Repository navigation
Anomaly Mirror
--use-anomaly-mirror makes the CLI download Stalker Anomaly 1.5.3 from the faith-beam.com mirror instead of ModDB. The mirror hosts a repack that extracts about 9× faster (about 20 s instead of about 3 min). The repack is the same game files, compressed with multithreaded LZMA2.
stalker-gamma full-install --use-anomaly-mirror
stalker-gamma anomaly install --use-anomaly-mirrorThe switch is off by default.
| Default (ModDB) | --use-anomaly-mirror |
|
|---|---|---|
| Source | https://www.moddb.com/downloads/start/277404 |
https://files.faith-beam.com/anomaly/1.5.3/Anomaly-1.5.3-Full.mt-mx5.7z |
| Archive | Anomaly-1.5.3-Full.2.7z |
Anomaly-1.5.3-Full.mt-mx5.7z |
| MD5 | d6bce51a4e6d98f9610ef0aa967ba964 |
123a0566485a77f8d44a659630cce856 |
With the switch, the download works like this:
- The mirror archive downloads to
<cache>/Anomaly-1.5.3-Full.mt-mx5.7z.part. If the connection drops, the download resumes, with up to 3 attempts. - When the download finishes, its MD5 is checked. If it matches, the file is renamed to
Anomaly-1.5.3-Full.mt-mx5.7z. - If the mirror can't be reached, keeps failing, or serves a file with the wrong hash, the partial file is deleted and the CLI falls back to ModDB. A cancelled download keeps its
.partfile so the next run can resume it.
-
A valid ModDB archive in your cache wins. If
Anomaly-1.5.3-Full.2.7zis already in your cache and its hash matches, the CLI uses it and doesn't download the mirror, even with the switch. To get the faster archive, delete the cached.2.7zfirst. -
A cached mirror archive is always used. Once a valid
mt-mx5.7zis in your cache, the CLI extracts it on every later run, with or without the switch. - It needs more RAM. Extracting the mirror archive takes about 4.8 GB of memory, compared with about 370 MB for the ModDB archive (see below).
-
cachepruning never deletes either Anomaly archive.
Both archives were extracted with the same 7-Zip command the CLI runs (7zz x -y <archive> -o<dir>). Each was extracted twice.
ModDB .2.7z
|
Mirror mt-mx5.7z
|
|
|---|---|---|
| Archive size | 9.74 GB | 9.88 GB (+1.5%) |
| Unpacked size | 17.36 GB (95 files) | 17.36 GB (95 files) |
| Compression | LZMA2, 64 MB dictionary, BCJ2 filter | LZMA2, 32 MB dictionary, BCJ filter, multithreaded |
| Time to extract | 189.7 s / 189.4 s | 20.5 s / 20.2 s |
| CPU time (user + sys) | ~189 s | ~231 s |
| Average cores busy | ~1.0 | ~11.4 of 18 |
| Speed (output written) | ~92 MB/s | ~860 MB/s |
| Peak memory (RSS) | ~370 MB | ~4.7–4.8 GB |
| Peak memory footprint | 369 MB | 4.81 GB |
Both archives unpack to the same 17,359,866,666 bytes.
Why they differ: the ModDB archive is one solid LZMA2 stream, which can only be decoded in order, so 7-Zip runs on a single core. The mirror repack was compressed with -mmt=on, which splits the stream into chunks that 7-Zip can decode on every core at once. Each thread holds its own buffers, which is why memory use is higher. The extra chunk boundaries cost about 1.5% of archive size.
The repack is equivalent to running this from inside an extracted Anomaly folder:
7zz a -t7z -mx5 -mmt=on ../Anomaly-1.5.3-Full.mt-mx5.7z ./*- Machine: Apple M5 Max, 18 cores, 48 GB RAM, macOS
-
Tool: 7-Zip (
7zz) 26.05, default thread settings -
Method: each archive extracted twice, alternating between them. Metrics are from
/usr/bin/time -l. - Caveat: both archives were already in the OS file cache, so the times cover decompression and writing only. On a slow drive, reading the archive adds time. Machines with fewer cores get less of a speedup.
Peak RSS is the most physical RAM the process used at any one point. Peak memory footprint is the figure macOS charges to the process (what Activity Monitor shows), which also counts memory that was compressed or swapped out.