for other 64 bit systems; as my example could be some kind of intermission/transition stage)

Re:Increased Pointer size

[tlindner]

I think IBM did. They allow host operating systems to implement a 32/64 bit hybrid machine where each proccess gets to choose what mode the processor is in upon execution.

So for applications that don't need a 64 bit address space they don't need to have it.

Re:Increased Pointer size

[bar-agent]

As an aside, did you know that Java has garbage-collection as a side-effect of using pure thoughts to hold its data structures together? And the kitten whiskers are what allows the JIT compiler to work so well.

On the other side of the fence, C++'s usage of magic pixie dust and moonbeams is what gives it its flexibility -- and by that I mean the ability to program in the procedural style or object-oriented style or template style, etc. The downside is that magic and moonbeams aren't really...reliable. So you end up shooting yourself in the foot, and every compiler is different.

Re:Increased Pointer size

[fitten]

Only for immediate data. Most data will reside in memory locations and/or registers. The op code (and operand) size for these operations shouldn't be any larger than the ones for 32-bit values.

Examples:
add r1,r2,r3 // should be the same size regardless of the size of the data manipulated
load r1, (r7) // same

Immeditates will possibly be effected, depending on whether the opcodes exist to use the smaller immediates.

Examples:
load r1, #1 // should be opcodes smart enough to load this value (1) using a small number of bits
add r1, #256 // same

Re:Increased Pointer size

[wavedeform]

Doesn't the sound of "un-gotten" jokes whizzing over your head get a little distracting, sometimes?

Re:Increased Pointer size

[rreyelts]

On a more serious note, almost all C++ applications prefer std::vector to std::list, and almost all Java applications prefer ArrayList over LinkedList - neither of which require any pointers.

These data structures tend to perform much better, because they are naturally cache friendly. It can take over 100,000 elements in a list to make it more expensive to do an insert or a remove in an array-based list over a link-based list.

In a language like C++, you can try to mitigate these factors by using a special-purpose allocator that greedily reserves large pools of contiguous memory to improve spatial locality.

Re:Increased Pointer size

[aphor]

I would think most modern apps consume most of their memory storing images, which aren't affected by the 32->64 change.

You may forget that a bitblit, or a bulk memory copy operation can be accomplished in half of the time using the same number of 64 bit registers as 32 bit registers. How do you think common operations like scaling and color transformation will be affected by the increased register size and memory IO path? In my experience (Ultrasparc real world apps like GIMP and OpenSSL) most bulk integer compute operations complete in 10-20% less time when run in 64 bit mode vs 32 bit mode on the same computer (probably potentiated by L1-L2 cache performance differences in each mode), and they consistently consume about 1/2 the userland CPU cycles during that time. The biggest payout in 64 bit computing I have found is using OpenSSL with the 64 bit assembly code for encryption routines and having the 'bn' (big number) math library in 64 bit mode: I could scp database dumps across the network at full speed without dipping into enough cpu cycles to affect normal operation.

Re:Increased Pointer size

[mmkkbb]

There is no way to load 32-bit immediate data with a 32-bit instruction word anyway!

Re:Heat Problem Back Ground

[jrockway]

I'll probably get modded down for saying this, but I have Karma to burn...

That article is some crybaby whining about how expensive the G5s are. "Apple is so dumb. Why would anyone pay that much for a CRAP computer," is what the article sounds like. I think that guy needs to take his superior knowledge elsewhere and try some benchmarks ("512K of cache isn't competitive for $3000")... apparently it is because it's winning benchmarks and people are buying them. Just because you can't afford it doesn't mean it's a bad computer (doesn't mean it's a good computer either). This guy needs to grow up and write an article with facts instead of emotions...

The Athlon64 is very cool

[Compact Dick]

I recently upgraded to a 754-pin Athlon64 3000+, and the hottest it's ever been is 51 C, a few degrees more than the room temperature of 43C. On a cooler night, with 100% CPU load for ~2.5 hours [2-pass XviD encoding], it peaked at 47 C. Quite impressive.

Re:The Athlon64 is very cool

[strictfoo]

Holy fuck dude... your room temp is 109.4 degrees Fahrenheit (thank you google)?

Open a window or buy an air conditioner or something!

Re:The Athlon64 is very cool

[pe1rxq]

If your room is 43C you better not stay in it to long....
Is the 43C really the room temperature or the temperature at some other point in the computer? (which sounds much more likely)

Jeroen

Re:The Athlon64 is very cool

[Your Pal Dave]

a few degrees more than the room temperature of 43C

Holy Crap! You must be really roductive !

Re:The Athlon64 is very cool

[Compact Dick]

I escaped by going to work that day :-) Yes, my room's ambient temperature hit 43 C and stayed there for a long time. It was hotter around my PC, but not by much.

Re:The Athlon64 is very cool

[Compact Dick]

Holy Crap! You must be really roductive !

And if you were to move to my place, spelling errors would decline! ;-)

Cheers!

-- CD

Re:The Athlon64 is very cool

[Your Pal Dave]

Dammit, I used the preview button (and even checked those URLs). I guess I better go turn up the old thermostat.

Re:The Athlon64 is very cool

[Jucius Maximus]

" I recently upgraded to a 754-pin Athlon64 3000+, and the hottest it's ever been is 51 C, a few degrees more than the room temperature of 43C. On a cooler night, with 100% CPU load for ~2.5 hours [2-pass XviD encoding], it peaked at 47 C. Quite impressive."

Are you aware that a 43C room temperature is incredibly hot? Standard room temperature in an office is 20C. A 47C room is like a scorching Texas day in the middle of summer.

Re:The Athlon64 is very cool

[RangerElf]

You live in the Kalahari desert?

He might live in Tucson or Phoenix, Arizona; the Sonora-Arizona desert is quite a hot place in summer. Around Hermosillo, Sonora, there's a permanent high-pressure system which pushes off any clouds coming our way, so there's uninterrupted sunlight some ... 14 hours a day :-)

All through July and August, the normal, everyday noon-time temperature is over 45 C, 47 C is not really all that surprising. Once I was out in the street at 50 C.

Being a desert, it cools down quickly at nighttime; but when there's no breeze to take away the heat, it sucks evilly to be at 10 PM and still at thirty-something C. So, being at 43 at night is awful. I know how you feel though.

The best part is the faces others make when they find out about summer in Hermosillo: "But, that's impossible!" Naa, just a bit warm.

-gus

Re:The Athlon64 is very cool

[sigaar]

In some parts in the world it does actually get that hot. Ever been to Africa? How is 32C on an avarage winter's afternoon for you? Now just imagine the summer...

Anyways, I've ony ever touched an Athlon64 once. It was a 3000+ and ran 28C at idle. Granted, room temperature was 25C (office aircon). Don't know what it would do in a hotter room.

Room temperature can make an enormous difference. My PC at home is a 2400+ AthlonXP, which ran at about 45C all winter under full load - room temperature is cold, below 15C. Now with summer approaching, it's already in the 50s. Guess I'll have to invest in a proper Thermaltake or something...

Re:The Athlon64 is very cool

[Fulcrum of Evil]

Ever been to Africa? How is 32C on an avarage winter's afternoon for you?

Which part? Africa has pretty much every environment present on the planet (possibly excluding tundra). If Ghana is 32C in December, I'd expect it to be about 34C in July.

Re:The Athlon64 is very cool

[LiquidCoooled]

How do you know he *isnt* in Texas?

Re:The Athlon64 is very cool

[Compact Dick]

Could be from my handle.

Re:The Athlon64 is very cool

[LiquidCoooled]

I've been on the web long enough to not blindly accept things.

Many people have links to websites of countries they dont live in. If I took everything at face value, Christmas Island would be overrun by teenage pranksters.

Anyway, your previous post mentions a 39 deg heatwave, I don't think england has ever been that warm in Autumn (Ever?)

Re:The Athlon64 is very cool

[Compact Dick]

I've been on the web long enough to not blindly accept things.

Wise observation :-)

Anyway, your previous post mentions a 39 deg heatwave, I don't think england has ever been that warm in Autumn (Ever?)

I am currently in the southern hemisphere, residing in au.bne. Here's a report about last Saturday's heatwave.

Re:The Athlon64 is very cool

[evilviper]

your room temp is 109.4 degrees Fahrenheit

Not all that incredible. 95F is common for my place, and that's only because I happen to have very good insulation. It's 120F+ outdoors a good 6 months out of the year.

Open a window or buy an air conditioner or something!

At up to 95F, I've never had a computer overheat. In fact, they never reach 130F, so air conditioning isn't really necessary at those temps.

Opening a window would be the stupidest thing to do.

Re:The Athlon64 is very cool

[evilviper]

If your room is 43C you better not stay in it to long....

Why not? I go out running in 130F degree weather. Surely sitting around in much lower tempuratures isn't likely to kill anyone.

Re:The Athlon64 is very cool

[sigaar]

My mom lives in the north Botswana. I often visit friends in Windhoek, Namibia. In both places, in the middle of the winter, I can sit outside at six in the morning, with nothing but my shorts on, and enjoy my coffee. Where my mom lives summer days are almost always over 40 C.

Around Cape Town summer days (dry summers) often reach above 40 C - not unusual at all.

Addressing

[vlad_petric]

Most importantly, it can address up 2^64 bytes of memory. And yes, that generally implies 64-bit integer GPRS. BTW, vector operations on x86 (MMX) also operate with 64 bit registers, but it can only access 4G (32G if you use the extended bits "hack").

Re:Addressing

[geg81]

Current x86 chips (like the AMD) also have 64bit pointers.

Re:Addressing

[javax]

The classic IA32 (aka x86) only has 32Bit for addressing memory (w/o using any evil hacks).
AA64 (aka AMD64/EMT64) uses an incompatible new addressing mode to use 64Bit pointers (though it can operate in 32Bit compatibility mode, too)

Re:So sick of it

[WormholeFiend]

nah, the proper solution is to develop the convenience store liquid nitrogen market.

WATT figures for G5 vs AMD-64?

[green pizza]

Does anyone have the numbers to compare how many watts of power the G5 uses vs a similar AthlonXP or AMD64? Ie, I'd like to see how a 2.0 or 2.5 GHz G5 compares to a 2.0 or 2.5 GHz AMD processor.

Re:WATT figures for G5 vs AMD-64?

[hattig]

Apart from the 2.6GHz FX55, but that is a top of the line 130nm part, the best of a process generation. And some people are overclocking these on air to 2.8, 2.9GHz. I expect that the FX55 uses around 100W though, because it isn't a 90nm part and it is the fastest

AMD's 90nm 3500+ uses under 67W of power however at 2.2GHz.

Re:WATT figures for G5 vs AMD-64?

[Anonymous Coward]

It's hard to make a comparison because for some reason IBM/Apple doesn't want to release official measurements for power usage. Which is strange because they should do really well in that measurement compared to AMD and Intel. Here's their official numbers:

2.4 GHz A64- 89 W
3.4 GHz P4(Northwood)- 89 W
3.4 GHz P4(Prescott)- 103 W

Best guess on the 2.5 GHz G5 is around 65 W.

Re:WATT figures for G5 vs AMD-64?

[nelsonal]

First off power consumption for a processor = .5*C*V^2*F where c is capacitance, V is voltage, and F is frequency. So if you can find capacitance you can get a pretty good estimate of the processor's power needs.
From Intel's datasheets: P4 90 nm (prescott) 520-550 models 84 W of design power (what Intel recommends the heatsink be able to pull).
550-560 models 115 W of design power.

From AMD's datasheets: design power (measured with max amplitude and nominal voltage) is 89 watts for all power ratings 3000+ to 3700+.

I couldn't find a PPC 970 data sheet at IBM but ee times claims it pulls 97 watts, but speed was not specified. That seems consistent with the water cooling on the G5, my air cooled P4 is plenty loud.

Re:WATT figures for G5 vs AMD-64?

[default luser]

I've scoured whitepapers and other sources looking for this, and IBM and Apple do guard their power numbers quite thoroughly.

I did find one graph for a 2.5GHz 970FX (90nm G5) processor that showed about 60 watts maximum power usage under full load. That tapered to below 20 watts when the machine automatically reduced voltage and frequency when load was low.

I imagine they're reluctant to hype the numbers because they're not particularly impressive. The Athlon 64 90nm shrink has shown similar power levels for the same performance (as you would expect, as they both use IBM's SOI), and also features similar on-the-fly voltage and frequency scaling.

Re:WATT figures for G5 vs AMD-64?

[svirre]

First off power consumption for a processor = .5*C*V^2*F where c is capacitance, V is voltage, and F is frequency. So if you can find capacitance you can get a pretty good estimate of the processor's power needs.

That is way simplified. That would be a first order approximation for the dynamic power in a device where each node toggles at f/2. The real activity in the device must be determined (weighted by net load) and used to replace the .5 figure in the formula you wrote. In addition we have static power consumption which is quite significant at 0.18um processes and below.

Re:WATT figures for G5 vs AMD-64?

[mczak]

Here's their official numbers: 2.4 GHz A64- 89 W 3.4 GHz P4(Northwood)- 89 W 3.4 GHz P4(Prescott)- 103 W

Careful. TDP numbers between AMD and intel are not directly comparable, as intels numbers are lower than the actual maximum. This is clearly stated in the datasheet, so you might hit instruction sequences which exceed TDP (by how much is unclear, voltage/amperage/wattage numbers compared with AMD would suggest by about 20%). So if you have a marginal cooling solution just good enough to get rid of these 103W without exceeding the cpu temperature, the cpu might still throttle down in some cases.

Best guess on the 2.5 GHz G5 is around 65 W.

Compared to the 2.2Ghz 90nm A64, which has a listed TDP of 67W, this isn't that impressive. Especially since it's expected that future 90nm A64 will have the same 67W listed (just as all A64 130nm have the same 89W power draw in the data sheet, from the 2800+ to the 4000+). It is however much better than the high-end prescott P4 (if that 65W is the real maximum unlike intel's numbers that would be only half the power draw of a high-end prescott).

Re:WATT figures for G5 vs AMD-64?

[Jameth]

That's a very misleading measurement. It's been noted many times before that the information AMD released was the max that the processor line would ever reach, not the current max. None of the current x86-64 processors actually hit that 89 watts.

Re:WATT figures for G5 vs AMD-64?

[sokoban]

Here is a good article telling about the Xserve G5's CPU Power usage. It uses the 970FX chips which have slightly lower power requirements, but gives some neat data about the g5 processors.

Re:WATT figures for G5 vs AMD-64?

[sokoban]

http://www.xlr8yourmac.com/G5/xserveG5.html
is the link I should have put. Dang Plain Old Text

Re:WATT figures for G5 vs AMD-64?

[10Ghz]

The TDP AMD quotes is not comparable to the figure Intel releases. What AMD does is to publish a figure for an entire family of CPU's. They are basically telling MoBo and heatsink-manufacturers to design their products so that they can handle a CPU with that TDP. The actual TDP of the CPU is significantly lower than the figure AMD says.

Basically, they say that "in some time in the future we might release a CPU that has this TDP. We advise you to prepare for it, by designing your products accordingly. Our current lineup is nowhere near that figure, however".

And like it was already pointed out. If G5 is so cool, why do they top out at just 2.5GHz, and that's with frigging liquid-cooling? Yes yes, I know: "They do it so make it silent". Couldn't it also just be that the G5 runs so hot, that it needs huge cooling-system?

Nitpicking...

[Tristandh]

Capable of addressing an astronomical 18 billion GB, or 18 exabytes, of memory,

I know the first 2 digits are 1 and 8, but 2^64 bytes is still 'only' 16 exabytes...

Re:Nitpicking...

[TonyZahn]

If you had read the sidebar you'd see the article defined the issues with base-2 and base-10 number names, and introduced the prefixes "mibi-" and "gibi", which should be familiar to /.ers.

When they say 18 exabytes, they're talking base-10, otherwise they would have used the "gibi-" equivalent (exibytes?)

Re:Nitpicking...

[AchilleTalon]

Not actually, almost language abuse to name 1024 bytes a kilobyte. Look at the National Institute of Standards and Technology for definition of international units. 1024 bytes is a kibibyte rather than a kilobyte. Kilo really mean 1000.

So, 2^64 = 1.84467e19 bytes = 1.84467e16 KB = 1.84467e13 MB = 1.84467e10 GB = 18.4467e09 GB = 18 billion GB

Re:Nitpicking...

[sedmonds]

If you use unambiguous prefixes, like the IEC has defined, 2^64 is 16 exbibytes. Which happens to be (roughly) 18 billion gigabytes, or 18 exabytes. I'm not sure why everyone insists on confusing the matter by using base 10 prefixes when there are base 2 prefixes available that are unambiguous.

Re:Nitpicking...

[IHateSlashDot]

This just shows how out of touch the standards bodies are with reality. Nobody has adopted that idiotic naming convention. A kilobyte is 1024 bytes. No question about it. What the standards body needs to do is to look at the common usage and use that. Making up words as they go along does not help as is evident by the confusion in this article.

Re:Nitpicking...

[Ynazar1]

2^64 = 1.84467441 × 10^19

Of course then we get into the argument that a kilobyte is 1024 bytes (or 1024 * 8 = 8192 bits), and that 4.7 GB DVD-R's only have about 4.5 GB of real capacity, etc.
Its all about decimal vs. binary numbering systems.

Re:Nitpicking...

[fitten]

I'm not sure why everyone insists on confusing the matter by using base 10 prefixes when there are base 2 prefixes available that are unambiguous.

Because we were using those prefixes in the computer world long before most of you were born and think it's quite rude for you folks to come along and try to change our vocabulary for us. Contrary to what you folks seem to think, WE know exactly what we are talking about, it's you folks who seem to have a problem understanding it and make up these rediculous sounding prefixes to "fix" a problem that doesn't exist other than through your own issues.

Re:Nitpicking...

[Fulcrum of Evil]

When they say 18 exabytes, they're talking base-10, otherwise they would have used the "gibi-" equivalent (exibytes?)

If they meant base10, they should have used existing conventions (10^x) rather than trying to shoehorn this kibi stuff into usage. Common usage is that an exabyte is 2^60 bytes. Nobody in there right mind uses exibyte for that. The standard is something that NIST is trying to force on the community without any sort of support - they're exceeding their mandate.

Re:Nitpicking...

[AchilleTalon]

That's just plain stupid. The prefixes are used in many other fields than computer science and to designate many other things than bits and bytes. So, redefining the kilo in a particular field because the scientists in that field are too lazy to learn the appropriate way to name things will just add to the confusion.

Why don't we call a chair a horse?

Re:Nitpicking...

[AchilleTalon]

I'm afraid you were not born yourself when the kilo, mega, giga and others prefixes came to life to designate things like: kilograms, kilovolts, kilometers, kilokelvins and the likes.

That doesn't mean because you insist to use then wrongly that make it legitimate to redefined them accordingly.

Standards is about making things clear for everyone, not just a click of initiated peoples thinking it's pretty cool to wrongly name a chair a horse.

Re:Nitpicking...

[Paradise Pete]

Why don't we call a chair a horse?

Because I don't think chairs can ride. But we could call a chair a cab. That might work.

Re:Nitpicking...

[Paradise Pete]

This just shows how out of touch the standards bodies are with reality. Nobody has adopted that idiotic naming convention. A kilobyte is 1024 bytes.

I think it's the other way around. A "kilo" byte was named that (I'm guessing) because it was a semi-clever way to refer to something that was pretty close to a thousand. But kilo meant 1000 long before that, and still means 1000 everywhere else.

Re:Nitpicking...

[zsau]

exbi-. The last two letters are always 'bi', for 'binary'.

I really hate the way computer users have usurped and redefined 'mega-', 'giga-' etc, and then complain when hard drive manufacturers use the original/real definitions.

Re:Nitpicking...

[Fulcrum of Evil]

The current convention (k==1024) has been in use for over 30 years by the entire industry. NIST declaring new prefixes is a solution to a problem that doesn't exist. What they should be doing is declaring that, in matters of bytes, convention rules. Then they could call the HD manufacturers on their semi-deceptive practices (not really deceptive because nobody believes them).

Re:Nitpicking...

[Kjella]

If they meant base10, they should have used existing conventions (10^x) rather than trying to shoehorn this kibi stuff into usage. Common usage is that an exabyte is 2^60 bytes.

"Common" usage is that exa = 10^18. Introducing exabytes as "almost but not quite the same" as 10^18 bytes was the computer industry's mistake, not SI or NIST.

The standard is something that NIST is trying to force on the community without any sort of support - they're exceeding their mandate.

Classic chicken and egg problem. Those that do know of it don't need to change it, those that don't aren't in a position. Basicly, I see two options:

1. Continue quibbling with endless debates if SI prefixes count or not, and if when (Is a 1Mbit line decimal or binary? 512MB RAM? 250GB HDD?)

2. Find two separate, different terms.

Guess which I prefer? NIST made some terrible mistakes though. Terrible names for one.

I use

1 kilobyte (kB)
1 kilobyte (KiB)
1 megabyte (MB)
1 megabyte (MiB)
1 gigabyte (GB)
1 gigabyte (GiB) ...as appropriate

If someone insists, I'll specify

1 decimal kilobyte (kB)
1 binary kilobyte (KiB)

Kjella

Re:So sick of it

[BobWeiner]

Obligatory PC Weenies link that fits the topic at hand.

Re:Heat Problem Back Ground

[the_2nd_coming]

umm, what heat problem? the 9 fans and liquid cooling are in their respective models to solve a SOUND problem. not a heat problem. but then if you bothered looking up statistics on the G5 you would not be able to bash the mac would you?

Add instruction sets size too

[Gopal.V]

The instruction sets are also generally 64 bit... so you end up with lesser disk space as well :) .. add the space in loading these into RAM ..

The notable exception is the Arm's thumb instruction set (it's cool).

The sad part "my address bus is bigger than you" is going the "I have more MHz than you" way soon as parallel CPUs (mulit-core or otherwise) become cheaper.. 90% of our tasks are better done parallel than using a single fast chip . Hell , half of the tasks really don't need anything beyond a 300/400 mhz clocks.

Re:Add instruction sets size too

[Wesley Felter]

What architectures use 64-bit instruction words? Not PowerPC, MIPS, or SPARC. Certainly not x86-64.

Re:Add instruction sets size too

[taradfong]

IA64 (Itanium) has 64 bit instructions, *but* each can hold up to 3 opcodes.

Re:Add instruction sets size too

[Jeff DeMaagd]

The instructions themselves are most often 32 bit. Someone noted x86. PPC and Alpha also has 32 bit instruction words, even for 64 bit mode instructions.

If they were 64bits for each individual instruction, I think it would have a huge undesirable impact on execution.

Re:Add instruction sets size too

[ArbitraryConstant]

In the case of PowerPC (and SPARC, MIPS, anything else I can think of), the opcodes is still 32 bits. The number of registers stays the same, the number of instructions stays (more or less) the same. There is no reason to extend the size of opcodes.

Re:Add instruction sets size too

[Fulcrum of Evil]

IA64 (Itanium) has 64 bit instructions, *but* each can hold up to 3 opcodes.

No, it has 41 bit instructions, packaged 3 at a time, with metadata about concurrency and crap.

Re:Add instruction sets size too

[shotfeel]

AFAIK the above posts are correct, the instruction sets are stil 32 bit.

Unlike some of the other posts though, I'm not worried about RAM or hard drive space, I'm worried about L1 and L2 cache space. If a sizable amount of the data cached goes from 32 to 64 bits, the processors could take a big performance hit unless the caches are bigger too.

Here's hoping programmers keep using 32 bit data structtures when they don't need 64 bits!

Re:Add instruction sets size too

[ArbitraryConstant]

"Unlike some of the other posts though, I'm not worried about RAM or hard drive space, I'm worried about L1 and L2 cache space. If a sizable amount of the data cached goes from 32 to 64 bits, the processors could take a big performance hit unless the caches are bigger too."

This is a valid concern, but the hit isn't that big. IIRC, 64-bit code and data tends to be about 10-15% bigger, doubling the cache is good enough to overcome this. Looking at the cache size of major x86 processors, the Prescott has gone from 8k to 16k L1 and 512k to 1024k L2. AMD64 processors are constant (starting to move to 1m L2 on the newest ones), but they had a lot of L1 to begin with.

Have you been watching cache sizes lately? They were constant at around 256k-512 k for close to 10 years (my P2 from 1996 has the same cache as my P4 from 2004). They're just beginning to increase now with 90 nm because cache increase is the most beneficial thing to do with the transistors becoming available.

"Here's hoping programmers keep using 32 bit data structtures when they don't need 64 bits!"

This is not a new thing. How many people now use a 32-bit int when an 8-bit char would do? Pretty much everyone, in my experience. They won't bother optimizing. When they do, it's because it needs a resource that's very scarce (network traffic, or huge numbers of the data structure being used). Also consider is that there's usually a penalty to loading values from a non aligned address. I'm not saying this dominates, but using the smallest type you can is almost never free.

Re:Add instruction sets size too

[addaon]

Of which 20% are nops.

you want heat data

[Anonymous Coward]

Wattage only tells you how much power the CPU consumes, it doesn't tell you how much heat it gives off.

Power consumption = Power In
Power In = Power Out + Heat

What you want to know is how much heat is given off.

Re:you want heat data

[Anonymous Coward]

The power coming out of a chip is marginal; it's pretty much all converted to heat.

Re:you want heat data

[imsabbel]

You are wrong. Believe me.

Heat wave

[Compact Dick]

It was a heat wave in my area last Saturday, hitting 39 C (103 F). My room acts as an oven with the heat from the roof adding extra fire. Lucky for me I was at work in air-conditioned comfort, leaving my poor A64 baby to suffer the gruelling heat [which didn't subside even after midnight -- I drank a lot of water that day, pleased at the A64's performance.]

PC vs Mac cooling.

[coobachey]

In the past few years where PC' Fanatics have gone to great measures in extreme clocking and tweaking to push the speeds to the limits, Mod'rs have ran into and pretty much conquered heat issues along with other roadblocks. Technology that has been adapted by PC and PC-part manufacturers, I'm no Mac expert but even thumbing through the latest magazines I have yet to stumble upon a "Mac" mod or a how-to on making them extreme, is this due to proprietary or licensing issues? Or can a Mac be reverse engineered the way PC's can be? Granted Macs are not worthless and have good reviews from most of its users that know the systems. Still I find it limiting and where PC has prevailed. Heat on a slower clock speed processor should not be an issue. How many fans does a G3, G4, G5 have these days (1 case)? Laptops would be even worse as for PC portables are highly concerned about heat themselves.

"Make it idiot proof, Someone will make a better idiot" -- Tweak

Coob

Re:PC vs Mac cooling.

[2nd Post!]

Oh, and I found this. What's interesting is that this mod is to make the Mac even quieter, and not more performant.

Re:PC vs Mac cooling.

[NardofDoom]

My Power Mac G4/400 runs around 80F, and that's with the PSU fan stopped. (It sounds like it's grinding coffee otherwise.

Oh, and it's got two fans. One in the PSU, the other blowing over the heat sink.

Re:PC vs Mac cooling.

[ndrw]

I'd say part of the reason (not trying to harsh Apple for prices here, they're premium systems for premium prices) is who wants to risk cutting up a $4000 system to put in more cooling?

Re:PC vs Mac cooling.

[owlstead]

Simple. You can buy a PC in parts. There is a well established market for it as well. Lots of upgrades are possible, while for the mac little to none are available. Not really a space most overclockers (who tend to buy their system from multiple vendors) are comformable with. Most of them will have grown into it as well. Start of with a DVD replacement, then multiple harddrives, and start overclocking sometime after. Also, the games industry for MAC's is way smaller than the PC games industry. This is a market where overclocking really counts. And since there is less of a market for mac internal parts, this will be double the case for mods.

Re:PC vs Mac cooling.

[XnetZERO]

Somewhere out there, a guy modded his G4 (maybe a G3) to include a beer tap. I've yet to see a PC modded to include a beer tap. Why are PCs so limited? Without knowing what the "latest magazines" are it'd be a little difficult to speculate why you haven't seen any "mac" mods other than to suggest that the "latest magazines" you're reading are PC centric and ignore the Mac platform (kind of like Tom's Hardware). There are Mac Modders out there, but be warned--they're a lot like Mad Hatters. Cheers-

For the record...

[lamz]

18 billion GB, or 18 exabytes, huh? That ought to be enough for everyone.

Fans and cooling

[2nd Post!]

Your iMac G5 has two fans. Not much space left for additional cooling, really, without interfering with the current cooling setup.

Your PowerMac G5 has nine fans. Again, not much space left for additional cooling without interfering.

And get this, the PowerMac G5 already uses a liquid cooling setup. The only possible additional mod is to hook the current setup to a resevoir and radiator on the outside of the case, as the inside already has a radiator per CPU and something like a 120mm fan per CPU.

Re:Fans and cooling

[photonagon]

The iMac G5 has three fans.

Re:Fans and cooling

[mh101]

And get this, the PowerMac G5 already uses a liquid cooling setup.

Partially correct... AFAIK only the Dual 2.5 GHz model has liquid cooling. My Dual 1.8GHz Mac has a couple massive heat sinks and a lot of fans.

Re:Teflon underside + Fans = Convection oven

[Graemee]

They could use the hot air to create a convection oven PC.

I doubt that 64 bit computing is that hot for PPC

[renoX]

The *BIG* thing for x86 for "64bit computing" is not in fact the 64 bitness but doubling the number of GPR!

As the PPC instruction set is sane (x86 is not, urgh), beside the extra-instruction needed for 64 bit computing, there are very few difference between a PPC running on 64bit code or a PPC running on 32bit unless of course you have an app which needs more than 4GB of memory or do lots of 64-bit integer calculation..

Re:Heat Problem Back Ground

[macaulay805]

RTFA? You must be new here .. (or me)

Re:So sick of it

[Lumpy]

I can buy .8-1.2ghz processors and motherboards that use extremely little power (tens of watts running full bore) and are certianly very useable with even windows XP running on it as a internet/ general use home or office pc.

But for some reason most people believe they need a 6.8ghz 2.2terabytes of ram and 15,000 rpm hard drives along with a hairdryer add on known as a 3d video card, bringing that machine up to the 400 watts+ power consumption mark.

and where does most of that energy go?? that 's right . HEAT.

Switch to mini-itx computer with a low-power processor, laptop hard drive and flat panel monitor, and you will generate very little heat, use 1/20th the electricity your current space heater under your desk called a comouter does and still get the same general computing done with the same speedy feel.

will you increase your lead in the seti@home, doom3, other CPU intensive tasks? nope. but office 2003, WinXP and all other office apps as well as most non 3d intensive games run perfecly.

Re:Heat Problem Back Ground

[coolgeek]

The G5 2.5 dualie I just received puts out waaaaay more heat than any computer I've ever seen. I had to move it out from under my desk to the side to allow the heat to escape. It was much more than a foot warmer.

It used to be...

[Balthisar]

It used to be that Mac fanatics would be proud about how little current the PPC used -- and consequently how little heat it gave off -- when compared with the Intel-style architecture. Guess we can't make that argument any more.

Now that G5's are liquid cooled, it makes me wonder if a 2.5GHz G5 is *really* a 2.5GHz G5, or if it's an overclocked 1.8GHz chip. You know, overclockers really pump things up with cool liquid cooling stuff. What's the fastest a 2.5GHz G5 could run with a traditional cooling system, like a fan and heatsink?

Oh, one more thing before I'm modded as a troll: my G4 PowerBook is my 8th Macintosh. What I'm asking is genuine curiosity.

Re:It used to be...

[MooseByte]

"Now that G5's are liquid cooled, it makes me wonder if a 2.5GHz G5 is *really* a 2.5GHz G5, or if it's an overclocked 1.8GHz chip."

If it's running reliably at 2.5GHZ, I'd say that's a 2.5GHz chip. ;-)

After all if your $FurryPet snuggles up against your machine's vents and it starts shutting down, it's still the same chip. Only difference is a friendly or hostile thermal environment. Whatever the manufacturer chooses to stamp on a given chip is at times not related to what it actually binned-out at.

Re:It used to be...

[arminw]

Increasing the speed of a G5 or any computer involves more than heat. The other system components also have to be faster. Like the article stated, the clock speed of a computer USED to be a measure of its performance but now the performance has to do more with how much total work a processor can do in a given amount of time. This varies greatly with the type of work to be done, how well the algorithm of the job is translated to the intructions of the processor and how well matched the architecture of a processor is to the algorithm.

Re:It used to be...

[GlassHeart]

What's the fastest a 2.5GHz G5 could run with a traditional cooling system, like a fan and heatsink?

That is most likely not Apple's only design criteria. They are asking, instead, "what's the fastest we can push this G5 chip without making too much noise?"

Re:It used to be...

[mritunjai]

Is it a troll ???

G5's have water cooling not because they produce hell lot of heat, but because Apple wanted to make them whisper quite!!! So instead of a honkin' turbo-jet fanm you have 9 (yeah, NINE) reaally sloow moving fans cooling the whole unit.

It takes a lot to make a good system and Apple sure knows how to do it right!! Finally a POWERFUL system that you can keep next to your bed and still be able to sleep soundly.

Re:It used to be...

[Tumbleweed]

You've got that backwards. You get used to the 'white noise' of the computer, and you should wake up when it goes missing - cuz somethin' bad's about ta happen.

Re:It used to be...

[dhovis]

I have mod points, but I figured I'd answer your question instead.

When you do a die shrink, you can lower the power required at particular clock rate, or you can run at a higher clock speed with the same power dissipated. So when IBM went from 130nm for the 970 to 90nm for the 970FX, the top clock speed went up from 2GHz to 2.5GHz. Other than the process change, I believe there were very few changes to the chip.

Now, when you go from 130 nm to 90nm, the linear dimension across the chip is ~70% of what is was, and the area of the chip is (70%)^2 or about 50% of the previous chip.

Lets use some numbers, these may not be 100% accurate, but they'll explain the basic concept. The 2GHz 970 had a die size of about 121mm^2 and put out a maximum of 42W. That is about 350mW/mm^2. If we assume that the 2.5GHz 970FX has that same power consumption, but has a die size of 60mm^2, then the 970FX will produce 700mW/mm^2. So you have the same amount of power, but you are trying to suck it out of a smaller piece of silicon. So you need much more efficient cooling to keep the chip temperature the same. Hence, the liquid cooling system in the dual 2.5GHz G5.

Re:It used to be...

[Have Blue]

A 1.8Ghz G5 is physically identical to a 2.0Ghz G5 is physically identical to a 2.5Ghz G5 (except for the die shrink). What IBM does is test the whole batch of chips at 2.5Ghz and see which ones are not reliable or producing bad results. These chips are demoted to 2.0Ghz and the test is repeated. Any chips still doing badly are tested at 1.8Ghz. Any chips that still don't work are truly defective and get discarded, but the rest have now been sorted into the 3 speed categories they will be sold as with warranties.

Apple took the chips IBM certified would run at 2.5Ghz for the high-end G5s, and added the cooling equipment necessary to keep them within the required temperature range. OCers can get away with violating the manufacturer's guidelines because they don't care about support or reliability.

Re:It used to be...

[Balthisar]

I guess I should have been a little clearer. I know that the 2.5GHz chip is rated as 2.5GHz. I also know that just the clockspeed alone does not a system make. I realize the same amount of power has to be pumped through smaller pipes and be disapated. I realize it's quiet and I miss whisper-quietness of my graphite iMac (Mac #6).

I guess the question I'm asking -- aside from being a question -- is really meant to be thought provoking. Does the use of a liquid cooling system fundamentally change the defacto cooling requirements for all future desktop computers? If you recall in the past, overclockers would use liquid cooling because they had to do so to keep the chip cool enough in order to increase the clock frequency. Higher frequency at this scale means more heat.

How do chip manufacturers specify cooling requirements anyway? So many joules per second must be removed, or simply do they leave it open and say the chip must always be below x-degrees? So are IBM or Motorola marketing a chip that's not possible to be used in any other manner aside from liquid cooling? And in the most base form, couldn't Intel just re-badge a 3Ghz chip with ridiculous cooling requirements and advertise it as a 4GHz chip?

Surely there must be something fundamental to the architecture of a chip (I don't know) that can account for its clockspeed (regardless of MIPS) aside from the ability to cool it, right? Well, up to reasonable frequency limits, right?

Good followup link from the article

[C.Batt]

Introduction to 64-bit computing

There's an informative link at the bottom of the article for those requiring a bit more insight into the effect of 64-bit computing. /wishes he had exa-bytes of memory right now... VS.NET on WinXP is a PIG!

Re:Good followup link from the article

[fitten]

VS.NET runs great for me on WinXP. What kind of system do you have? If you have less than 512M memory, upgrade.

Also, don't do like one site I was at did... they had/have a bunch of P4 Celerons (1.5GHz Willies) running on PC133 memory... that's just horrible to begin with before actually trying to do any work.

Re:Good followup link from the article

[C.Batt]

Oh, I know it's a memory problem. Only 256M. I just joined the development team on this project and they re-puposed an older machine (still P4 1.8, or something) for me. Apparently I'm scheduled for an upgrade to 1G in the next couple of weeks, but until then, it's hellish.

The solution is pretty big. A rebuild-all takes about 20 minutes under the current scheme. The disk just thrashes as the system looks for swap-space.

Ah well. That's life.

Re:anyone else noticed how COOL the AMD-64 chips r

[ManxStef]

Yeah, I noticed that too. Most likely it's because they 0.13 micron process (as with all the current Athlon chips), which is a lot more efficient than the original 0.18 micron "fry an egg on me" Athlons. The next step will be the 0.09 process, which will result in faster & cooler (or as cool) chips again :)

So let me get this straight...

[emil]

Thus, the fabrication technology used for the PowerPC 970 was designed to eke out higher performance by trading away reliability; for these markets, the trade-off between reliability and performance is different.

This is what Apple is putting in their servers? And they are expecting market traction?

Re:So let me get this straight...

[Anonymous Coward]

Well, a very similar trade-off applies to other processors in the same league e.g. high-end workstation and lower-end server (Opteron, Xeon). The comparison made in the article is between the 970(fx) and Power4/5.

Re:So let me get this straight...

[Thu25245]

Okay, so I haven't R'ed TFA. But it seems they're talking about the fab process, not the CPUs. Which means that IBM is gunning for higher clock rate CPUs, at the expense of having a higher percentage of dies fail testing. The unreliable dies that fail will never be released to be put into servers, or will be clocked down enough that the do run reliably.

Re:Heat Problem Back Ground

[WuphonsReach]

Just because you can't afford it doesn't mean it's a bad computer (doesn't mean it's a good computer either).

Hear hear.

I've priced out equivalent machines to the current Apple offerings, and you do indeed get what you pay for. A dual-CPU 1.8Ghz Powermac is around $2500 sans monitor. Price out a dual-Xeon or dual-Opteron and you end up at around $2000-$2500 for a comparable system.

Where Apple might be missing the boat is in the ultra low end where you can buy a system for $600. (But why should they try to compete down there where margins are razor-thin?)

Re:hot apple g5

[malfunct]

Can you give me the recipie for making potato chips from hot apples because I might be able to make money on that.

Re:Too bad MacOS X can't use it!

[Anonymous Coward]

Apple acknowledged this when the released the G5 - Panter is a 32bit OS - meaning the G5 runs in 32bit mode under Panter. Tiger, on the other hand, will be a full 64bit OS and is slated to be released in March 2005 (and Apple seems to be pretty good at making OS release dates).

CPU power defeated by bloated software?

[shed]

From the article:

The quest for CPU power has been largely defeated by bloated software in applications and operating systems. Some programs I wrote in Basic on an Apple II ran faster than when written in a modern language on a G4 Dual-processor Mac with hardware 1,000 times faster.

Come on. What language are we talking about here? My basement collection includes a II+, a IIe, two IIcs and a Franklin compatible. I challenge anyone to come up with a program in Applesoft that runs faster on one of my museum pieces than on a modern Mac using C++, Java or even Perl. I mod his article -1 for troll.

While software has become bloated and to some extent inefficient, people often forget that we expect a lot more from our computers now than the single-tasking 80 column display days.

Re:CPU power defeated by bloated software?

[Tumbleweed]

the single-tasking 80 column display days

_80_ columns?! Why, you young punks had it good! When I was a boy, we only had 40 columns, and we felt grateful, as our fathers had no video display at all. Grandfathers? They used their fingers or an abacus.

You've come a long way, baby.

Re:Mac's are doomed!

[repetty]

I agree, big time.

I used to be very pleased that my Macs lacked a CPU fan while x86 users where bolting on everything but the kitchen sink in efforts to cool their systems down.

More important, Macs had just one fan -- the power supply fan -- to cool the entire box while x86 systems were fucking festooned with them.

Nowadays, though, a prudent Mac owner should at least consider his home air-conditioning system's cooling capacity since G4's and G5's will definitely heat up a room.

And I don't even want to get into fan counts.

I'd like Apple to make a nice, fast, cool, quiet system... and here's how to do it: underclock.

Modern CPU's are grotesquely overpowered for the needs of 99.9% of the general public. Crank those babies back.

The fruit of research and progress are not merely systems that can run faster and hotter, but faster and cooler... quieter.

Never going happen while there's this pissing contest going on, though.

--Richard

Re:atoms in one universe

[pclminion]

So, what happens if you want to use your fancy computer to model two universes?

Simple. Since we've got a computer capable of simulating the entire universe, we'll just use it to simulate a universe which contains a computer which is capable of simulating more than one universe.

Re:anyone else noticed how COOL the AMD-64 chips r

[Bruha]

bah the idiot is probably not running any applications.. load up seti at home or prime95 and let it burn in for a few minutes..

Then touch your heatsink and tell me it's cool.

Think he forgot about coolnquiet tech.. that proc probably runs sub 500mhz when idle.. so passive cooling works.

Hot?

[bogado]

If it is that hot, maybe it need water cooling or some other crazy refrigerating scheme so easily found here in slashdot. :-)

Re:anyone else noticed how COOL the AMD-64 chips r

[ruiner5000]

that is an incredible assumption with no basis. perhaps one would realize that in fact there is less heat dissapated by AMD CPUs than Tom's Hardware Guide has led everyone to believe. Perhaps if you are paying attention you would probably say because they don't put off that much heat, that is why you can put your hand on the heatsink. That is the actual case. It it was not dissapating heat correctly then there would be stability issues.

Re:Heat Problem Back Ground

[timeOday]

I think that guy needs to take his superior knowledge elsewhere and try some benchmarks ("512K of cache isn't competitive for $3000")... apparently it is because it's winning benchmarks and people are buying them.

Would you mind pointing me to such benchmarks? I've consistently found it very difficult to find any decent comparos of Macs vs PCs. Every PC CPU review includes pages of benchmarks on a wide variety of games, media encoding, office apps, and scientific applications. I'm really curious how fast the G5's are, but simply timing a Photoshop filter won't cut it.

Re:Mac's are doomed!

[Zemplar]

"I'd like Apple to make a...fast...system... and here's how to do it: underclock."

So that's the secret!

I'm so surprised that processor engineers never thought of that one. Oh wait...

Re:Heat Problem Back Ground

[the_2nd_coming]

it is cooler than a frigen P4 or AMD 64 CPU

Let me be the 2^32 to say..

[crymeph0]

Plot straight from the game? What secret brick was I supposed to push to find that?

Re:Mac's are doomed!

[GlassHeart]

It's not a pissing contest, by which I presume you mean a pointless pursuit. Apple was faced with an important market segment - audio and graphics professionals - jumping ship both because PC software are beginning to catch up and because PC hardware are not only cheaper but actually perform much better. For many of these customers, render time is money.

If you don't need that kind of power, you should stay on the lower end. The G5 iMac or the eMac don't seem to run all that hot.

As for fan count, presumably you're referring to the PowerMac G5. Apple preferred more low-speed fans over a few high-speed fans as a noise reduction measure. What was your complaint?

It's not too hot!

[artemis67]

Steve Jobs tells me that the real problem is that the planet isn't cold enough!

Re:I doubt that 64 bit computing is that hot for P

[Logger]

You're point is accurate, but I wouldn't dismiss the importance of >4GB memory access. I would venture that the >4GB memory access is more important to Apple than the Windows market.

Apple's customer base consists of a larger percentage of video editors and Photoshop users than Windows users do. Not that Windows doesn't have these customers, but they make up a smaller percentage of the total Windows PC market. These users buy the premium products, paying top dollar. Apple cannot afford to lose them to a 64 Windows machine.

When your using something like Final Cut Pro (http://www.apple.com/finalcutpro/) every Gigabyte of memory helps. So does that liquid cooled CPU.

Re:I doubt that 64 bit computing is that hot for P

[Tumbleweed]

Wrong. The _BIIG_ thing in moving the Mac to 64 bits was massively increasing the size of the RDF; it's now big enough to encompass the entire galaxy. Global domination? Pfeh. Child's play.

Re:Heat Problem Back Ground

[toddestan]

Where Apple might be missing the boat is in the ultra low end where you can buy a system for $600. (But why should they try to compete down there where margins are razor-thin?)

Because poor college students like me, who buy $600 computers is going to buy a PC. And when I get out of college and have the means to drop $2500 on a computer, guess what I'll probably buy.

Actually, the two real reasons why Apple doesn't sell low end machines is that it would undercut sales of their more expensive machines, and totally destroy the second hand Mac market (where used Macs are way overvalued, IMHO). Without a strong second hand market to sell a used Mac for a good price, people will be less likely to buy their high end machines.

it's hot. really. (a little perspective)

[Mr. Slurpee]

for literal heat, this puppy is pretty hot.

my dual 2.5GHz PowerMac G5 idles at 52C (125F) on CPU A and 50C (122F) on CPU B. the memory controller is actually one of the hotter things, it idles at 62C (143F). however, it's not the hottest thing, of course: at full load (DVD rip+encode or playing 15 videos at once + MP3 + tasks + flicking around Exposé) both CPUs have hit a max of 83C (181F) (the computer is supposed to automatically sleep around 90C or so).

so why so effing hot? i mean, this idles at the max temp my athlon 2500 peaks at! it certainly idles at a hotter temp than it needs to, but i have no problem with that: the system runs the fans dynamically to keep the noise down, so at idle it's not as cool as it could be. the difference in noise in my room when i sleep the athlon is ridiculous - the G5 sounds like a slightly loud external hard drive that's spun up. the system also has a liquid cooling system to quench the processors. this seems to just keep the processors within their range. the value that i see in it is response to new heat - the CPU temps flick around a lot and are very responsive to load and the loss of load. after ramping up the CPUs to >80C, it take about three or four seconds after the load drops for the CPU temps to drop 15-20C, then maybe a total of ten or twelve seconds to drop to idle temp.

for some real-world perspective... a DVD rip+encode with HandBrake with using ffmpeg engine, MP3 audio, 2-pass encoding, and gunning for your average 700MB movie time (800-1300kbps?) takes slightly less than the length of the DVD. an hour and a half long movie took about and hour and fifteen minutes to get on to my hard drive. MP3 ripping in iTunes will run up to 28x, but it's not fully loading the processors so i wonder about a drive read bottleneck. the first night i got it, i was at a loss for how to really test the speed on it, so i just decided to open up a shitload of videos. basically i played a DVD (fluff, the GPU does that), opened up something in VLC, opened up about 13 videos in QuickTime of various sizes and formats, played some MP3 music (fluff again, that's ball sweat of a cutting edge proc), and still had enough processing power to comfortably navigate files, chat, browse web pages, and flick around Exposé. around all of these things plus one is when a few of the videos would start stuttering and expose would start dropping frames to keep collapse speed uniform. anything past this would really start robbing time from videos.

all in all? it's fast. it's quiet. it gets hot, but it takes care of itself. coming from a 375MHz G3-upgraded PowerMac 7600 (vintage '98), i'm not doing too shabby. i just decided i'd scramjet at mach 7 to the top of the pack and then sit there for another few years.

Re:I doubt that 64 bit computing is that hot for P

[TheRaven64]

I don't often need more than 4GB of memory, but more than 4GB of address space is very nice. When I'm working with very large datasets (i.e. a lot of the time) I can just mmap the whole thing. Makes things a lot nicer...

Re:anyone else noticed how COOL the AMD-64 chips r

[Dumass]

Intel and AMD report thermal characteristics differently. AMD reports max thermal dissipation for the entire line (notice that 3200+ has the same 89W as a 3800+), whereas Intel gives thermals for each chip, so a 105W P4 will run at 105W.

No AMD chip currently in the market gets near the 89 Watts quoted.

I'm not sure about your clock throttling statement, but I wont deny that it could be true.

Ari

memory bandwidth

[morcheeba]

The other half of that coin is that, because pointers take twice the memory, they'll consume twice the memory bandwidth. Adding more sticks doesn't help the problem -- you need to up the front-side-bus just to keep even.

The solution is to make your own 16 and 32-bit pointers & do a little math at the last second to convert to a 64-bit pointer (e.g. add a base address). Depending one what your mix of calculation/memory access, that conversion could be totally masked by the time it takes to access memory.

Re:Heat Problem Back Ground

[OwnedByTwoCats]

Yeah. Because Apple's $800 computer that includes a 17" display is just so much more money.

There is an anomoly in the time line.

[barfy]

This article is a dupe of an article that will be posted 4 hours in the future.

It is important that we repair the time line now... ;Encrypt
54339 96784 23470 08472 37932 87412
12327 12300 09483 71953 31994 73821
12086 67544 42139 48901 93753 18632 ;end

Re:There is an anomoly in the time line.

[kps]

You mean, the IBM PowerPC 970 can emulate an IBM 5100?

Apple and Cooling

[green pizza]

Do you really think that Apple would use liquid cooling if the processors only released 65W ?
I think the liquid cooling was implemented for the same reason that the G5 minitower case has 9 fans -- to keep the machine cool, even in a warm room, while running no louder than the hard drives or gfx card fan.

Apple also seems to be very conservative with their heatsinks. The heatsink in my PowerMac 8500 was HUGE. The CPU itself was a 120 MHz 604, which according to Motorola/Freescale, was only a 16.5 watt proc. The heatsink on the CPU card was at least as large as a 4" x 6" photograph, had a solid 1/4" base, and lots of huge fins. A smaller "ramsink" heatsink covered the L2 cache chips.

I saw something similar at CompUSA on the blue-and-white G3 minitowers -- though the 300 MHz G3 consumed only 7 watts, the heatsink Apple used to cover the G3 and the L2 cache chips was about the size of a 12oz beer can!

Uh huh.

[Ayanami Rei]

When have you _ever_ had a CPU fail with proper cooling, motherboard and power supply?

It's mostly marketing fluff, IMHO. CPUs don't spontaneously fail if they aren't put in machines assembled by idiots.

Re:Uh huh.

[John Whitley]

It's mostly marketing fluff, IMHO. CPUs don't spontaneously fail if they aren't put in machines assembled by idiots.

Er, whatever. I've encountered CPU failures in production systems. Nor do I have any reason to believe that the servers were "assembled by idiots", or that there was anything problematic about the datacenter environment (cooling, power, etc). Work in an environment with enough 24x7 production boxen and you're bound to see all manner of component failures.

Moreover, IBM wouldn't be using current fab lines to make POWER series processors at design rules with comfortable margins (lowering potential clock rate) just for "marketing". That makes no sense. Instead they'd do just what was done for the PowerPC chip lines -- they'd shrink the design rule and up the clock rate to get improved performance.

Re: uh oh...

[Cryptnotic]

What if our universe is only a simulation running on a computer in another universe?

Maybe it's possible to disprove this based on the idea that the other universe would need to be physically larger?

Re: uh oh...

[pclminion]

What if our universe is only a simulation running on a computer in another universe?

We could potentially detect it. Consider the types of computers we use in this universe. A cosmic ray or other energetic particle is capable of flipping a bit in RAM. If we are actually a simulation in a similar kind of computer, we should start looking for inconsistencies of the type that could be explained by memory corruption.

For example, an object might change color suddenly. This could be an indication that we are actually in a simulation, and that something has corrupted a small piece of state.

We might also look for effects which could be explained if we were running in a discretized simulation of continuous spacetime. The Planck length, for example, might be explained by computations using a limited number of bits -- lengths shorter than a certain size would simply be unrepresentable in the simulation's numbering system.

Re: uh oh...

[rekoil]

That would go a long way to explain the "disappearing sock" phemonenon...

It's the die size that requires additional cooling

[willy_me]

They use liquid cooling because it's harder to disapate the heat from the small 90nm core. The CPU's don't use much power, it's just that the heat they do produce is concentrated in such a small area. Better cooling is required to disapate this heat.

90nm transiters require less power then their larger counterparts. The problem is, for the same die size they use more power. So you end up with a relatively low power CPU that requires massive cooling.

Apple and IBM had a lot of problems because they expected their new CPUs that consume less power to be easy to cool. They were wrong. For each square mm, more heat must be disapated.

Re: uh oh...

[BeerCat]

We could potentially detect it. Consider the types of computers we use in this universe. A cosmic ray or other energetic particle is capable of flipping a bit in RAM. If we are actually a simulation in a similar kind of computer, we should start looking for inconsistencies of the type that could be explained by memory corruption.

Flipping one bit in an 8-bit number, and you might see a change; A becomes B, for example. Flip one bit in a 24-bit number, and it might be hard to tell that R255 G242 B155 has now become R254 G242 B155.
So, how do you know that such changes haven't happened?


Of course, if that was the case, then cosmic rays could actually be like a gigantic DIP switch to change one bit deliberately, to keep the universe "on track" (Asimov's Second Foundation springs to mind)

Re:it's hot. really. (a little perspective)

[mh101]

For comparison, my dual 1.8GHz PowerMac G5's idle temps are around 48C and 44C, with my memory controller at 54C. Your processors are about 700 MHz faster than mine, so those temperatures sound about right compared to mine.

And when you look at the size of the heatsinks and fans required to keep them at even this temperature, it's no wonder there's no G5 Powerbooks yet.

We can call it Deep Thought...

[GrahamCox]

or maybe "The Earth". Oh. What a dull name.

Re:Mac's are doomed!

[repetty]

Where do I begin...

First, I want to reinterate that I'm referring to what's in the interested of the vast, general population and not in nitch groups.

Since you recommended the lower end of the line to me, let me describe the features of the lower end computer: fast, cheap, and no moving parts (besides the hard drive).

To answer your specific question (since you asked), my complaint is, as it originally was, heat. But now that you bring it up, my secondary complaint is moving part count.

You seem very satisfied with things as they are and I'm not.

Re:Mac's are doomed!

[pkhuong]

More fans running at less than max speed means more redundancy. It's more reliable AND less noisy. What are you complaining about?

Re:It's the die size that requires additional cool

[evilviper]

They use liquid cooling because it's harder to disapate the heat from the small 90nm core.

Even with a very small core, you just need a slightly better heatsink to spread it around, and aircooling will continue working fine...

You certainly don't need water-cooling just because the core is a little bit smaller. If that was the case, the P4 should be much easier to cool than an Athlon XP. In fact, it's quite the opposite.

Re:Heat Problem Back Ground

[jrockway]

But that's why people buy macs. To use photoshop quickly. So how long a filter application takes is very relevant.

I don't have a mac here, but opterons and g5s are probably similar. Pick the one you like, they're both expensive :)

Re:atoms in one universe

[mxpengin]

ok , who will start the coding of vuware ?

Re:Heat Problem Back Ground

[andreyw]

"I'll probably get modded down for saying this, but I have Karma to burn..."

The second rule of /. moderation is that you ask to be modded down if you want to be modded up. So if I ask you to mod me up, I'm sarcastically asking to be modded down. Which means I really want to be modded up! Meta-sarcasm!