Test with Burn in test v8.1 fail issue

[andylu]

Operating Intel CPU P4500 + QM57 platform, we test with PassMark software Burn in test v8.1 exception occurs.
Model used CPU Type:
Intel CPU.Celeron.P4500.Mobile.Processor.1.86GHz/2MB.PGA988.CP80617004803AA.SLBUX
However, if the CPU replacement by others: (ex: CPU type i5 or i7)
Intel CPU.I7-620M.Mobile.Processor.2.66GHz/4MB.PGA988.CP80617003981AH.SLBTQ
Intel CPU.I5-520M.Mobile.Processor.2.4GHz/3MB.CP80617004119AE.SLBU3
Test result is normal.
Dual LAN Chipset on the platform that we use the following
LAN1: WG82574L SLBA8
LAN2: WG82577LM.SLGWS
Is this part of a relevant recommendation can provide?

[Tim (PassMark)]

When is the crash occurring, at startup or during a test?
If during a test are you able to narrow the crash down to a specific test by running them one at a time?

If the crash is happening during the startup it's possible that some of the CPU information or temperature information is not being collected correctly. In this case could you please launch BurnInTest in debug mode and send us a copy of the log files that are generated (and if there is a crash dump file them a zipped coy of that would be great).

[andylu]

Hi Tim,
The crash occurring during a test.
Different test conditions(select different test items or Lan percentage 70%~90%), the test results did not differ.

[andylu]

This crash is occurring during test.
Different test conditions results did not differ
Provide relevant archives for your reference

[Tim (PassMark)]

So there is a network test error, not a software crash as we originally though you meant.

If you keep decreasing the duty cycle for the network test do you reach a level where the test will run without errors?

If you run the network test by itself do you still see the errors?

[andylu]

Hi Tim,
Yes, we run network test by itself.
We only choose this single test item and tested different BIT Lan loading settings percentages.
Different percentages 50%, 55% and 60%.
The test result pass when tested in only 50% can be verified.
Can you explain what cause this ?

[Tim (PassMark)]

It looks like you are using the "bad packet generates error" setting so a single missing packet will log an error. The simple network test uses ICMP_ECHO (ping) packets, as such these packets are not always guaranteed to arrive so it is not unusual for a single packet to go missing or be ignored by the receiving hardware.

For some reason it looks like the receiving network device is not returning one of the packets above a certain duty cycle. You could try using the "bad packet ration generates error" option to see if the errors are consistent (a large number of packets are being ignored or lost rather than a single one). It could be that the hardware on the receiving end starts ignoring the packets once they are being received above a certain packets/sec threshold.