|
|
@@ -1344,7 +1344,8 @@ tries to minimize is @code{alpha * T_execution + beta * T_data_transfer}, where
|
|
|
@code{T_execution} is the estimated execution time of the codelet (usually
|
|
|
accurate), and @code{T_data_transfer} is the estimated data transfer time. The
|
|
|
latter is however estimated based on bus calibration before execution start,
|
|
|
-i.e. with an idle machine. When StarPU manages several GPUs, such estimation
|
|
|
+i.e. with an idle machine. You can force bus re-calibration by running
|
|
|
+@code{starpu_calibrate_bus}. When StarPU manages several GPUs, such estimation
|
|
|
is not accurate any more. Beta can then be used to correct this by hand. For
|
|
|
instance, you can use @code{export STARPU_BETA=2} to double the transfer
|
|
|
time estimation, e.g. because there are two GPUs in the machine. This is of
|
|
|
@@ -1390,6 +1391,7 @@ next chapter.
|
|
|
@menu
|
|
|
* On-line:: On-line performance feedback
|
|
|
* Off-line:: Off-line performance feedback
|
|
|
+* Codelet performance:: Performance of codelets
|
|
|
@end menu
|
|
|
|
|
|
@node On-line
|
|
|
@@ -1600,6 +1602,45 @@ evolution of the number of tasks available in the system during the execution.
|
|
|
Ready tasks are shown in black, and tasks that are submitted but not
|
|
|
schedulable yet are shown in grey.
|
|
|
|
|
|
+@node Codelet performance
|
|
|
+@section Performance of codelets
|
|
|
+
|
|
|
+The performance model of codelets can be examined by using the
|
|
|
+@code{starpu_perfmodel_display} tool:
|
|
|
+
|
|
|
+@example
|
|
|
+$ starpu_perfmodel_display -l
|
|
|
+file: <malloc_pinned.hannibal>
|
|
|
+file: <starpu_slu_lu_model_21.hannibal>
|
|
|
+file: <starpu_slu_lu_model_11.hannibal>
|
|
|
+file: <starpu_slu_lu_model_22.hannibal>
|
|
|
+file: <starpu_slu_lu_model_12.hannibal>
|
|
|
+@end example
|
|
|
+
|
|
|
+Here, the codelets of the lu example are available. We can examine the
|
|
|
+performance of the 22 kernel:
|
|
|
+
|
|
|
+@example
|
|
|
+$ starpu_perfmodel_display -s starpu_slu_lu_model_22
|
|
|
+performance model for cpu
|
|
|
+# hash size mean dev n
|
|
|
+57618ab0 19660800 2.851069e+05 1.829369e+04 109
|
|
|
+performance model for cuda_0
|
|
|
+# hash size mean dev n
|
|
|
+57618ab0 19660800 1.164144e+04 1.556094e+01 315
|
|
|
+performance model for cuda_1
|
|
|
+# hash size mean dev n
|
|
|
+57618ab0 19660800 1.164271e+04 1.330628e+01 360
|
|
|
+performance model for cuda_2
|
|
|
+# hash size mean dev n
|
|
|
+57618ab0 19660800 1.166730e+04 3.390395e+02 456
|
|
|
+@end example
|
|
|
+
|
|
|
+We can see that for the given size, over a sample of a few hundreds of
|
|
|
+execution, the GPUs are about 20 times faster than the CPUs (numbers are in
|
|
|
+us). The standard deviation is extremely low for the GPUs, and less than 10% for
|
|
|
+CPUs.
|
|
|
+
|
|
|
@c ---------------------------------------------------------------------
|
|
|
@c MPI support
|
|
|
@c ---------------------------------------------------------------------
|
