Apple's Website Wrong About Intel SPEC Scores?

Apple lists the Intel Spec scores here for the P4 3.0 here:

http://www.apple.com/powermac/

as 889 on SPECint and 693 on SPECfp.

The SPEC site lists the scores here:

http://www.spec.org/osg/cpu2000/resu...0422-02135.pdf

http://www.spec.org/osg/cpu2000/resu...0505-02156.pdf

1164 and 1213 respectively. Not to mention the fact that thats a huge difference, but there's also the fact the 3.2 P4 just came out (which Apple can be excused from comparing against since it was just announced). Any explanations?

It's because they did their own compiling using GCC 3.3, which is horribly slow. I'd like to see some independant benchmarks on the rest because Apple has a history or exaggeration and using SPEC in this way just throws the rest of their results into question as well.

Its not that I didn't expect this from them, but the combination of a much hyped CPU and the first time I've seen an industry standard benchmark on their site almost led me to believe it, then I remembered the P4s were over 1000 in both categories. The design is still very elegant (the interior in particular is beautiful) and the performance is at least comprable to a PC for the first time in some time, but they're still upto their old misleading information.

The specs on the system look great and match the PC world, but they always have to go push the edge of believability in some area. Pricing is typical Apple of course, and they need some better options on graphics cards, but I assume they are trying to keep the cost down.

One interesting thing is that the different processor speeds all have different bus speeds. There is always a 2:1 ratio (or 4:1 if you compare to the frequency).

Quote:

---

Originally posted by Marsolin
The specs on the system look great and match the PC world, but they always have to go push the edge of believability in some area. Pricing is typical Apple of course, and they need some better options on graphics cards, but I assume they are trying to keep the cost down.

One interesting thing is that the different processor speeds all have different bus speeds. There is always a 2:1 ratio (or 4:1 if you compare to the frequency).

---

Well they offer the Radeon 9600/9800 and GeForce FX5200 on the low end. As far as Apple's history is concerned, thats a decent selection. I don't think the bigger GeForce FX designs would have fit in with the Apple design. I suppose pricing isn't that bad, but only if you disregard the fact that PCs prices are dirt cheap right now.

I do like how they moved away from the clear/frosted Lucite of the older designs to new industrial look. Apple's one of those technology companies that is a trend setter in design with the likes of Sony etc., and I have little doubt this new look will also be immitated in the future.

It keeps in line with the Apple philosophy...lots of hype, lots of great features, a sizable pricetag and no backwards compatibility. I did find some interesting ommisions, the first being the lack of SCSI availability (though this might have made a wirefree interior harder than with Serial ATA). Secondly, I was surprised that Bluetooth was not integrated. It would've presented them with an opportunity to introduce a wireless keyboard/mouse etc. Also, the new high end speakers are black Logitechs, which should sound great, but won't have the asthetics of a set of custom Harman Kardons.

Not really all that impressive, Granted it's a big step forward from the last mac. A couple of things do impress me but, PCI-X and the case looks good.

edit: You should stop by more often bob :)

I think the reason Apple used GCC 3.3 on the P4 is simply because that's what they use on their own platform. It's a more apples-to-apples comparison, although I still disagree with it -- use the fastest available for each platform. It's not Intel's fault that Apple doesn't make their own compiler :)

Do you guys think that Intel's compiler alone could account for the near doubling of performance in the SPECfp test? I've never heard GCC talked down that much for x86. Everyone raves about the Intel compiler, but I was surprised (and still am) that it alone could create that much of a performance increase. The guys at the video card forum have already torn apart the Quake 3 benches as well. I suppose the lesson rehashed here is that again Apple has their niche and they execute well in several areas, particularly product design and usability, but will resort to any marketing methods neccesary.

BTW FaTs, I'm here time to time -- don't see you much either :p . I'm interning full time and I have several other things to do as well including preparing for graduate exams etc.

Quote:

---

It keeps in line with the Apple philosophy...lots of hype, lots of great features, a sizable pricetag and no backwards compatibility.

---

What do you mean by "no backwards compatibility"? The G5 will run 32 bit OS X and apps just fine. In fact, it'll still do OS 9 (though I can't imagine why you'd want to ;) ).

Quote:

---

Originally posted by Ramuman
Its not that I didn't expect this from them, but the combination of a much hyped CPU and the first time I've seen an industry standard benchmark on their site almost led me to believe it, then I remembered the P4s were over 1000 in both categories. The design is still very elegant (the interior in particular is beautiful) and the performance is at least comprable to a PC for the first time in some time, but they're still upto their old misleading information.

---

ROFL, almost leading you to believe it. Welcome to the life of an uneducated Mac user who gets suckered into buying something he/she does not know about. I remember looking at some of their "benchmarks" on that site, and amazingly they had the Mac about 30% faster than anything Intel had. Not that they suck beyond belief (close though), but looking at the commercials should tell anyone that they are desperate. I agree though, some of their stuff looks pretty good, but I am not fond of their other gear. I won't even mention the price, ohh wait I just did :cool:. Anyways, that definitely would not be my first site to check if I wanted to get information on benches. Be sure to head over there to get a good laugh.:) I even remember the flame war that went on when between shadowfire and some idiot Mac user on another forum. Man out of the 50 guys who flamed against shadowfire, I think one actually had used a PC or was rational. Where am I going with this? Who knows, I'm out :cool:.

Quote:

---

trend setter in design with the likes of Sony etc.,

---

That's funny, when I think Sony, I think of crap made in a Chinese prison camp. Then I think about reliability problems and bad engineering.

You can't even attempt to put Sony in Apples class. Without Apple, there would be very, very little style in which is still a very, very boring market. I like their new deisgn, the whole lucite thing was getting old.

Quote:

---

Originally posted by Ramuman
Do you guys think that Intel's compiler alone could account for the near doubling of performance in the SPECfp test? I've never heard GCC talked down that much for x86. Everyone raves about the Intel compiler, but I was surprised (and still am) that it alone could create that much of a performance increase. The guys at the video card forum have already torn apart the Quake 3 benches as well. I suppose the lesson rehashed here is that again Apple has their niche and they execute well in several areas, particularly product design and usability, but will resort to any marketing methods neccesary.

BTW FaTs, I'm here time to time -- don't see you much either :p . I'm interning full time and I have several other things to do as well including preparing for graduate exams etc.

---

I don't think so. I'm sure I've seen much better results from gcc then this. SPEC FP also has a lot of Fortran stuff, and the Fortran compiler used is obscure to say the least. From reading other forums the Fortran compiler used actually converts the Fortran to C then uses gcc to compile, but that could potentially introduce some performance issues IMO, just witness the Sun facerec (???) sub benchmark in SPEC. It was a port from Fortran to C, and Sun came up with a way to improve the final performance by a factor of 10X by eliminating some of the problems introduced.


Anyway, compilers still do not explain why Apple is only getting 275 FPS on quake using a 3.0 GHz P4, + 875 chipset, and a 9800 pro.

Good read:

http://www.haxial.com/spls-soapbox/apple-powermac-G5/

Quote:

---

Originally posted by SiliGoose
What do you mean by "no backwards compatibility"? The G5 will run 32 bit OS X and apps just fine. In fact, it'll still do OS 9 (though I can't imagine why you'd want to ;) ).

---

Actually, it probably won't run OS 9. Starting a few months ago on their G4 models, Apple starting crippling their computers' BIOS so that the computer won't load an OS older than the one it shipped with. You can't even install an older revision of the same OS: If a Mac came with OS X 10.2.3, you can't load 10.2.1.

As for why you'd want to use OS 9... If you've got a computer lab or some other type of standardized environment, it's so much more convenient to be able to use the same OS and apps on every machine. Another reason is that there are a few apps that don't run well in the "Classic" environment within OS X.

Intel compilers are supposed to do very well against GCC (not sure if anything changed in 3.3 though).

Kinda dated and from Intel: http://developer.intel.com/software/...r_gnu_perf.pdf

It was probably a combination of compiler and the settings. They used GCC and probably didn't even try to optimize anything vs most likely optimizing everything for their G5.