Grandstanding: " To perform ostentatiously so as to impress an audience." = you.
You dont read, you don't learn, you don't develop software, don't don't know what you are talking about, you sound like you are 13 years old (+/-2 years).
Hand coding - let's see, one example from over 4 years ago...
Read before responding. I gave you 4 examples,not 1, out 1000s from the first page of a google search. The last of which was 2 months ago. You are talking out your butt.
And now with Altivec is a "more capable vector unit" delusions. If it was superior, why isn't the G5 faster? Oh yes, because no compiler auto-vectorizes. That would seem to have been a crappy design choice, thereby making it less capable, in effect, crippling itself.
Complier availability is different than how good/efficient a CPU architecture is.
Yes altivec is better. It has about 10 times the diversity of instructions of SSE2.
The G5s floation point performance is due to the fact that each FP unit can do and multiply+add in a single cycle. So a single 970 can do 4 FP instructions every cycle!
Linpack is one type of test compared to spec's 26, and secondly, it's a test of SUPERCOMPUTERS,
No, its a test of floating point, not how 'super' the name is. Further it is administered by professional computer scientists.
I have already answered your SPEC question please read last post.
To sum up my posts: Opteron has an extra integer unit (3 vs. 2) giving it an int advanatage. G5 has better floating point and vector performance, but uses 1/2 the transistors. So the overall advantage, per transistor (i.e. equal cost), is well in favor of if the G5 which could have 2 cores in the same real estate, easily beating even the opterons integer performance.
If you don't or won't understand anything we've discussed, and you're intent on being a smartass... then toilet paper trees til you get it out of your system. In a few years, you will grow up and go to college and take a few CS courses. Perhaps you will finally understand the fundimentials of how computers and compilers work.
You are grandstanding. Your use of "crap", "stupid", "moron", and idiot don't improve your arguments.
"SPEC is subject to all kinds of problems."
All benchmarks are subject to problems. If you are testing for the CPU, somehow you must control for variables of compiler, OS, system architecture, and the amount of time and expertise of the tester.
Uh, how exactly do you get "GCC, is said to generate code that less well optimised than Intel's" from "GCC, is said to generate code that less well optimised for x86"?
Look at these redhat GCC 3.3/4.0 benchmarks. Notice how the 2-way PPC970 is twice as fast as the 4-way P4 on many tests and at close to par on the others. Now this is not the end all, am I'm sure you could come up with a different test that shows the P4 beating the G5, but certianly the G5 is not a "peice of crap".
You arguement about standardizing compilers is equivalent...
Standardizing of compilers is scientific method. Ideally you'd do a bank of tests, and unroll the variables: Standard compilers, standard OS, standard CPUs. Or you could tune each system to the max and then compare, that was LinPack and you didn't like that one either.
Hmm, does this appear to be vector processing done by a compiler?
Exactly my point! Intels compiler does auto-vectorization. GCC doesn't. If you test C code, P4+intel against GCC+G5, you are crippling the G5 by leaving out the altivec unit, which is a more capacble vector unit than SSE2
"hand coding...becomes completely out of reach for humans"
Hand coding is still done frequently on high performance algorithms:
You have no idea how compilers, CPU architectures, system architectures are designed. Your arguments are simplistic to say the least.
"GCC, is said to generate code that less well optimised for x86"
If you actually read the articles you link, and supporting articles, you would realize a little context. The article is saying that GCC is less well optimised than intels compiler, not less well optimised than GCC for PowerPC.
How is spec testing the compiler? I don't see any spec compiler benchmarks.
If you don't understand this, its not worth my time to argue with you. Let me walk you through it: If GCC and intels compiler generate SPEC scores 40% different from one another on the same CPU, than by definition it is testing the compiler! So if you wanted to do a head to head CPU test, one way of normalizing the test is to standardize the compiler to take the compiler variable out of the equation - what apple/veritest did, and generally good scientific method. Whether there are better compilers is arguably not useful if they aren't available for both platforms.
I certainly know that they're better equipped to produce optimal code for a modern CPU than you.
So far people are still smarter than compilers, much slower at writing asm, but still smarter. Hand coding is still done all the time by programmers to optimise parts of their code. Some CPU functions are still not handled very well by compilers like vector processing and are often done by hand coding.
SPEC is C & Fortran code. IT'S ALL ABOUT THE COMPILER! if you don't understand this, you don't understand how compilers work.
2 way spec measures throughput, while the uniprocessor measures raw speed. System architecture doesn't account for the vast majority of the G5's crappy throughput.
throughput is all about system architecture: buses bridges, caches, and RAM. Different CPU versions may support different buses/speeds and/or multi-way shared memory interfaces, this is separate from how well the CPU performs or what the "goodness" of an CPU architecture is. (ie you can tack a hypertransport interface onto any of the CPUs and change its memory bandwidth)
The point of the benchmark is to use test the system under optimal conditions, and Apple used GCC, which is bad at optimizing for x86.
Nonsense. GCC is bad for the G5 too. GCC is far more optimized for X86 (just think about the number of man hours difference based on X86 market share - same for all open source projects). IBM's compiler for G5 is far better. While intels compiler auto vectorizes, GCC does not. When you do a head-to-head test how do you do it fairly? One way is to tweak everything to the max, the other way is to normalize on compilers, OSs, or programs. Apple chose to normalize on GCC, which is a standard and open source (nothing hidden). That is a fair type of test, and certainly isn't "cheating". The controversy was about why the test had hyperthreading turned off, which was done to make the P4 perform better, not worse. If you google further you'll see that others replicating the test, found the system also to be faster with hyperthreading turned off (which is now general knowledge that hyperthreading can make certain tasks slower). But all that is besides the point.
Fact is almost every CPU architecture is better than X86: PowerPC, Alpha, MIPs, HP-RISC. The problem is competing in the marketplace against something with so much market mass. Sad really. We end up with the worst of all worlds, X86, because its got a corner on the market.
Because the G5 gets par performance with half the silicon real estate, you could get a dual core G5 for the same cost as opteron. If technological goodness and price won the day, instead of hegemony, you'd be telling me how your dual-core G5 thrashes everything from AMD and intel.
the G5 is so much better, why didn't Apple use them?
Ahh. See you like everyone else are blinded by marketing. The 970 is one of the most efficient chip for obvious reasons (it isn't carriing 30 years of x86 baggage). Lets just say for argument that the G5 is somewhat equal to the opteron (averaging linpack and SPEC) The 970 does it with half the transistors (58M verses 106M). Less power, less heat, less cost. You could make a double core G5 for the same price as an opteron.
Apple is switching because one company cannot push a whole chip architecture cost effectively, no matter how good the architecture is
Don't believe me? Why are XBox, Sony, and nitendo all using PowerPC chips? The most performance, the lowest price. Microsoft can get a 3 core 970 for the same price as a P4 (58M verses 169M transistors)
First, your SPEC marks are misleading using two-way systems that are highly dependant on the system architecture, no the chip.
The numbers from the same source are:
2600 Opteron: 1796 int
2200 970: 1040 int
2700 970: 1276 int (scaled)
SO even if we used SPECint. the difference is 29%. NOT 3-1.
Apple's dishonest G5 benchmarks... Hand coding?!?! Are you insane?
Ahh. What you seem to not understand is SPEC benchmarks test not only a processor but a compiler. Apple's benchmarks were not unfair at all, they just used GCC on both platforms, normalizing the compiler (which argualbly still favored intel which has received more GCC development work). The intel set called foul because intel compiler does better. OK, so? IBM compiler does better too. Intel compiler SPEC mark optimisations are well known, says Peter Glaskowsky, editor-in-chief of Microprocessor Report (google it). Hand coding... I guess you're to young to know what thats like.
"LINPACK has been largely superceded by LAPACK which has been designed to run efficiently on shared-memory, vector supercomputers."
My point exactly. Whats yours? Linpack is less efficient. Efficiency, does not a benchmark make (one could argue the opposite)
...accountable SPEC
Thats the problem, SPEC isn't accounable. The manufacturers who submit to spec.org, don't detail test conditions nor do they get verified. Arguabily SPEC is somewhat more broad than linpack (though not as different as you would like to make out). But linpack is all these things and the 970 gets twice the performance in a set of tests using complex algorithms (which use PF, int, branches, rotates, matrixes, etc.)
the G5 is so much better, why didn't Apple use them?
Ahh. See you like everyone else are blinded by marketing. The 970 is one of the most efficient chip for obvious reasons (it isn't carriing 30 years of x86 baggage). Lets just say for argument that the G5 is somewhat equal to the opteron (averaging linpack and SPEC)
SPEC is subject to all kinds of problems. One of which is compilers. Google the intel contraversy of ~2years ago. Intels complier seems to auto recognize SPEC algorithms and insert hand coded optimized routines. OK fair enough. What is AMD using? And what is the G5 using, GCC? GCC is horrible at optimizing code for the G5. Using the IBM complier would be more interesting, and a hand coded runoff even more interesting.
If you'd read the link I provided, you'd see that Lapack was created because of changes in supercomputer architecture, providing a more accurate benchmark.
No, you read it. The link you provided said Lapack was more efficient, not more accurate.
Linpack is type of test: linear algebra, and again, if you'd bothered to read the links, or even my post, you'd notice that 26 different types of test > 1 type.
Linpack is a series of different and algorithms (read last post) just like SPEC is a series of different algorithms.
Finally, you're using a very muddled secondhand source based on 1 test to desperately hang onto your delusion that the G5 is anywhere near the Opteron.
Where is your more trustable and unbiased source? The fact is per Proc, the 970 bests the opteron in verified supercomputer benchmarks 2-1. If the Opteron was really 3 times faster, how come the 970 beats it 2-1?
I'd haven't ever seen a unbiased SPEC comparison. There are many questions. Who is doing the test and what are their motives, what compiler are they using, what optimizations, does the compiler recognize SPEC and optimize for it (remember the intel scandal?), 3rd party validation, etc.
I like linpack, because each supercomputer group is hot-rodding their system to the max performance however they can, with smart CS people dedicated to each system to get the most out of it. And it is verified by a third party.
Linpacks age has nothing to do with validity. Lapack supersedes linpack because its more efficient, not more accurate at benchmarking. Math problems don't change with time.
It is not one test, but a collection of different tests: "analyze and solve linear equations and linear least-squares problems. The package solves linear systems whose matrices are general, banded, symmetric indefinite, symmetric positive definite, triangular,and tridiagonal square. In addition, the package computes the QR and singular value decompositions of rectangular matrices and applies them to least-squares problems."
I use Rpeak to eliminate network latency specific to that cluster implementation. This is valid because Linpack is a dense set calculations, its not just getting cached in the CPU! Furthermore Rmax wouldn't change the fact the G5 tromps the Opteron in FP.
So the Opteron trades FP for int. hmmm, not very exciting.
Most operations you'll do on your desktop are integer-based unless you're into heavy CAD or image manipulation.
I'd suggest something different. The only Apps these days that don't use FP are the programs where you can't tell the difference between 1GHz and 4GHz (ie. word processing). FP is used intensivly thoughout a modern computer experience (graphics/sound/media/video/MP3s/games/3D).
Take a look at the Linpack scores of the Top 500 supercomputer list. It's one of the fairest comparisons around, every processor optimised to the max, done by professional CS people proving only they're machine. You'll see that they are totally consistant between listing of the same processor type too.
Take the Rpeak and divide by number of procs.
What do you get?
2.3 GHz 970 (G5) = 9.2 GFlops (#7 on list)
2.2 GHz Opteron = 4.2 GFlops (#17 on list)
Take a look at top500.org
This is the top 500 supercomputers benchmarked on linpack. I use it because it is one of the fairest benchmarks, since everyone on the list is optimised to the max. Now take the Rmax number and divide by number of procs. You'll find from system to system using the same proc that they're very consistant.
I gave the ibm paper before. Look at the power curve (second graph down), follow the speed down to 1.6 GHz, now what is its power? ~22Watts.
Now from the curve you can see that cranking the GHz on the 970 makes the power increase exponentially (like all processors). So at 2.7 GHz it really is hot, not P4 hot, but very hot.
The reason for water cooling is low noise, not that it is sucks more heat than a P4. The reason no G5 in a laptop is??? If you look at the same power curve the G5 at 1.25 GHz is around 15 watts not much different than a G4. Perhaps Apple thinks paying more for a G5 without a really big performance or GHz boost isn't worth it.
Steve Jobs summed it up at the conference by noting that IBM's PowerPC road map would only deliver about a fifth the performace per watt as a comparable Intel chip. How is that a weaker product?
His answers don't jive with the numbers. I think this quote is:
1. Steves marketing spin
2. IBM doesn't have a roadmap, so what do you do?
3. Intels SpeedStep can give low average power for the perceived performance, even though at full bore the PPC has better performance/watt. This is a features issue, put a speedstep like feature into PPC and it would do even better (in fact I think IBM has already licenced such tech from transmeta, but hasn't used it). But again if IBM won't make it, you can't use it.
I think you confuse market forces verses better technology. Apple is a single customer for 970FX, they do not have the market strength alone to push the PPC architecture.
XBox and playstation do. The proof is here in applications were cost REALLY counts. Where are ALL the game systems? Thats right PPC.
As for the notebook/power issue lets take a look:
P4 3.2 GHz = 6.3 GFLOPS (82 watts)
Pentium M 1.6 GHz = 3.6 GFLOPS (24 watts) PPC970FX 1.6 GHz = 6.3 GFLOPS (22 Watts)
So what is apple talking about? Well Pentium M Speedstep probably has better power management to turn the clock down dynamically during occational use (like a word processor). Other than this add-on feature the pentium family has NO advantage other than market clout. The PPC gets P4 performance with pentium M power draw.
Lots of good tech has died because of stronger competitors with weaker products.
because faster clock speed can mroe than make up for it, and die size isn't that important at all.
Wrong. Fabed silicon is much like any other market resource. 3 times the silicon real estate = 3 times the cost. Die size (or the inferred transistor count) is also proportional to heat and power.
So for the same costs you could have a 3 core PPC 970 verses a 1 core P4, assuming IBM was willing to make them. The PPC would be almost 3 times faster. This is exactly why microsoft went PPC for Xbox.
As for faster clock rate, I think both AMD and PPC prove that's not the case, and power dissipation is the limiting factor "to crank up the MHz" - which is why P4 clock rates have virtually stalled out.
Apple didn't chose X86 because its a good idea... they did it cause IBM left them no choice.
PowerPC is a better platform than X86. But than again who isn't?
X86 is kludges and full of baggage. Just goes to show with a lot of money behind it, kludges are pretty good.
But as for faster and more efficient, its not. The PPC970 acheives similar benchmarks to a P4 at 2/3 the Mhz and 1/3 the die size, and less power as well. So for the same cost you could have a 3 core PPC (think Xbox). The laptop issue is about dynamic power features not inherent power consumption where the PPC whoops X86.
Apple is switching not because the X86 is better is just has a stable marketshare with a lot of backing.
Honestly it is a sad day for compution to see thing go backwards.
The problem is a market issue, not a technology one - as PowerPC is a far superior design to X86. It sad to see technology digression.
The PowerPC architecture has so much less baggage it can match intel performance with 1/2 to 1/3 the transistor count. Making it innately cheaper, less heat, and lower power. Also means you could make a 3 core chip for the same price as intel (a la Xbox).
The problem is IBM willing to make them? Apparently not.
There are many myths. 68000 weren't a dead-end. In fact the 68000 was far better prepared to scale than X86. It just didn't have the market. In fact the 68060 was faster than the initial PowerPC 601 it was replaced with.
Low power. The PowerPC 970Fx is lower power/performance than intel equivalents. IBM just hasn't invested in adding all of the power scaling features for laptops that the intel does.
$99 computers are different issue from CPU prices. If a company in china can make money with 50 cent margins on commodity computers good for them. Apple doesn't care about the walmart buyer. Walmart offers cheap chinese goods in every catagory, but they aren't competitive nor threaten the quality mid or high end goods offered by others.
A $500 Mac mini or $1000 iMac tell you nothing about the cost of processors. To a large degree silicon is a commodity priced per square area. The fact that PowerPC = P4 performace with 1/3 the real estate tell you most everything you need to know.
"Performance-wise, the dual-processor Xserve G5 compares well with Linux-based x86 servers. This is not surprising, considering that Mac OS 10.x is based on FreeBSD UNIX. Using the included Apache server, we ran the Xserve G5 through our standard WebBench tests. It did quite well on the static WebBench test, outperforming a competitor's dual processor x86-64 server running Apache server on SuSe Linux-64."
So who do you beleive? I'll take eWeek over some college guys.
This is nonsense. However good transitive is, PowerPC performance is on par with everthing intel is producing. If intel currently happens to be in the lead by a small margin, translation will still be slower. Apple would Apple switch platform because Intel has been in the lead for 3 months?
An why do it?
More performance? No.
Lower cost? No. Look at the specs
The PPC 970FX is on par with what Intel is making while using 3 TIMES fewer transistors. Transistor count = die size = cost. Why do you think playstation/Xbox/nintendo are ALL using PowerPC? Because it is the best cost&performance around!
Lower Power? No. This is one of the myths of the G5. It is a very low power device for the performance. Why is apple not making a G5 laptop? because unlike their windows counterparts who make big clunky laptops to accomodate a P4, Apple is more concerned about sleak design. Why water cooling? Cause they want quiet computers. Get this: at the same clock speed the G5 uses less power than the G4.
There is no reason for apple to switch. These are dumb rumors.
Your argument for evolution is unfortunately wrong.
Sexual variation, both physical and psycho-physiological, are the most common variations in humans. The reason? Because sexual variations are of the few physical variations that aren't life threatening. (ie. you could live without a liver, but could live as a gay person or hermaphrodite just fine).
Because they are not unadaptive to life, there is no reason to evolution throw them out of the genetic code. In fact, many hypothesis' have very adaptive explanations for such outcomes.
The sole evolutionary reason for a single persons existence is not to "replenish the earth", that is a uninformed, simplistic view influencd by modern Christian dogma (itself a simplistic view of the bible)
The demo design is terrible. (did I meantion ugly)
Zooming is somewhat interesting, though it could be implemented into other OSs easily enough (OS X has a zooming feature that could be mapped to mouse buttons).
This guy is so in love with text mode, that his fractal program in the demo uses ASCII graphics as if it was 1978!
The dell has a fast processor, but its hobbled with shared ram intel extreme video.
The Mac mini has at least a real video GPU. As apples Mac Mini website puts "it try running Halo on the cheapo PC".
Firewire and DVI are something you won't find on the Dell $500 computer.
The monetary values are equivalent. Apple has chosen to put money into features other than a fast processor, because it adds more real value to the customer (Who cares how many GHz it has, what does it do?)
Slashdot Mirror
http://www.folklore.org/StoryView.py?project=Macin tosh&story=On_Xerox,_Apple_and_Progress.txt&topic= Origins&sortOrder=Sort%20by%20Date&detail=medium
Of course, I am only the fool for even responding to your childish nonsense.
No, just young, absurd, and unteachableYou dont read, you don't learn, you don't develop software, don't don't know what you are talking about, you sound like you are 13 years old (+/-2 years).
Read before responding. I gave you 4 examples,not 1, out 1000s from the first page of a google search. The last of which was 2 months ago. You are talking out your butt. Complier availability is different than how good/efficient a CPU architecture is.Yes altivec is better. It has about 10 times the diversity of instructions of SSE2.
The G5s floation point performance is due to the fact that each FP unit can do and multiply+add in a single cycle. So a single 970 can do 4 FP instructions every cycle!
No, its a test of floating point , not how 'super' the name is. Further it is administered by professional computer scientists. I have already answered your SPEC question please read last post.To sum up my posts: Opteron has an extra integer unit (3 vs. 2) giving it an int advanatage. G5 has better floating point and vector performance, but uses 1/2 the transistors. So the overall advantage, per transistor (i.e. equal cost), is well in favor of if the G5 which could have 2 cores in the same real estate, easily beating even the opterons integer performance.
If you don't or won't understand anything we've discussed, and you're intent on being a smartass... then toilet paper trees til you get it out of your system. In a few years, you will grow up and go to college and take a few CS courses. Perhaps you will finally understand the fundimentials of how computers and compilers work.
All benchmarks are subject to problems. If you are testing for the CPU, somehow you must control for variables of compiler, OS, system architecture, and the amount of time and expertise of the tester.
Sigh. read: "Dell's own figures were calculated using different compilers and host operating system: Windows XP Pro, Intel's own C++ and Fortran compilers, and the MicroQuill SmartHeap Library 6.01. Secondly, the compiler used by VeriTest, GCC, is said to generate code that less well optimised for x86." QED less well than intels compiler in the previous sentence
GCC for PowerPC is not as mature: "The gcc scheduler is not really designed ideally for a processor like the 970 and the Power4...that was one of the things that we're continuing to work on to try to get the best performance out of the processor."
GCC on intel is far more mature with a long history, read a little of the history: "...When Intel released the Pentium some of their team produced a version of gcc with enhancements which gave 30% speed improvements on some benchmarks..."
Look at these redhat GCC 3.3/4.0 benchmarks. Notice how the 2-way PPC970 is twice as fast as the 4-way P4 on many tests and at close to par on the others. Now this is not the end all, am I'm sure you could come up with a different test that shows the P4 beating the G5, but certianly the G5 is not a "peice of crap".
Standardizing of compilers is scientific method. Ideally you'd do a bank of tests, and unroll the variables: Standard compilers, standard OS, standard CPUs. Or you could tune each system to the max and then compare, that was LinPack and you didn't like that one either.
Exactly my point! Intels compiler does auto-vectorization. GCC doesn't. If you test C code, P4+intel against GCC+G5, you are crippling the G5 by leaving out the altivec unit, which is a more capacble vector unit than SSE2
Hand coding is still done frequently on high performance algorithms:
Fact is almost every CPU architecture is better than X86: PowerPC, Alpha, MIPs, HP-RISC. The problem is competing in the marketplace against something with so much market mass. Sad really. We end up with the worst of all worlds, X86, because its got a corner on the market. Because the G5 gets par performance with half the silicon real estate, you could get a dual core G5 for the same cost as opteron. If technological goodness and price won the day, instead of hegemony, you'd be telling me how your dual-core G5 thrashes everything from AMD and intel.
Apple is switching because one company cannot push a whole chip architecture cost effectively, no matter how good the architecture is
Don't believe me? Why are XBox, Sony, and nitendo all using PowerPC chips? The most performance, the lowest price. Microsoft can get a 3 core 970 for the same price as a P4 (58M verses 169M transistors)
2600 Opteron: 1796 int
2200 970: 1040 int
2700 970: 1276 int (scaled)
SO even if we used SPECint. the difference is 29%. NOT 3-1.
Ahh. What you seem to not understand is SPEC benchmarks test not only a processor but a compiler. Apple's benchmarks were not unfair at all, they just used GCC on both platforms, normalizing the compiler (which argualbly still favored intel which has received more GCC development work). The intel set called foul because intel compiler does better. OK, so? IBM compiler does better too. Intel compiler SPEC mark optimisations are well known, says Peter Glaskowsky, editor-in-chief of Microprocessor Report (google it). Hand coding... I guess you're to young to know what thats like. My point exactly. Whats yours? Linpack is less efficient. Efficiency, does not a benchmark make (one could argue the opposite) Thats the problem, SPEC isn't accounable. The manufacturers who submit to spec.org, don't detail test conditions nor do they get verified. Arguabily SPEC is somewhat more broad than linpack (though not as different as you would like to make out). But linpack is all these things and the 970 gets twice the performance in a set of tests using complex algorithms (which use PF, int, branches, rotates, matrixes, etc.) Ahh. See you like everyone else are blinded by marketing. The 970 is one of the most efficient chip for obvious reasons (it isn't carriing 30 years of x86 baggage). Lets just say for argument that the G5 is somewhat equal to the opteron (averaging linpack and SPEC)I like linpack, because each supercomputer group is hot-rodding their system to the max performance however they can, with smart CS people dedicated to each system to get the most out of it. And it is verified by a third party.
Linpacks age has nothing to do with validity. Lapack supersedes linpack because its more efficient, not more accurate at benchmarking. Math problems don't change with time.
It is not one test, but a collection of different tests: "analyze and solve linear equations and linear least-squares problems. The package solves linear systems whose matrices are general, banded, symmetric indefinite, symmetric positive definite, triangular,and tridiagonal square. In addition, the package computes the QR and singular value decompositions of rectangular matrices and applies them to least-squares problems."
I use Rpeak to eliminate network latency specific to that cluster implementation. This is valid because Linpack is a dense set calculations, its not just getting cached in the CPU! Furthermore Rmax wouldn't change the fact the G5 tromps the Opteron in FP.
So the Opteron trades FP for int. hmmm, not very exciting.
I'd suggest something different. The only Apps these days that don't use FP are the programs where you can't tell the difference between 1GHz and 4GHz (ie. word processing). FP is used intensivly thoughout a modern computer experience (graphics/sound/media/video/MP3s/games/3D).Take a look at the Linpack scores of the Top 500 supercomputer list. It's one of the fairest comparisons around, every processor optimised to the max, done by professional CS people proving only they're machine. You'll see that they are totally consistant between listing of the same processor type too.
Take the Rpeak and divide by number of procs.
What do you get?
2.3 GHz 970 (G5) = 9.2 GFlops (#7 on list)
2.2 GHz Opteron = 4.2 GFlops (#17 on list)
G5 smokes the Opteron by 2-to-1.
Take a look at top500.org This is the top 500 supercomputers benchmarked on linpack. I use it because it is one of the fairest benchmarks, since everyone on the list is optimised to the max. Now take the Rmax number and divide by number of procs. You'll find from system to system using the same proc that they're very consistant.
I gave the ibm paper before. Look at the power curve (second graph down), follow the speed down to 1.6 GHz, now what is its power? ~22Watts.
Now from the curve you can see that cranking the GHz on the 970 makes the power increase exponentially (like all processors). So at 2.7 GHz it really is hot, not P4 hot, but very hot.
The reason for water cooling is low noise, not that it is sucks more heat than a P4. The reason no G5 in a laptop is??? If you look at the same power curve the G5 at 1.25 GHz is around 15 watts not much different than a G4. Perhaps Apple thinks paying more for a G5 without a really big performance or GHz boost isn't worth it.
1. Steves marketing spin
2. IBM doesn't have a roadmap, so what do you do?
3. Intels SpeedStep can give low average power for the perceived performance, even though at full bore the PPC has better performance/watt. This is a features issue, put a speedstep like feature into PPC and it would do even better (in fact I think IBM has already licenced such tech from transmeta, but hasn't used it). But again if IBM won't make it, you can't use it.
XBox and playstation do. The proof is here in applications were cost REALLY counts. Where are ALL the game systems? Thats right PPC.
As for the notebook/power issue lets take a look:
P4 3.2 GHz = 6.3 GFLOPS (82 watts)
Pentium M 1.6 GHz = 3.6 GFLOPS (24 watts)
PPC970FX 1.6 GHz = 6.3 GFLOPS (22 Watts)
P4 and 970 benchs from Top500.com. Pentium M from scaled comparison with P4.
So what is apple talking about? Well Pentium M Speedstep probably has better power management to turn the clock down dynamically during occational use (like a word processor). Other than this add-on feature the pentium family has NO advantage other than market clout. The PPC gets P4 performance with pentium M power draw.
Lots of good tech has died because of stronger competitors with weaker products.
So for the same costs you could have a 3 core PPC 970 verses a 1 core P4, assuming IBM was willing to make them. The PPC would be almost 3 times faster. This is exactly why microsoft went PPC for Xbox.
As for faster clock rate, I think both AMD and PPC prove that's not the case, and power dissipation is the limiting factor "to crank up the MHz" - which is why P4 clock rates have virtually stalled out.
Apple didn't chose X86 because its a good idea... they did it cause IBM left them no choice.
PowerPC is a better platform than X86. But than again who isn't? X86 is kludges and full of baggage. Just goes to show with a lot of money behind it, kludges are pretty good. But as for faster and more efficient, its not. The PPC970 acheives similar benchmarks to a P4 at 2/3 the Mhz and 1/3 the die size, and less power as well. So for the same cost you could have a 3 core PPC (think Xbox). The laptop issue is about dynamic power features not inherent power consumption where the PPC whoops X86. Apple is switching not because the X86 is better is just has a stable marketshare with a lot of backing. Honestly it is a sad day for compution to see thing go backwards.
The problem is a market issue, not a technology one - as PowerPC is a far superior design to X86. It sad to see technology digression.
The PowerPC architecture has so much less baggage it can match intel performance with 1/2 to 1/3 the transistor count. Making it innately cheaper, less heat, and lower power. Also means you could make a 3 core chip for the same price as intel (a la Xbox).
The problem is IBM willing to make them? Apparently not.
There are many myths.
68000 weren't a dead-end. In fact the 68000 was far better prepared to scale than X86. It just didn't have the market. In fact the 68060 was faster than the initial PowerPC 601 it was replaced with.
Low power. The PowerPC 970Fx is lower power/performance than intel equivalents. IBM just hasn't invested in adding all of the power scaling features for laptops that the intel does.
$99 computers are different issue from CPU prices. If a company in china can make money with 50 cent margins on commodity computers good for them. Apple doesn't care about the walmart buyer. Walmart offers cheap chinese goods in every catagory, but they aren't competitive nor threaten the quality mid or high end goods offered by others.
A $500 Mac mini or $1000 iMac tell you nothing about the cost of processors. To a large degree silicon is a commodity priced per square area. The fact that PowerPC = P4 performace with 1/3 the real estate tell you most everything you need to know.
eWeek benchmarked the same Xserve and said this:
"Performance-wise, the dual-processor Xserve G5 compares well with Linux-based x86 servers. This is not surprising, considering that Mac OS 10.x is based on FreeBSD UNIX. Using the included Apache server, we ran the Xserve G5 through our standard WebBench tests. It did quite well on the static WebBench test, outperforming a competitor's dual processor x86-64 server running Apache server on SuSe Linux-64."
So who do you beleive? I'll take eWeek over some college guys.
An why do it?
More performance? No.
Lower cost? No. Look at the specs The PPC 970FX is on par with what Intel is making while using 3 TIMES fewer transistors. Transistor count = die size = cost. Why do you think playstation/Xbox/nintendo are ALL using PowerPC? Because it is the best cost&performance around!
Lower Power? No. This is one of the myths of the G5. It is a very low power device for the performance. Why is apple not making a G5 laptop? because unlike their windows counterparts who make big clunky laptops to accomodate a P4, Apple is more concerned about sleak design. Why water cooling? Cause they want quiet computers. Get this: at the same clock speed the G5 uses less power than the G4. There is no reason for apple to switch. These are dumb rumors.
Your argument for evolution is unfortunately wrong.
Sexual variation, both physical and psycho-physiological, are the most common variations in humans. The reason? Because sexual variations are of the few physical variations that aren't life threatening. (ie. you could live without a liver, but could live as a gay person or hermaphrodite just fine).
Because they are not unadaptive to life, there is no reason to evolution throw them out of the genetic code. In fact, many hypothesis' have very adaptive explanations for such outcomes.
The sole evolutionary reason for a single persons existence is not to "replenish the earth", that is a uninformed, simplistic view influencd by modern Christian dogma (itself a simplistic view of the bible)
The demo design is terrible. (did I meantion ugly)
Zooming is somewhat interesting, though it could be implemented into other OSs easily enough (OS X has a zooming feature that could be mapped to mouse buttons).
This guy is so in love with text mode, that his fractal program in the demo uses ASCII graphics as if it was 1978!
The dell has a fast processor, but its hobbled with shared ram intel extreme video.
The Mac mini has at least a real video GPU. As apples Mac Mini website puts "it try running Halo on the cheapo PC".
Firewire and DVI are something you won't find on the Dell $500 computer.
The monetary values are equivalent. Apple has chosen to put money into features other than a fast processor, because it adds more real value to the customer (Who cares how many GHz it has, what does it do?)