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

Update README.md

Browse Source
This commit is contained in:
moebiusband73
2020-10-12 07:53:07 +02:00
committed by GitHub
parent 064453b9f9
commit f88b557482
1 changed files with 3 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@@ -37,7 +37,7 @@ TAG ?= GCC
ENABLE_OPENMP ?= false
ENABLE_LIKWID ?= false
OPTIONS = -DSIZE=40000000ull
OPTIONS = -DSIZE=120000000ull
OPTIONS += -DNTIMES=10
OPTIONS += -DARRAY_ALIGNMENT=64
#OPTIONS += -DVERBOSE_AFFINITY
@@ -47,6 +47,8 @@ OPTIONS += -DARRAY_ALIGNMENT=64
The verbosity options enable detailed output about affinity settings, allocation sizes and timer resolution.
*Notice:* OpenMP involves significant overhead through barrier cost, especially on systems with many memory domains. The default problem size is set to almost 4GB to have enough work vs overhead. If you suspect that the result should be better you may try to further increase the problem size. To compare to original stream results on X86 systems you have to ensure that streaming store instructions are used. For the ICC toolchain this is now the default (Option `-qopt-streaming-stores=always`.
2. Build with:
```
make