The biggest problem is they have to re-write not because it's slower. Plus if VMX was so great why didn't IBM include it in there Power-5 CPU.
IBM's position was always too negative with respect to technologies like Altivec. Remember, it was IBM that damaged the AIM alliance with its disagreement on Altivec development. With PPC970 they finally accepted to use it but it was too late.
Quote:
The benchmark test your probly basing your info on is RC5. Distributed.net FAQ warns that the RC5 benchmark is a poor thing to use to characterize the performance of a CPU. It relies heavily on rotate instructions, which not all processors implement equally well (they're not used that often). Altivec has a vector permute unit that has a 128-bit vector rotate instruction. SSE does not have such an instruction. That's why the G5 performs so much better in that benchmark. It's a nice thing if you're software is really dependent on bit manipulation (crypto, stuff like that), but what most apps use the vector unit for is FMACs (multiply-accumulate instructions), and AltiVec and SSE do those equally fast.
It is true that RC5 is a highly biased test and should not be used without a clear comparison context. However, it is a common knowledge between SIMD programmers that SSE2 is roughly 2 times slower than Altivec at the same clock speed. I don't know though how Altivec compares to SSE3.
So, while Pentium M processors are great for general use, their SIMD capabilities leave much to be desired. If there is something that saves the day, this is their highest clock speed (2.1 GHz for P-M vs. 1.67 for G4).
All I know is that Yonah will be widely available around march 2006 as a dual core with speed up to 2.16GHz, and a 667MHz FSB.
I believe it will be an UPGRADE from any G4 based PowerBook, which we know won't be faster that 1.8GHz ( slightly overclocked 7448 ) at best.
If Apple has (had) trouble designing the Intel PowerBook (who said it was gonna be easy?) they may use Freescale's 7448 in a last update during this fall/winter.
I'm sure that the first Intel based Macs will be good machines, maybe not flawless, but as good as any PPC based.
I also believe that we may see this chip (yonah) across the iBook and mini lines a few month after the release of the 1st Intel based Mac (if it's the PowerBook), maybe in a single core format though.
That's because frame rate is a much bigger function of GPU power than of CPU power these days and there is really nothing in Halo that needs 64 bits.
If you take something like Gaussian (quantium chemistry computational system), using dual core 64 bit processor more than double the speed of calculations in a lot of cases compared to their 32 bit simgle core counterparts.
Video encoding such as DIVX or XVID or H.264 speed effectively doubles by using dualcore.
Halo was just an example, however, could you point me to the benchmarks which show that, I'd be interested in looking at them.
Yonah's fine for an iBook or MacMini, however the 667Mhz bus of Intels upcomming Yonah is half as wide as AMD's EXISTING mobile cpu architecture. Remember when the PowerBook used to be the most powerfull laptop on the market? Now it's simply going to be on PAR with the average PC notebook, in terms of performance.
Yonah's fine for an iBook or MacMini, however the 667Mhz bus of Intels upcomming Yonah is half as wide as AMD's EXISTING mobile cpu architecture. Remember when the PowerBook used to be the most powerfull laptop on the market? Now it's simply going to be on PAR with the average PC notebook, in terms of performance.
You cant compare FSB and thinga across different processes and architechtures - that is simply no basis for comparison - find yourself some benchmarks, then come whining - not to me, but to apple.
Yonah's fine for an iBook or MacMini, however the 667Mhz bus of Intels upcomming Yonah is half as wide as AMD's EXISTING mobile cpu architecture. Remember when the PowerBook used to be the most powerfull laptop on the market? Now it's simply going to be on PAR with the average PC notebook, in terms of performance
Do you have any technical data , what so ever, to suggest that 667 MHz bus will limit performance of Yonah?
BTW, I just got a 1.73 GHz PM (Dothan) (533 MHz FSB, dual DDRII) Fujitsu notebook and it cleanly outruns a 3.2 GHz Northwood 800 FSB HT Dell desktop in terms of video encoding, Matlab, and other such CPU intensive tasks. Just to give you an idea of current state of Pentium M power.
Comments
Originally posted by Relic
The biggest problem is they have to re-write not because it's slower. Plus if VMX was so great why didn't IBM include it in there Power-5 CPU.
IBM's position was always too negative with respect to technologies like Altivec. Remember, it was IBM that damaged the AIM alliance with its disagreement on Altivec development. With PPC970 they finally accepted to use it but it was too late.
The benchmark test your probly basing your info on is RC5. Distributed.net FAQ warns that the RC5 benchmark is a poor thing to use to characterize the performance of a CPU. It relies heavily on rotate instructions, which not all processors implement equally well (they're not used that often). Altivec has a vector permute unit that has a 128-bit vector rotate instruction. SSE does not have such an instruction. That's why the G5 performs so much better in that benchmark. It's a nice thing if you're software is really dependent on bit manipulation (crypto, stuff like that), but what most apps use the vector unit for is FMACs (multiply-accumulate instructions), and AltiVec and SSE do those equally fast.
It is true that RC5 is a highly biased test and should not be used without a clear comparison context. However, it is a common knowledge between SIMD programmers that SSE2 is roughly 2 times slower than Altivec at the same clock speed. I don't know though how Altivec compares to SSE3.
So, while Pentium M processors are great for general use, their SIMD capabilities leave much to be desired. If there is something that saves the day, this is their highest clock speed (2.1 GHz for P-M vs. 1.67 for G4).
I believe it will be an UPGRADE from any G4 based PowerBook, which we know won't be faster that 1.8GHz ( slightly overclocked 7448 ) at best.
If Apple has (had) trouble designing the Intel PowerBook (who said it was gonna be easy?) they may use Freescale's 7448 in a last update during this fall/winter.
I'm sure that the first Intel based Macs will be good machines, maybe not flawless, but as good as any PPC based.
I also believe that we may see this chip (yonah) across the iBook and mini lines a few month after the release of the 1st Intel based Mac (if it's the PowerBook), maybe in a single core format though.
Originally posted by skatman
That's because frame rate is a much bigger function of GPU power than of CPU power these days and there is really nothing in Halo that needs 64 bits.
If you take something like Gaussian (quantium chemistry computational system), using dual core 64 bit processor more than double the speed of calculations in a lot of cases compared to their 32 bit simgle core counterparts.
Video encoding such as DIVX or XVID or H.264 speed effectively doubles by using dualcore.
Halo was just an example, however, could you point me to the benchmarks which show that, I'd be interested in looking at them.
Originally posted by ngmapple
Yonah's fine for an iBook or MacMini, however the 667Mhz bus of Intels upcomming Yonah is half as wide as AMD's EXISTING mobile cpu architecture. Remember when the PowerBook used to be the most powerfull laptop on the market? Now it's simply going to be on PAR with the average PC notebook, in terms of performance.
You cant compare FSB and thinga across different processes and architechtures - that is simply no basis for comparison - find yourself some benchmarks, then come whining - not to me, but to apple.
Yonah's fine for an iBook or MacMini, however the 667Mhz bus of Intels upcomming Yonah is half as wide as AMD's EXISTING mobile cpu architecture. Remember when the PowerBook used to be the most powerfull laptop on the market? Now it's simply going to be on PAR with the average PC notebook, in terms of performance
Do you have any technical data , what so ever, to suggest that 667 MHz bus will limit performance of Yonah?
BTW, I just got a 1.73 GHz PM (Dothan) (533 MHz FSB, dual DDRII) Fujitsu notebook and it cleanly outruns a 3.2 GHz Northwood 800 FSB HT Dell desktop in terms of video encoding, Matlab, and other such CPU intensive tasks. Just to give you an idea of current state of Pentium M power.