vickybat
I am the night...I am...
Its because gpu's are good in heavily parallel processes. Hashing sees individual sp's in a gpu as compute units and thus divides the task among them and utilizes their compute power.
VLIW architecture of amd gpu's has an upper hand in hashing and all gpu's are well ahead than the latest cpu's. See it this way:
Lets say a core i5 2500 has 4 physical cores having a fixed set of execution units and radeon 5750 has 700 streaming multiprocessors. You see almost a 40 times difference between their scores and its because each set of sp's in 5750 are doing the same thing as a single core in 2500. So the sheer performance difference due to the huge sp count and performing the task in a parallel manner.
Hope this explanation holds true. Others can definitely mention their points and correct any wrong info.
VLIW architecture of amd gpu's has an upper hand in hashing and all gpu's are well ahead than the latest cpu's. See it this way:
Lets say a core i5 2500 has 4 physical cores having a fixed set of execution units and radeon 5750 has 700 streaming multiprocessors. You see almost a 40 times difference between their scores and its because each set of sp's in 5750 are doing the same thing as a single core in 2500. So the sheer performance difference due to the huge sp count and performing the task in a parallel manner.
Hope this explanation holds true. Others can definitely mention their points and correct any wrong info.