Check a prototype of 2nd version of this platform
being developed by cKnowledge.org in collaboration with
MLCommons.
crowdsource experiment.tune.opencl.clblast
ck-action:crowdsource experiment.tune.opencl.clblast (v3.0.0)
License: BSD 3-clause (code) and CC BY-SA 4.0 (data)
Creation date: 2017-05-23
Source: GitHub
cID: 081173242a88bc94:4d32ac1a044a71ad
Don't hesitate to get in touch if you encounter any issues or would like to discuss this community project!
Creation date: 2017-05-23
Source: GitHub
cID: 081173242a88bc94:4d32ac1a044a71ad
Description
- Automation framework: CK
- Workflow: CLBlast crowd-tuning workflow
- Development repository: ck-ml
- Source: GitHub
- How to get the stable version via the CK client:
pip install cbench
cb download module:experiment.tune.opencl.clblast --version=3.0.0 --all
ck crowdsource experiment.tune.opencl.clblast --help - How to get the development version:
pip install ck
ck pull repo:ck-ml
ck crowdsource experiment.tune.opencl.clblast --help
- How to run from Python:
import ck.kernel as ck
r=ck.access({'action':'',
'module_uoa':'',
... See JSON API below ...
})
if r['return']>0: return r
...
- Module description: auto/crowd-tune OpenCL-based CLBlast (GFLOPs)
- API Python code: Link
- JSON API:
" Input: { (local) - if 'yes', local crowd-benchmarking, instead of public (user) - force different user ID/email for demos (choices) - force different choices to program pipeline (repetitions) - statistical repetitions (default=1), for now statistical analysis is not used (TBD) (no_compile) - if 'yes', skip program compilation (for Android) (m) - dataset dimension M (n) - dataset dimension N (k) - dataset dimension K } Output: { return - return code = 0, if successful > 0, if error (error) - error text if return > 0 } "