I got three questions and one inline comment about this MP:
1. I am curious why we're using not using "ioengine=libaio" in DGX2/dataset/global-include.fio, in this MP it's being used for DGXA100/dataset/global-include.fio and DGXH100/dataset/global-include.fio.
2. For DGXA100 and DGXH100, why most of the numjobs are 1/4 of the number stated in [testname], e.g.
+[rd-0,wr-100,rand,blk-128k,jobs-256]
+numjobs=64
It appears only ramdisk rd-100,wr-0 and rd-75,wr-25 are using the same number.
3. Why do we need unattended-upgrades package?
Hi Ian,
Overall it's looking good.
I got three questions and one inline comment about this MP:
1. I am curious why we're using not using "ioengine=libaio" in DGX2/dataset/ global- include. fio, in this MP it's being used for DGXA100/ dataset/ global- include. fio and DGXH100/ dataset/ global- include. fio. wr-100, rand,blk- 128k,jobs- 256]
2. For DGXA100 and DGXH100, why most of the numjobs are 1/4 of the number stated in [testname], e.g.
+[rd-0,
+numjobs=64
It appears only ramdisk rd-100,wr-0 and rd-75,wr-25 are using the same number.
3. Why do we need unattended-upgrades package?