- Mar 24, 2005
- 135
- 0
- 0
My real question is 'does transistor count scale linearly with respect to die size?'
For instance: if I had a 100mm^2 die area on 130 nm process with n number of transistors, I could have Xn the number of transistors with the same 100mm^2 die area on a 90 nm process?
I understand that scaling like this isn't exact, but there seems to be a general way to figure it out.
The reason why I ask is because of a thread from the Video portion of the forums.
NVIDIA's NV40 GPU is said to contain roughly 223 million transistors and have roughly a 300mm^2 die size on, I believe, IBM's 130 nm process.
NVIDIA's G70 GPU is said to contain roughly 304 million transistors and have roughly a 300mm^2 die size on TSMC's 110 nm process.
If NVIDIA were to use a 90 nm process, how many more transistors could NVIDIA add to their GPU assuming the same 300mm^2 die size?
Thanks in advance.
For instance: if I had a 100mm^2 die area on 130 nm process with n number of transistors, I could have Xn the number of transistors with the same 100mm^2 die area on a 90 nm process?
I understand that scaling like this isn't exact, but there seems to be a general way to figure it out.
The reason why I ask is because of a thread from the Video portion of the forums.
NVIDIA's NV40 GPU is said to contain roughly 223 million transistors and have roughly a 300mm^2 die size on, I believe, IBM's 130 nm process.
NVIDIA's G70 GPU is said to contain roughly 304 million transistors and have roughly a 300mm^2 die size on TSMC's 110 nm process.
If NVIDIA were to use a 90 nm process, how many more transistors could NVIDIA add to their GPU assuming the same 300mm^2 die size?
Thanks in advance.