Copyright by William Lloyd Bircher 2010 - The Laboratory for ...
Copyright by William Lloyd Bircher 2010 - The Laboratory for ...
Copyright by William Lloyd Bircher 2010 - The Laboratory for ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
I/O device. <strong>The</strong> transfer size, source and destination are specified through the memory<br />
mapped I/O space. <strong>The</strong> disk controller per<strong>for</strong>ms the transfer without further intervention<br />
from the microprocessor. Upon completion or incremental completion (buffer<br />
full/empty) the I/O device interrupts the microprocessor. <strong>The</strong> microprocessor is then able<br />
to use the requested data or discard local copies of data that was sent. This study uses the<br />
number of interrupts originating from the disk controller. This approach has the<br />
advantage over the other metrics in that the events are specific to the subsystem of<br />
interest. This approach is able to represent fine-grain variation with low error. In the<br />
case of the synthetic disk workload, <strong>by</strong> using the number of disk interrupts/cycle an<br />
average error rate of 1.75% is achieved. <strong>The</strong> model is provided in Equation 5.4. An<br />
application of the model to the memory-intensive mcf is shown in Figure 5.6. Note that<br />
this error rate accounts <strong>for</strong> the large DC offset within the disk power consumption. This<br />
error is calculated <strong>by</strong> first subtracting the 21.6W of idle (DC) disk power consumption.<br />
<strong>The</strong> remaining quantity is used <strong>for</strong> the error calculation.<br />
NumCPUs<br />
+<br />
∑<br />
i =<br />
1<br />
21.<br />
6<br />
+<br />
DMAAccess<br />
Cycle<br />
Interrupts<br />
i<br />
Cycle<br />
×<br />
9.<br />
18<br />
−<br />
i<br />
× 10.<br />
6 ⋅10<br />
Cycle<br />
78<br />
7<br />
DMAAccess<br />
−<br />
2<br />
i<br />
Interrupt<br />
×<br />
Cycle<br />
45.<br />
4<br />
2<br />
i<br />
× 11.<br />
1 ⋅10<br />
15<br />
(5.4)