Do you mean 3.06GHz 800Mhz FSB against 3.2GHz 800Mhz FSB?
Perhaps on SPEC becasue on real world, this guys (http://www.pcworld.com/news/article/0,aid,111270,00.asp) say there is only a 2-7% increase! I duno
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
PowerMac G5: World's Fastest Desktop Computer
- Thread starter MacRumors
- Start date
- Sort by reaction score
- Status
Do you mean 3.06GHz 800Mhz FSB against 3.2GHz 800Mhz FSB?
Perhaps on SPEC becasue on real world, this guys (http://www.pcworld.com/news/article/0,aid,111270,00.asp) say there is only a 2-7% increase! I duno
Originally posted by hacurio1
Do you mean 3.06GHz 800Mhz FSB against 3.2GHz 800Mhz FSB?
Perhaps on SPEC becasue on real world, this guys (http://www.pcworld.com/news/article/0,aid,111270,00.asp) say there is only a 2-7% increase! I duno
Their is no 800 MHz FSB model for the 3.06 GHz Pentium 4b, as I've mentioned before, it uses a 533 MHz FSB (the entire reason why it's clocked at such a frequency). On real world benchmarks (and SPEC for that matter) the 3.2 GHz Pentium 4 is on average 5% faster than the 3.0 GHz Pentium 4c which is significantly faster than the 3.066 GHz Pentium 4b on real world benchmarks and SPEC due in much part to it's much faster FSB (800 MHz to 533 MHz), superior motherboard, faster and lower latency memory, superior memory subsystem, and minor core revisions.
geezzas525-
This is your first and only warning: cut out the caps and the spamming. I've had to delete way too many of your posts from this thread. Do it again and you're banned.
I'm really disappointed no-one reported any of those posts.
This is your first and only warning: cut out the caps and the spamming. I've had to delete way too many of your posts from this thread. Do it again and you're banned.
I'm really disappointed no-one reported any of those posts.
Plumbercrackboy
$ 4200 for a dual opteron boxx??? Funny their config page says $2968 for me.
Boxx M4
Admittedly this is for the cheap machine with dual 240's but this config still delivers significantly higher spec int and fp scores than the G5.
AMD Opteron spec scores
It allso includes pro Quadro graphics rather than the G5's game card and twice as much ram.
no firewire? add $20
no audio? add an audigy2 $100
more disk? well theres a ton of room, add as many drives as you like at non premium (ie non vendor) prices, looks like it has room for about 6.
You allso contend this is a server only chip, not workstation. AMD dont agree with you, here is a couple of quotes from the horses mouth.
AMD Opteron Workstation Processor
The mass perception of marketting will is one thing, but it is best not to confuse this will with reality.
The claims made by apple that they where first and that they are fastest are demonstrably false, no amount of marketting want can make it otherwise.
Widely swallowed fudd is still fudd, a tool of conmen and fodder for fools.
$ 4200 for a dual opteron boxx??? Funny their config page says $2968 for me.
Boxx M4
Admittedly this is for the cheap machine with dual 240's but this config still delivers significantly higher spec int and fp scores than the G5.
AMD Opteron spec scores
It allso includes pro Quadro graphics rather than the G5's game card and twice as much ram.
no firewire? add $20
no audio? add an audigy2 $100
more disk? well theres a ton of room, add as many drives as you like at non premium (ie non vendor) prices, looks like it has room for about 6.
You allso contend this is a server only chip, not workstation. AMD dont agree with you, here is a couple of quotes from the horses mouth.
AMD Opteron Workstation Processor
The mass perception of marketting will is one thing, but it is best not to confuse this will with reality.
The claims made by apple that they where first and that they are fastest are demonstrably false, no amount of marketting want can make it otherwise.
Widely swallowed fudd is still fudd, a tool of conmen and fodder for fools.
Having the "fastest" personal computer on my desktop isn't really all that important to me. It will just be nice to know that we are "on par" with other professional desktop systems. Hopefully, Apple and IBM will be able to keep up. All this chest-beating and crotch-grabbing regarding "world's fastest" is an unpleasant distraction.
Expandability still Lacking...
Only TWO PCI slots and ONE PCI-X slot. Now that the next-generation PowerMac will have only TWO PCI slots, anyone feel two is too few? Also, not seems to have much room for extra drives, especially optical drives.... (but I can't imagine anyone needing more than two optical drives in a box... there's always Firewire to pick up the slack when it comes to attaching more drives). Lacking a "big tower" option, two PCI slots seem too little.
Only TWO PCI slots and ONE PCI-X slot. Now that the next-generation PowerMac will have only TWO PCI slots, anyone feel two is too few? Also, not seems to have much room for extra drives, especially optical drives.... (but I can't imagine anyone needing more than two optical drives in a box... there's always Firewire to pick up the slack when it comes to attaching more drives). Lacking a "big tower" option, two PCI slots seem too little.
Originally posted by ddtlm
Rower_CPU:
Yeah I wasn't sure what to say... "Hey this guy annoys me... ban him!"
Just use those handy-dandy "Report post to moderator" links and we'll sort it out.
I'm not ignoring his claims I'm just trying to tell you that his comparrisons are crap. The simple fact is he is trying to compare benchmarks from one test ran with one compiler to another test done with another compiler. You can't do that. It's like trying to compare a Photoshop test to a Bryce test to see which one finishes a photo render first. It doesn't work. The only way to compare is to use the same compiler. That's what Apple did and yes it did show slower on the Intel based systems but guess what it in reality should have slower results on the G5 also since it wasn't specifically optimized for either cpu. The results Apple got were actually lower then the ones IBM has been able to achieve so I think if you want fair comparissons you have to do real world testing. If you want falsely based results from benchmarking fine then let Intel run optimized tests on there machines and let Apple do the same. Apple chose the path of least resistance in using a compiler that would not be better on the G5 or the Intel hardware. So in other words Apple didn't try to make the G5 look good. If they had wanted to do that they would have made there own 64bit compiler and then used the GCC only with the Intel hardware and made there machines run dual in all the tests. In this way they probably could have shown at least a 1.5 to 2x speed advantage in all categories rather then being slower in one or two categories. It all comes down to equal testing which this guy is not showing.
I agree with 99% of what you just said - but workstations do not equal desktop PC's. Honestly, the line between them is now blurring though.
To Apple's target audience, this is the first 64 bit PC - at least it's the first one that anyone is trying to sell them.
In other news, I think Apple may be doing AMD a HUGE favor by laying the 64 bit card on the table here. This alows them to: a) bring up the fact that they are already shipping, and b) if they are faster, rub Apple and Intel's noses in it.
Unfortuanetly, for AMD, they don't seem to have a good enough PR / marketting department. Most consumers see them as that other company who is always bitting at Intel's heals. I hope they can shake that reputation, but I doubt it.
In the end it will come down to who sells more boxes, Apple/IBM, Dell-Compaq-HP/Intel or UnkownPCMaker/AMD. Actually, more boxes is not right - it's who makes more profit selling their boxes. Only time will tell, but until a big name starts selling Opteron PC's - they're going nowhere. (Have HP or Dell even anounced any Opteron plans?)
geezzas525 - spamming
Hi,
This is my first ever post, hello to everyone.
I'm glad that the posts from 'geezzas525' were removed, as they were quite depressing.
I love Mac's and am planning on getting a G5 soon... Any recommendations on the model?
At least I get the ADC discount, mmm nice
Best regards,
Andrew
PB G4 - 667 - 768mb
10Gb Ipod
Hi,
This is my first ever post, hello to everyone.
I'm glad that the posts from 'geezzas525' were removed, as they were quite depressing.
I love Mac's and am planning on getting a G5 soon... Any recommendations on the model?
At least I get the ADC discount, mmm nice
Best regards,
Andrew
PB G4 - 667 - 768mb
10Gb Ipod
No it's just based on the fact that linux continues to not be considered a desktop or consumer OS. Nor would I consider a single processor G5 to be a workstation. That's just a desktop machine that has been stupidly overpriced.
Go find me an Athlon 64 with a 64 bit version of Windows and you can say AMD won. By their own admission that's their desktop CPU.
IBM, Sun, Compaq, HP all had 64 bit hardware long before AMD conceived the idea for their 64 bit series so they still haven't won a thing in the server or workstation markets.
Really you can debate that one until you're blue in the face but AMD didn't win the race to anything here. Part is their own fault and part is Microsofts but that's life and they still weren't first.
Those saying the PPC970 needs a 90 nm process to go beyond 2 GHz are incorrect.
Moore's law is distorted enough to begin with without further distortion. In fact all it's derived from is that process manufacturing technology advanced roughly every 12 months and there's a rough relationship between that and transistor count. It was then changed to 18 - 24 months, which still hasn't been too accurate until recently, because it was wrong.
Just out of interest when you say Intel and AMD have been doubling their clock speeds are you referring to the fact that Intel has moved 1 GHz (2.2 - 3.2 GHz or 45%) from Jan 2002 - June 2003 or the fact that AMD moved 600 MHz (1.67 - 2.25 GHz or 35%, which they aren't even at anymore)? Please fill me in on which manufacturer moves anything close to 100%.
Finally Apple compared their hardware to off the shelf available hardware at the time of their testing. They cannot compare to a processor that is released on the same day as they announce their own hardware. They chose Dell because Dell dominates the market.
Hyperthreading problematic
Okay, why is this: Hyperthreading was enabled on both Intel machines for SPECxx_base2000 tests, but disabled for SPECxx_rate_base2000 tests (pp. 7-8 of Veritest's apple_performance.pdf). So it was only disabled for about half the tests and both were reported.
There seem to be reports of Hyperthreading causing problems when number crunching: http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=pan.2003.06.07.12.47.50.695978.2430%40polybus.com&rnum=4&prev=/groups%3Fhl%3Den%26lr%3D%26ie%3DISO-8859-1%26q%3D%2522disable%2Bhyperthreading%2522%26btnG%3DGoogle%2BSearch
Perhaps it was necessary to turn it off to run the test software.
Okay, why is this: Hyperthreading was enabled on both Intel machines for SPECxx_base2000 tests, but disabled for SPECxx_rate_base2000 tests (pp. 7-8 of Veritest's apple_performance.pdf). So it was only disabled for about half the tests and both were reported.
There seem to be reports of Hyperthreading causing problems when number crunching: http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=pan.2003.06.07.12.47.50.695978.2430%40polybus.com&rnum=4&prev=/groups%3Fhl%3Den%26lr%3D%26ie%3DISO-8859-1%26q%3D%2522disable%2Bhyperthreading%2522%26btnG%3DGoogle%2BSearch
Perhaps it was necessary to turn it off to run the test software.
MacBidoule our leaders in the rumor world have some interesting input on the testing debacle.
http://translate.google.com/transla...www.macbidouille.com/&hl=en&lr=&ie=UTF-8&sa=G
Scroll down to 'Apple Cheater'.
I personally as I have said think all the benchmarks shown at Apple and at the guys site beating on Apple are a bunch of junk. The truth comes from the side by side comparrisons of apps as done in the keynote and as real world people will do as soon as the G5 is released.
http://translate.google.com/transla...www.macbidouille.com/&hl=en&lr=&ie=UTF-8&sa=G
Scroll down to 'Apple Cheater'.
I personally as I have said think all the benchmarks shown at Apple and at the guys site beating on Apple are a bunch of junk. The truth comes from the side by side comparrisons of apps as done in the keynote and as real world people will do as soon as the G5 is released.
Originally posted by ddtlm
Rower_CPU:
Yeah I wasn't sure what to say... "Hey this guy annoys me... ban him!"
Yeah, me neither (sorry this is a late rejoiner - had to nip off and do some work, godammit). I thought you tried to handle it quite well actually (subtle rather than confrontational)- and he did calm down after a bit, so...
Oh well, you live and learn.
It's actually quite nice to see that a moderator will step in to keep things civilised.
So I guess that's yet *another* reason to switch?
Oh, and on that theme:
agdickinson - if you want a G5 (rather than the now fairly cheap G4's) then I'd suggest you get either the 1.8 Gig or, ideally (and if you do dual processor work, definitely), the 2.0 Gig (faster memory, bigger drive, better PCI than the 1.6 Gig, essentially). I don't know what everyone else feels, but these two top-end machines seem really good value (I priced up a similar spec dual Xeon 3.06 from Armari (who *really* know how to make computers fly) and the Apple was actually cheaper (there - I've said it)).
When Steve Jobs was running through the prices, I saw the 1.6 at $1999 and thought "Oh bollocks, here comes a $4,000+ dual 2.0 Gig). Which just goes to show I shouldn't be such a misery guts...
Anyway, just my 1.3 pence (approximately)
Brother Mugga
ram prices
So to juice up this machine from the standard to the *RAM out the gills* maximum cost more than the machine itself... $3,750
18004memory.com will max out the g5 for $1,362.
http://www.18004memory.com/category.asp?catid=9&subcatid=911
Just observing and posting!
So to juice up this machine from the standard to the *RAM out the gills* maximum cost more than the machine itself... $3,750
18004memory.com will max out the g5 for $1,362.
http://www.18004memory.com/category.asp?catid=9&subcatid=911
Just observing and posting!
Re: Hyperthreading problematic
The rate tests (being used to test multiple cpus) are the only SPECmarks that Hyperthreading affects, single SPEC CPU2000 isn't threaded, the reason they didn't disable Hyperthreading for both comparisons was probably because the Pentium 4 actually performs better with Hyperthreading turned off in single cpu SPEC.
Other suspicious changes:
Choice of compilation flags for the Pentium 4 isn't the best performance-wise
PowerMac G5 were used a single-threaded malloc implementation optimized for speed, Pentium 4 tests did not
This page provides a few more.
https://forums.macrumors.com/showthread.php?s=&threadid=30402&perpage=25&pagenumber=3
The rate tests (being used to test multiple cpus) are the only SPECmarks that Hyperthreading affects, single SPEC CPU2000 isn't threaded, the reason they didn't disable Hyperthreading for both comparisons was probably because the Pentium 4 actually performs better with Hyperthreading turned off in single cpu SPEC.
Other suspicious changes:
Choice of compilation flags for the Pentium 4 isn't the best performance-wise
PowerMac G5 were used a single-threaded malloc implementation optimized for speed, Pentium 4 tests did not
This page provides a few more.
https://forums.macrumors.com/showthread.php?s=&threadid=30402&perpage=25&pagenumber=3
By the way.
Just before I toddle off to bed...
Am I correct in presuming that the antennae for bluetooth and airport are external because of the G5 case acts as a Faraday Cage?
I'm wary of bringing this up, because my last passing reference to decibels produced a slew (a word I've always longed to use in polite conversation) of physicsesque discussion, thereby revealing our true geeky natures to the watching world.
Just a thought.
Nighty night
Brother Mugga
PS: Apologies if this has already been discussed, but I didn't see it previously?
Just before I toddle off to bed...
Am I correct in presuming that the antennae for bluetooth and airport are external because of the G5 case acts as a Faraday Cage?
I'm wary of bringing this up, because my last passing reference to decibels produced a slew (a word I've always longed to use in polite conversation) of physicsesque discussion, thereby revealing our true geeky natures to the watching world.
Just a thought.
Nighty night
Brother Mugga
PS: Apologies if this has already been discussed, but I didn't see it previously?
Pentiums back in the dirt
Hate to be so immodest as to quote myself, but:
Just found this link to Slashdot on MacNN: http://apple.slashdot.org/article.pl?sid=03/06/24/2154256&mode=thread&tid=126&tid=181
This confirms what I suspected, that Hyperthreading was turned off because actually hurt Intel's test performance! Check out the link for Greg Joswiak's explanation of the test configs.
Hate to be so immodest as to quote myself
, but:Just found this link to Slashdot on MacNN: http://apple.slashdot.org/article.pl?sid=03/06/24/2154256&mode=thread&tid=126&tid=181
This confirms what I suspected, that Hyperthreading was turned off because actually hurt Intel's test performance! Check out the link for Greg Joswiak's explanation of the test configs.
Re: Pentiums back in the dirt
Read my previous post, Hyperthreading hurts single cpu SPEC which was why it was turned on for those tests, Hyperthreading was turned off for the SPECrate tests, where it would boost by a significant amount. Hyperthreading boosts performance in threaded programs and multiprocessing, it will no doubt boost SPECrate (which is really just SPEC to test multi-processor configurations). That the VP would say such a obviously wrong thing really annoys me.
I mean honestly, if Apple really cared about hyperthreading hurting performance, why would they leave it on for single cpu SPEC?
Originally posted by ralphh
Hate to be so immodest as to quote myself
Just found this link to Slashdot on MacNN: http://apple.slashdot.org/article.pl?sid=03/06/24/2154256&mode=thread&tid=126&tid=181
This confirms what I suspected, that Hyperthreading was turned off because actually hurt Intel's test performance! Check out the link for Greg Joswiak's explanation of the test configs.
Read my previous post, Hyperthreading hurts single cpu SPEC which was why it was turned on for those tests, Hyperthreading was turned off for the SPECrate tests, where it would boost by a significant amount. Hyperthreading boosts performance in threaded programs and multiprocessing, it will no doubt boost SPECrate (which is really just SPEC to test multi-processor configurations). That the VP would say such a obviously wrong thing really annoys me.
I mean honestly, if Apple really cared about hyperthreading hurting performance, why would they leave it on for single cpu SPEC?
You know, I think Apple did fudge their ad campaign a little. Kinda reminiscent of the 1 Billion gigaflops (or whatever)- supercomputer line.
But geez
I mean I'm an underdog supporter, that's why I like Apple and AMD, but let's be honest here. Add 'em up and they still don't outsell Intel. So maybe they should be teaming up instead of bickering.
Let's show Intel where they can shove these 64 Bits!
But geez
I mean I'm an underdog supporter, that's why I like Apple and AMD, but let's be honest here. Add 'em up and they still don't outsell Intel. So maybe they should be teaming up instead of bickering.
Let's show Intel where they can shove these 64 Bits!
Ok.. I'll add one more last bit into the SPEC debate... other than this, everyone should hold their decisions until they get their hands on one.
Anyways, Apple Hardware VP Defends Benchmarks
Interesting read, really. The VP said they disabled the HyperThreading so that the Intel CPUs would look better. Also, SSE2 was enabled, but it was done in the compiler.
I read the article that person wrote. I agreed with some things and I didn't agree with others. I'm just surprised the person took the time out to write something like that, but hey, people have their own opinions.
- Status