On Sunday 08 April 2012 20:52:06 Sieghard wrote:
>
> Anyway, as this list is meant for discussion about MSEIDE and MSEGUI, I
> don't really want to stretch this topic. At least not if there's no common
> consent that the discussion was worthwhile.
It is worthwile IMO. I consistently have to think
On Sunday 08 April 2012 19:57:17 Sieghard wrote:
> > Not 100% correct - depends on hardware implementation. Multithreading
> > means the ability to achieve parallelism within a single app -
>
> In the narrowest sense, that's right. But what about a microcontroller
> running several execution thread
On Monday 09 April 2012 01:46:17 IvankoB wrote:
> > Please, could you avoid resending the whole previous communication with
> > every new article?
>
> Me don't see anything.
> Does my Opera do that ? Hmm..
>
Check the achive. :-)
http://article.gmane.org/gmane.comp.ide.mseide.user/25468
Martin
--
> That would mean that the AMD shaders had very
> poor performance, and alltogether the shaders wouldn't even caom close to
> the performance of your main CPU.
>
No, the test means that work of both 100% loaded CPUs may (on easily
parallelized task) can be easily (13% load for AMD & 20% load for
> Please, could you avoid resending the whole previous communication with
> every new article?
Me don't see anything.
Does my Opera do that ? Hmm..
--
For Developers, A Lot Can Happen In A Second.
Boundary is the first to
Hallo Ivanko,
Du schriebst am Sun, 8 Apr 2012 16:54:30 +0500:
> Opps, the timings were VERY wrong (there was a process in background
...
> 1) core2duo CPU: 20 secs (at 100% load of both 3.06GHz cores,)
> 2) nVidia CUDA : 17 secs (GPU freq= 405MHz, shader freq= 810MHz, GPU
> load = 20%, CPU load =
Hallo Ivanko,
Du schriebst am Sun, 8 Apr 2012 13:17:12 +0500:
> Multithreading is based on interleaved processing, while multiprocessing
> means truely parallel actions.
> ==
> Not 100% correct - depends on hardware implementation.
Well, it _is_ correct, though not exclusive. You may ha
There's a math library for such purposes:
==
OpenCL version of MAGMA now available
A beta version of OpenCL MAGMA is now available from The Innovative
Computing Lab at The University of Tennessee, Knoxville. The beta
supports LU, QR, and Cholesky factorizations on Linux. The follow
Opps, the timings were VERY wrong (there was a process in background
eating the CPU & nVidia PhysX slices, though HD6850 aimed but this
powerful card didn't even notice that task )
1) core2duo CPU: 20 secs (at 100% load of both 3.06GHz cores,)
2) nVidia CUDA : 17 secs (GPU freq= 405MHz, shader fre
nVidia shaders are somewhat more powerful than their AMD counterparts
(which is probable), this might even be counted as a near propotionality.
==
nVidia uses 2 times less shaders but at doubled frequency than AMD.
So,we can say that AMD cards are more parallelism-in-mind than nVidia
which
Hallo IvankoB,
Du schriebst am Sun, 08 Apr 2012 00:16:11 +0500:
> Me've just tried converting a 18M FLV video, the results are as follows:
> 1) dual core CPU (Core2Duo E8400 3.06GHz 6ML2 - it's a serious CPU) - 1
> min 15 sec
> 2) ATI/AMD APP on 960 shaders (AMD HD6850 - the main graphic board
Multithreading is based on interleaved processing, while multiprocessing
means truely parallel actions.
==
Not 100% correct - depends on hardware implementation. Multithreading
means the ability to achieve parallelism within a single app -
HyperThreading P4 CPUs (predecessors of DualCore on
12 matches
Mail list logo
