I call Fanboy! By the time Woodcrest machines appear - sometime in 2007, mark my words - the Opterons available then will soundly trounce Intel's best.
FTM, I doubt Apple will use Woodcrest processors at all. They'll be too expensive.
Fanboy? Onlooker's post was balanced and considered ("looking more promising", "possibly even in the lead"). It was hardly raving fanboism. If anything, your post was closer to fanboism than his.
What evidence do you have that Woodcrest won't appear until sometime 2007? It doesn't depend on a new process (it's 65 nm like Yonah), and it is based on the same core architecture as Yonah. There's little reason to suspect that it will be late.
Call me naive, but I really think that Apple, with it's access to detailed future CPU roadmaps, it's long history of building computers, their intimate knowledge of dealing with CPU suppliers, and the resources they have dedicated to making these sorts of decisions, will probably make a good decision. When I read someone complaining how stupid Apple was for choosing Intel over AMD I have to think that Apple probably had a good reason for choosing Intel.
Maybe pricing is a factor or predictability of supply or maybe Intel offers more bang per buck. I don't care as long as it's faster and just works.
I think I said something, a few months ago, about how Intel would inevitably integrate a proper vector processor more similar to Altivec. This is great news, since it means that going from the G5 to the Intel powermac won't make my 3D rendering & mathematical simulations run slower.
Many similar instructions to AltiVec, but different design. Still doesn't have FMADDs, and not sure if a FMADD is possible as fused macro-op. Doesn't have the permute instructions either.
But as everyone knows, one's mileage may vary.
Quote:
now, if only there were two vector cores per cpu core. . .
If you look at this diagram:
You see that there are 3 128-bit SSE units and 2 packed 128-bit FPUs, one and add and the other a divide. It can theoretically do 4 double precision floating ops per cycle. That's Itanium and Power4/5 territory, not that good in reality though. It will likely be the integer king though.
What evidence do you have that Woodcrest won't appear until sometime 2007? It doesn't depend on a new process (it's 65 nm like Yonah), and it is based on the same core architecture as Yonah. There's little reason to suspect that it will be late.
There isn't any evidence but one's "feelings."
Woodcrest is on the roadmap to be the first out of the gate sometime in early Q3, July or August. Conroe in late Q3 either August or September, and Merom in early Q4 in October or November.
If that happens, Apple can update the Pro desktop Mac in August with a "Quad" 3 GHz Woodcrest system which would probably, on average, be equivalent to a 3.5 GHz Power Mac G5 Quad system. They probably should save a speed grade for upgrades though.
There's always a chance that a delay could happen, but all signs are pointing to shipping in Q3. There's been working silicon for at least 6 months, Intel is letting media used the systems, albiet in a closed environment, and there doesn't need to be a hugely complicated new chipset that needs to be developed. Signs are good.
Once Conroe and Woodcrest are shipping AMDs unfettered dominance is over. Intel won't win every battle but they'll share the crown in different areas.
Umm, in 2H 06, AMD could see a competitive onslaught the likes the world has never seen before. Intel will be releasing so many incredible products it's going to make people's head spin. They may indeed win every market.
1. It is looking like AMD will not ship 65 nm processors in 2006, but early Q1 2007. That is their curse, for being late, and their saving grace because it will enable them to ship more competitive products. Tactics may win a battle, but logistics win wars. Fabs are logistics in this war. There is very little they can really do until their 65 nm fab starts shipping CPUs.
2. AMD will release a Turion X2 in May (?) to compete with Yonah, but it's highly doubtful they can release a top end version at more than 2 GHz and 35 Watt TDP because it will be at 90 nm, either that or AMD is willing to ship at very low volumes. Intel will have a 2.33 GHz Yonah which should be about 16% faster than a 2 GHz Turion X2. in Q2 and 2.33 GHz Merom in Q4 which should be about what 40% faster than a 2 GHz Turion X2. AMD will not be able to respond until the 65 nm fab starts shipping product.
3. AMD will release mainstream AM2 processors with up to DDR2-800 support and 2.8 GHz dual-core, maybe 3 GHz in June. These will be 100+ Watt TDP processors. From what we know so far, a 2.66 GHz Conroe will be competitive to that. Intel can ship 3 and 3.33 GHz Conroe XE chips. Intel is already slated to have 3 GHz Woodcrests CPUs. AMD will not be able to respond until the 65 nm fab starts shipping product.
The server front is where the real battle will happen:
4. AMD may be able to release a quad-core in Q4 on 90 nm. They also maybe able to release 4 MB L3 cache on dual and quad cores in Q4. DDR2-1066 may be a possibility as well. They'll need it.
Intel greatly alleviates it's 2 and 4 socket FSB problems with the Bensley platform: 2 independent FSB (dual independent bus (DIB) along the same lines as the PowerMac Quad), 2 to 4 channel FB-DIMM memory system (up to 17 GB/s bandwidth), and some needed cache snooping features.
Intel should be competitive with AMD 2 and 4 socket systems with Netburst-based 3.46 GHz Dempsey CPUs in Q2. In Q3 and Q4, Woodcrest hits the streets at up to 3 GHz. If the current benches hold, 2.66 GHz Woodcrest about equal to a 3 to 3.2 GHz Athlon 64 or Opteron. Don't think AMD can ship anything higher than 3 GHz on 90 nm, so, the L3 cache, DDR2-1066 and or quad core would be needed to compete.
But the most interesting chip could be the Netburst-based "Tulsa" chip with its 16 MB of on-chip L3 SRAM cache running at half CPU clock. A 4 socket Tulsa system (8 cores) would make a very interesting system.
There will also be a new Itanium. The dual-core Montecito Itanium will have a flabbergasting 24 MB of L3 SRAM cache running at CPU clock (1.9 GHz). Fabbed at 90 nm too!
On the low watt end of the scale, Sossaman (essentially Yonah with chipset support for 2 sockets) in Q2 at 31 Watt TDP. In Q4, ~35 TDP Woodcrest (essentially Merom with multi-socket support) CPUs.
The only game left in town where AMD maybe has a definite advantage is the 8+ socket x86 systems, and even there, it's more a function of the system builders architecture rather than the inherent FSB or memory design of the CPU.
To people that have that processor as an option. I prefer the idea them falling behind, and Apple having allied access to the best, and fastest processors made.
To people that have that processor as an option. I prefer the idea them falling behind, and Apple having allied access to the best, and fastest processors made.
really if you had your own self-interest at heart you would want AMD to do their best and come out with a faster processor. Intel would respond; you would be better off.
IS bragging about the company that makes your CPU so important, or healthy competition that lowers prices and raises speeds for everyone?
really if you had your own self-interest at heart you would want AMD to do their best and come out with a faster processor. Intel would respond; you would be better off.
IS bragging about the company that makes your CPU so important, or healthy competition that lowers prices and raises speeds for everyone?
Your interpretation wasn't actually my train of thought when I posted that. I was merely stating that I would be happy that Apple could finally compare Apples vs. PC standing on the same footing with the fastest processors available so there would be no bickering over the processor differences. I
Comments
Originally posted by cubist
I call Fanboy! By the time Woodcrest machines appear - sometime in 2007, mark my words - the Opterons available then will soundly trounce Intel's best.
FTM, I doubt Apple will use Woodcrest processors at all. They'll be too expensive.
Fanboy? Onlooker's post was balanced and considered ("looking more promising", "possibly even in the lead"). It was hardly raving fanboism. If anything, your post was closer to fanboism than his.
What evidence do you have that Woodcrest won't appear until sometime 2007? It doesn't depend on a new process (it's 65 nm like Yonah), and it is based on the same core architecture as Yonah. There's little reason to suspect that it will be late.
Maybe pricing is a factor or predictability of supply or maybe Intel offers more bang per buck. I don't care as long as it's faster and just works.
Woodcrest
5160, 3.00GHz, 4Mb, $850
5150, 2.66GHz, 4Mb, $700
5140, 2.33GHz, 4Mb, $470
5130, 2.00GHz, 4Mb, $330
5120, 1.86GHz, 4Mb, $270
5110, 1.60GHz, 4Mb, $230
Would I take a SMP Woodcrest 2Ghz system. Yup...I'd love that.
Once Conroe and Woodcrest are shipping AMDs unfettered dominance is over. Intel won't win every battle but they'll share the crown in different areas.
Originally posted by Splinemodel
I think I said something, a few months ago, about how Intel would inevitably integrate a proper vector processor more similar to Altivec. This is great news, since it means that going from the G5 to the Intel powermac won't make my 3D rendering & mathematical simulations run slower.
Many similar instructions to AltiVec, but different design. Still doesn't have FMADDs, and not sure if a FMADD is possible as fused macro-op. Doesn't have the permute instructions either.
But as everyone knows, one's mileage may vary.
now, if only there were two vector cores per cpu core. . .
If you look at this diagram:
You see that there are 3 128-bit SSE units and 2 packed 128-bit FPUs, one and add and the other a divide. It can theoretically do 4 double precision floating ops per cycle. That's Itanium and Power4/5 territory, not that good in reality though. It will likely be the integer king though.
Originally posted by Mr. H
What evidence do you have that Woodcrest won't appear until sometime 2007? It doesn't depend on a new process (it's 65 nm like Yonah), and it is based on the same core architecture as Yonah. There's little reason to suspect that it will be late.
There isn't any evidence but one's "feelings."
Woodcrest is on the roadmap to be the first out of the gate sometime in early Q3, July or August. Conroe in late Q3 either August or September, and Merom in early Q4 in October or November.
If that happens, Apple can update the Pro desktop Mac in August with a "Quad" 3 GHz Woodcrest system which would probably, on average, be equivalent to a 3.5 GHz Power Mac G5 Quad system. They probably should save a speed grade for upgrades though.
There's always a chance that a delay could happen, but all signs are pointing to shipping in Q3. There's been working silicon for at least 6 months, Intel is letting media used the systems, albiet in a closed environment, and there doesn't need to be a hugely complicated new chipset that needs to be developed. Signs are good.
Originally posted by hmurchison
Once Conroe and Woodcrest are shipping AMDs unfettered dominance is over. Intel won't win every battle but they'll share the crown in different areas.
Umm, in 2H 06, AMD could see a competitive onslaught the likes the world has never seen before. Intel will be releasing so many incredible products it's going to make people's head spin. They may indeed win every market.
1. It is looking like AMD will not ship 65 nm processors in 2006, but early Q1 2007. That is their curse, for being late, and their saving grace because it will enable them to ship more competitive products. Tactics may win a battle, but logistics win wars. Fabs are logistics in this war. There is very little they can really do until their 65 nm fab starts shipping CPUs.
2. AMD will release a Turion X2 in May (?) to compete with Yonah, but it's highly doubtful they can release a top end version at more than 2 GHz and 35 Watt TDP because it will be at 90 nm, either that or AMD is willing to ship at very low volumes. Intel will have a 2.33 GHz Yonah which should be about 16% faster than a 2 GHz Turion X2. in Q2 and 2.33 GHz Merom in Q4 which should be about what 40% faster than a 2 GHz Turion X2. AMD will not be able to respond until the 65 nm fab starts shipping product.
3. AMD will release mainstream AM2 processors with up to DDR2-800 support and 2.8 GHz dual-core, maybe 3 GHz in June. These will be 100+ Watt TDP processors. From what we know so far, a 2.66 GHz Conroe will be competitive to that. Intel can ship 3 and 3.33 GHz Conroe XE chips. Intel is already slated to have 3 GHz Woodcrests CPUs. AMD will not be able to respond until the 65 nm fab starts shipping product.
The server front is where the real battle will happen:
4. AMD may be able to release a quad-core in Q4 on 90 nm. They also maybe able to release 4 MB L3 cache on dual and quad cores in Q4. DDR2-1066 may be a possibility as well. They'll need it.
Intel greatly alleviates it's 2 and 4 socket FSB problems with the Bensley platform: 2 independent FSB (dual independent bus (DIB) along the same lines as the PowerMac Quad), 2 to 4 channel FB-DIMM memory system (up to 17 GB/s bandwidth), and some needed cache snooping features.
Intel should be competitive with AMD 2 and 4 socket systems with Netburst-based 3.46 GHz Dempsey CPUs in Q2. In Q3 and Q4, Woodcrest hits the streets at up to 3 GHz. If the current benches hold, 2.66 GHz Woodcrest about equal to a 3 to 3.2 GHz Athlon 64 or Opteron. Don't think AMD can ship anything higher than 3 GHz on 90 nm, so, the L3 cache, DDR2-1066 and or quad core would be needed to compete.
But the most interesting chip could be the Netburst-based "Tulsa" chip with its 16 MB of on-chip L3 SRAM cache running at half CPU clock. A 4 socket Tulsa system (8 cores) would make a very interesting system.
There will also be a new Itanium. The dual-core Montecito Itanium will have a flabbergasting 24 MB of L3 SRAM cache running at CPU clock (1.9 GHz). Fabbed at 90 nm too!
On the low watt end of the scale, Sossaman (essentially Yonah with chipset support for 2 sockets) in Q2 at 31 Watt TDP. In Q4, ~35 TDP Woodcrest (essentially Merom with multi-socket support) CPUs.
The only game left in town where AMD maybe has a definite advantage is the 8+ socket x86 systems, and even there, it's more a function of the system builders architecture rather than the inherent FSB or memory design of the CPU.
AMD's 65 nm fab really can't come soon enough.
Originally posted by THT
AMD's 65 nm fab really can't come soon enough.
To people that have that processor as an option. I prefer the idea them falling behind, and Apple having allied access to the best, and fastest processors made.
Originally posted by onlooker
To people that have that processor as an option. I prefer the idea them falling behind, and Apple having allied access to the best, and fastest processors made.
really if you had your own self-interest at heart you would want AMD to do their best and come out with a faster processor. Intel would respond; you would be better off.
IS bragging about the company that makes your CPU so important, or healthy competition that lowers prices and raises speeds for everyone?
Originally posted by wwwork
really if you had your own self-interest at heart you would want AMD to do their best and come out with a faster processor. Intel would respond; you would be better off.
IS bragging about the company that makes your CPU so important, or healthy competition that lowers prices and raises speeds for everyone?
Your interpretation wasn't actually my train of thought when I posted that. I was merely stating that I would be happy that Apple could finally compare Apples vs. PC standing on the same footing with the fastest processors available so there would be no bickering over the processor differences. I
then I was mistaken. Sorry about that.