The Go runner does not have a problem with supporting parallel
requests for most multimodal models. Now that we won't be potentially
falling back to server.cpp, this restriction can be lifted.
However, the new mllama model can't support parallel requests, so we
will need to keep a restriction for that.
* Fix embeddings memory corruption
The patch was leading to a buffer overrun corruption. Once removed though, parallism
in server.cpp lead to hitting an assert due to slot/seq IDs being >= token count. To
work around this, only use slot 0 for embeddings.
* Fix embed integration test assumption
The token eval count has changed with recent llama.cpp bumps (0.3.5+)
In mult-brand GPU setups, if we couldn't fully load the model we
would fall through the scheduler and mistakenly try to load across
a mix of brands. This makes sure we find the set of GPU(s) that
best fit for the partial load.
This change fixes the handling of keep_alive so that if client
request omits the setting, we only set this on initial load. Once
the model is loaded, if new requests leave this unset, we'll keep
whatever keep_alive was there.
Users may not realize the siny new model they're trying to load
fits on their disk, but can't load into system+GPU memory. Today
we crash, but with this fix, we'll give them a better error message
before even trying to load it.
Previously, some costly things were causing the loading of GGUF files
and their metadata and tensor information to be VERY slow:
* Too many allocations when decoding strings
* Hitting disk for each read of each key and value, resulting in a
not-okay amount of syscalls/disk I/O.
The show API is now down to 33ms from 800ms+ for llama3 on a macbook pro
m3.
This commit also prevents collecting large arrays of values when
decoding GGUFs (if desired). When such keys are encountered, their
values are null, and are encoded as such in JSON.
Also, this fixes a broken test that was not encoding valid GGUF.
Until ROCm v6.2 ships, we wont be able to get accurate free memory
reporting on windows, which makes automatic concurrency too risky.
Users can still opt-in but will need to pay attention to model sizes otherwise they may thrash/page VRAM or cause OOM crashes.
All other platforms and GPUs have accurate VRAM reporting wired
up now, so we can turn on concurrency by default.
This adjusts our default settings to enable multiple models and parallel
requests to a single model. Users can still override these by the same
env var settings as before. Parallel has a direct impact on
num_ctx, which in turn can have a significant impact on small VRAM GPUs
so this change also refines the algorithm so that when parallel is not
explicitly set by the user, we try to find a reasonable default that fits
the model on their GPU(s). As before, multiple models will only load
concurrently if they fully fit in VRAM.
While models are loading, the VRAM metrics are dynamic, so try
to load on a GPU that doesn't have a model actively loading, or wait
to avoid races that lead to OOMs
Our default behavior today is to try to fit into a single GPU if possible.
Some users would prefer the old behavior of always spreading across
multiple GPUs even if the model can fit into one. This exposes that
tunable behavior.
The GPU drivers take a while to update their free memory reporting, so we need
to wait until the values converge with what we're expecting before proceeding
to start another runner in order to get an accurate picture.
This moves all the env var reading into one central module
and logs the loaded config once at startup which should
help in troubleshooting user server logs
Prior refactoring passes accidentally removed the logic to bypass VRAM
checks for CPU loads. This adds that back, along with test coverage.
This also fixes loaded map access in the unit test to be behind the mutex which was
likely the cause of various flakes in the tests.
