Re: [RFC 0/2] Delay initializing of large sections of memory

[Mike Travis <travis@sgi.com>]

On 6/21/2013 12:00 PM, Yinghai Lu wrote:
> On Fri, Jun 21, 2013 at 11:44 AM, Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
>> On Fri, Jun 21, 2013 at 11:36:21AM -0700, Yinghai Lu wrote:
>>> On Fri, Jun 21, 2013 at 9:25 AM, Nathan Zimmer <nzimmer@sgi.com> wrote:
>>>> This rfc patch set delays initializing large sections of memory until we have
>>>> started cpus.  This has the effect of reducing startup times on large memory
>>>> systems.  On 16TB it can take over an hour to boot and most of that time
>>>> is spent initializing memory.
>>>
>>> One hour on system with 16T ram? BIOS or OS?
>>>
>>> I use wall clock to check bootime on one system with 3T and 16 pcie cards,
>>> Linus only takes about 3m and 30 seconds from bootloader.
>>>
>>> wonder if you boot delay is with so many cpu get onlined in serialized mode.
>>>
>>> so can you try boot your system with "maxcpus=128" to get the boot time with
>>> wall clock ?
>>
>> Why use the "wall clock" when we have the wonderful bootchart tools and
>> scripts that do this all for you, and can tell you exactly what part of
>> the kernel is taking what time, to help with fixing issues like this?
> 
> bootchart is not completed.
> 
> printk timestamp come after mem get initialized.
> ....
> [    0.004000] tsc: Fast TSC calibration using PIT
> 
> before that stamp are all 0.
> 
> Yinghai
> 

On UV the system console function has an option to include
timestamps, both sequential in HH:MM:SS and deltas to 100ms.
So we get both the BIOS and system times before printk time
is active.  We also have a custom "script" command that adds
timing info.