scheduler: add <gpu_type_regex> plan class option.#7047
Merged
Conversation
Contributor
There was a problem hiding this comment.
Pull request overview
Adds support for selecting “other” (non-NVIDIA/AMD/Intel/Apple) OpenCL GPU types in scheduler plan classes using a <gpu_type_regex> match against the coprocessor-reported type/model, enabling projects to target emerging GPU vendors/models without client/server code changes.
Changes:
- Extend
PLAN_CLASS_SPECto store and parse a compiled<gpu_type_regex>for GPU type matching. - Update plan class selection logic to pick an OpenCL coprocessor by regex when applicable.
- Adjust OpenCL GPU JSON summarization to report vendor + model (instead of a generic
"opencl"type).
Reviewed changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| sched/plan_class_spec.h | Adds have_gpu_type_regex and compiled regex_t storage in plan class spec. |
| sched/plan_class_spec.cpp | Parses <gpu_type_regex> and uses it during plan class compatibility checks. |
| lib/opencl_boinc.h | Clarifies that OPENCL_DEVICE_PROP::name represents the device model. |
| lib/coproc.cpp | Changes JSON summary for non-core OpenCL devices to emit vendor as type and device name as model. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+87
to
+88
| bool have_gpu_type_regex; | ||
| regex_t gpu_type_regex; |
Comment on lines
+875
to
+889
| } else if (have_gpu_type_regex) { | ||
| for (int i=1; i<sreq.coprocs.n_rsc; i++) { | ||
| if (!regexec(&(gpu_type_regex), sreq.coprocs.coprocs[i].type, 0, NULL, 0)) { | ||
| cpp = &sreq.coprocs.coprocs[i]; | ||
| break; | ||
| } | ||
| } | ||
| if (!cpp) { | ||
| if (config.debug_version_select) { | ||
| log_messages.printf(MSG_NORMAL, | ||
| "[version] plan_class_spec: No gpu_type match found\n" | ||
| ); | ||
| } | ||
| return false; | ||
| } |
Comment on lines
+901
to
909
| if (cpp) { | ||
| if (config.debug_version_select) { | ||
| log_messages.printf(MSG_NORMAL, | ||
| "[version] plan_class_spec: Custom coproc %s found\n", gpu_type | ||
| "[version] plan_class_spec: OpenCL coproc %s found\n", gpu_type | ||
| ); | ||
| } | ||
| if (cpp->bad_gpu_peak_flops("Custom GPU", msg)) { | ||
| if (cpp->bad_gpu_peak_flops("OpenCL coproc", msg)) { | ||
| log_messages.printf(MSG_NORMAL, "%s\n", msg.c_str()); | ||
| } |
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
2 issues found across 4 files
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="sched/plan_class_spec.cpp">
<violation number="1" location="sched/plan_class_spec.cpp:875">
P1: Regex-matched GPUs bypass the general GPU validation block because that block is still gated by `strlen(gpu_type)`. Plan classes using only `gpu_type_regex` may skip RAM/driver-related constraints.</violation>
<violation number="2" location="sched/plan_class_spec.cpp:904">
P2: When the `have_gpu_type_regex` branch (or NVIDIA/AMD/Intel/Apple branches) sets `cpp`, `gpu_type` is empty, so this log message will print "OpenCL coproc found" with a blank name. Use `cpp->type` or `cpp->opencl_prop.name` instead to produce a meaningful diagnostic.</violation>
</file>
Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.
AenBleidd
approved these changes
Apr 29, 2026
If the client detects (via OpenCL) a GPU that's not one of the original 4 (NVIDIA, AMD, Intel, Apple) it reports it to the scheduler with its model name as its 'type'. A GPU plan class can now specify its GPU type with a regular expression that's matched against this. This allows BOINC to use new GPU types - such as integrated GPUs on ARM chips - with no changes to client or server software. All that's needed is for projects to develop applications, and to create plan classes with appropriate <gpu_type_regex>
AenBleidd
force-pushed
the
dpa_plan_class2
branch
from
6a2c1bb to
41785ec
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
Note
Copilot was unable to run its full agentic suite in this review.
Adds support for matching “other” OpenCL GPU devices via a new <gpu_type_regex> plan-class option, and adjusts OpenCL device reporting to expose vendor/name information for better scheduling on new GPU families.
Changes:
- Introduce
gpu_type_regexto plan class parsing and selection logic for OpenCL GPUs. - Apply the same GPU RAM/FLOPS checks to regex-matched devices as for typed devices.
- Modify OpenCL summary JSON emission to output vendor/name rather than the previous generic type.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 5 comments.
| File | Description |
|---|---|
| sched/plan_class_spec.h | Adds stored state for compiled GPU type regex. |
| sched/plan_class_spec.cpp | Parses/compiles <gpu_type_regex> and uses it during GPU selection and validation. |
| lib/coproc.cpp | Changes the OpenCL JSON summary fields to report vendor and model name. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+1274
to
+1283
| if (xp.parse_str("gpu_type_regex", buf, sizeof(buf))) { | ||
| if (regcomp(&(gpu_type_regex), buf, REG_EXTENDED|REG_NOSUB) ) { | ||
| log_messages.printf(MSG_CRITICAL, | ||
| "BAD GPU TYPE REGEXP: %s\n", buf | ||
| ); | ||
| return ERR_XML_PARSE; | ||
| } | ||
| have_gpu_type_regex = true; | ||
| continue; | ||
| } |
Comment on lines
+876
to
+881
| for (int i=1; i<sreq.coprocs.n_rsc; i++) { | ||
| if (!regexec(&(gpu_type_regex), sreq.coprocs.coprocs[i].type, 0, NULL, 0)) { | ||
| cpp = &sreq.coprocs.coprocs[i]; | ||
| break; | ||
| } | ||
| } |
| if (!cpp) { | ||
| if (config.debug_version_select) { | ||
| log_messages.printf(MSG_NORMAL, | ||
| "[version] plan_class_spec: No gpu_type match found\n" |
Comment on lines
+901
to
910
| if (cpp) { | ||
| if (config.debug_version_select) { | ||
| log_messages.printf(MSG_NORMAL, | ||
| "[version] plan_class_spec: Custom coproc %s found\n", gpu_type | ||
| "[version] plan_class_spec: OpenCL coproc %s found\n", cpp->type | ||
| ); | ||
| } | ||
| if (cpp->bad_gpu_peak_flops("Custom GPU", msg)) { | ||
| if (cpp->bad_gpu_peak_flops("OpenCL coproc", msg)) { | ||
| log_messages.printf(MSG_NORMAL, "%s\n", msg.c_str()); | ||
| } | ||
| } |
Comment on lines
239
to
246
| if (!strlen(cp.opencl_prop.name)) continue; | ||
| if (!out.empty()) out += ",\n"; | ||
| summary_json(buf2, | ||
| "opencl", | ||
| cp.type, | ||
| cp.opencl_prop.vendor, | ||
| cp.opencl_prop.name, | ||
| cp.count, | ||
| (int)((double)cp.opencl_prop.global_mem_size/MEGA), | ||
| cp.opencl_prop.opencl_device_version, |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
If the client detects (via OpenCL) a GPU that's not one of the original 4 (NVIDIA, AMD, Intel, Apple)
it reports it to the scheduler with its model name as its 'type'.
A GPU plan class can now specify its GPU type with a regular expression that's matched against this.
This allows BOINC to use new GPU types -
such as integrated GPUs on ARM chips -
with no changes to client or server software.
All that's needed is for projects to develop applications, and to create plan classes with appropriate <gpu_type_regex>
Summary by cubic
Add
gpu_type_regexto GPU plan classes so the scheduler can match OpenCL device model names and schedule apps on new GPU types (e.g., ARM iGPUs) without client updates. Also update the OpenCL summary JSON to reportvendorandname(model).New Features
gpu_type_regex; version selection tries it first, thengpu_type, with clearer debug logs.vendorandnameinstead of a generic type.Migration
gpu_type_regex(e.g.,Mali|Adreno) to plan classes to enable new GPU families.Written for commit 41785ec. Summary will update on new commits. Review in cubic