Seriously though. If you have read the previous posts, several of them have been discussing the macdoobie rumors authenticity. So are we convinced? Not entirely. They may very well be fake. But then again, they may be real. We don't know. That's why we are discussing the rumors right now. I personally will be convinced first when Steve is showing them in a PS bake-off at WWDC in SF trouncing a P4-3ghz of course
Or maybe those people accidentally swapped the Cinema and Bryce benchmarks
maybe so 'cause something is wrong. I know for certain that 1.85 or a bit less is the magic number for predicting render times/bench scores using Cinema/CineBench when moving to single to dual processors. I just looked up some old scores for single vs dual 500 G4s (older Cinebench version) and single score (this is not seconds, mind you, this is a bench score) is 6.8, and dual is 12.35. The 1.85x multiplier would predict 12.58. Oh well, it's all just damn numbers anyway. (yes the 1.85 is supposed to work on Cinebench scores AND render seconds)
Seriously though. If you have read the previous posts, several of them have been discussing the macdoobie rumors authenticity. So are we convinced? Not entirely. They may very well be fake. But then again, they may be real. We don't know. That's why we are discussing the rumors right now. I personally will be convinced first when Steve is showing them in a PS bake-off at WWDC in SF trouncing a P4-3ghz of course
I hope he is trouncing a dual 3GHz Xeon. That would be soo sick!
If they use Cinebench I will be *very* curious about the OpenGL score. Want to see how much improvement can the new cpu and motherboard (new bus) can provide
If they use Cinebench I will be *very* curious about the OpenGL score. Want to see how much improvement can the new cpu and motherboard (new bus) can provide
you can say that again! I am so sick to death of seeing Intel and Athlon C4D scores climb into the atmosphere while I'm stuck at 2 feet off the ground. I thought about getting a Radeon 9700 and decided it would be a waste without the system to run it on. Hopefully we'll know soon (WWDC) what is REALLY coming and what it will do. Meanwhile, where's my piggy bank?
you can say that again! I am so sick to death of seeing Intel and Athlon C4D scores climb into the atmosphere while I'm stuck at 2 feet off the ground. I thought about getting a Radeon 9700 and decided it would be a waste without the system to run it on. Hopefully we'll know soon (WWDC) what is REALLY coming and what it will do. Meanwhile, where's my piggy bank?
MacRumors: One reader notes that the graph times for the PPC G4 Dual 1.42 and Pentium IV 3GHz are identical to old benchmarks posted at Bare Feats
So they ARE fake then \
Or, in a highly unlikely scenario, MB's source simply compared the 970s to the already established benchmarks. A member of the MB team registered and posted at the MR forums and stated that their source is reliable and that they do not have access to these machines themselves. Sigh...I retired from the MR forums a while back....all this talk about MR makes me feel nostalgic.
Who decides? reliability? MB? We all know that the PPC 970 is the next one.... By September.... Why MB in France is more accurate in its forecast than yours in US?
Having finally calmed down about the benches. They seem to fall in line with the expected leap from the G4 anyhow.
So, the 1.4 would outperform the P4 twice per clock. That puts it within P4 beating range. I can see that. Add a better bus and altivec on that and it's conceivable they could be neck and neck.
No surprises there.
Similarly, that being the case, I'd expect a dual 970 at 1.8 to pound the crap out of a single 3 gig P4. No shock there.
I have a pinch of faith in those benches. My gut seems to think there's nothing outlandish about them. Let's face it, after the G4 debacle, they actually look shockingly good. I just think we need to get used to the idea that Intel can have competition, can be beaten and that, given half the chance, PPC will clobber the outdated Wintel codec.
A 1.8 gig performs twice at the clock of a G4. Factor in an extra fpu unit and altivec all on a superior bus...are we really surprised at these numbers? I'm not. I'm wowed at the benches. Just to see something(!) after the drought...but...logically, there's nothing offensive here to me. What should they be?
Maybe PPC is about to finally realise it's potential against wintel.
Maybe Intel shouldn't have pished off IBM all those years back. Revenge sorbet coming up.
maybe so 'cause something is wrong. I know for certain that 1.85 or a bit less is the magic number for predicting render times/bench scores using Cinema/CineBench when moving to single to dual processors. I just looked up some old scores for single vs dual 500 G4s (older Cinebench version) and single score (this is not seconds, mind you, this is a bench score) is 6.8, and dual is 12.35. The 1.85x multiplier would predict 12.58. Oh well, it's all just damn numbers anyway. (yes the 1.85 is supposed to work on Cinebench scores AND render seconds)
Unfortunately you have made a bad assumption -- the 1.85 number will change between different processors and different memory architectures. The 970 and Apple's chipset & memory system could result in a higher or lower multiplier.
Comments
Originally posted by Markus
Are you really all convinced by MB rumors?
It's all we've got right now...
Originally posted by Markus
Are you really all convinced by MB rumors?
Yes, of course. Why would they not be true
Seriously though. If you have read the previous posts, several of them have been discussing the macdoobie rumors authenticity. So are we convinced? Not entirely. They may very well be fake. But then again, they may be real. We don't know. That's why we are discussing the rumors right now. I personally will be convinced first when Steve is showing them in a PS bake-off at WWDC in SF trouncing a P4-3ghz of course
Or maybe those people accidentally swapped the Cinema and Bryce benchmarks
maybe so 'cause something is wrong. I know for certain that 1.85 or a bit less is the magic number for predicting render times/bench scores using Cinema/CineBench when moving to single to dual processors. I just looked up some old scores for single vs dual 500 G4s (older Cinebench version) and single score (this is not seconds, mind you, this is a bench score) is 6.8, and dual is 12.35. The 1.85x multiplier would predict 12.58. Oh well, it's all just damn numbers anyway. (yes the 1.85 is supposed to work on Cinebench scores AND render seconds)
Originally posted by NETROMac
Yes, of course. Why would they not be true
Seriously though. If you have read the previous posts, several of them have been discussing the macdoobie rumors authenticity. So are we convinced? Not entirely. They may very well be fake. But then again, they may be real. We don't know. That's why we are discussing the rumors right now. I personally will be convinced first when Steve is showing them in a PS bake-off at WWDC in SF trouncing a P4-3ghz of course
I hope he is trouncing a dual 3GHz Xeon. That would be soo sick!
Originally posted by scottiB
would an 80% increase it AltiVec provide an explanation?
The PPC 970 has not only a better FPU, but if I remember correctly it has two FPU's.
Originally posted by artcat
OTOH, I just saw a posting in another forum by a person in-the-know, that these short renders can really skew the results, so, who knows? not me
You went to Postforum?
imashination (Matthew) is actually a programmer in MAXON UK. He knows what he's talking
imashination (Matthew) is actually a programmer in MAXON UK. He knows what he's talking
yeah, I think he just went out on his own (away from Maxon). Too bad MB didn't use CineBench, then, eh?
Originally posted by Markus
OK ... Do you you trust every Benchmarks ??
Hmm, in the past PowerPC often performed better in real world than the benchmarks would let guess...
Originally posted by artcat
Too bad MB didn't use CineBench, then, eh?
If they use Cinebench I will be *very* curious about the OpenGL score. Want to see how much improvement can the new cpu and motherboard (new bus) can provide
If they use Cinebench I will be *very* curious about the OpenGL score. Want to see how much improvement can the new cpu and motherboard (new bus) can provide
you can say that again! I am so sick to death of seeing Intel and Athlon C4D scores climb into the atmosphere while I'm stuck at 2 feet off the ground. I thought about getting a Radeon 9700 and decided it would be a waste without the system to run it on. Hopefully we'll know soon (WWDC) what is REALLY coming and what it will do. Meanwhile, where's my piggy bank?
So they ARE fake then \
Originally posted by artcat
you can say that again! I am so sick to death of seeing Intel and Athlon C4D scores climb into the atmosphere while I'm stuck at 2 feet off the ground. I thought about getting a Radeon 9700 and decided it would be a waste without the system to run it on. Hopefully we'll know soon (WWDC) what is REALLY coming and what it will do. Meanwhile, where's my piggy bank?
Just rob a bank and you will be rich
Originally posted by NETROMac
MacRumors: One reader notes that the graph times for the PPC G4 Dual 1.42 and Pentium IV 3GHz are identical to old benchmarks posted at Bare Feats
So they ARE fake then \
Or, in a highly unlikely scenario, MB's source simply compared the 970s to the already established benchmarks. A member of the MB team registered and posted at the MR forums and stated that their source is reliable and that they do not have access to these machines themselves. Sigh...I retired from the MR forums a while back....all this talk about MR makes me feel nostalgic.
So, the 1.4 would outperform the P4 twice per clock. That puts it within P4 beating range. I can see that. Add a better bus and altivec on that and it's conceivable they could be neck and neck.
No surprises there.
Similarly, that being the case, I'd expect a dual 970 at 1.8 to pound the crap out of a single 3 gig P4. No shock there.
I have a pinch of faith in those benches. My gut seems to think there's nothing outlandish about them. Let's face it, after the G4 debacle, they actually look shockingly good. I just think we need to get used to the idea that Intel can have competition, can be beaten and that, given half the chance, PPC will clobber the outdated Wintel codec.
A 1.8 gig performs twice at the clock of a G4. Factor in an extra fpu unit and altivec all on a superior bus...are we really surprised at these numbers? I'm not. I'm wowed at the benches. Just to see something(!) after the drought...but...logically, there's nothing offensive here to me. What should they be?
Maybe PPC is about to finally realise it's potential against wintel.
Maybe Intel shouldn't have pished off IBM all those years back. Revenge sorbet coming up.
Lemon Bon Bon
Originally posted by artcat
maybe so 'cause something is wrong. I know for certain that 1.85 or a bit less is the magic number for predicting render times/bench scores using Cinema/CineBench when moving to single to dual processors. I just looked up some old scores for single vs dual 500 G4s (older Cinebench version) and single score (this is not seconds, mind you, this is a bench score) is 6.8, and dual is 12.35. The 1.85x multiplier would predict 12.58. Oh well, it's all just damn numbers anyway. (yes the 1.85 is supposed to work on Cinebench scores AND render seconds)
Unfortunately you have made a bad assumption -- the 1.85 number will change between different processors and different memory architectures. The 970 and Apple's chipset & memory system could result in a higher or lower multiplier.