- Mar 27, 2009
- 12,968
- 221
- 106
I noticed the Oland and Litho/Mars 384sp GCN 1.0 GPUs are listed with different die sizes (90mm2 vs 77mm2). This even though the listed specs (Shading Units, TMUs, ROPs, PCIe bus, memory bus) are the same.
Oland (used on the desktop) is 90mm2 on 28nm.
Example:
(R7 250) https://www.techpowerup.com/gpudb/2459/radeon-r7-250
Litho and Mars (used on mobile) are 77mm2 on 28nm.
Examples:
(R7 M370) https://www.techpowerup.com/gpudb/2695/radeon-r7-m370
(HD 8750M) https://www.techpowerup.com/gpudb/1968/radeon-hd-8750m
Does anyone know why this is this?
Is one fabbed at Global Foundries while the other set is fabbed at TSMC (or vice versa)? Or is there a different reason for the die size difference?
Oland (used on the desktop) is 90mm2 on 28nm.
Example:
(R7 250) https://www.techpowerup.com/gpudb/2459/radeon-r7-250
Litho and Mars (used on mobile) are 77mm2 on 28nm.
Examples:
(R7 M370) https://www.techpowerup.com/gpudb/2695/radeon-r7-m370
(HD 8750M) https://www.techpowerup.com/gpudb/1968/radeon-hd-8750m
Does anyone know why this is this?
Is one fabbed at Global Foundries while the other set is fabbed at TSMC (or vice versa)? Or is there a different reason for the die size difference?
Last edited:
