On 5/29/20 4:59 PM, Igor Mammedov wrote:
On Fri, 29 May 2020 15:33:47 +0200
Michal Privoznik wrote:
Currently, when defining a HMAT cache for a NUMA node (in
parse_numa_hmat_cache()) there is this check that forces users to
define HMAT latency/bandwidth first. There is no real need for
this,
On Fri, 29 May 2020 15:33:47 +0200
Michal Privoznik wrote:
> Currently, when defining a HMAT cache for a NUMA node (in
> parse_numa_hmat_cache()) there is this check that forces users to
> define HMAT latency/bandwidth first. There is no real need for
> this, because nothing in the parse
Currently, when defining a HMAT cache for a NUMA node (in
parse_numa_hmat_cache()) there is this check that forces users to
define HMAT latency/bandwidth first. There is no real need for
this, because nothing in the parse function relies on that and
the HMAT table is constructed way later - when