Differences

documentation:iiswc2010_tutorial_flexus [2010/11/28 13:43] mferdman		documentation:iiswc2010_tutorial_flexus [2010/12/03 22:25] (current) mferdman
Line 88:		Line 88:
	* Hit ESC and type **:wq** to save the file and exit.		* Hit ESC and type **:wq** to save the file and exit.
	* Type **mkdir /host**		* Type **mkdir /host**
-	* This is usually a good time to save out a checkpoint right before you mount the host file system.  At the Simics console, type **CTRL-C** followed by something like **write-configuration <ckpt_dir>/<your_checkpoint_name_b4_sfsmount>**	+	* This is usually a good time to save out a checkpoint right before you mount the host file system.  At the Simics console, type **CTRL-C** followed by something like **write-configuration ~/images/b4_sfsmount**
	* Type **run** at the Simics console to resume.		* Type **run** at the Simics console to resume.
	* Within the simulated console, type **mount /host**		* Within the simulated console, type **mount /host**
	* Type **ls /host** to see the underlying host machine's root directory		* Type **ls /host** to see the underlying host machine's root directory
-	At this point, you should copy microbenchmark files from **~/tutorial_files/microbenchmarks** into the target machine (by copying it from **/host** to a location on the simulated disk).  Save out a NEW checkpoint called **~/images/benchloaded** and quit out of Simics. Now open the checkpoint you saved with vi by typing **vi ~/images/benchloaded** and locate and delete the following lines:	+	At this point, you should copy microbenchmark files from **/host/home/pf_user/tutorial_files/microbenchmarks** into the target machine (by copying it from **/host** to a location on the simulated disk).  Save out a NEW checkpoint called **~/images/benchloaded** and quit out of Simics. Now open the checkpoint you saved with vi by typing **vi ~/images/benchloaded** and locate and delete the following lines:
	<code>		<code>
Line 109:		Line 109:
	- You can see how the source code inserts the magic instructions by looking at **spinlock.c**		- You can see how the source code inserts the magic instructions by looking at **spinlock.c**
-	- Create a new Simics script called break.simics and fill it in with this:	+	- Create a new Simics script called break.simics and fill it in with this: (this should already be available for you under ~/simics-3.0.22/targets/serengeti)
	<code>		<code>
	@def hap_callback(user_arg, cpu, arg):		@def hap_callback(user_arg, cpu, arg):
Line 122:		Line 122:
	</code>		</code>
-	- Launch Simics by typing **start-simics break.simics**	+	- Launch Simics by typing **../../scripts/start-simics break.simics**
	- Within the simulated console, navigate to the directory where you copied over the microbenchmark files.		- Within the simulated console, navigate to the directory where you copied over the microbenchmark files.
	- Type: **./spinlock 4 1000000000 10 10 0**  (this indicates we want 4 threads and run for effectively an infinite number of iterations)		- Type: **./spinlock 4 1000000000 10 10 0**  (this indicates we want 4 threads and run for effectively an infinite number of iterations)
Line 128:		Line 128:
	- Type **run** again and wait until the first thread starts to execute and triggers the magic breakpoint		- Type **run** again and wait until the first thread starts to execute and triggers the magic breakpoint
	- Save a final checkpoint by typing **write-configuration ~/images/spinlock**		- Save a final checkpoint by typing **write-configuration ~/images/spinlock**
		+	- **FINAL STEP** (to prepare the checkpoint we will be using in the ProtoFlex part of the tutorial).  This final step is needed to maximize the performance of the underlying simulated I/O system. Simics is typically the initiator of DMA transactions, which occur at some bulk-sized granularity.  This granularity is set by default to a very low value (64 Bytes) in default Simics checkpoints.  Since Simics is a software-based simulator, issuing many small bulk transfers imposes no simulation overhead.  In our system, large bulk transfers are far more desirable. To change this default setting, you will need to **EDIT** the checkpoint file and make one small change. Copy theType the following commands:
		+	<code>
		+	write-configuration ~/checkpoints/final
		+	quit
		+	perl -pi -e 's/dma_block_size: 64/dma_block_size: 8192/' ~/checkpoints/final
		+	</code>
	\\		\\
Line 134:		Line 140:
	======3. Working with Flexus======		======3. Working with Flexus======
-	From the workload we just created, you will get to chance to run some sample jobs with Flexus and create a Flexpoint library. By this point you should have a valid initial checkpoint stored as **~/images/spinlock**.	+	From the simics checkpoint you just created, you will get a chance to run some sample jobs with Flexus. By this point you should have a valid initial checkpoint stored as **~/images/spinlock**.
-	- Before starting, you should create a few initial directories in the home (which we will explain in the next steps):<code>	+	- Before starting, you should have a few initial directories in the home (which we will explain in the next steps):<code>
-	mkdir ~/ckpts	+	~/ckpts
-	mkdir ~/images	+	~/specs
-	mkdir ~/specs	+
	</code>		</code>
-	- The flexus simulator is stored as ~/tutorial_files/flexus_tutorial.tgz.  Copy this file to your home directory and extract the tarball.  You should have a directory called ~/flexus.	+	- We created the **ckpts** and **specs** directory in your home.
		+	- The flexus simulator is stored as ~/tutorial_files/flexus_v4.
	====Getting familiar with the run_job script====		====Getting familiar with the run_job script====
Line 147:		Line 153:
	++++CLICK - Expand/Collapse|		++++CLICK - Expand/Collapse|
-	The run_job script should be run from the ~/tutorial_files/flexus_v4 directory and requires for the home directory of the user to contain a .run_job.rc.tcl file.  Additionally, a **~/specs** directory must contain at least an interactive job configuration.	+	The run_job script should be run from the ~/tutorial_files/flexus_v4 directory and requires for the home directory of the user to contain a .run_job.rc.tcl file.  Additionally, a **~/specs** directory must contain at least a job configuration (We placed a job configuration in this directory).
	- Copy the example RC file from **~/tutorial_files/flexus_v4/scripts/.run_job.rc.tcl** into **~/**		- Copy the example RC file from **~/tutorial_files/flexus_v4/scripts/.run_job.rc.tcl** into **~/**
-	- Create a ~/specs/interactive/ directory and place a user-preload.simics file there (empty file is OK for the tutorial)	+	- Execute the **run_job** script from the ~/tutorial_files/flexus_v4 directory to confirm correct setup (the command-line help will be displayed when the prerequisites are met)
-	- Execute the run_job script from the ~/flexus/ directory to confirm correct setup (the command-line help will be displayed when the prerequisites are met)	+
	The .run_job.rc.tcl file contains "rungen" sections with directives for each workload.  When executing run_job, the rungen is selected with the "-run" parameter.  Typical rungens are "phase" for phase generation, "flexpoint" for flexpoint generation, "trace" for functional simulation jobs, and "timing" for the detailed cycle-accurate simulations.		The .run_job.rc.tcl file contains "rungen" sections with directives for each workload.  When executing run_job, the rungen is selected with the "-run" parameter.  Typical rungens are "phase" for phase generation, "flexpoint" for flexpoint generation, "trace" for functional simulation jobs, and "timing" for the detailed cycle-accurate simulations.
	The run_job script has already been configured for you.		The run_job script has already been configured for you.
-	* Take a look at the various paths and options that are specified in it by examining the ~/flexus/scripts/global.run_job.rc.tcl file.	+	* Take a look at the various paths and options that are specified in it by examining the ~/tutorial_files/flexus_v4/scripts/global.run_job.rc.tcl file.
	Flexus scripts expect a specific directory hierarchy for the checkpoints.		Flexus scripts expect a specific directory hierarchy for the checkpoints.
Line 166:		Line 171:
	* load the initial checkpoint in Simics (using the **start-simics** script)		* load the initial checkpoint in Simics (using the **start-simics** script)
	* simics> **read-configuration ~/images/spinlock**		* simics> **read-configuration ~/images/spinlock**
-	* simics> **run-command-file ~/flexus/scripts/create_mem_and_io_proxy.simics**	+	* simics> **run-command-file ~/tutorial_files/flexus_v4/scripts/mem_io_proxy.simics**
	* simics> **write-configuration ~/ckpts/spinlock/baseline/phase_000/simics/phase_000**		* simics> **write-configuration ~/ckpts/spinlock/baseline/phase_000/simics/phase_000**
	To verify that the basic run_job settings are correct and that the spinlock workload is properly set up, use run_job to launch Simics with the spinlock workload (NONE indicates that no Flexus simulator library should be loaded):		To verify that the basic run_job settings are correct and that the spinlock workload is properly set up, use run_job to launch Simics with the spinlock workload (NONE indicates that no Flexus simulator library should be loaded):
-	* **~/flexus/scripts/run_job NONE spinlock**  (error message about "flexus" missing is OK)	+	* **~/tutorial_files/flexus_v4/scripts/run_job NONE spinlock**  (error message about "flexus" missing is OK)
	Add configuration for the "spinlock" benchmark to the "trace" rungen of ~/.run_job.rc.tcl		Add configuration for the "spinlock" benchmark to the "trace" rungen of ~/.run_job.rc.tcl
Line 176:		Line 181:
	* configure statistics region interval at 50000000 (50M) cycles		* configure statistics region interval at 50000000 (50M) cycles
-	Run a "spinlock" trace job with TraceCMPFlex.	+	Run a "spinlock" trace job with CMP.L2Shared.Trace
-	* Example trace configuration can be found in the **scripts/trace/user-*load.simics** files.	+	* **run_job -run trace -cfg 4cores -local CMP.L2Shared.Trace spinlock**
-	* **~/flexus/scripts/run_job -run trace -cfg test_cfg_trace -local TraceCMPFlex spinlock**	+	* Explanation of "local": -local requests to run a batch of jobs locally.  without -local an interactive run is assumed which waits at the simics> prompt instead of running.
-	* Explanation of "local": -local requests to run a batch of jobs locally.  without -local an interactive run is assumed which waits at the simics> prompt instead of running.	+	* Explanation of "remote": -remote will submit jobs to a remote cluster (e.g., Condor, PBS, etc...) [not available for the tutorial].
-	* Explanation of "remote": -remote will submit jobs to a remote cluster (e.g., Condor, PBS, etc...) [not available for the tutorial].	+	* You can **run** simulation, interrupt it with **ctrl+c**, and change debug severity with **flexus.debug-set-severity iface**.
	++++		++++
-	====Displaying statistics through the stat-manager tool====	+	====Displaying statistics with the stat-manager tool====
	++++CLICK - Expand/Collapse|		++++CLICK - Expand/Collapse|
	Find the run directory for the trace job in ~/results/ and examine the resulting statistics database:		Find the run directory for the trace job in ~/results/ and examine the resulting statistics database:
-	* **~/flexus/stat-manager/stat-manager list-measurements**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager list-measurements**
	* See the cache hit/miss statistics, branch predictor stats, and instruction mix breakdown.		* See the cache hit/miss statistics, branch predictor stats, and instruction mix breakdown.
-	* **~/flexus/stat-manager/stat-manager print "Region 000" | less**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager print 'Region 000' | less**
-	* **~/flexus/stat-manager/stat-manager print "Region 001" | less**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager print 'Region 001' | less**
	* By default, stat-manager aggregates statistics across all cores.  You can override this behavior with the -per-node flag.		* By default, stat-manager aggregates statistics across all cores.  You can override this behavior with the -per-node flag.
-	* **~/flexus/stat-manager/stat-manager -per-node print "Region 001" | less**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager -per-node print 'Region 001' | less**
	++++		++++
		+
		+	====Running timing simulations====
		+	++++CLICK - Expand/Collapse|
		+	Run a "spinlock" timing job with CMP.L2SharedNUCA.OoO
		+	* **run_job -run timing -cfg 4cores -ma CMP.L2SharedNUCA.OoO spinlock**
		+	* NOTE: When running timing simulations, one must pass the **-ma** parameter to Simics.
		+	* You can **run** simulation, interrupt it with **ctrl+c**, and change debug severity with **flexus.debug-set-severity iface**, **run 10** will run 10 cycles on all CPUs.
		+	* Rebuild the simulator with **vverb** debug output (CMP.L2SharedNUCA.OoO-vverb) and try running simulation with **flexus.debug-set-severity vverb** to see the detailed debug output.
		+	++++
		+
		+	======4. Using Statistical Sampling with Flexus======
	====Creating a flexpoint library====		====Creating a flexpoint library====
Line 199:		Line 215:
	++++CLICK - Expand/Collapse|		++++CLICK - Expand/Collapse|
	* Configure the "flexpoint" rungen in .run_job.rc.tcl to create 20 flexpoints, spaced 200000 (200K) instructions apart.		* Configure the "flexpoint" rungen in .run_job.rc.tcl to create 20 flexpoints, spaced 200000 (200K) instructions apart.
-	* **~/flexus/scripts/run_job -ckpt-gen -postprocess "$HOME/flexus/scripts/postprocess_ckptgen.sh flexpoint 20 mystate" -local -cfg test_cfg_trace -run flexpoint TraceCMPFlex spinlock**	+	* **~/tutorial_files/flexus_v4/scripts/run_job -ckpt-gen -postprocess "$HOME/tutorial_files/flexus_v4/scripts/postprocess_ckptgen.sh flexpoint 20 mystate" -local -cfg 4cores -run flexpoint CMP.L2Shared.Trace spinlock**
	* **-ckpt-gen** ensures that state is written out at the end of simulation		* **-ckpt-gen** ensures that state is written out at the end of simulation
	* **-postprocess** specifies the script to run after each job		* **-postprocess** specifies the script to run after each job
Line 211:		Line 227:
	++++CLICK - Expand/Collapse|		++++CLICK - Expand/Collapse|
	Add configuration for the "spinlock" benchmark to the "timing" rungen of ~/.run_job.rc.tcl		Add configuration for the "spinlock" benchmark to the "timing" rungen of ~/.run_job.rc.tcl
-	* configure simulation to stop at 15000 (15K) cycles	+	* configure simulation to stop at 150000 (150K) cycles
-	* configure statistics region interval at 5000 (5K) cycles	+	* configure statistics region interval at 50000 (50K) cycles
-	Run a "spinlock" timing job with CMPFlex.OoO.	+	Run a "spinlock" timing job with CMP.L2SharedNUCA.OoO.
-	* Example timing configuration can be found in the scripts/timing_v9/user-*load.simics files.	+	* **~/tutorial_files/flexus_v4/scripts/run_job -run timing -cfg 4cores -local -ma -state mystate CMP.L2SharedNUCA.OoO spinlock**
-	* **~/flexus/scripts/run_job -run timing -cfg test_cfg_timing -local -ma -state mystate CMPFlex.OoO spinlock**	+
	* NOTE: When running timing simulations, one must pass the **-ma** parameter to Simics.		* NOTE: When running timing simulations, one must pass the **-ma** parameter to Simics.
	* NOTE: Don't forget to specify **-state** to load the microarchitectural state created with the trace simulator, otherwise each flexpoint is run from cold microarchitectural state, severely biasing the results!		* NOTE: Don't forget to specify **-state** to load the microarchitectural state created with the trace simulator, otherwise each flexpoint is run from cold microarchitectural state, severely biasing the results!
Line 223:		Line 238:
	* Notice much more detailed statistics for timing simulator compared to the trace simulator.		* Notice much more detailed statistics for timing simulator compared to the trace simulator.
	* Find the IPC of some of the flexpoints' results using stat-manager:		* Find the IPC of some of the flexpoints' results using stat-manager:
-	* **~/flexus/stat-manager/stat-manager format-string "<EXPR:{Nodes-uarch-TB:User:Commits:Busy}/({Nodes-uarch-TB:User:AccountedCycles}+{Nodes-uarch-TB:System:AccountedCycles})>" "Region 001"**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager format-string "<EXPR:{Nodes-uarch-TB:User:Commits:Busy}/({Nodes-uarch-TB:User:AccountedCycles}+{Nodes-uarch-TB:System:AccountedCycles})>" "Region 001"**
	The default postprocess.sh script (which runs after each job if a -postprocess override is not specified) automatically creates a stats_db.out.selected.gz file that contains only statistics between 100K and 150K instructions.		The default postprocess.sh script (which runs after each job if a -postprocess override is not specified) automatically creates a stats_db.out.selected.gz file that contains only statistics between 100K and 150K instructions.
	Use stat-sample to combine all the stats_db.out.selected.gz files into a single statistics file.		Use stat-sample to combine all the stats_db.out.selected.gz files into a single statistics file.
-	* **~/flexus/stat-manager stat-sample stats_db.out.gz */stats_db.out.selected.gz**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-sample stats_db.out.gz */stats_db.out.selected.gz**
	* Examine the resulting stats_db.out.gz file that contains the combined results of all flexpoints.		* Examine the resulting stats_db.out.gz file that contains the combined results of all flexpoints.
	* Examine the IPCs of the various flexpoints:		* Examine the IPCs of the various flexpoints:
Line 234:		Line 249:
	* If bringing UIPCs into Excel, compute =STDEV() and =CONFIDENCE() for 95% confidence.		* If bringing UIPCs into Excel, compute =STDEV() and =CONFIDENCE() for 95% confidence.
	* Bring time Breakdowns into Excel:		* Bring time Breakdowns into Excel:
-	* **~/flexus/stat-manager/stat-manager print sum | grep ":Bkd:" > breakdown.tsv**	+	* **~/tutorial_files/flexus_v4/stat-manager/stat-manager print sum | grep ":Bkd:" > breakdown.tsv**
	* Use Excel to split data by the colon (":") character into columns.  Apply the Pivot Chart feature to plot the time breakdown.		* Use Excel to split data by the colon (":") character into columns.  Apply the Pivot Chart feature to plot the time breakdown.
	\\		\\