summaryrefslogtreecommitdiffstats
path: root/.gitignore
diff options
context:
space:
mode:
authorSimon Rettberg2020-08-05 11:37:55 +0200
committerSimon Rettberg2020-08-05 11:37:55 +0200
commitd80110c71f15abce3a4d7d87550b05a0baf35120 (patch)
tree6ebf3f3cdf409eb6440ebadc3943fbb87062277a /.gitignore
parentChange loop%d strings to xloop%d (diff)
downloadxloop-d80110c71f15abce3a4d7d87550b05a0baf35120.tar.gz
xloop-d80110c71f15abce3a4d7d87550b05a0baf35120.tar.xz
xloop-d80110c71f15abce3a4d7d87550b05a0baf35120.zip
Cache last decompressed cluster for compressed QCOW
The naive implementation repeatedly decompressed the same cluster for every 4k read, which was incredibly wasteful and resulted in ~20MB/s read speeds if the underlying qcow2 was read via GBit network. This approach yields 200MB/s+
Diffstat (limited to '.gitignore')
0 files changed, 0 insertions, 0 deletions