Vc-XeonPhi

[Kulakov, Igor]

ok. here some tests:

1. I get 13.6% without VC_IMLP definition at all. But in this case float_v::Size == 4.
2. Same with VC_IMPL=SSE.
3. VC_IMPL=AVX gives same result. I check - float_v::Size == 8.
4. on MIC I have seg fault with Release. and Assertion `hitIndex < NHits'  in Debug.
5. on MIC with VC_IMPL=Scalar it works nice.
6. We have only gcc 4.4.7 on the node. Vc-0.99 requires at least GCC 4.6.

> Maybe I can see a pattern if you give me more results. Also, if you can 
> identify small sections of code where it goes wrong we could try to enhance 
> the Vc unit tests.
I will look into this later.

Regards,
Igor

On Aug 26, 2013, at 9:28 AM, Matthias Kretz wrote:

> I'm currently looking into the MIC failures. The tests actually pass, but 
> libstdc++ segfaults after main returned...
> 
> Let me make sure I understand your remaining issue:
> You get 13.6% clones with Vc-XeonPhi using ICC and VC_IMPL=SSE, right?
> What you you get if you use GCC 4.7/4.8 instead?
> Any difference for AVX?
> And how does it look on MIC?
> 
> Maybe I can see a pattern if you give me more results. Also, if you can 
> identify small sections of code where it goes wrong we could try to enhance 
> the Vc unit tests.
> 
> -Matthias
> 
> On Monday 26 August 2013 13:20:45 Kulakov, Igor wrote:
>> Matthias, thank you.
>> 
>> It has improved the situation.
>> 
>> 1. I have only 4 test fails with Vc (not 18 as before), all of them are mic
>> tests, none are sse. 2. The tracker runs now in Release (do not crashes)
>> with sse. But with same 13.6% clones.
>> On Aug 26, 2013, at 8:00 AM, Matthias Kretz wrote:
>>> In the meantime I found and fixed two issues relating to misuse of
>>> aliasing in the ICC workarounds in Vc (don't get me started what
>>> workarounds those are). Right now I have quite reasonable unit test
>>> results with the XeonPhi branch in git. I suggest you update to that one
>>> and give it a try. If it works better I can make a new preview release.
>>> 
>>> Regards,
>>> 
>>> 	Matthias
>>> 
>>> On Monday 26 August 2013 11:38:01 Kulakov, Igor wrote:
>>>> Hi Matthias,
>>>> 
>>>> Thank you for the notice.
>>>> So you think that the needed compiler is found?
>>>> 
>>>> Then we still have exactly same problems with it and Vc-0.99 ...:
>>>> 1. Vc mask_sse test fails.
>>>> 2. Tracker crashes in release.
>>>> 3. Tracker gives 13.6% clones in debug.
>>>> 
>>>> Regards,
>>>> Igor
>>>> 
>>>> On Aug 26, 2013, at 5:51 AM, Matthias Kretz wrote:
>>>>> Hi,
>>>>> 
>>>>> On Friday 23 August 2013 21:27:20 Kulakov, Igor wrote:
>>>>>> how is it possible that Vc first write about icc 13.1.0, then about
>>>>>> 13.1.3?
>>>>>> Which version will be used?
>>>>> 
>>>>> Just noticed this now: Both are talking about the same compiler (see the
>>>>> build-date 20130607). Apparently the cmake-internal identification
>>>>> method
>>>>> fails to determine the patch-level correctly.
>>>>> 
>>>>>> P.S. composer_xe_2013.5.192 should have icc 13.1.3
>>>>>> 
>>>>>> Thank you,
>>>>>> Igor
>>>>>> 
>>>>>> -- The C compiler identification is Intel 13.1.0.20130607
>>>>>> -- The CXX compiler identification is Intel 13.1.0.20130607
>>>>>> -- Check for working C compiler:
>>>>>> /opt/intel/composer_xe_2013.5.192/bin/intel64/icc -- Check for working
>>>>>> C
>>>>>> compiler: /opt/intel/composer_xe_2013.5.192/bin/intel64/icc -- works --
>>>>>> Detecting C compiler ABI info
>>>>>> -- Detecting C compiler ABI info - done
>>>>>> -- Check for working CXX compiler:
>>>>>> /opt/intel/composer_xe_2013.5.192/bin/intel64/icpc -- Check for working
>>>>>> CXX
>>>>>> compiler: /opt/intel/composer_xe_2013.5.192/bin/intel64/icpc -- works
>>>>>> --
>>>>>> Detecting CXX compiler ABI info
>>>>>> -- Detecting CXX compiler ABI info - done
>>>>>> -- Detected Compiler: Intel 13.1.3
>>>>>> -- Performing Test c_compiler_can_offload
>>>>>> -- Performing Test c_compiler_can_offload - Success
>>>>>> -- Performing Test cxx_compiler_can_offload
>>>>>> -- Performing Test cxx_compiler_can_offload - Success
>>>>>> -- C/C++ Compiler can offload to MIC.
>>>>>> -- MIC ICC Version: "13.1.3.192 Build 20130607"
>>>> 
>>>> _______________________________________________
>>>> Vc mailing list
>>>>[please enable javascript to see the address]
>>>> https://compeng.uni-frankfurt.de/mailman/listinfo/vc
>> 
>> _______________________________________________
>> Vc mailing list
>>[please enable javascript to see the address]
>> https://compeng.uni-frankfurt.de/mailman/listinfo/vc
> -- 
> Dipl.-Phys. Matthias Kretz
> 
> Web:   http://compeng.uni-frankfurt.de/?mkretz
> 
> SIMD easy and portable: http://compeng.uni-frankfurt.de/?vc