[Gllug] HugePages

[Steve Parker]

On 28/10/10 19:06, James Courtier-Dutton wrote:
> On 28 October 2010 18:02, Steve Parker<steve at steve-parker.org>  wrote:
>    
>> All the documentation I can find about huge pages talks about reserving
>> a small fraction of the total RAM for huge pages - eg,
>> Documentation/vm/hugetlbpage.txt has an example of "echo 20>
>> /proc/sys/vm/nr_hugepages";
>>
>>      
> My understanding of hugepages is that you only use them if you really have to.
> The reason to have hugepages is that the x86/x64 CPU has a fixed size
> page table. I.e. How many pages it can refer to.
> If you have more physical memory than the CPU can reference with
> normal sized pages, one has to then use hugepages.
> So, I would approach this problem by trying to keep the amount of
> hugepages as low as you can.
> I cannot remember what the limit point is, but your large RAM figures
> are probably there.
> I would probably try to fix this problem by changing the hardware to a
> different CPU architecture that handles large RAM sizes better than
> the x86 CPU type can.
> There are other CPU types that do not have the limits that the x86
> architecture does.
>    
I think that that is a different thing; x86_64 can access 16Tb natively. 
hugepages are (on x86_64) 2Mb pages instead of the default 4Kb pages, so 
you can allocate (and keep track of) fewer pages in kernel space
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug