NASA Benchmarks the New G5 Powermac 751
sockit2me9000 writes "Well NASA's Langley Research Center recently benchmarked the new G5 dual 2ghz Powermac against a dual 1ghz Xserve, a dual 1.25 ghz Powermac, a Pentium4 2 ghz, and a Pentium4 2.66 ghz. To make things fair, the second processor in the G5 was switched off, as well as the other dual sysytems. Then, they all ran Jet3d. Even with un-optimized code and one processor, the G5 performance is impressive."
Single Processor Mode (Score:5, Informative)
For those of you who were wondering, you too can switch off one of your Mac's dual CPU's with the Apple CHUD Tools [apple.com]. Look near the bottom of the page. It'll make you appreciate your second processor
Personally though, I want to see how well it runs Seti@Home [berkeley.edu].
Re:Single Processor Mode (Score:1, Informative)
fortran compiler (Score:5, Informative)
Re:Summary (Score:3, Informative)
Real Translation: 0.4% slower, at 75% of the clock speed.
Re:Costs - correction (Score:5, Informative)
Re:Single Processor Mode (Score:5, Informative)
You can also do this simply with the cpus= boot argument; here's a reference [apple.com].
Re:Turn the optimizations on first. (Score:5, Informative)
It should be noted that Apple uses gcc to compile Mac OS X and most of their applications, so it would be appropriate to use gcc on the G5. Intel's compiler might be a more appropriate choice for the Xeon.
Re:Interesting choice of processors (Score:4, Informative)
Budget had nothing to do with it; the PowerMac G5 isn't shipping yet. NASA had to have obtained theirs through a special arrangement with Apple.
Re:Summary (Score:1, Informative)
Disclaimer: I'm still an x86 fan, just impressed by the new G5.
If I remember right... (Score:4, Informative)
So, based on what was said at the keynote (and my interpretation), the G5s are 10dB quieter. Twice as quiet sounds more impressive. Note that saying "half as loud" still implies "loud" so psychologically it's not as impressive.
If I'm wrong, I'm sure someone will jump on me soon enough. I'm holding on tight.
Re:MFLOPS/MHz? No AMD, Old P4, Old Redhat. (Score:4, Informative)
And surely the version of RH Linux hardly matters. Maybe they benchmarked using this OS because (shock, horror) it is what they use daily.
Steve Jobs lucked out -- again. (Score:1, Informative)
But Jobs wasn't at Apple during that time. What's the timeline:
1983: Scully joins Apple.
1985: Scully fires Jobs. Jobs, no longer at Apple , begins Next.
1991: PowerPC alliance between Apple and IBM.
1993: Scully leaves Apple, Spindler becomes CEO.
1996: Gil Amelio becomes CEO. Jobs approaches Amelio with idea for Apple to buy Next. Apple buys Next.
1997: Amelio fired from Apple.
Re:G5 is really a full-blown workstation (Score:3, Informative)
Re:Single Processor Mode (Score:0, Informative)
And before anyone asks... (Score:0, Informative)
Dell Dimension 4600
Pentium 4 2.66GHz
512MB RAM
40GB hard drive
48x CD-ROM
XP Home
$749. Ships today. Oh, and that comes with a free 15" flat-panel monitor. (Link [dell.com])
Configured with a CD-RW and a 120GB hard drive, the above system is $919.
Apple Power Macintosh G5
1.8GHz (I'll be fair and not pick the dual 2GHz model)
512MB RAM
160GB hard drive
Superdrive
$2399. Not shipping until September. (By then, the above Dell will be $500 or less.) No monitor included.
Even if the G5 has better performance (and honestly, I still haven't seen reliable benchmarks with the dual 2.0GHz facing a P4/3.2GHz with Hyperthreading on), it still can't compete on price.
No matter how cool Mac OS X is... no matter how awesome these new G5s will be... if they are three times the cost of a PC, buyers will have a hard time justifiying it. In this economy, Apple will be hard-pressed to sell those $2300-$3000 desktops when people can get an equivalent-or-just-slightly-slower Dell for $800... including a flat-panel monitor.
Re:I'll wait for a real comparison. (Score:5, Informative)
Re:I'll wait for a real comparison. (Score:3, Informative)
Hyper threading does not give you the performance of a multi-processor setup. Hyper threading speeds things up when you have lots of independent threads. Lets say You get into a situation where you have, say, a cache miss and the CPU has to wait like 100 cycles to read crap out of ram. With HT, the CPU can use those cycles to run programs running in other threads.
It's like having two CPUs, but only one can run at a time, so while one waits, the other runs.
It just lets the chip come closer to it's theoretical maximum speed.
huh? (Score:2, Informative)
Re:And before anyone asks... (Score:1, Informative)
Go spec a Dell with a Xeon processor, a DVD burner, Digital Optical Audio I/O and a Huge hard drive (doesn't _have_ to be SATA) and a good graphics card and you will have a more realistic comparison.
Factor in the amazing case design (yes, as a tech it's so much easier to work on Apple machines than anything else) and things turn the other way...
Try adding 8GB of RAM to the PC and... oh, wait, you can't!
As for the benchmarks with HyperThreading tutned OFF they did that to speed up the PC, it was slower in those particular benchmarks with it turned off than it was with it enabled. HyperThreading isn't two real CPUs, it's not magic that speeds up _everything_
- k
Another message from the Benchmark author (Score:5, Informative)
Date: Mon, 13 Jan 2003 23:29:38 -0500
From: Craig Hunter
Subject: G4 vs. P4 performance
I have been following the discussion of Rob Galbraith's benchmarks with much interest, as I have spent a good deal of time testing, optimizing, and benchmarking software for the G4 (OS X) and P4 (Linux).
The first thing to realize is that there are numerous benchmarks that show the P4 is faster, and there are numerous benchmarks that show the G4 is faster. What matters? Well, probably the benchmarks that apply to the kind of work you do. For people doing photo processing with the software Rob tested, his results are extremely relevant. But, someone working with a program optimized for AltiVec and dual processors might have a completely opposite experience.
Just to give an example of a benchmark that goes the other way, see this chart.
(You're welcome to mirror this benchmark image, since my web site may not handle a lot of traffic). These real-world results come from the Jet3D computational fluid dynamics noise prediction software, which I developed for my doctoral thesis and currently use in my work at NASA. Jet3D is written in a combination of FORTRAN 77, FORTRAN 90, and C, and is optimized for AltiVec and dual processors on G4 hardware. When compiled on Linux using Intel's ifc compiler tools, Jet3D also becomes optimized for the P4 (using the various SIMD extensions available on the P4).
As you can see, the G4 does quite well here. A dual processor 1.25GHz G4 system is more than 3.5X faster than a single processor 2GHz P4 system. Though it's not shown on the chart, a single 1.25GHz G4 processor benchmarks at about 1589 MFLOPS, 1.9X faster than the P4. If you look at MFLOPS per MHz for a single processor, the G4 comes in at 1.27 MFLOPS/MHz, while the P4 comes in at 0.42 MFLOPS/MHz. If you want a good example of the MHz myth, look at the Cray, which comes in at 1.78 MFLOPS/MHz with only a 500MHz processor, beating both the G4 and P4.
Without AltiVec, the Jet3D benchmark would be about 794 MFLOPS on the dual-1.25GHz G4, which erases the performance lead over the P4. And then, using only a single processor, the 1.25GHz G4 benchmarks at about 418 MFLOPS, which is about half as fast as the P4. And all of a sudden, the G4 doesn't look very compelling. For the Jet3D benchmark, AltiVec and dual processors are key (AltiVec more so than dual procs). This is true for most benchmarks I have looked at; thus numerically intensive applications that can't use AltiVec and/or dual processors are likely to suffer on the G4.
In the case of Jet3D, it was easy to optimize for AltiVec. I was able to hand-vectorize about 10 lines of code within the guts of the FORTRAN algorithm and convert the computations to C for easy access to AltiVec hardware instructions. It had a huge effect for not a lot of work. For other more complicated cases, it may be possible to use the VAST compiler tools to automatically vectorize and tie in with AltiVec (VAST has parallel tools also). But in some cases, vectorization is not possible or feasible. In those instances, you're stuck with the processor's scalar performance, and the P4 generally has better scalar performance than the G4 in my experience. One final note: these are my personal views, and do not represent the views of NASA Langley Research Center, NASA, or the United States Government, nor do they constitute an endorsement by NASA Langley Research Center, NASA, or the United States Government
Re:And before anyone asks... (Score:3, Informative)
Those systems aren't *exactly* what I would call comparable. A HD that is 4x bigger, a superdrive, and thats just the stats you posted (I'm sure I could draw it out with things like the Airport Antenna).
"I still haven't seen reliable benchmarks with the dual 2.0GHz facing a P4/3.2GHz with Hyperthreading on"
Veritest disabled HT for tests where the system would be slower, left it on where it was faster. They also enabled SSE2. You can check all of that in their report off of their website.
"if they are three times the cost of a PC, buyers will have a hard time justifiying it."
Apple doesn't sell in the low-range (exempting iBooks), they sell mid-range and up. For those of us who purchase Apple systems, we don't want the cheapest system we can get, we want a system that
Re:NASA Verifies Apple Benchmarks? (Score:5, Informative)
More relevant, perhaps, are the figures in the raw MFLOPS graph:
Alas, difficulties in cross-platform benchmarking rear their ugly head:Scalar Code:
G4 using Absoft F90 v8: f90 -s -O -lU77 -N11
P4 using Portland Group F90 v4.0-3: pgf90 -byteswapio -tp p7 -O1
The author did apparently make an effort to use the compiler and flags best suited for each architecture if I read this correctly....
I don't know how much I trust NASA tho. Afterall, they only do RealMedia and WindowsMedia streaming media. Perhaps there's some bias there in favor of Windows (yes, I realize that the testbed P4 system ran Red Hat. Lighten up)
Damn Dude, Read What I Wrote (Score:4, Informative)
Apple claims 15.7 for the Dual 2GHz G5, and the 3GHz P4 getting an 8.07. NASA gives the Dual 2GHz G5 498MFLOPS and the 2.66GHz P4 255MFLOPS.
If you use your math skills: 15.7 / 8.07 about equals 498 / 255. So therefore we can draw the conclusion that they have similar results.
Now, NASA only used a 2.66MHz P4 while Apple used a 3GHz P4. Although remember NASA's figure that the P4 had 0.096 MFLOPS/MHz? Give the P4 333 more MHz, and you find it has about 286.968 MFLOPS. NASA also suggests a 20% performance increase can be expected with compilers that take advantage of the G5.
Although, even without this increase Apple's benchmark and NASA's benchmarks are very close. Which would lead one to draw the conclusion that Apple's benchmarks were in fact valid.
I should also note that a P4 would not perform as well in a dual system as the G5 does. So your 500 MFLOPS number is a little rediculous. The G5 which is an amazing dual proc chip saw it's 254 MFLOPS for a single processor (508 when doubled) drop to 498 MFLOPS in a dual system. And the P4 isn't designed for a dual system, doesn't support HyperTransport, etc.
Dak
Re:OS X 10.2.7 (Score:4, Informative)
MacOS X 10.2.7 - codename Smeagol [thinksecret.com] - is a stop-gap solution to provide JUST ANY working OS to the G5's until Panther is ready for prime time. Your Sofware Update is right not to install it on your machine, as most probably it is not a G5, sir.
Re:If I remember right... (Score:5, Informative)
Re:Single Processor Mode (Score:2, Informative)
Turning off a processor does not give the other one any performance increase at all.
Re:And before anyone asks... (Score:5, Informative)
How about a more fair comparison? Namely, between similarly configured high-end single-processor systems:
Apple PowerMac G5:
1.8GHz PowerPC G5
250GB Serial ATA - 7200rpm
SuperDrive (DVD-R/CD-RW)
512MB DDR400 SDRAM (PC3200)
Mac OS X
AppleWorks
ATI Radeon 9800 Pro
56k V.92 internal modem
No Monitor
$2874
Dell Dimension XPS:
3.2GHz Pentium 4
200GB Ultra ATA - 7200rpm
DVD+RW/DVD+R/CD-RW
512MB DDR400 SDRAM
Microsoft® Windows® XP Professional w/ Microsoft® Plus!
Microsoft® Works Suite 2003
ATI Radeon 9800 pro
No Monitor
$3062
And if you are to believe the benchmarks, it seems that Apple is selling the faster system for a lesser price than a similarly configured Dell.
Apple has never competed at the low end. It is not starting now.
-Mike
Re:Single Processor Mode (Score:5, Informative)
Re:MFLOPS/MHz? No AMD, Old P4, Old Redhat. (Score:3, Informative)
/ MFLOP \
| ----- |
\ S
-----------
/ MCYCLE \
| ------ |
\ S
Multiply by the reciprocal...
MFLOP S
----- * ------
S MCYCLE
Millions cancel, seconds cancel...
FLOP
-----
CYCLE
So it seems that this unit is equal to 1 floating point operation per CPU cycle. That makes a little bit more sense as a unit.
Re:5177 MFLOPS 288 MFLOPS (Score:5, Informative)
Re:Single Processor Mode (Score:4, Informative)
No shit. You can also not be a total ramrod and realize that one needs CHUD installed for that particular System Pref. (fyi : CHUD = Computer Hardware Understanding Developement Tools)
You should check your facts before you flame.
(tig)
"We do not inherit the land from our ancestors"
"We borrow it from our children"
Re:NASA Verifies Apple Benchmarks? (Score:5, Informative)
1) One of the Mac's processors was disabled
2) 195.3% advantage was on an MFLOP/MHz basis
That is how they are comparing the architechture of the chip and it's performance outside of a MHz pissing race. They are in the same ballpark now MHz wise so why shouldn't they take a look at how the actual chip performs. Not to mention how much more will likely come out of the chip with maturing compilers to take advantage of the arch.
Re:Costs (Score:3, Informative)
Re:MFLOPS/Mhz. - Useless Metric (Score:5, Informative)
Originally, the pipelining was segmented based on the I-Fetch, D-Fetch (register/etc), Exec, Reg-Write-Back, with expensive floating point doing with different timing considerations (externalized or delay-locking multi-stage execution). Then they started sub-dividing each of those stages (especially in CISC archetectures). Now its common to see 15 integer execution pipeline stages - either with shared resources, such that you can only have one divide occuring at any given time (early P-I, P-II, P-III), or with fully independent/concurrent resources (AMD's Athlon).
The addition of the pipelinable-stages between I-Fetch, D-Fetch, exec, and WB was somewhat trivial, because prior to pipelining, there were still seperate events on seperate clock-ticks with inter-stage latching. However, in CPU's with exec-stages that are pipelined, you are introducing additional latches that cause additional undesirable propagation delays.
So a 15 stage integer multiply unit (excluding fetch/WB) has 15 x [guestimating] 4 propagations of additional latency over a single-stage I-unit. If there are resource-based stage-interlocks, or worse data-dependencies, then the pipelining is useless and you're totally hit by the excess propagation delays.
Still, marketing being what it is these days, adding more stages means less propagations per stage, thus less worst-case propagation time, and thus higher clockability (all else being equal; temperature, etc).
The P4, however, compensated by double-clocking the core integer stages, so the number of advertised stages is somewhat misleading.
On a side note, due to the latching in pipelining, you're definately doing more total work for a given instruction. And more importantly, the designers have to think of totally different logic-algorithms to efficiently pipeline than to single-stage. My guess is that the pipelined versions will always be less efficient (especially considering that not all stages will fully utilize their allotted clock-time), and thus there's an additional loss.
Ok, so this supports your post, but here's the part about power/heat.
There are two types of transistors used in modern CPUs (everything past the Pentium). BiPolar and Field-Effect. The CMOS-FET refers to Complementary Metal-Oxide-Semiconductor Field-Effect-Transistor. This acts similarly to a capacitor in that there is a charge and discharge time with little waste current, and power dessipation is typical V=IR, Pwr = I^2 * R. The gate capacitor charge-time is the killer, and what limits switching speed (and thereby clock-speed). Shrinking the area of the capacitor (related to the micron-size stated,
There's another way of reducing switching speed.. Increasing the amount of current running through the wires that ultimately charge/discharge the gate-capacitors. FETs are poor amplifiers, but BiPolar (while more complex and harder to make small) are phenominal. In addition to their complexity, Bipolar also are power hogs. While a FET only consumes power while turning on or off, BiPolars are always on, consuming power (there is current bleeding from the switch). So what often happens is that designers sprinkle BJT's here and there to amplify the current (at the expense of cost/complexity and power-dessipation), and continue using FETs everywhere else.
The bigger and greater number of BJTs that are used, the faster some heavily loaded FET gates will charge and the quicker their switching time will be.
If you up the voltage on a CPU, you're enhancing the amplifier's ability to charge the capacitors and thus gives you more safety-room to up the external clock-speed.
Again, this deviates somewhat from my knowledge domain, but you can often merely co
+3db Doubles Power, +10dB Doubles Loudness (Score:3, Informative)
Re:The G5 (Score:3, Informative)
Re:The G5 (Score:0, Informative)
Alternatively, you can run WinXP on a $1200 whitebox PC from my local computer store down the road (again, Canadian dollars - about $900 US) with a P4-2.66GHz and a gig of DDR-RAM, 800MHz FSB, WinXP Pro, etc.
Hardware savings alone negate the US$160 paid for the OS, and PC hardware is easily upgradeable with off-the-shelf parts for the next ~3-5 years or so.
summary:
WinXP plus the next 3 service packs == US$160
OSX 10.3, 10.4, 10.5 == US$387 (plus inflation)
Of course, this is all academic unless you can actually afford the hardware, and anybody who wants Apple hardware badly enough will ignore the pricetag completely.
G5 has much slower SpecFP than Power4 (same clock) (Score:3, Informative)
Wrong Conclusion!
According to this pdf (page 13) [ibm.com] G5 @ 1.8GHz has 1051 SpecFP.
At the same time Power4 @ 1.7GHz has 1598 SpecFP !!! [spec.org]
It is very clear that Power 970 (G5) is much-much slower in floating point than it's Power4.
Bow wow wow yippie yo yippie yay... (Score:3, Informative)
Re:Single Processor Mode (Score:5, Informative)
Also, they were getting baseline tests on performance, against the G4. They also broke it down to performance per MHz, which the G5 took a huge lead in.
I suspect a dual G5, with an optimized compiler, would prove more than a match for the dual Xeon setup (That would cost significantly more, similar spec dual-xeon dell's are in the $4000 range), at least for this application (which heavily benefits from Altivec, and Altivec is still king of the SIMD world, SSE2 isn't even close in performance)
Re:Single Processor Mode (Score:5, Informative)
Re:Apple had little say in the Power4 CPU (Score:3, Informative)
Ah, but Power4 != G5. The G5 was called the GigaProcessor UltraLite (i know) in development and is quite a different beast from a full-blown Power4; it is scaled down to 'desktop tolerances', has not as many cores, and an AltiVec unit. Apple probably had a hand in what is now known as the 'G5' since the very beginning.
Re:Single Processor Mode (Score:3, Informative)
A benchmark will only show what is benchmarked. How many here care about how fast the cpu is in signle percentage points? Today's cpus are not as much as a bottleneck as they once were. Hardware has advanced beyond software for most apps.
What really matters is what you plan to do with it.
A disagree on the smp thing if your testing out chip performance. The reason being is that the OS and the particular benchmark have a huge influence on what the numbers will look like.
If your running a single threaded app that requires alot of cpu time, the smp benchmark would obfuscate the result and not tell you which cpu is the fastest.
For running a web server or testing i/o on the new motherboard then yes smp should be included. Alot factors change performance including which OS has the best i/o scheduler.
I use pc's as workstations for coding and browsing the web. A benchmark for me would be bootime and latency. MacOSX has been rumoured to be quite sluggish in disk latency. A faster processor certainly would not help this. My current athlonX+ takes awhile to load W2k with mysqladmin, interbase, OpenOffice, and Mozilla at startup. Would a twice as fast G5 mac help? Or would a new Serial ATA controller and drive help?
MaximumPC has bootimes in their benchmarks and I would like to see this and i/o tested on these new macs. They have hypertransport pci buses besides Serial ATA support. Sweet! This should help make the system alot faster overall and more responsive.
Re:G5 is really a full-blown workstation (Score:2, Informative)
Mac OS X automatically splits execution threads among multiple CPUs. Even something as basic as a progress bar or a network service daemon will run in separate threads. The Mach microkernel makes heavy use of threads so the basic OS itself will experience noticeable improvements in message passing between tasks (e.g. crossing the kernel boundary), especially given the dedicated CPU buses.
Also, there are a whole bunch of tasks running at any given time. I've got 64 right now on my Beige G3. Each of those tasks is running one or more threads (most only one). But if iTunes can run on one CPU while my compiler runs on the other, that's going to be a big performance gain. Not to mention how the new XCode is going to benefit from SMP.
Windows is probably in the same boat as far as distributing threads, but Mac OS X makes heavier use of threads/tasks (at least, we think so because Windows is too proprietary for us to know for sure). Plus there are a bunch of very important applications where SMP really matters, e.g. Adobe products, scientific research.
Re:Single Processor Mode (Score:2, Informative)
$394 P4 3.0 Ghz (800mhz bus, fastest they have at the moment) (pricewatch.com)
$165 1GB DDR 400 (pricewatch.com)
$376 Radeon 9800 Pro (pricewatch.com)
$159 Seagate 7200rpm 160GB SATA HD (via newegg.com)
$185 Albatron PX865PE Pro II Mobo (w/ USB 2.0 and SATA support) (newegg.com)
$242 LiteOn LDW-400D DVD-RW (newegg.com)
$80 Firewire 800 1394b PCI Host Adapter (firewiredirect.com)
$225 Creative Labs Sound Blaster Audigy 2 Platinum Ex (newegg.com) (Has the optical audio in/out)
$90 Liberal estimate for a decent case and power supply
Total (not including shipping and tax): $1916
*I don't know much about Intel systems honestly, so not sure if the motherboard I chose is decent. Was just doing a quick browsing of Anandtech for "good" mobos.
Also, some of my choices probably aren't the best. I've heard mixed reviews about SB Audigy's, but it was just a brand name I went for that I knew would have optical audio in/out.
Anyways, some of the specs are actually better than the G5, some aren't. Really depends on what you are going to use it for and what is available. I don't mind either platform honestly.
Oh. And just to spite the AC above, we'll go ahead and add an Apple Cinema Display to our system
Grand Total!
$2650! (Add your tax and shipping charges and you probably get pretty close to $3k. I'm sure with more research you can bring this price down further... or wait 3 months
Re:Single Processor Mode (Score:3, Informative)
For this type of work the CPU usually runs flat out and the bottlenecks that apply to things like opening MSWord documents hardly come into play.
If it's properly written, then HDD access speed is irrelevant, and even main memory access is hardly ever the bottleneck.
This is one of those applications where the system speed is determined by the speed and efficiency of the CPU.
Re:Single Processor Mode (Score:2, Informative)
Not well, IIRC the seti folks were reluctant to do certain optimizations for the Mac. The distributed.net stuff works much better.
Re:NASA Verifies Apple Benchmarks? (Score:3, Informative)
Re:Single Processor Mode (Score:3, Informative)
You're right. However the poster didn't ask for a dual-proc system
And if you did read my post, I noted that you would have to factor in shipping costs, so I'm not claiming ignorance to the prices.
Whatever though!
IHBT.
Re:Single Processor Mode (Score:3, Informative)
Altivec is single-precision, SSE2 is double precision. The latter is invaluable for scientific computations of many types of matrix problems, and being wrong twice as fast is of little use.
Altivec is nice, for what it is meant for (mainly media type calculations, signal processing, etc.) But scientists will prefer SSE2.
Re:Single vs Double Precision (Score:2, Informative)
You said:
Altivec is nice, for what it is meant for (mainly media type calculations, signal processing, etc.) But scientists will prefer SSE2.
Well I was under the impression, that if I need double-precision floating point arithmetic I use the FPU (which is quite fast with the G5 as you can see in Fig. 1 [cox.net] of the Article). ;-)
Concerning fast vector operations - some problems can be "dumbed down" to take advantage of the faster single precision units and lastly you can do an additional Newton-Raphson refinement step where double precision is needed.
Your claim "scientists will prefer SSE2" was proven wrong by the article, where the Jet3D test suite was only ported to take advantage of the Altivec SIMD unit, not the SSE2 SIMD unit - unless of course you argue that NASA engineers are not scientists
IMHO real scientists use whatever they think is best suited to solve their specific problem.
Re:Single vs Double Precision (Score:4, Informative)
I agree that AltiVec is superior to SSE (ie. single precision), but you compared it to to SSE2, which is a bit apple-to-oranges (no pun intended, btw). If the G5 FPU is faster than current SSE2 at double precision, it just proves the well thought out design of the PowerPC architecture (and the unfortunate legacy of Intel's FPU instruction set, which is still a handicap even with SSE/SSE2, due to the need to mode switch).
But SSE2 is still immature, and I expect compilers to improve, as well as chip implementations. Once they do, a more meaningful comparision can be made.
The Intel chips NEED stuff like SSE/SSE2 to achieve faster floating point speeds, whereas the PowerPC can get by without it, thanks to a much better FPU design, and thus, PowerPC makers will probably not spend the silicon to make a double precision SIMD instruction set anytime soon.
I stand by my claim that while most consumer and media software can get by with single precision, scientific computing (ie. large matrix calculations, to be blunt) quite often needs double precision (hell, you can get libraries that use 128 bit long doubles, these days), and will ultimately prefer SSE2. Scientists fuss with single precision SIMD simply because many of their applications can benefit so much from SIMD that it is worth the pain to use single precision (with proper conditioning and verification, etc.) Now that double precision SIMD is available, I can only predict they will want to jump to it, once tools for using it are there.
Granted, if Intel can't make a double precision SIMD unit that outperforms a double precision general FPU like the G5's, for matrix problems, then they don't deserve to design chips for scientific computing.
Re:Damn Dude, RTFA (Score:3, Informative)
The thing is, this usually works best when you have two separate types of task running concurrently on the processor, because they can (hopeully) take advantage of more execution units at one time. But, if you run two of the same scientific number crunching programs, on the hyper-threaded processor, they will both be competing for the floating point, and other, execution units at the same time. That competition has to be serialized by the chip, and the result may well be slower than if you just ran one process on the chip.
So, for this kind of computation, the hyper-threading may not be such a win (and it is probably marketed more for things like web-servers, where you might have a database program, and web content delivery program, both executing, and they may parallelize with each other a bit.