Inter-core cooperative TLB prefetchers
A chip multiprocessor includes a plurality of cores each having a translation lookaside buffer (TLB) and a prefetch buffer (PB). Each core is configured to determine a TLB miss on the core's TLB for a virtual page address and determine whether or not there is a PB hit on a PB entry in the PB for the virtual page address. If it is determined that there is a PB hit, the PB entry is added to the TLB. If it is determined that there is not a PB hit, the virtual page address is used to perform a page walk to determine a translation entry, the translation entry is added to the TLB and the translation entry is prefetched to each other one of the plurality of cores.
1. A chip multiprocessor comprising:
at least one prefetch buffer (PB); and
a plurality of cores each communicatively coupled to the at least one prefetch buffer and comprising a translation lookaside buffer (TLB), wherein each of the plurality of cores is configured to:
determine a TLB miss on the TLB for a virtual page address;
determine whether or not there is a PB hit on a PB entry in the at least one PB for the virtual page address;
add the PB entry to the TLB if it is determined that there is a PB hit; and
use the virtual page address to perform a page walk to determine a translation entry, add the translation entry to the TLB and prefetch the translation entry to each other one of the plurality of cores if it is determined that there is not a PB hit.
2. The chip multiprocessor of claim 1 wherein each of the plurality of cores comprises one of the at least one prefetch buffer.
3. The chip multiprocessor of claim 1 wherein each of the plurality of cores is further configured to:
receive a translation entry prefetched from another of the plurality of cores; and
add the received translation entry to a PB of the core.
4. The chip multiprocessor of claim 1 wherein the chip multiprocessor is selected from the group comprising at least one of a dual-core multiprocessor and a quad-core multiprocessor.
5. A method for performing Leader-Follower prefetching comprising:
determining a TLB miss on a TLB for a virtual page address;
determining whether or not there is a PB hit on a PB entry in a PB for the virtual page address;
adding the PB entry to the TLB if it is determined that there is a PB hit; and
using the virtual page address to perform a page walk to determine a translation entry, adding the translation entry to the TLB and prefetching the translation entry to each of one or more cores if it is determined that there is not a PB hit.
6. The method of claim 5 further comprising:
receiving a translation entry prefetched from another one of the plurality of cores; and
adding the received translation entry to the PB.