We have an insight into what AMD has in store with its Barcelona platform and SSE 128. According to the included .pdf:
*SSE128 won't make scalar SSE code (and x87 code) faster than it is in K8.
*Matrix math will be 85% faster per core, other math intensive apps gain 10% to 50%.
Feel free to take a read yourself
Link
*SSE128 won't make scalar SSE code (and x87 code) faster than it is in K8.
*Matrix math will be 85% faster per core, other math intensive apps gain 10% to 50%.
Feel free to take a read yourself
Link