Re: [Gems-users] Question about opal decode failure

[Xuan Qi <xqi@xxxxxxxxxxx>]

Hi guoqiang,

please check this old thread. It explains why that happens.

https://lists.cs.wisc.edu/archive/gems-users/2008-February/msg00035.shtml

Best regards,
Xuan Qi


On Wed, 16 Apr 2008, Guoqiang Yang wrote:

> Hi there,
>  I'm trying to run spec openmp benchmark in GEMS, but when I tried
> benchmark "apsi", opal complained that it can't decode some of the
> instructions. I searched the mailing digest and found that someone also
> met the same problem before. My question here is : even though opal can
> not decode some instructions, can it still progress the simulation? what
> is the effect of those undecoded instruction? Simulation correctness?
> Simulation Speed?
>
>  And here is some of output information about decode failure:
>  ### statici::decodeFails. line:2862. inst=0x85b00e0d cycle[ 2186353 ]
>  1000 0101 1011 0000:0000 1110 0000 1101
>  ### statici::decodeFails. line:2862. inst=0x85b08f8f cycle[ 2187129 ]
>  1000 0101 1011 0000:1000 1111 1000 1111
>  ### statici::decodeFails. line:2862. inst=0x99b30f8f cycle[ 2187400 ]
>  1001 1001 1011 0011:0000 1111 1000 1111
>  ### statici::decodeFails. line:2862. inst=0x99b28e0d cycle[ 2187145 ]
>  1001 1001 1011 0010:1000 1110 0000 1101
>  ### statici::decodeFails. line:2862. inst=0xadb50e0d cycle[ 2187417 ]
> 1010 1101 1011 0101:0000 1110 0000 1101
>  ### statici::decodeFails. line:2862. inst=0xadb58f8f cycle[ 2187691 ]
> 1010 1101 1011 0101:1000 1111 1000 1111
>  ### statici::decodeFails. line:2862. inst=0x91b20e11 cycle[ 2188008 ]
> 1001 0001 1011 0010:0000 1110 0001 0001
> ### statici::decodeFails. line:2862. inst=0xa5b48e11 cycle[ 2188304 ]
> 1010 0101 1011 0100:1000 1110 0001 0001
> ### statici::decodeFails. line:2862. inst=0xb9b70e11 cycle[ 2188314 ]
> 1011 1001 1011 0111:0000 1110 0001 0001
> ### statici::decodeFails. line:2867. inst=0x8db18a48 cycle[ 2189102 ]
> 1000 1101 1011 0001:1000 1010 0100 1000
>
>  If the simulation still progresses, can I just comment out these
> debugging information and let the simulation continue?
>
> Many thanks,
>
> Guoqiang Yang
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.