- problem arose from a shift being done on an unsigned value which did not
keep the signed characteristics of the represented signed value
- introduce an arithmetic_shift on the UnsignedInteger trait with a blanket
implementation
- add the edge case which revelead the issue
- the asm has been verified to only change for the shift operation being
applied, meaning no performance regression will occurr
Done to circumvent criterion limitation regarding automatic
truncation of long benchmark ID.
Using a println() call we ensure the complete name is displayed
before benchmark execution to ease manual parsing and debugging.
The steps responsible for setting the OP_FLAVOR and ALL_PRECISION
variables were never executed due to usage of non-existing env
variable.
This causes OP_FLAVOR value to be null and thus would trigger
error on benchmarks that doesn't handle unknown values for
BENCH_OP_FLAVOR.
Also fixes filename to parse for additional boolean benchmark.
Following the same pattern as GPU benchmarks, HPU benchmarks rely
on a common workflow. All the manual launches via
workflow_dispatch event are now done in one place. That way, one
doesn't have to browse the workflow tree to find the right HPU
benchmark to trigger.
Following the same pattern as GPU benchmarks, CPU benchmarks rely on a common workflow. Weekly benchmarks are all gathered in one place. Also, all the manual launches via workflow_dispatch event are now done in one place. That way, one doesn't have to browse the workflow tree to find the right CPU benchmark to trigger.
Signed-off-by: David Testé <david.teste@zama.ai>
- encode the position of bits proven to be 0 in the hashes
- hash the infinite norm instead of the euclidean one
- hash the value of k with the statement
This is done to avoid confusion for operations that might have the same between layer. For example, 'bitand' operation have the same name for shortint and integer layers
When doing regression generation, one can provide a global parameters set name pattern to filter head branch benchmark results.
This fixes the issue encountered when there are more than one parameters' set used to benchmark an operation, for example, in core_crypto or shortint tfhe-rs layer.
