Slashdot Mirror
G5 Benchmark Roundup
An anonymous reader writes "In an ironic twist to the recent benchmark wars, Intel referred the Mac site MacFixIt to an analyst at Gartner Group who actually backed the PowerPC G5 platform with this assertion: 'These models certainly equal Intel's advanced 875 platform and should allow Apple to go until 2005 without a major platform refresh.'"
Another anonymous user writes, "While browsing the Xbench benchmark comparison site, I discovered some G5 benchmarks! The 'G5 Lab Machine at WWDC' got an overall score of 164.78, but much higher scores in certain areas. All of the tests are calibrated to give 100 on an 800MHz DP Quicksilver G4."
vitaboy writes "Sound Technology, one of the "leading UK distributors specialising in musical instruments, music software and pro-audio equipment," seems to have some data regarding the real-world performance of the G5 compared to the high-end PC. They state, 'The dual 2GHz Power Mac G5 with Logic Platinum 6.1 can play 115 tracks, compared with a maximum of 35 tracks on the Dell Dimension 8300 and 81 tracks on the Dell Precision 650 each with Cubase SX 1.051 ... More impressively, the 1.6GHz single-processor Power Mac G5 played 50 percent more tracks than the 3GHz Pentium 4-based system.'"
if you had read anything in the last week you would know that the only area the G5 is slow in is integeres. It dominates on floats. it's basically a tie on integers: in Single processor mode it loses by perhaps 10% on the SPEC tests and in Dual processor mode it beats the Dual xeons by a margin of maybe 20%.
Some drink at the fountain of knowledge. Others just gargle.
People obviously shouldn't form an opinion on a new platform in the first week following its much hyped anouncement.
Of course they should. That opinion is perfectly valid. And it is, "Wow. Those are going to be really fast. They look cool. I'm excited."
Or else, why would these PC-centric doofus post early benchmarks and make asses out of themselves if not to try to defuse an apparent threat?
In my experience, PC doofuses have always been big with the benchmarks. It's like a bragging right to them. "I tweaked my dual Smockron 4500 and got it up to 313.3 on SPECdickweed_base!"
Meanwhile, us Mac doofuses (and I use the term with the greatest affection) spend that same time actually working. Because we need the extra cash to feed our $4000-a-year Mac habit.
What I want are options.
Oh, come on now. No you don't. What you really want is a computer that satisfies all of whatever your personal criteria for goodness are. If there were only one computer in the world but it were perfect, you'd be happy.
The whole "what we really want is choice" thing just ain't so.
Actually, a higher FPS rating in Quake3-based games (I think the magic number is around 125) lets you jump higher and run a little faster. The key is that the engine physics are computed per frame, and something about the way they're written (maybe a rounding problem somewhere in there, don't ask me) allows for higher jumps and faster movement when you hit around 115 - 125 FPS.
Every cloud has a silver lining (except for the mushroom shaped ones, which have a lining of Iridium & Strontium 90)
ahem... the Dell wasn't running Logic.
They showed four top-shelf apps: Photoshop, Mathematica, Emagic, and one other I'm spacing on. In each case the apps were not demoed by mac but rather by someone from the app company.
Emagic is the software company, not the program, and the fact that their Logic program one was demoed by Gerhard from Emagic rather than someone from 'mac' ( I think you meant Apple!) is a rather dubious disinction when you consider that Emagic is actually a subsidiary of Apple.
Having said that, my contacts in the pro-audio community are hugely impressed by the specs that were being thrown around. Apple's decision to but Emagic and discontinue development on the PC version of Logic was widely criticised, but I think the pay-off of having Logic optimised for G5 will win Apple a lot of sales.
A pizza of radius z and thickness a has a volume of pi z z a
The dimms used are standard 64bit parts, so you need a matched pair for 128-bit access. muchlike an i875 chipset needs dimms to be added as identical pairs.
Apple did not cripple the other machines. Read the testing procedures, the reply from Apple, and the numerous comments about the issue. Apple provided the best results possible using GCC and it really all comes down to the compiler. Spec.org even supports this fact as they have commented on the controversy.
You'll notice on spec.org that there are G5 benchmarks provided by IBM that do extremely well against the Opteron and P4. Oh, and by the way, these same results that are provided by IBM are significantly better than the results Apple is using. Just goes to show you how much SPEC relies on compiler. The important thing is the G5 is competitive and so close it probably doesn't matter too much. Doesn't every company claim to have the fastest CPU out? Come one now, repeat after me. M..A..R..K..E..T..I..N..G.
As for the Quake benchmarks, there's not much documentation provided, but from what I understand, the results are consistent with that model Dell running a fresh install of Quake with no tweaks or optimizations. Still, I agree the numbers could be a lot higher and we'll have to wait until a G5 is released to get the real verdict.
Finally, Apple states they are the first 64-bit desktop. Yeah it's marketing and it relies heavily on the definition of a desktop, but I tend to agree with Apple. Why do I agree that Apple has the first 64-bit desktop? Because any 64-bit machine that was released before the G5 had to be ordered from a manufacturer of workstations or built using parts intended and marketed for workstations. If I had billions of dollars and I buy a an ASCII White to use at home, does that mean ACSII white becomes the fastest desktop ever?
Yeah, a consumer could always buy a 64-bit machine, but it's not marketed toward them and is more difficult to obtain. There was no mass market/consumer availability. Can you go into your local CompUSA or Best Buy and get a 64-bit computer? No. Can you order a 64-bit computer from the Home & Home Office section of Dell? No. You will, however, be able to walk into an Apple Retail store or CompUSA and buy a 64-bit G5 once they start shipping in Aug/Sept. In fact, you can buy one right now, except you won't get it for another month or two.
So yes, Apple may very well have the first 64-bit desktop... as long as Dell Home doesn't begin selling a 64-bit machine before September. They also conducted their own benchmarks because unlike the results provided by Apple, the results posted on spec.org are not well-documented and are usually inflated quite a bit.
Apple states they are the first 64-bit desktop.
Actually, Apple never has said that it is the first 64-bit desktop, as they know that they would be wrong due to previous 64-bit workstations. They have repeatedly said that it is the first 64-bit personal computer, as shown on their PowerMac page.
...is that these things don't have an optimized operating system yet. It's like running benchmarks of Photoshop on Windows ME on a dual Opteron or something.
Once 10.3 comes out, and once 64 bit apps get optimized, this system will kick even more butt...
Orange
The G5 results are
SPECfp_base2000: 840
SPECint_base2000: 800
SPECfp_rate_base2000: 15.7
SPECint_rate_base2000:17.2
All of these are documented in the Veritest report, which includes a complete breakdown of results.
As for the intel compiler, the fact that icc produces good code for AMD processors has been known for some time. For those of us who prefer to use free compilers, the gcc results are still of some interest.
Of course, to really compare "gcc performance" one might choose to subtract out the Fortran programs-- those were compiled by the non-free NagWare Fortran. Or you could choose to compare those propriatary results with scores published on SpecBench.
The F90 programs are galgel, facerec, lucas, and fma3d. The F77 programs are wupwise, swim, applu, mgrid, sixtrack, and apsi.
Let's massage the data into submission...
The machine that a dual 2GHz G5 trounced in all the real-world app tests was a Dell with dual 3.06GHz Xeons. Notice, I said "real-world app tests," not the questionable benchmarks. You can dispute the benchmarks, but it's hard to argue the performance differences I saw with Photoshop, Mathematica, etc. The Dell was flat-out dusted.
/. discussions at ~$4000 in configurations comparable to the G5's. I just did it myself. I configured my Dell PWS 450 by selecting two 3.06GHz Xeons, downgrading to 512MB of RAM, upgrading to a 120GB hard drive (still smaller than the G5's 160MB), upgrading to the cheapest drive that could write DVDs, adding a modem, adding a FireWire card, and subtracting a monitor. Components not specifically listed here were left at their default settings. Final price: $3772.
If a dual 3.06GHz Xeon system was shown to be slower than the dual 2.0GHz G5, please explain how a Dell with only dual 2.4GHz Xeons (which is what I presume you meant) is faster.
The Dell dual 3.06GHz Xeon system has been repeatedly spec'd out in recent
Since the bone-stock G5 is $3000, please explain how the dual Xeon costing $3772 is cheaper.
BTW, the exact Dell system above configured with 2.4GHz dual Xeons is $2522, not "under $2000" as you seem to have claimed.
~Philly
Why don't you read the benchmarks before speculating? Some of the benchmarks are written in Fortran 90. Neither f2c nor g77 (available from fink) support this code, so Apple used NAG Fortran f95 v4.2
Yes, there is a GNU Fortran95 compiler, but it's "in a pupal state."
Let's call a Spade a Spade.
Dell is no saint here.
The MicroQuill SMP HEAP library it used costs a cool 1200 Bucks. I didn't think Dell is shipping this library with each system it sells.
So, for Spec number's you can't look at Dells and believe them either.
I want to see a pure test between systems and library's I can buy.
Q.