"that is applicable to a remotely significant percentage of users then"
Well, if that's your criteria then 70% is a massive overestimation. ALMOST ALL applications that one uses day to day on a dp Mac are single threaded, any performance benefits come in the main from load balancing, and we all know how under utilised a single CPU is most of the time.
" Estimated initial clocks are 1.4-1.8. An dual 1.8 (3.6 total) 970 wouls be CONSERVATIVELY equal to a 9 GHz P4. Make sense yet why this is important?"
This is absolutely ridiculous. IBM have already published provisional SPEC scores for the PPC 970 @ 1.8Ghz, if I remember rightly, the scores were about equivalent to the top of the range Opteron. If Apple use 2x 1.8 Ghz 970s in their top machine, it'll be very fast, but hardly bettr than it's x86-64 equivalent.
it wouldn't have helped much if they HAd asked parliament - look what happened in the UK. 2 million people marched against the war but it happened anyway.
The fuckers, what do I care if BP get richer? The profits all end up in tax havens rather than paying for British schools and hospitals.
" I know it won't happen but I'd love to see the posts after something like that."
I don't think it would help much. The kind of nobhead who uses "gay" as a blanket insult is the kind of idiot who a) can't afford a Mac anyway, being as he is a 13 year old who still has to wash his dad's car to get enough pocket money to fund his XBOX habit b) would hardly post his approval, and even if he did, no-one would see it because he can't earn enough karma to get over the threshold.
"There's no doubt that the L3 helps, but actual memory bandwidth would help a whole lot more."
I wonder if we'll ever get a chance to find out? Imagine that the PPC 970 comes along with it's 800 or 900Mhz bus, and it posts performance improvements over the MPC 7455 of 1.5 - 2X clock for clock. The 970 is - certainly in terms of integer and fp resources - a massively more capable chip than the 7455, so will will ever be able to untangle the many factors that go to make up it's higher performance. I'm not sure we ever will. However, if Moto would just give us a DDR FSB capable 7455...
Incidentally, Moto's DDR mode on the interface for the 7455's L3 cache gives a pretty marginal real world boost over the SDR mode - www.powerlogix.com have an interesting whitepaper on just this subject.
I can certainly see that it might, but if we're talking LARGE amounts of video (beyond a few tens of seconds worth, in real life) then we've got to be thinking about getting our data from disk, and main memory bandwidth again becomes less relevant. If I'm encoding 30 mins from a D1-esque QuickTime movie to MPEG2 using Apple's MPEG2 encoder on a G4, how's the FSB bandwidth slowing me down? The idea that it isn't seems to be born out by the class leading throughput that Apple's (highly Altivec optimised) encoder delivers. It may not be the world's best quality encoder, but it's so very very fast.
Slashdot Mirror
you've SEEN BENCHMARKS?
I take it that you don't actually use a Mac professionally, then...
God save us from know-nothing wannabes.
"that is applicable to a remotely significant percentage of users then"
Well, if that's your criteria then 70% is a massive overestimation. ALMOST ALL applications that one uses day to day on a dp Mac are single threaded, any performance benefits come in the main from load balancing, and we all know how under utilised a single CPU is most of the time.
I'll just post this again, then:-
"Dual rpocessors give a 70% speed increase at best."
Err... nope. Dual processors give a 100% speed increase at BEST.
ti?
"Besides which, dnetc isn't hugely applicable to real world computer use."
So what? We're talking BEST CASE here, not common case - the vast majority of apps aren't multi-threaded at all.
look at your dnetc scores...
"Dual rpocessors give a 70% speed increase at best."
Err... nope. Dual processors give a 100% speed increase at BEST.
" Estimated initial clocks are 1.4-1.8. An dual 1.8 (3.6 total) 970 wouls be CONSERVATIVELY equal to a 9 GHz P4. Make sense yet why this is important?"
This is absolutely ridiculous. IBM have already published provisional SPEC scores for the PPC 970 @ 1.8Ghz, if I remember rightly, the scores were about equivalent to the top of the range Opteron. If Apple use 2x 1.8 Ghz 970s in their top machine, it'll be very fast, but hardly bettr than it's x86-64 equivalent.
Yeah! But would a 128-way '040 actually be much of an advance over a dual G4?
I think they threatened on AppleInsider's Powermac G5 write-up, too.
they HAVE threatened legal action on quite a few rumour sites recently - Think Secret's still got 2 pulled stories on it's front page.
separation of business and state? you mean they're NOT the same thing?
sounds like you need a separation of church and state to me.
we (in the UK) could do with one too...
if you're suggesting that there are 100 million racially European people in the US who weren't born there then you're an idiot.
it's spelled immigrants
and that might not be a bad idea, especially as we're probably talking about less than a hundred thousand people.
it wouldn't have helped much if they HAd asked parliament - look what happened in the UK. 2 million people marched against the war but it happened anyway.
The fuckers, what do I care if BP get richer? The profits all end up in tax havens rather than paying for British schools and hospitals.
Jebus, American cell networks are fucked up.
" I know it won't happen but I'd love to see the posts after something like that."
I don't think it would help much. The kind of nobhead who uses "gay" as a blanket insult is the kind of idiot who a) can't afford a Mac anyway, being as he is a 13 year old who still has to wash his dad's car to get enough pocket money to fund his XBOX habit b) would hardly post his approval, and even if he did, no-one would see it because he can't earn enough karma to get over the threshold.
I've always had trouble keeping my trousers on.
definitely
"Everyone who bothers to observe you"
have you thought about trading up to a life?
oh, you should have said
"There's no doubt that the L3 helps, but actual memory bandwidth would help a whole lot more."
I wonder if we'll ever get a chance to find out? Imagine that the PPC 970 comes along with it's 800 or 900Mhz bus, and it posts performance improvements over the MPC 7455 of 1.5 - 2X clock for clock. The 970 is - certainly in terms of integer and fp resources - a massively more capable chip than the 7455, so will will ever be able to untangle the many factors that go to make up it's higher performance. I'm not sure we ever will. However, if Moto would just give us a DDR FSB capable 7455...
Incidentally, Moto's DDR mode on the interface for the 7455's L3 cache gives a pretty marginal real world boost over the SDR mode - www.powerlogix.com have an interesting whitepaper on just this subject.
No, that's NOT what I wrote. I wrote that APPLE weren't reducing clock speeds generation by generation.
Not only anonymous, but stupid as well.
I can certainly see that it might, but if we're talking LARGE amounts of video (beyond a few tens of seconds worth, in real life) then we've got to be thinking about getting our data from disk, and main memory bandwidth again becomes less relevant. If I'm encoding 30 mins from a D1-esque QuickTime movie to MPEG2 using Apple's MPEG2 encoder on a G4, how's the FSB bandwidth slowing me down? The idea that it isn't seems to be born out by the class leading throughput that Apple's (highly Altivec optimised) encoder delivers. It may not be the world's best quality encoder, but it's so very very fast.
" Attention moderators! "Alan Partridge" is a known troll"
Known to whom, exactly?
Mod me down by all means, but it's easier to do so if the post is shit, rather than because some tosser AC told you to do so.
My posting on Slashdot is trivial enough, but to actually concern yourself with what I post is just fucking sad.