Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Differential expression performance improvements #234

Merged
merged 26 commits into from
May 12, 2022
Merged

Differential expression performance improvements #234

Show file tree
Hide file tree
Changes from 21 commits
Commits
Show all changes
26 commits
Select commit Hold shift + click to select a range
30e6250
Diffex perf
blrnw3 Feb 18, 2022
f169ab2
Diffex algo tweaks
blrnw3 Mar 10, 2022
6ef1a82
algo work and shift be gone
blrnw3 Mar 11, 2022
06a92b6
revert xarray change
blrnw3 Mar 11, 2022
0189df6
unrevert fine change
blrnw3 Mar 12, 2022
ae583f1
tdb bug
blrnw3 Mar 12, 2022
cf01a6e
merge with main
Mar 28, 2022
104e513
remove debugging print
Mar 28, 2022
ea76455
add note about tiledb workaround
Mar 28, 2022
462b01f
remove obsolete configuration hooks for diffexp parallelism
Mar 28, 2022
a414895
lint
Mar 28, 2022
05b9e1e
replace deprecated np.where use with recommended nonzero()
Mar 28, 2022
750eee4
simplify X slicing
Mar 28, 2022
c0f4f46
Merge branch 'main' into brodgers/diffex
Mar 28, 2022
2074d08
Merge branch 'main' into brodgers/diffex
Apr 4, 2022
191edfd
remove tiledb 0.13.1 work-arounds
Apr 4, 2022
e1e47ba
revert tiledb.Ctx handling to correctly use configuration subsystem
Apr 4, 2022
075c00c
clean up tiledb config params
Apr 4, 2022
63f5db5
smaller size for default python tiledb query buffer
Apr 4, 2022
f294595
try even smaller default python buffer
Apr 4, 2022
002120b
1GB default tiledb python buffer
Apr 4, 2022
b395933
Add high performance diffex routes (#244)
May 2, 2022
ab9de54
Merge branch 'main' into brodgers/diffex
ebezzi May 2, 2022
46b37ce
Merge branch 'main' into brodgers/diffex
ebezzi May 12, 2022
3b86eae
Infra changes
ebezzi May 12, 2022
00e9930
Replace modules with symlinks
ebezzi May 12, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 2 additions & 13 deletions hosted/create_rdev_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,17 +47,6 @@
# A list of allowed matrix types. If an empty list, then all matrix types are allowed
allowed_matrix_types: []

diffexp:
alg_cxg:
# The number of threads to use is computed from: min(max_workers, cpu_multipler * cpu_count).
# Where cpu_count is determined at runtime.
max_workers: 64
cpu_multiplier: 4

# The target number of matrix elements that are evaluated
# together in one thread.
target_workunit: 16_000_000

data_locator:
api_base: http://{env}-backend.internal.rdev.single-cell.czi.technology/dp/v1
s3:
Expand All @@ -74,11 +63,11 @@
# data_locator / s3 / region_name.
tiledb_ctx:
sm.tile_cache_size: 60129542144 # 56 GB
sm.num_reader_threads: 32
py.init_buffer_bytes: 17179869184 # 16GiB

limits:
column_request_max: 32
diffexp_cellcount_max: 50000
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the setting for prod is in single-cell-infra, PR incoming

diffexp_cellcount_max: 1500000

dataset:
app:
Expand Down
2 changes: 1 addition & 1 deletion hosted/requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ umap-learn==0.4.6
sentry-sdk[flask]==0.14.3
six==1.14.0
sqlalchemy==1.3.18
tiledb==0.13.1
tiledb==0.13.2
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Stick with 0.13.2 or should we try 0.14.x now? @MDunitz has already found issues with 0.14.x, so maybe hold off if 0.13.2 is well-tested?

Either way, we need to keep Explorer at-or-ahead of Portal's TileDB version. And we should keep this version in sync with server/requirements.txt (in this repo).

urllib3==1.26.5
Werkzeug==1.0.1
zipp==3.1.0
163 changes: 0 additions & 163 deletions server/common/compute/diffexp_generic.py
View file
Open in desktop

This file was deleted.

17 changes: 0 additions & 17 deletions server/common/config/server_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,6 @@
from server.common.errors import ConfigurationError, DatasetAccessError
from server.common.utils.data_locator import discover_s3_region_name
from server.common.utils.utils import is_port_available, find_available_port, custom_format_warning
from server.compute import diffexp_cxg as diffexp_tiledb
from server.dataset.matrix_loader import MatrixDataLoader, MatrixDataType


Expand Down Expand Up @@ -48,10 +47,6 @@ def __init__(self, app_config, default_config):
self.single_dataset__about = default_config["single_dataset"]["about"]
self.single_dataset__title = default_config["single_dataset"]["title"]

self.diffexp__alg_cxg__max_workers = default_config["diffexp"]["alg_cxg"]["max_workers"]
self.diffexp__alg_cxg__cpu_multiplier = default_config["diffexp"]["alg_cxg"]["cpu_multiplier"]
self.diffexp__alg_cxg__target_workunit = default_config["diffexp"]["alg_cxg"]["target_workunit"]

self.data_locator__s3__region_name = default_config["data_locator"]["s3"]["region_name"]
self.data_locator__api_base = default_config["data_locator"]["api_base"]
self.adaptor__cxg_adaptor__tiledb_ctx = default_config["adaptor"]["cxg_adaptor"]["tiledb_ctx"]
Expand All @@ -69,7 +64,6 @@ def complete_config(self, context):
self.handle_adaptor() # may depend on data_locator
self.handle_single_dataset(context) # may depend on adaptor
self.handle_multi_dataset() # may depend on adaptor
self.handle_diffexp()
self.handle_limits()

self.check_config()
Expand Down Expand Up @@ -255,17 +249,6 @@ def handle_multi_dataset(self):
except ValueError:
raise ConfigurationError(f'Invalid matrix type in "allowed_matrix_types": {mtype}')

def handle_diffexp(self):
self.validate_correct_type_of_configuration_attribute("diffexp__alg_cxg__max_workers", (str, int))
self.validate_correct_type_of_configuration_attribute("diffexp__alg_cxg__cpu_multiplier", int)
self.validate_correct_type_of_configuration_attribute("diffexp__alg_cxg__target_workunit", int)

max_workers = self.diffexp__alg_cxg__max_workers
cpu_multiplier = self.diffexp__alg_cxg__cpu_multiplier
cpu_count = os.cpu_count()
max_workers = min(max_workers, cpu_multiplier * cpu_count)
diffexp_tiledb.set_config(max_workers, self.diffexp__alg_cxg__target_workunit)

def handle_adaptor(self):
# cxg
self.validate_correct_type_of_configuration_attribute("adaptor__cxg_adaptor__tiledb_ctx", dict)
Expand Down
Loading