Hi Ahmad,
First, there seems to be a bug in the prechargeAllPercent calculation. Could
you provide some more details on the settings you use for the DRAM, page policy
etc.
Second, I am mostly concerned there might be some weird effects if the stats
are not reset, and you suddenly switch to atomic, then back to timing. It could
be that it is all fine, I’m merely not 100% convinced that it is not causing
problems. Please just make sure that you see stats per interval, and that you
only actually count the intervals that use the o3 CPU (the easiest way of
achieving this would be to not switch once you reach the ROI).
Thanks,
Andreas
From: Ahmad Hassan <[email protected]<mailto:[email protected]>>
Reply-To: gem5 users mailing list
<[email protected]<mailto:[email protected]>>
Date: Monday, 20 January 2014 13:19
To: gem5 users mailing list <[email protected]<mailto:[email protected]>>
Subject: Re: [gem5-users] GEM5 power stats for Micron spreadsheet
Hi Andreas,
Thanks. I am using 'repeat-switch' to pass only 50% of the instructions or
cycles through detailed cpu and memory hierarchy e.g. O3. My simulation only
has 1 CPU. With repeat-swtich option I provided switching CPU as 'Atomic' cpu
so that switching happens between O3 and Atomic CPU and my simulation finishes
quickly with 50% sampling rate.
As you are suggesting that I shouldn't use 'Atomic' CPU then how shall I do
sampling and get realistic memory stats please?
kind Regards, Ahmad
On 20 January 2014 13:09, Andreas Hansson
<[email protected]<mailto:[email protected]>> wrote:
Hi Ahmad,
When the stats are reset they should also be dumped, so I don’t think that’s
the issue here. If you want temporal memory-system stats, you should not use
Atomic though, as the only information you will get is the bytes read/written.
I suggest to stick to the O3 CPU once you reach the ROI.
The prechargeAllPercent seems completely off in your example (558.79), and I’ll
have a look at the code. Perhaps this is also a result of the atomic/o3
switching.
Andreas
From: Ahmad Hassan <[email protected]<mailto:[email protected]>>
Reply-To: gem5 users mailing list
<[email protected]<mailto:[email protected]>>
Date: Monday, 20 January 2014 12:00
To: gem5 users mailing list <[email protected]<mailto:[email protected]>>
Subject: [gem5-users] GEM5 power stats for Micron spreadsheet
Hi,
I am using gem5 rev 9975 that implements the stats for calculating power
through micron spreadsheet. For my benchmark, I get the following numbers:
system.mem_ctrls.busUtilRead 0.07 #Data bus utilization
in percentage for reads
system.mem_ctrls.busUtilWrite 0.05 #Data bus
utilization in percentage for writes
system.mem_ctrls.pageHitRate 89.04 #Row buffer hit rate,
read and write combined
system.mem_ctrls.prechargeAllPercent 558.79 #Percentage of time for
which DRAM has all the banks in precharge state
The bus utilization number are really low as per my expectations. i am using
sampling. is it possible that the stats are getting reset when CPU switches
back and forth between O3 and Atomic. I use '--repeat-switch=10000000'
kind Regards, Ahmad
-- IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended recipient,
please notify the sender immediately and do not disclose the contents to any
other person, use it for any purpose, or store or copy the information in any
medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered
in England & Wales, Company No: 2557590
ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ,
Registered in England & Wales, Company No: 2548782
_______________________________________________
gem5-users mailing list
[email protected]<mailto:[email protected]>
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
-- IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended recipient,
please notify the sender immediately and do not disclose the contents to any
other person, use it for any purpose, or store or copy the information in any
medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered
in England & Wales, Company No: 2557590
ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ,
Registered in England & Wales, Company No: 2548782
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
