derek/gem5
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

mem-cache: Optimize strided prefetcher address generation

Browse Source 
This commit optimizes the address generation logic in the strided
prefetcher by introducing the following changes

(d is the degree of the prefetcher)

* Evaluate the fixed prefetch_stride only once (and not d-times)
* Replace 2d multiplications (d * prefetch_stride and distance *
prefetch_stride) with additions by updating the new base prefetch
address while looping

Change-Id: I3ec0c642bc9ec7635b0d38308797e99b645304bb
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
This commit is contained in:
Giacomo Travaglini
2023-11-09 16:57:01 +00:00
committed by Richard Cooper
parent a5fece3b91
commit 1d5be8d9e5
1 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/mem/cache/prefetch/stride.cc vendored
View File

@@ -1,6 +1,6 @@
/*
* Copyright (c) 2018 Inria
* Copyright (c) 2012-2013, 2015, 2022 Arm Limited
* Copyright (c) 2012-2013, 2015, 2022-2023 Arm Limited
* All rights reserved
*
* The license below extends only to copyright in the software and shall
@@ -169,16 +169,16 @@ Stride::calculatePrefetch(const PrefetchInfo &pfi,
return;
}
// Round strides up to atleast 1 cacheline
int prefetch_stride = new_stride;
if (abs(new_stride) < blkSize) {
prefetch_stride = (new_stride < 0) ? -blkSize : blkSize;
}
Addr new_addr = pf_addr + distance * prefetch_stride;
// Generate up to degree prefetches
for (int d = 1; d <= degree; d++) {
// Round strides up to atleast 1 cacheline
int prefetch_stride = new_stride;
if (abs(new_stride) < blkSize) {
prefetch_stride = (new_stride < 0) ? -blkSize : blkSize;
}
Addr new_addr = pf_addr + distance * prefetch_stride
+ d * prefetch_stride;
new_addr += prefetch_stride;
addresses.push_back(AddrPriority(new_addr, 0));
}
} else {