Intel will probably change their IA64 plans cause the 32bit x86 market is TOO STRONG. Developers will go with what they know. If they can get great performance out of existing apps hardware makers are likely to choose hammer over itanium. If devs can get hybrid 64/32 bit apps, or code once (and maybe only do minor tweaking) then they'll not bother with itanium versions. Intel, I think, thought it could dictate the future of the PC platform outright (and that others would have no choice but to follow), but that has never been truly possible on the PC side. X86 only moves forward while maintaining compatibility to the last generation. X86 is very big now, perhaps too big. People aren't going to just abandone it because Intel has decided it's time for a new platform. Intel started it, but I think they now realize they can't just step off it without hurting their sales. No, no. X86 is going to be with us for a few more years yet, and in most cases it's probably going to be a 32bit variant (read, all laptops and consumer PC's, and even most workstations and small business servers)
I wouldn't worry about Itanium (at all) or Hammer (much) the bread winners are still going to be Pentium and Athlon 4,5 etc etc... for at least the next two years. Guaranteed.
I agree with Moogs on this one. The Register article was one of the most bizarre bits of rambling I've read in a long time, especially the bit about pipelines, branch mispredictions, and clock frequency.
[quote]So potentially, there's no need for L3 on-die cache.<hr></blockquote>
Even at the same MHz rating as DDR, the L3 cache should have much lower latency than main DDR SDRAM. There are memperf claimed results mentioned at <a href="http://www.xlr8yourmac.com/systems/dual_1ghz_performance_test.html#storytop"; target="_blank">http://www.xlr8yourmac.com/systems/dual_1ghz_performance_te st.html#storytop</a> . 9actual results may be in te xlr8 forums) The dual 1GHz G4 beats a DDR Athlon in some of the memory benchmarks, implying that L3 cache could still be useful in a DDR Mac. (I'd reckon that the non L3 cached G4s would get stomped the same bandwidth test ).
My take: Remember when the Moto PPC roadmap showed the G5 as being a 7500, which was changed to 8500? I reckon that Apple told Motorola to get rid of the 7500 desktop CPU and the 8500 really is an embedded CPU.
IF the register is right. IF the register was right last time, we would have G5s NOW.
Barto is correct. The old road maps used to list the 7500 as the G5. So, it's back to the future. We've official come full circle. Jobs must be laughing his a$$ off. To summarize:
I still find this development depressing. Although I don't give the register article much weight, if true, it then it spells out the truth: we're going to be two steps behind x86 for the foreseeable future.
Moving the current G4 design over to a 130 nm process won't give us more than 1.4, 1.5 GHz, and even with a 266 MHz buys (how old is that tech now?), this isn't going to compete with the offerings from AMD and Intel (which will be offering bus speeds like that sooner).
Apple must do more than "keep up", they must thrive, and to grow marketshare they've got to offer hardware that's considerably better than the competition.
Now the 7500 is an interesting chip and with the longer pipeline it could signify a significant enough change in architecture to give Macs the speed they need. But again the problem is timing...if the 7500 isn't out until Feb 2003, then by this time even 2 GHz will be two steps behind the x86 CPUs. That's what makes this article so believable, is that it's predictions suck so hard. Just like Apple normally does.
I don't know if you all have figured this out yet, but the G4 and the G5 thing being that the G4 represents the 7400 series (G being the 7th letter in the alphabet and the 4 being the generation of the chip). This would make since why the G5 would be the 7500 PPC. As you know this rumor of the embedded chip that was renamed the 8500 was seemingly done to separate it from the desktop series. To which we see the road map presented in The Register article makes since as to the progression to the G5. The only question I have is who put this rumor that the G5 would be a 32/64 bit chip? I know that MOT said that the 8500 chip would be this 64 bit, but isn't this admittedly a embedded design? I can't say I know what Apple plans to do and I'm pretty sure that none of the people on this or any other forum can either.
<strong>I don't know if you all have figured this out yet, but the G4 and the G5 thing being that the G4 represents the 7400 series (G being the 7th letter in the alphabet and the 4 being the generation of the chip). </strong><hr></blockquote>
HOLY CRAPMONKIES! That means my PowerMac 8600 is a SPEED DEMON!!!
<strong>I still find this development depressing. Although I don't give the register article much weight, if true, it then it spells out the truth: we're going to be two steps behind x86 for the foreseeable future.</strong><hr></blockquote>
That's the cold hard truth that anyone without Apple-colored happy-goggles on has understood since 1999.
1 gHz G4 for "some months", then a traditional 10-15% speedbump at MWNY or Seybold. Then another in January 2003 bringing us up to 1.5 gHz or so, maybe with a new motherboard design.
Meanwhile the consumer lineup MIGHT get up to 1 GHZ by that time.
<strong>I don't know if you all have figured this out yet, but the G4 and the G5 thing being that the G4 represents the 7400 series (G being the 7th letter in the alphabet and the 4 being the generation of the chip). This would make since why the G5 would be the 7500 PPC.</strong><hr></blockquote>
Hmm, sounds good. But--will we see it THIS Q2? Though it's true that the DP1G is just about the fastest thing on the planet for video and 2D work, and probably 'fast enough' for 'most everything else, but, a lot of this is about keeping-up with the Jones's. I'm not sure why Intel are moving to a 533MHz system bus, but I think it's mostly playing the numbers game to keep one step ahead of AMD's 333DDR busses [soon ish?]. However, a doubling in bus speed would certainly help the Q-S boxes to be faster in bandwidth hungry situations-i.e.-3D. With 512K of cache, DDR, and faster chips [on the .13 process] Apple would have removed the main gripes about most users have with the G4s' in comparison to the PCs. ...Now, if only they'd make the lewer end, FULL versions of the GF3/4 standerd, note some MX rubbish! Macs' are NOT low-proced comps!
The last redesign of the <a href="http://e-www.motorola.com/"; target="_blank">http://e-www.motorola.com/</a>; site changed the structure of the 'PowerPC ISA' index to list the family of processors called the MPC7XXX. That's what led me to believe in the existence of the MPC7500, but now that The Register has jumped on the bandwagon, I'm skeptical... <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
By 2003 their will be Intel and AMD processors up to 3 even 4 GHz, with increase in performance of the motherboards. Moto has to get there goodies done by MWNY, then we can be satisfied for a month or 6.
If those predictions are true, then what about Apple's claims about "we want to close the MHz gap?" It's a GHz gap by now, and if they go onlike this, is going to be a 2GHz gap by MWSF 03. Then we'll see that new chip 7500 supposedly, that's going tohave a few errata, as always with new chips, and soon we'll have a 5GHz gap.
That shall be a challenge for the marketing team to make up an excuse "MHz myth for 5GHz".
It will be a good new to see 1,5 ghz G4 for MWNY with DDRAM,
For me the specifications of the 7500 are the one of a G5 :
rapid I/O architecture, 14 stage pipeline bus, (the statement of the first rumors about the G5) 11 for integer, 512 K of L2 cache, SOI 0,13 microns.
Compared to the original G4 : the whole architecture is different, much more difference than between a G3 and a 603.
Perhaps this rumor is false and crap, but it sounds logical. I would be personnaly happy to see 1,5 ghz 7470 for MWNY. It should be an incrase in performance of 60 to 70 % (considering that we where stuck at 500 mhz for 18 months this would be f the highest jump in performance since the G3)
Yeah I agree on that. (It's funny in fact I dreamt of a 1.5GHz G4 with DDR about 2 days ago at MWNY, somehow found 1.5GHz a strange rate though)
What is making me pessimistic is not the outlook for MWNY, but the outlook beyond that, the way it looks we're not going to hit 2GHz this year, and it even looks like no 64bit chip until 04 or even later. I'd have seen this as a great opportunity for Apple to be an industry's first, bringing 64bit to consumer PCs. even if the performance wouldn't be much better than the 32bit chips, "having more bits" is certainly a good thing for market attraction. If there's any way to make people forget about the lack of MHz, it's the higher "bit" rate on a chip. For examples, watch the console market during the last 10 years.
People LOVE numbers they can toss around to boast themselves. That's a point Apple isn't going to be able to exploit anytime soon, and that sucks. (as it won't help their sales).
But I said to myself, I'm waiting till MWNY, if we get DDR (ie a new mobo, faster dual chips (not 66MHz faster, I mean FASTER) and a new case) I'm getting one. If Apple fails to deliver once more, I'll wait till MWSF 03. Until then they HAVE to deliver (so I'd get that), if they don't deliver there, they're dead, and I'm certainly not going to spend money on a dead product.
Comments
I wouldn't worry about Itanium (at all) or Hammer (much) the bread winners are still going to be Pentium and Athlon 4,5 etc etc... for at least the next two years. Guaranteed.
<strong>I think the Register is full of crap on this one (again).
[ 02-11-2002: Message edited by: Moogs ? ]</strong><hr></blockquote>
I agree with Moogs on this one. The Register article was one of the most bizarre bits of rambling I've read in a long time, especially the bit about pipelines, branch mispredictions, and clock frequency.
Even at the same MHz rating as DDR, the L3 cache should have much lower latency than main DDR SDRAM. There are memperf claimed results mentioned at <a href="http://www.xlr8yourmac.com/systems/dual_1ghz_performance_test.html#storytop"; target="_blank">http://www.xlr8yourmac.com/systems/dual_1ghz_performance_te st.html#storytop</a> . 9actual results may be in te xlr8 forums) The dual 1GHz G4 beats a DDR Athlon in some of the memory benchmarks, implying that L3 cache could still be useful in a DDR Mac. (I'd reckon that the non L3 cached G4s would get stomped the same bandwidth test
[ 02-11-2002: Message edited by: Stoo ]</p>
IF the register is right. IF the register was right last time, we would have G5s NOW.
Barto
G5 = 7500 w/ .13 construction and SOI
G5 = 8500 w/ .13 construction, SOI, and on-chip memory controller
G5 = 7500 w/ .13 construction, SOI, RapidIO, on-chip goodies
[ 02-11-2002: Message edited by: CodeWarrior ]</p>
Moving the current G4 design over to a 130 nm process won't give us more than 1.4, 1.5 GHz, and even with a 266 MHz buys (how old is that tech now?), this isn't going to compete with the offerings from AMD and Intel (which will be offering bus speeds like that sooner).
Apple must do more than "keep up", they must thrive, and to grow marketshare they've got to offer hardware that's considerably better than the competition.
Now the 7500 is an interesting chip and with the longer pipeline it could signify a significant enough change in architecture to give Macs the speed they need. But again the problem is timing...if the 7500 isn't out until Feb 2003, then by this time even 2 GHz will be two steps behind the x86 CPUs. That's what makes this article so believable, is that it's predictions suck so hard. Just like Apple normally does.
<strong>I don't know if you all have figured this out yet, but the G4 and the G5 thing being that the G4 represents the 7400 series (G being the 7th letter in the alphabet and the 4 being the generation of the chip). </strong><hr></blockquote>
HOLY CRAPMONKIES! That means my PowerMac 8600 is a SPEED DEMON!!!
-----------
RosettaStoned
<strong>
HOLY CRAPMONKIES! That means my PowerMac 8600 is a SPEED DEMON!!!
-----------
RosettaStoned</strong><hr></blockquote>
Don't be an idiot. The 8600 is a model of a Power Mac not the chip. Or was that a joke? Never mind.
[ 02-12-2002: Message edited by: Arty50 ]</p>
<strong>I still find this development depressing. Although I don't give the register article much weight, if true, it then it spells out the truth: we're going to be two steps behind x86 for the foreseeable future.</strong><hr></blockquote>
That's the cold hard truth that anyone without Apple-colored happy-goggles on has understood since 1999.
1 gHz G4 for "some months", then a traditional 10-15% speedbump at MWNY or Seybold. Then another in January 2003 bringing us up to 1.5 gHz or so, maybe with a new motherboard design.
Meanwhile the consumer lineup MIGHT get up to 1 GHZ by that time.
<strong>I don't know if you all have figured this out yet, but the G4 and the G5 thing being that the G4 represents the 7400 series (G being the 7th letter in the alphabet and the 4 being the generation of the chip). This would make since why the G5 would be the 7500 PPC.</strong><hr></blockquote>
What about the G3? It's called PPC750
And the 601, 603 and 604? Where they F0s?
Hmm, sounds good. But--will we see it THIS Q2? Though it's true that the DP1G is just about the fastest thing on the planet for video and 2D work, and probably 'fast enough' for 'most everything else, but, a lot of this is about keeping-up with the Jones's. I'm not sure why Intel are moving to a 533MHz system bus, but I think it's mostly playing the numbers game to keep one step ahead of AMD's 333DDR busses [soon ish?]. However, a doubling in bus speed would certainly help the Q-S boxes to be faster in bandwidth hungry situations-i.e.-3D. With 512K of cache, DDR, and faster chips [on the .13 process] Apple would have removed the main gripes about most users have with the G4s' in comparison to the PCs. ...Now, if only they'd make the lewer end, FULL versions of the GF3/4 standerd, note some MX rubbish! Macs' are NOT low-proced comps!
Peace and Truth -- SammyF <><
The Register is NEVER right.
Voila
Fieldor
Geforce4 Ti 4600 rocks
G1: 601 chip
G2: 603/e/v and 604/e/v chips
G3: 7xx chips
G4: 74xx chips so far
If those predictions are true, then what about Apple's claims about "we want to close the MHz gap?" It's a GHz gap by now, and if they go onlike this, is going to be a 2GHz gap by MWSF 03. Then we'll see that new chip 7500 supposedly, that's going tohave a few errata, as always with new chips, and soon we'll have a 5GHz gap.
That shall be a challenge for the marketing team to make up an excuse "MHz myth for 5GHz".
The future used to be bright.
G-News
It will be a good new to see 1,5 ghz G4 for MWNY with DDRAM,
For me the specifications of the 7500 are the one of a G5 :
rapid I/O architecture, 14 stage pipeline bus, (the statement of the first rumors about the G5) 11 for integer, 512 K of L2 cache, SOI 0,13 microns.
Compared to the original G4 : the whole architecture is different, much more difference than between a G3 and a 603.
Perhaps this rumor is false and crap, but it sounds logical. I would be personnaly happy to see 1,5 ghz 7470 for MWNY. It should be an incrase in performance of 60 to 70 % (considering that we where stuck at 500 mhz for 18 months this would be f the highest jump in performance since the G3)
What is making me pessimistic is not the outlook for MWNY, but the outlook beyond that, the way it looks we're not going to hit 2GHz this year, and it even looks like no 64bit chip until 04 or even later. I'd have seen this as a great opportunity for Apple to be an industry's first, bringing 64bit to consumer PCs. even if the performance wouldn't be much better than the 32bit chips, "having more bits" is certainly a good thing for market attraction. If there's any way to make people forget about the lack of MHz, it's the higher "bit" rate on a chip. For examples, watch the console market during the last 10 years.
People LOVE numbers they can toss around to boast themselves. That's a point Apple isn't going to be able to exploit anytime soon, and that sucks. (as it won't help their sales).
But I said to myself, I'm waiting till MWNY, if we get DDR (ie a new mobo, faster dual chips (not 66MHz faster, I mean FASTER) and a new case) I'm getting one. If Apple fails to deliver once more, I'll wait till MWSF 03. Until then they HAVE to deliver (so I'd get that), if they don't deliver there, they're dead, and I'm certainly not going to spend money on a dead product.
G-News
"I don't see in this rumor any reason to complain. ...It will be a good new to see 1,5 ghz G4 for MWNY with DDRAM, "<hr></blockquote>
I agree.
[quote]powerdoc
" It should be an incrase in performance of 60 to 70 % "<hr></blockquote>
I hope your right, where did you get 60-70%?
<strong>
What about the G3? It's called PPC750
And the 601, 603 and 604? Where they F0s?</strong><hr></blockquote>
Yep! you got it.