Skip to content

Disaggregated serving #365

New issue

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

#

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

Already on GitHub? # to your account

Jump to bottom
Merged
merged 15 commits into from
Apr 18, 2025
Merged

Disaggregated serving #365

merged 15 commits into from
Apr 18, 2025

Conversation

quic-amitraj
Copy link
Contributor

@quic-amitraj quic-amitraj commented Apr 16, 2025 •
edited
Loading

Adding support of-

  1. prefill_only
  2. mdp_ts_json_path

quic-rishinr and others added 9 commits April 16, 2025 05:30
@quic-rishinr @quic-amitraj
Support for dual compilation on VLMs (#361)
d8887b0 
Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
@quic-amitraj
Dis serving (#364)
cd38dfc 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Minor-fix-1
f41a0d7 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Minor changes
b13695b 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Changed approch and refactoring
7e679c3 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Ruff check and format
78dff2c 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Added test for prefill_only
fd2a612 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Minor-fix
26e50c5 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Merge branch 'main' into dist_serve
1626cda
ochougul
ochougul reviewed Apr 17, 2025
View reviewed changes
QEfficient/base/modeling_qeff.py Outdated
@@ -300,7 +308,7 @@ def _compile(
command.append(f"-custom-IO-list-file={custom_io_yaml}")

# Write mdp_config.json file
if mdp_ts_num_devices > 1:
if not mdp_ts_json_path:
Copy link
Contributor

Choose a reason for hiding this comment

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

if not mdp_ts_json_path and mdp_ts_num_devices>1

Copy link
Contributor

Choose a reason for hiding this comment

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

not addressed yet?

ochougul
ochougul reviewed Apr 17, 2025
View reviewed changes
ochougul
ochougul reviewed Apr 17, 2025
View reviewed changes
ochougul
ochougul reviewed Apr 17, 2025
View reviewed changes
ochougul
ochougul requested changes Apr 18, 2025
View reviewed changes
QEfficient/base/modeling_qeff.py Outdated
@@ -300,7 +308,7 @@ def _compile(
command.append(f"-custom-IO-list-file={custom_io_yaml}")

# Write mdp_config.json file
if mdp_ts_num_devices > 1:
if not mdp_ts_json_path:
Copy link
Contributor

Choose a reason for hiding this comment

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

not addressed yet?

ochougul and others added 4 commits April 18, 2025 14:25
@ochougul
simplified _compile code
3516bb6 
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
@ochougul
Merge branch 'main' into dist_serve
eb92774 
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
@quic-amitraj
Comments addressed
e82eaca 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj
Lint and formated
35bfd35 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@ochougul ochougul marked this pull request as ready for review April 18, 2025 11:02
@ochougul ochougul requested a review from quic-rishinr as a code owner April 18, 2025 11:02
@ochougul
fixed modelling qeff
1c372cf 
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
@ochougul ochougul mentioned this pull request Apr 18, 2025
Closed
@quic-amitraj
test bug fixed
8243ea2 
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
@quic-amitraj quic-amitraj merged commit 3de4072 into main Apr 18, 2025
5 checks passed
@quic-xiyushi
Copy link

Is the prefill_only flag available only for QEFFAutoModelForCausalLM? Why don't we support it for other classes as well, such as multimodal models?

eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @quic-rishinr
@ochougul @eplatero97
Disaggregated serving (quic#365)
51b1cf1 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @quic-rishinr
@ochougul @eplatero97
Disaggregated serving (quic#365)
91eeb68 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @quic-rishinr @ochougul
Disaggregated serving (quic#365)
c45c3e9 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Signed-off-by: eplatero <quic_eplatero@quicinc.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @eplatero97
Disaggregated serving (quic#365)
cd06158 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>

Signed-off-by: eplatero <quic_eplatero@quicinc.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @eplatero97
Disaggregated serving (quic#365)
8f0c080 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>

Signed-off-by: eplatero <quic_eplatero@quicinc.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @eplatero97
Disaggregated serving (quic#365)
c5e9513 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>

Signed-off-by: eplatero <quic_eplatero@quicinc.com>
eplatero97 pushed a commit to eplatero97/efficient-transformers that referenced this pull request Apr 29, 2025
@quic-amitraj @eplatero97
Disaggregated serving (quic#365)
ac539ab 
Adding support of-
1. `prefill_only`
2. `compile_for` for VLM
3. `mdp_ts_json_path`

---------

Signed-off-by: Rishin Raj <quic_rishinr@quicinc.com>
Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Onkar Chougule <quic_ochougul@quicinc.com>
Signed-off-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>
Co-authored-by: Rishin Raj <quic_rishinr@quicinc.com>
Co-authored-by: Onkar Chougule <quic_ochougul@quicinc.com>
Co-authored-by: Onkar Chougule <168134249+ochougul@users.noreply.github.com>

Signed-off-by: eplatero <quic_eplatero@quicinc.com>
@quic-rishinr quic-rishinr deleted the dist_serve branch June 13, 2025 08:39
# for free to join this conversation on GitHub. Already have an account? # to comment
Reviewers

@ochougul ochougul ochougul requested changes

@quic-rishinr quic-rishinr Awaiting requested review from quic-rishinr quic-rishinr is a code owner

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@quic-amitraj @quic-xiyushi @ochougul @quic-rishinr