[Beowulf] AMD performance (was 500GB systems)

[Vincent Diepeveen]

Not a single test in HPC tested memory with more than 1 core at the  
time.

There were a few pingpong latency tests getting performed single core  
by some manufacturers using a raw sort of implementation which wasn't  
very realistic.

For  software engineers to optimize his software it's crucial to have  
accurate technical information
on performance of the hardware.

I had to write therefore tests to test the basic hardware performance  
in order to know where to design
the software better.

If you call that 'full of yourself', i realize what sort of  
'scientist' you are.

On Jan 12, 2013, at 11:10 PM, Mark Hahn wrote:

>> Yes i was the inventor of that test to jump using a RNG randomly.
>> Paul Hsieh then modified it from calling the RNG and correcting for
>> the RNG, to the direct pointer math
>> as you show here.
>
> Vincent, you are hillariously full of yourself.
>
>> If you are using 32 bits ints however, you'll not be able to address
>> more than 16GB ram.
>> You need 64 bits ints for P.
>
> wow, you're so smart!