Re: [Gems-users] ATMTP: Problem with strange Transaction-Aborts

[BYONG WU CHONG <bernard.chong@xxxxxxxx>]

Hi, Philipp

I ran ATMTP with no problems on microbenchmarks provided from GEMS.
$GEMS/microbenchmarks/atmtp_examples

Did your compiled benchmark have TLB warm up or TLB miss handling?
Did you check your compiled benchmark (332_ammp)
did not call malloc/new from a transaction?
As you know well, syscall is not allowed from GEMS TM (inc. ATMTP).
Most of the failures I had belong to above two problems.

Though ATMTP seemed more advanced than baseline Ruby HTM,
I used normal Ruby TM instead of ATMTP for HTM simulations
because ATMTP does not support unbounded transactions.
Making a benchmark that runs ATMTP successfully seemed much harder for me, because
programmers have to check the transaction size and other restrictions by hand.

- Bernard


-----Original Message-----
From: gems-users-bounces@xxxxxxxxxxx [mailto:gems-users-bounces@xxxxxxxxxxx] On Behalf Of Philipp Tolke
Sent: Monday, July 13, 2009 12:49 AM
To: Gems Users
Subject: Re: [Gems-users] ATMTP: Problem with strange Transaction-Aborts

Hello,

| 6826445428   0 [0,0] TID -1 XACT ABORT 0 caused by 0 [ 0, 0 ] xid: 0
| address: [0x0, line 0x0] delay: 6826445428  PC [0x0, line 0x0]  *PC 
| 0x0 'illtrap 0'

Since I patched ruby as you recommended, at least this does not cause an error anymore. However, after this message the simulation stops. If I press ^C twice, I can see, that the Simulation hangs again on the instruction "mov %g0, %o0", which compiles out of the microbenchmark rock_tm.il.

How are those zero-riddled transaction-aborts caused? Is it an error in my program?

Regards,
--
Philipp Tölke