You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -218,7 +237,7 @@ When run on the Intel® PAC with Intel Arria10® 10 GX FPGA hardware board
218
237
219
238
Setting the `private_copies` attribute to 0 (or equivalently omitting the attribute entirely) produced good throughput, and the reports show us that the compiler selected 3 private copies. This does produce the optimal throughput, but in this case it probably makes sense to save some area in exchange for a very small throughput loss by specifying 2 private copies instead.
220
239
221
-
When run on the FPGA emulator, the `private_copies` attribute has no effect on kernel time. You may actually notice that the emulator achieved higher throughput than the FPGA in this example. This is because this trivial example uses only a tiny fraction of the spatial compute resources available on the FPGA.
240
+
When run on the FPGA emulator or simulator, the `private_copies` attribute has no effect on kernel time. You may actually notice that the emulator achieved higher throughput than the FPGA in this example. This is because this trivial example uses only a tiny fraction of the spatial compute resources available on the FPGA.
0 commit comments