|
Message-ID: <CAFYn=yD18zb1cCA3M2U5tHE1iHV6Ug6HWaUhZidV9n-uJbqvag@mail.gmail.com>
Date: Mon, 16 Sep 2013 18:42:19 -0400
From: Yaniv Sapir <yaniv@...pteva.com>
To: john-dev <john-dev@...ts.openwall.com>
Subject: Re: bcrypt-parallella on 64-core (was: Katja's weekly
report #13)
Please do not read 25 words from 0xf0400. The higher addresses are reserved
and the behaviour is undefined. Highest word in this group is 0xf0448,
which means 19 words.
I will restart the system.
On Mon, Sep 16, 2013 at 4:47 PM, Katja Malvoni <kmalvoni@...il.com> wrote:
> On Mon, Sep 16, 2013 at 8:20 PM, Yaniv Sapir <yaniv@...pteva.com> wrote:
>
>> BTW, the restriction is not just with fetch, but with load in general.
>> The only safe method for a core on this row to read data from ERAM is using
>> eDMA. Writing to ERAM, or writing data from host (or external space in
>> general) onto these cores is no problem.
>>
>
> This partially explains why test failed. Some cores never read start flag,
> row 3 cores are among them. But not only row 3. Row 7 as well and some
> cores from other rows. After reading start flag, each core should put it to
> zero and wait for new one. In total, 38 cores do not read start flag
> although they should (file cores). In other runs there were 25 (cores2), 23
> and 19 cores which didn't read start flag. First row worked fine except in
> case when 38 cores failed to read start flag and proceed with computation.
> Because of this host is stuck in polling loop.
>
> I tried reading registers but I don't know what to conclude from them. And
> unfortunately before I copied contents of registers into this email, e-read
> 0 0 0xf0400 25 locked up the system.
>
> Katja
>
--
===========================================================
Yaniv Sapir
Adapteva Inc.
1666 Massachusetts Ave, Suite 14
Lexington, MA 02420
Phone: (781)-328-0513 (x104)
Email: yaniv@...pteva.com
Web: www.adapteva.com
============================================================
CONFIDENTIALITY NOTICE: This e-mail may contain information
that is confidential and proprietary to Adapteva, and Adapteva hereby
designates the information in this e-mail as confidential. The information
is
intended only for the use of the individual or entity named above. If you
are
not the intended recipient, you are hereby notified that any disclosure,
copying,
distribution or use of any of the information contained in this
transmission is
strictly prohibited and that you should immediately destroy this e-mail and
its
contents and notify Adapteva.
==============================================================
Content of type "text/html" skipped
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.