i want to run parsec blackscholes in fs mode with simsmall inputs in gem5
with these configurations :
--num-cpus=64 --num-dirs=64 --cpu-type=O3CPU --cpu-clock=2GHz --caches
--l1d_size=16kB --l1i_size=16kB --l2cache --num-l2cache=64
--mem-type=SimpleMemory --mem-size=2GB --ruby --network=garnet
--t
Dear all,
I wonder does anyone successfully run FS mode on ARM_MESI_Three_Level_HTM
protocol? If so is it possible to share the gem5 version and the running
command? I tried multiple versions and commands but it doesn't work and the
details are below.
First, the newest gem5 v21 changes some stats
Hi all,
Running an ARM cpu based system in Full System mode, I want to reset stats
during simulation using M5Ops " m5 reset_stats".
While almost all stat counters are resetted, some seem to keep there previous
values. Among them, all system.cpu.dcache stats.
Is this behavior expected? And if it
Hello everybody,
I'm trying to run an ARM full-system sImulation in gem5 using the following
parameters:
build/ARM/gem5.opt --outdir=/home/frank/gem5/m5out/test_16-04_umtp_issue
configs/example/fs.py --num-cpus=1 --sys-clock=2GHz --mem-size=2048MB
--caches --l2cache --cpu-type=DerivO3CPU
--kernel
Hi,
Everyone, I want to implement the FS mode in powerpc ISAs , and does anyone
do the same things? If you are just doing something related to it or
learning well about FS mode in any other ISAs , please give me some help!
I will be grateful!
mj
___
On Monday 24 November 2014 01:46:53 Thom Popovici via gem5-users wrote:
> Hello again.
>
> So I have tried making the caches as small as possible, therefore trying to
> remove the contention at the cache level. The command line that i hit was:
> ./build/X86/gem5.opt -d ./test1 ./configs/example/fs.
Hello again.
So I have tried making the caches as small as possible, therefore trying to
remove the contention at the cache level. The command line that i hit was:
./build/X86/gem5.opt -d ./test1 ./configs/example/fs.py -n 2
--cpu-type=detailed --caches --l1d_size=128B --l1i_size=128B --l1d_assoc=
I'm not questioning your motivations, just pointing out why we never
bothered to fix the problem even though we were aware of it :).
Steve
On Sat, Nov 22, 2014 at 11:39 AM, Thom Popovici
wrote:
> I agree with the fact that it is unrealistic not having caches and
> separated ones, however for wh
I agree with the fact that it is unrealistic not having caches and
separated ones, however for what I want to experiment I do not want to have
caches :). I can make the data cache and the instruction cache 64 B each
(the size equal with the size of a line) and hope for the best.
I did not use that
Have you tried running with the O3CPUAll debug flag? That may shed some
more light on whats happening. Steve's suggestion sounds like a
possibility.
On Sat, Nov 22, 2014 at 12:24 PM, Steve Reinhardt via gem5-users <
gem5-users@gem5.org> wrote:
> I don't recall the details, but there's some issu
I don't recall the details, but there's some issue with data accesses and
instruction fetches sharing the same port to memory with O3 that leads to a
livelock (or deadlock?) situation... something like you need to do an
ifetch to make forward progress, but every time the port is free you
re-issue a
Hi, all!
I want to remove the caches from the full system configuration. I want
multiple processors that are directly connected to the membus and the
memory. As far as I saw the O3CPU demands caches, why? I managed to modify
the function request_caches from True to False. I managed to instantia
Hello,
I want to run some *CPU intensive benchmark in FS mode on ARM* in gem5.
The intent is to first make sure that DVFS implementation and
configuration is functional followed by evaluation of DVFS policies. I
had a look at SPEC benchmark, BBechmark pages but facing issues while
running the
Hi all,
Don't mistake this for hassling, but I really want to learn about
ruby.stats file.
*
All Non-Zero Cycle Demand Cache Accesses
miss_latency: [binsize: 2 max: 216 count: 136446636 average: 3.45033 |
standard deviation: 5.75141 | 0 134373207 0 0 0 0 0
Hi all,
I am trying to get the specific cache & memory lines accessed by an
application. I am using ruby fs mode and my command is;
./build/X86/gem5.fast configs/example/ruby_fs.py --num-cpus="8"
--kernel="x86_64-vmlinux-2.6.22.9" --debug-flags=Cache --l1i_size="32kB"
--l1d_size="32kB" --l2_size=
Yes it does. Normally people take a checkpoint after Linux boots and they start
their application of interest. When you restore from the checkpoint you'll only
get stats from the code that is executed after that. You can also use the m5
binary on the disk image (m5 resetstats m5 dumpresetstats
Never mind! I just found the answer and is YES, gem5 includes
linux boot in statistics.
Pavlos
On Sat, 2012-10-13 at 02:29 +0300, Pavlos Maniotis wrote:
> Hello everyone,
>
> Does anybody know when exactly gem5 starts collecting statistics?
> I wonder if it includes linux boot process or not.
>
Hello everyone,
Does anybody know when exactly gem5 starts collecting statistics?
I wonder if it includes linux boot process or not.
Thanks in advance,
Pavlos
___
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/ge
Hi,Tony,
A further question about .rcS script
Is there any special commands of running SPEC CPU2000 in FS mode?
I only got some reference of Syscall Emulation Mode Command Lines
Also, if possible, can you share an example of rcS script you created for
running SPEC in FS mode.
Thank you very much
Got it.
Thank you.
On Wed, Jun 20, 2012 at 11:03 PM, Anthony Gutierrez wrote:
> That's just an example. You can put the binary and input anywhere you
> like. And run it in the script the same way you'd run it from the command
> line.
>
>
> On Wed, Jun 20, 2012 at 10:56 PM, Bojun Ma wrote:
>
>> A
That's just an example. You can put the binary and input anywhere you like.
And run it in the script the same way you'd run it from the command line.
On Wed, Jun 20, 2012 at 10:56 PM, Bojun Ma wrote:
> A question of ammp.rcS
>
> #!/bin/sh
>
> cd /benchmarks/spec/ammp00/
>
>
> /sbin/m5 checkpoint
A question of ammp.rcS
#!/bin/sh
cd /benchmarks/spec/ammp00/
/sbin/m5 checkpoint 0 0
/sbin/m5 checkpoint 1 2
/sbin/m5 loadsymbol
/sbin/m5 resetstats
./ammp < input/mdred.in
/sbin/m5 exit
--
"./ammp < in
look at the ammp.rcS file that already exists in configs/boot.
I would put the binary and input in some place designated for benchmarks.
Perhaps a directory called "benchmarks" within the home directory, then cd
into that directory from your script before you attempt to launch the
binary. Also you
Any suggestion of how to write .rcS file.
I am not familiar with that. I just make a little modification of the
ammp.rcS script in configs/boot.
By the way is that right to copy the benchmark to the mount point of the
disk image?
Regards,
Bojun
On Wed, Jun 20, 2012 at 10:36 PM, Anthony Gutierrez
usage: /sbin/m5 exit [delay]
/sbin/m5 resetstats [delay [period]]
/sbin/m5 dumpstats [delay [period]]
/sbin/m5 dumpresetstats [delay [period]]
/sbin/m5 readfile [filename]
/sbin/m5 execfile
/sbin/m5 checkpoint [delay [period]]
/sbin/m5 loadsymbol
Here is the system.terminal file. (in attachment)
It seems cannot open the input file of ammp.
Yes I compile it statically.
I mean I put it in the /mnt of my Ubuntu OP which is the mount point I used
for the disk image.
The Structure is like this:
bojun@bojun-desktop:/mnt$ ls
164.gzip bin dev
The stats file isn't really useful here. What do the simulator and
system.terminal output say? Is it exiting because of m5 exit? If so I have
a feeling that your binary isn't able to execute. Did you compile it
statically? Why are you placing it in the /mnt directory of the disk image?
-Tony
On W
Hi, All,
I am running SPEC CPU2000 benchmark in ARM FS mode.(ammp)
I copy ammp binary and input file to the /mnt of the disk image which
download from the gem5 website.
My .rcS file:
#!/bin/sh
/sbin/m5 checkpoint 0 0
/sbin/m5 checkpoint 1 2
/sbin/m5 loadsymb
28 matches
Mail list logo
