Making sense of rPerf nums..

[LeePatto]

Hi All,

I need to compare rperf numbers for a couple of servers, can someone please tell me how to read these numbers and compare please? What I have are:-

72.86 z
82.23 y
...for one
and:-

9.86h
...for the other..

Any help of guidance would be greatly appreciated.

Many thanks

 

[unclebrian]

The y and z distinguish different size processors y=1.9GHz z=1.7GHz assuming the server is a 690.

Information taken from site below

http://www-03.ibm.com/systems/p/hardware/reports/factsfeatures_p4.pdf

I can only assume the h is another footnote reference. possibly for a p5 570 2 way

 

[mrn]

Take a look at this to compare rperfs

http://www-03.ibm.com/systems/power/hardware/reports/system_perf.html

Mike

"Whenever I dwell for any length of time on my own shortcomings, they gradually begin to seem mild, harmless, rather engaging little things, not at all like the staring defects in other people's characters."

 

[LeePatto]

Hi there, thanks for your responses.

I have already checked out the links provided and they are where I got some of my initial information. I don't know if it exists... but, is there a document that says:-

p520 x4(P5) rperf = xxxx

p680 x24(P4) rperf = xxxx

(I know one of these links proves this.. but it's not entirely clear... example is: xxxx , xxxx , xxxx , xxxx

A load of different numbers?!

...and what these rperf numbers actually mean is.... etc.. etc.. and here's how to perform the math required for the comparison... etc.. etc..

I can't find anything that informs me of the comparison process. I can see that the two servers... the p690 is appx 4.5 times faster than the p520, purely on the total Mhz figure. I would like to crack the rperf code ;-> I know that this is probably looking more compilcated than it actually is... which is the case for I.T.

Thanks again

 

[unclebrian]

If you look at the information in the link the 4 sets of rperf numbers are for a 8 16 24 and 32 way machines respectfully

A 4 way p520 p5 has a rperf of 31.48
A 24 way p690 p4 has a rperf of 72.86 with 1.7 GHz

The p690 is therefore about 2.3 time faster than the p520 but is a very large box and consumes a lot more power.

 

[unclebrian]

sorry 31.48 is for a p6 p520 4 way figure for p5 p520 is about half this so 4.5 times more through put on p690

 

[LeePatto]

Thanks for clearing that up for me. I gathered that the numbers were respective to the CPU's listed, but again, just needed this confirming. So as my original process confirmed, doing the math on the total Mhz for the boxes is equivalent to the rperf number?

Thanks again