Do P4s stay cooler because they have a heatspreader?

[Sahakiel]

If I had to hazard a guess, I would say it's due to different architectures. Internally, both CPU's translate x86 ISA to be used with what's been called a "RISC-like" processor. Unless both design teams are twins, it's highly unlikely both implementations are anywhere near similar to each other. If I remember correctly, the Athlon also has more execution engines to begin with, which would mean a lot more die space and more silicon in use at any given time.

There's also the issue with familiarity with the .13u process. Intel may have been able to tweak their design to use less power on average, turning off transisters not in use, reducing current flow, etc. AMD adding an extra metal layer to TBreds indicates difficulty maintaining signal integrity, which isn't surprising for such an old design, but also shows they're running out of tweaks.

I'm sure I missed some points, or misreported, but this is what I think at the moment.

 

[AtomicDude512]

Originally posted by: Sahakiel
If I had to hazard a guess, I would say it's due to different architectures. Internally, both CPU's translate x86 ISA to be used with what's been called a "RISC-like" processor. Unless both design teams are twins, it's highly unlikely both implementations are anywhere near similar to each other. If I remember correctly, the Athlon also has more execution engines to begin with, which would mean a lot more die space and more silicon in use at any given time.

There's also the issue with familiarity with the .13u process. Intel may have been able to tweak their design to use less power on average, turning off transisters not in use, reducing current flow, etc. AMD adding an extra metal layer to TBreds indicates difficulty maintaining signal integrity, which isn't surprising for such an old design, but also shows they're running out of tweaks.

I'm sure I missed some points, or misreported, but this is what I think at the moment.

Good info, thank you.

 

[akapadia]

athlon64's will definitely be cooler with a strained silicion, soi .13micron process...though, the die will be signficantly larger than the xp's, it should still run much cooler.
However, the reason why athlon xp's are warmer than p4's is not completely because the of the die size or the heat spreader, but it is important to notice that the other than the 3.06 the p4's can't get even close to its theoretical maximum power above 100W (such as turning off transistors). Also, P4's have a large amount of cache, which runs much cooler than the fpu for example, compared to the xp's (other than barton). Hopefully on the athlon64's, the hypertransport controller won't get too got, because from what i've heard its not going to have too much cache (128L1 and 256L2?), since the athlon design doesnt benefit as much as intel's from increased cache. Also I'm not sure where the two companies currently place their thermal monitors, because obviously its a big difference...does anybody know where they are on the die - i assume near the ALU or something?