Serial Loopback differences

[Cosmo]

I just noticed some odd things when trying to do loopback operation on some serial ports.

I have setup both COM1 and COM2 to loopback at 115K. When I use BIT 5.3Pro it shows the throughput as 5340.7 Bytes/sec.

I was tipped off by this while working with a vendor of ours and they were using Version 4.0 Pro. This version, when testing at 115K shows a throughput of 11359.7 Bytes/sec.

Has something drastically changed with the translation of the throughput? Any insight?

[passmark]

It might be becuase you have selected a lower duty cycle on the serial test? Or becuase you have other tests running at the same time using up the CPU time?

[Cosmo]

I don't think this is the case. I am using a 100% idential machine, ONLY difference is the version of BIT (V4.0 compared to V5.3).

The cycle is at 100%, and NO other tests are running.

[passmark]

Maybe it is the flow control setting in the preferences window?

[BrooksL]

I just started looking at the serial loopback test and see the same or similar issue (currently running B.I.T. 5.1):

When testing one serial port, throughput seems to match the test speed (considering estimate of 10bits/byte-sent).

However, when I select both serial ports for test in the preferences, the throughput for each port drops to about 1/2 the expected throughput, regardless of the test speed selected or test duty-cycle.

What's the deal? - I know there is more than enough processor and bus bandwidth available to keep both UARTs operating at max throughput.

So, where is the bottle-neck that's affecting the B.I.T. throughput results?

[Ian (PassMark)]

We will do some testing and post our results when available.
Ian