Slashdot Mirror
Better Business Bureau Targets Apple's G5 Ads
deathazre writes "The Council of Better Business Bureaus has suggested Apple Computer withdraw its claims of the world's fastest, and first 64-bit, PC after a complaint by Dell. However, even having one of their ads banned in the U.K. didn't stop them here in the States."
Apple has already stated the ad has run its course and it will be 'mindful' of the request in the future. Meaning we've already made out on that one, better luck next time :)
You are aware that those numbers are usually LinPack numbers, not just theoretical specs, right? the Xbox probably does run at 80 GFlops in some theoretical sense, when you consider the graphics chip. But, that won't translate into LinPack numbers. The Apple supercomputer, on the other hand, actually has a LinPack score worth mentioning.
I'm sure that's it. That concern for the consumers' well-being would be why Dell advertises the Celeron 2.4GHz as only suitable for word processing and e-mail. ``Ooh, ma'am, if you're going to be surfing the 'web, you're going to need top-notch power for that kinda number-crunching!''
psst, linux has been 64 bit for a while, and it runs on a G5!
Apple Manipulating the Results
From http://spl.haxial.net/apple-powermac-G5/
Before we examine the SPEC results that Apple/Veritest claims, it must be noted that Apple/Veritest have used a few "cheats" to make the G5 look better. So whenever you see a Apple/Veritest result in the following tables, be aware that it has been affected by the following "cheats".
Apple/Veritest used a special fast malloc library on the G5 benchmark, but did not use it on the Dell/Intel benchmark, thus giving the G5 an unfair advantage. Here is the relevant quote from the Veritest report:
"Installed a high performance, single threaded malloc library. This library implementation is geared for speed rather than memory efficiency and is single-threaded which makes it unsuitable for many uses. Special provisions are made for very small allocations (less than 4 bytes)."
(Page 5, also see Appendix E, Page 26, Veritest PDF)
For both the Dell Dimension 8300 and the Dell Precision 650, Apple/Veritest performed the multi-processor "Rate" benchmarks with hyperthreading DISABLED. They had hyperthreading ENABLED for the single-processor benchmarks, but DISABLED for the multi-processor benchmarks, despite the fact that hyperthreading would have improved the performance of the multi-processor "Rate" benchmarks, while having little or no effect on the single-processor benchmarks. In either case, this performance-enhancing feature of the Intel processors should not have been disabled. Here is the quote:
Dell Dimension 8300 Configuration for SPEC CPU2000 Rate Base Testing
[...]
Use the system setup utility to disable hyperthreading in the system.
[...]
Dell Precision 650 Configuration for SPEC CPU2000 Rate Base Testing
[...]
Use the system setup utility to disable hyperthreading in the system.
(Pages 7-8, Veritest PDF)
Apple/Veritest used a special fast "relaxed IEEE math operations" compiler option on the G5 benchmark, but did NOT use it on the Dell/Intel benchmark, thus giving the G5 an unfair advantage in floating-point operations. If you are going to use a performance-enhancing trick on one computer, then to be fair you must use it on the other computer as well. The equivalent option in GCC for the Dell/Intel computers is "-ffast-math", but Apple/Veritest did not use it. Here is the relevant quote from the Veritest report:
"Appendix E. Apple Power Mac G5 GCC 3.3 Compiler and Linking Option Descriptions
-fast [...] also enables the use of C99 aliasing rules and relaxed IEEE math operations.
-fastf [...] also enables the use of C99 aliasing rules and relaxed IEEE math operations."
(Page 26, Veritest PDF)
OK, guys. You heard it here first.
"Real Angus Beef" means that the cow was more than 50% black in color. It tells you NOTHING about the quality of the meat. (Not to say that they're not using good beef in their Six Dollar Burger...just watch out when you go to the grocery store)
Angus's Gold and Silver labels (or maybe it's platinum and titanium...whatever) are excellent beef. But "Certified Angus" alone doesn't say anything about quality. You'll be much better off looking for USDA Prime beef, preferably grain (not grass) fed, from the Midwest (not Texas).
That is all. : )
Why yes, I AM a rocket scientist!
I have a G5.
For my code, it's faster clock-for clock than a Xeon, and (usually) slower clock-for-clock than an opteron. Benchmarks can be made to say just about anything, but I bet the G5 is the fastest thing around for some people running their software.
To date, Apple has not released a 64-bit OS for the G5, and not only has not announced any intention to do so, but simply avoids admitting this to be the case. Think what you like, but even if you think "64-bit" doesn't mean "a single application can access more than 4 GB of RAM", you certainly have to be perplexed by the sense in which Apple claims to have "broken the 4 GB barrier", given that their latest OS provides your app access to the RAM just the way an Intel-based 32-bit system can.
Since there is not 64-bit OS for this machine (although Linux is very close), I cannot prove that my G5 has 64-bit hardware, tho I guess I believe it.
I will now accept my troll-mod, since I have posted this atrocity in an apple.stlashdot.org story.
The bottle neck of a Modem is SO SMALL that on any modern machine processing power is irrelevant. So called 'web accelerators' do two things to improve perceived performance: they cache popular web content (this reduces dns lookup latency, routing latency) and in many cases recompress images with more compressed jpegs and if the browser supports it compresses the html as well. The net effect is that you squeeze 40 to 50k out of the website on the isp end (with all the processing overhead that entails on THEIR end) before sending over the modem.
The last machines that decoded a jpeg at less than 5 KBps were 386's -- and how many of those had 32bit displays?
So anyhow, the short version is when talking about 'netburst' and 'webcellerator' and 'aolhighspeed' the 5 KBps bottle neck of an average modem is such a huge limiting factor that processor is not an issue.
If you are talking about huge bandwidth like gig-e and greater, then you have to have a processor and bus fast enough to support the throughput. This is why the advertisements that lean toward "faster computer faster downloads!" are misleading.
And to quickly address soft modems: any g4 or pentium3 is beyond powerful enough for v.92 softmodems. The speed of the processor does not effect connection speed, line quality does.
Read Heinlein's 1953 Revolt in 2100, now more than ever.
The BBB is nothing but a protection racket for businesses that traditionally garner lots of complaints (e.g. door-to-door sales, home improvement, predatory lending) to avoid escalation of a large number of complaints to people who would actually take some enforcement action.
The BBB is esentially useless after the fact if you've been screwed, but I personally check any local tradesman, etc. If I see a "blemish," knowing how easy they are to avoid, I do no business with that company. It's the same kind of due diligence as checking Google for references to a mail order company before you place an order.
One CPU cycle wasted on digital restrictions management is ONE TOO MANY.
>Apple uses the same components as every single other PC vendor.
True for:
- RAM
- HD
Sort-of true for:
- video cards (different BIOS, but same otherwise AFAIK)
- optical drives (different firmware in some cases, special supplier agreements in some cases)
False for:
- CPU
- Mobo
- I/O chips (many of them are Apple ASICs)
- Power supply
- LCD screens (if applicable)
- Mouse & Keyboard
- Case
Apparently you've never ever looked inside a Mac before, or you'd know this. A G5 desktop is not an Opteron machine with a Gigabyte mobo and Antec power supply in a generic white-box case with an off-the-shelf Logitech KB and mouse, or something like that. A Powerbook is not a Dell Latitude with a different badge on it.
>Apple uses the same "cheap, generic components" and charges you more for them.
I guess cheap is a matter of opinion, but hardly generic. Take a Mac apart sometime. The stuff has brand names on it. Sony, IBM, ATI, Matsushita, Apple, etc. etc.
sigh... Photoshop has been 64 bit enabled since the G5 was introduced, Finalcut followed suit shortly after the introduction, and AfterEffects will no doubt be discontinued in the very near future, since Apple will clearly be pushing Shake (64 bit).
Just as irrigation is the lifeblood of the Southwest, lifeblood is the soup of cannibals. -- Jack Handy
The impressive fact is not the overall performance of the VT G5 cluster, but that the price-performance ratio was better if they bought G5s at the standard Apple educational price (not some insane 1-time giveaway deal cooked up just for VT) than if they bought systems from HP, IBM, Intel, or AMD.
From How Virginia Tech built a supercomputer:
"Intel, HP, IBM, and AMD were all trying to come up with ways to work with us," says Lockhart."But the prices were out of reach and IBM's 970 chip would not be available in time to allow the new Virginia Tech cluster to be ranked."
From Confessions of the World's Largest Switcher:
He looked at various architecture options and was in the process of buying Dells when the deal fell through. He also worked with IBM and AMD and couldn't get the price to match. The budgets were coming in at $9 to $12 million dollars.
When Dell built a similar cluster for more than half the price ($3M vs. $5.2M for VT's), they got a cluster with less than 1/4 of the performance.
Of course, this "performance" is measured by a benchmark, and all benchmarks lie, and single-computer desktop usage doesn't look like large-scale cluster usage, but the fact is, this was not a matter of somebody deciding to buy Apple and blindly throwing a bunch of cash at it. The Apple offering had better price/performance for their needs.
They are now working to get 2 terabytes of IS.
>Most computers, even servers, don't get put into supercomputing clusters, so they're not built for that.
And yet, VT found that the G5 desktop had better price/performance at standard educational prices than any of the offerings from the PC vendors they were talking to at the same time (HP, Intel, and AMD).
I beg to differ about video and audio encoding. I won't bother wasting time repeating apple's benchmarks in this where it was almost twice as fast as the fastest xeon's encoding a DV stream to MPEG2. I have experienced it myself first hand. The Dual G5 can encode full DV to VBR mpeg2 faster than real time. In contrast, my older G4 took about 3-4x real time.
I've found that cheap RAM can often adversely affect the stability of Apple's computers. In two instances I've used non-Apple RAM and found the system to crash randomly, and behave inconsistently on the whole, until the non-Apple RAM was removed and replaced with higher quality RAM.
In one instance, I just picked the wrong RAM, even though it said it was compatible with my computer, it wasn't. I got the manufacturer to send me a more expensive, higher quality chip instead. In the second instance (which was a 2 GHz G5, by the way), the retailer included a free GB of RAM with the purchase, but the RAM was cheap. In that case, they offered to exchange the RAM for the cost of the new RAM minus retail of what they gave us (which was a decent deal).
So, this may have nothing to do with your problem, since a lot of people have perfectly decent RAM. However, if you have purchased RAM from someone else, or had it included with your purchase, take it out and see if the problems disappear. The hardware test disk will not find problems in cheap RAM; it didn't in either of the cases above.
"Centrino" is not a processor, it's simply Intel's "label of approval" for a configuration that fulfills certain new mobility-related requirements. Currently the only such configuration is Intel's Pentium-M (note: not Pentium 4-M) on a Intel 855 chipset based motherboard. The 855 chipset has WiFi among other things.
:-)
So the processor you are thinking is the (very delicious) Pentium-M, and all "Centrino certified" laptops have onboard WiFi. You were partly correct, hope this clarified the rest
I don't believe the Opterons were available at that time either, making the G5 even better. Apple really was the fastest at that time. I don't know why people keep arguing with us (you and I and others) on this point.
And again...
Just to exercise my fingers a little more I'll repeat myself. No individual customers complained to the BBB about Apple. Dell did. This is comparable to Ford making a tip to the BBB about Chevy's claim that their mid-sized SUV gets the best overall mileage of all other currently available mid-sized SUVs.
Both were deemed satisfactory, the decision hinged on other factors.
That's according to insiders from VA Tech IT, at least.
"It is our blasphemy which has made us great, and will sustain us, and which the gods secretly admire in us." - Zelazny
In response to both you and the parent, this is not the case. By the time G5s shipped in september, intel was shipping 3.2GHz parts. Additionally, opterons were available early on in the summer, 3 months before the G5. Granted, Dell didn't offer opterons (and still doesn't), but that's because Dell doesn't deal with AMD. They were most certainly available before G5s though, by anyone's definition.
Slashdot | Apple Hardware VP Defends Benchmarks
spl is an idiot and known Mac community troll. Proof for the idiot part can be found here.
Oh, and here is what Luxology had to say benchmarkwise ...
Micron or Falcon Northwest shipped an Opteron gaming machine targeted at home users. This was the first 64-bit personal computer and most benchmarks showed the opterons faster than the G5, so the G5 wasn't the fastest, wasnt the first 64, it really wasn't anything but the best Mac you could buy.
At the time of Apple's G5 announcement (3 months before they even started shipping) Boxx had already been shipping dual Opteron machines for about a month.
I'm out of my mind right now, but feel free to leave a message.....
Actually, the 533 MHz Xeon bus means a dual Xeon system is likely memory bound for a lot of video tasks, which are bandwidth-critical, or a mix of bandwidth and CPU bound. Single-processor P4 is a lot faster than Xeon for video decoding, for example, since the bus is 2/3rds faseter. Dual G5 and Dual Opteron provide way more bandwidth per processor than Xeon in dual configurations, and so win for a lot of media processing tasks.
My video compression blog
Some readers might interpret this as meaning that 64-bit pointers are the only benefits of a 64-bit CPU. I'd like to point out the advantages of single-instruction (u_)int64_t operations.
There are a bunch of algorithms that will run twice as fast on 64-bit CPUs and 32-bit CPUs. String comparisons where the string length is known a priori (as in Java or Pascal strings) can be handled 8 bytes at a time rather than 4 at a time. There are also some tricks that can be done with null-terminated strings, but these Multi-precission arithmatic and memory comying routines also benifit greatly from 8 byte words.
On 64-bit systems, you could also do things like re-writing the O'caml virtual machine so that it internally uses 63-bit integers and doesn't box 32-bit integers.
Copyright Violation:"theft, piracy"::Anti-Trust Violation:"thermonuclear price terrorism"<-Overly dramatic language.
Well, moron, it was the G4 that was a "SuperComputer". At least it fell under the export restrictions for supercomputers of the time. Not those of 1980 but of 1999.
Lars T.
To the guy who modded me down from perfect to terrible Karma - Apple haters still suck
I don't mod but... Let me tell... SPEC benchmarks are bullshit.
Real life, 1600 mhz G5 here, just mem added (768mb) I can play Unreal 2004 demo at 100 fps levels (1280*960) while system gets a fax at background, without a hitch.
Just 1 question. SPEC benchmarks use Altivec extensions?
IMHO, everyone stay happy in platform they selected and don't bitch about others.
Tell you a formula, if I see a single guy using "Dell" against Apple G5, I won't moderate it insightful. Same moron probably compared Dell P4's 3200 Mhz levels to SGI/Cray 600Mhz. Sony PS2 was? 300mhz? so it must suck compared to xbox which is 733 Mhz and so on...