fix TRACEFILE names
This commit is contained in:
parent
9fc45204ae
commit
270e0226c8
|
@ -100,8 +100,9 @@ class RuntimeHWConfig:
|
|||
pre-built runtime config? It kinda contains a mix of pre-built and
|
||||
runtime parameters currently. """
|
||||
|
||||
tracefile = "+tracefile0=TRACEFILE" if trace_enable else ""
|
||||
autocounterfile = "+autocounter-filename0=AUTOCOUNTERFILE"
|
||||
# TODO: supernode support
|
||||
tracefile = "+tracefile0=TRACEFILE0" if trace_enable else ""
|
||||
autocounterfile = "+autocounter-filename0=AUTOCOUNTERFILE0"
|
||||
|
||||
# this monstrosity boots the simulator, inside screen, inside script
|
||||
# the sed is in there to get rid of newlines in runtime confs
|
||||
|
|
|
@ -84,7 +84,7 @@ By default, the read-rate is set to 0 cycles, which is equivalent to disabling A
|
|||
|
||||
|
||||
Now when you run a workload, an AutoCounter output file will be placed in the
|
||||
`sim_slot_<slot #>` directory on the F1 instance under the name AUTOCOUNTERFILE.
|
||||
`sim_slot_<slot #>` directory on the F1 instance under the name AUTOCOUNTERFILE0.
|
||||
|
||||
.. Note:: AutoCounter is designed as a coarse-grained observability mechanism. It assumes the counters will be read at intervals greater than O(10000) cycles. If you intend on reading counters at a finer granularity, please consider using synthesizable printfs (otherwise, simulation performance may degrade more than necessary)
|
||||
|
||||
|
|
|
@ -37,11 +37,16 @@ instead of ``no``:
|
|||
enable=yes
|
||||
|
||||
Now when you run a workload, a trace output file will be placed in the
|
||||
````sim_slot_<slot #>```` directory on the F1 instance under the name ``TRACEFILE0``.
|
||||
You can add ``TRACEFILE0`` to your ``common_simulation_outputs`` or
|
||||
``simulation_outputs`` in your workload ``.json`` file to have this automatically
|
||||
copied back to your manager. See the :ref:`defining-custom-workloads` section
|
||||
for more information about these options.
|
||||
``sim_slot_<slot #>`` directory on the F1 instance under the name ``TRACEFILE0-C0``.
|
||||
The first ``0`` in this filename disambiguates between multiple SoCs on one FPGA
|
||||
if you're running in supernode mode and will always be ``0`` if you're not running
|
||||
in supernode mode. The ``C0`` represents core 0 in the simulated
|
||||
SoC. If you have multiple cores, each will have its own file (ending in ``C1``,
|
||||
``C2``, etc). To copy all TracerV trace files back to your manager, you can
|
||||
add ``TRACEFILE*`` to your ``common_simulation_outputs`` or
|
||||
``simulation_outputs`` in your workload ``.json`` file. See the
|
||||
:ref:`defining-custom-workloads` section for more information about these
|
||||
options.
|
||||
|
||||
Selecting a Trace Output Format
|
||||
---------------------------------
|
||||
|
|
|
@ -103,7 +103,7 @@ tracerv_t::tracerv_t(
|
|||
if (tracefilename) {
|
||||
// giving no tracefilename means we will create NO tracefiles
|
||||
for (int i = 0; i < NUM_CORES; i++) {
|
||||
std::string tfname = std::string(tracefilename) + std::to_string(i);
|
||||
std::string tfname = std::string(tracefilename) + std::to_string("-C") + std::to_string(i);
|
||||
this->tracefiles[i] = fopen(tfname.c_str(), "w");
|
||||
if (!this->tracefiles[i]) {
|
||||
fprintf(stderr, "Could not open Trace log file: %s\n", tracefilename);
|
||||
|
|
Loading…
Reference in New Issue