Skip to contentSkip to conversation

ia64 emulator or test machine available

DISCUSSION Nick Piggin 2 participants 3 messages Nov 27, 2004 – Nov 29, 2004

lore
Share & tools
lore
#0 lore raw

Hi,

I have been doing some work in the area of page table lock reduction in collaboration with Christoph Lameter. The main aim of this is to improve multi-threaded page fault performance, but may also benefit other areas, like multi threded futex performance, and page reclaim.

I have working implementations on i386 and x86-64, and now I would like to get an ia64 version running. For two reasons: first, SGI is pushing most of the work in this area, obviously for their ia64 systems;

second, ia64's implementation of pte access faults in software means there is a possibility to implement mutual exclusion for pte access with pte lock bits rather than cmpxchg, and with no downside. pte lock bits may prove to be more scalable than cmpxchg in some areas.

With the help of Keith and Christoph, we have a version that is close to booting - however I can't ask or expect them to debug my bugs. So if anyone is interested in giving me access to an emulator or (preferably) machine with serial console and power access that boots recent 2.6 kernels then I would be much obliged.

I wouldn't require anything fancy, nor for (hopefully) too long a time.

Thanks,
Nick

#1 lore raw

Hi Nick

Quoted 16 lines · Nick Piggin

Hi,

I have been doing some work in the area of page table lock reduction in collaboration with Christoph Lameter. The main aim of this is to improve multi-threaded page fault performance, but may also benefit other areas, like multi threded futex performance, and page reclaim.

I have working implementations on i386 and x86-64, and now I would like to get an ia64 version running. For two reasons: first, SGI is pushing most of the work in this area, obviously for their ia64 systems;

second, ia64's implementation of pte access faults in software means there is a possibility to implement mutual exclusion for pte access with pte lock bits rather than cmpxchg, and with no downside. pte lock bits may prove to be more scalable than cmpxchg in some areas.

Quoted 7 lines

With the help of Keith and Christoph, we have a version that is close to booting - however I can't ask or expect them to debug my bugs. So if anyone is interested in giving me access to an emulator or (preferably) machine with serial console and power access that boots recent 2.6 kernels then I would be much obliged.

I wouldn't require anything fancy, nor for (hopefully) too long a time.

Possibly the best place to start debugging would be on the Ia64 SKI simulator, http://www.hpl.hp.com/research/linux/ski/index.php.

Current 2.6 kernel boots and runs. Use 'sim_defconfig' for initial configuration.

Quoted 7 lines

Thanks,
Nick
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

--------------------------------------------------
Darren Williams <dsw AT gelato.unsw.edu.au>
Gelato@UNSW <www.gelato.unsw.edu.au>
--------------------------------------------------

#2 lore raw
In reply to Nick Piggin
Quoted 7 lines · Darren Williams

Possibly the best place to start debugging would be on the Ia64 SKI simulator, http://www.hpl.hp.com/research/linux/ski/index.php.

Current 2.6 kernel boots and runs. Use 'sim_defconfig' for initial configuration.

Thanks Darren, and everyone else who kindly replied.

The simulator looks interesting - I will look into that.

I have been given access to an ia64 system that has been sufficient for some basic testing. Thank you all for your generosity.

Nick