Sign in with
Sign up | Sign in
Your question

Question about die shrink

Last response: in CPUs
Share
September 27, 2009 4:00:34 AM

ok recently i heard that intel is going to release a 32nm chip which is going to be used instead of the 45nm, but how does a smaller die effect the speed? and how does a smaller die produce less heat can someone help me out.

More about : question die shrink

a b à CPUs
September 27, 2009 6:00:15 AM

First, 32nm is not the die size. It is related, but die size also has to do with the number of transistors. 32nm is the size of a single transistor. Smaller transistors can typically switch faster using less voltage, allowing an identical architecture CPU to run faster with less power. Often though, the manufacturer will make a similar size die on the smaller process, but using more transistors. This allows for more features and more cores while staying in roughly the same power envelope.
m
0
l
September 27, 2009 6:40:51 AM

o so the 32nm is the transistors size not the cpu die, now that makes sense, shrinking the transistors so they can add more in the same chip.

so smaller transistors use less volts and power which is why 32 run cooler than 45, and they switch faster, which is why a core 2 duo 32 is faster than a core 2 duo 45 clock per clock

is what im saying correct or im not making any sense?
m
0
l
Related resources
a b à CPUs
September 27, 2009 3:04:05 PM

Quote:
core 2 duo 32

There are NO C2D/C2Q/i7/i5 32nm CPUs yet.

Quote:
so smaller transistors use less volts and power which is why 32 run cooler than 45

Yes.

Quote:
they switch faster, which is why a core 2 duo 32 is faster than a core 2 duo 45 clock per cloc

Yes, they switch faster, but architecture (as in Nehalem, Conroe,etc) also changes between the shrinks.
m
0
l
September 27, 2009 7:49:28 PM

in the last one about the architecture, your talking about how they moved from p4 netbrust to core 2 then to nahlem right? and how they changed cpu's stages for reading and writing.

so for example the p4 used more stages but with less work per clock, core 2 used less stages with more work per clock and reading in parallel etc.. right?
m
0
l
a b à CPUs
September 27, 2009 8:33:12 PM

Quote:
in the last one about the architecture, your talking about how they moved from p4 netbrust to core 2 then to nahlem right?

Not only that, but the move from Conroe(65nm) to the Wolfdales(45nm),etc. This is not a massive change, but there is a change.
m
0
l
a b à CPUs
September 28, 2009 6:59:20 AM

Not to mention new instruction sets!
m
0
l
September 28, 2009 7:29:55 AM

intel goes by tick tock right? tick is when they shrink the transistors size so they can add more in the same chip to gain more performance at lower volts/power, and tock for a new processor microarchitecture that is going from Pentium 4 to Core 2 to Core I serious.

m
0
l
a b à CPUs
September 28, 2009 10:26:02 AM

Tick Tock = New die shrink, New Arch, New Shrink etc.
m
0
l
!