Merge pull request #72 from VijayKandiah/release-accelwattch

AccelWattch v1.0 Integration

LICENSE

@@ -1,25 +1,30 @@
 BSD 2-Clause License
 
-Copyright (c) 2020, Tim Rogers
+Copyright (c) 2018-2021, Vijay Kandiah, Junrui Pan, Mahmoud Khairy, Scott Peverelle, Timothy Rogers, Tor M. Aamodt, Nikos Hardavellas
+Northwestern University, Purdue University, The University of British Columbia
 All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions are met:
 
-* Redistributions of source code must retain the above copyright notice, this
-  list of conditions and the following disclaimer.
-
-* Redistributions in binary form must reproduce the above copyright notice,
-  this list of conditions and the following disclaimer in the documentation
-  and/or other materials provided with the distribution.
+1. Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer;
+2. Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution;
+3. Neither the names of Northwestern University, Purdue University,
+   The University of British Columbia nor the names of their contributors
+   may be used to endorse or promote products derived from this software
+   without specific prior written permission.
 
 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
-DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
-FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
-DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
-SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
-CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
-OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
+LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGE.

README.md

@@ -1,4 +1,4 @@
-# Welcome to the top-level repo of Accel-Sim
+# Welcome to the top-level repo of Accel-Sim and AccelWattch
 
 The [ISCA 2020 paper](https://conferences.computer.org/isca/pdfs/ISCA2020-4QlDegUf3fKiwUXfV0KdCm/466100a473/466100a473.pdf)
 describes the goals of Accel-Sim and introduces the tool. This readme is meant to provide tutorial-like details on how to use the Accel-Sim
@@ -10,6 +10,16 @@ Accel-Sim: An Extensible Simulation Framework for Validated GPU Modeling,
 in 2020 ACM/IEEE 47th Annual International Symposium on Computer Architecture (ISCA)
 ```
 
+This repository also includes AccelWattch: A Power Modeling Framework for Modern GPUs. The [MICRO 2021 paper](http://paragon.cs.northwestern.edu/papers/2021-MICRO-AccelWattch-Kandiah.pdf) introduces AccelWattch. Please look at our [AccelWattch MICRO'21 Artifact Manual](https://github.com/accel-sim/accel-sim-framework/blob/release-accelwattch/AccelWattch.md) for detailed information on various AccelWattch components. For information on just running AccelWattch, please look at the [AccelWattch Overview](https://github.com/accel-sim/accel-sim-framework/blob/release-accelwattch/README.md#accelwattch-overview) section in this read-me.
+If you use any component of AccelWattch, please cite:
+
+```
+Vijay Kandiah, Scott Peverelle, Mahmoud Khairy, Amogh Manjunath, Junrui Pan, Timothy G. Rogers, Tor Aamodt, Nikos Hardavellas,
+AccelWattch: A Power Modeling Framework for Modern GPUs,
+in 2021 IEEE/ACM International Symposium on Microarchitecture (MICRO)
+```
+
+
 ## Dependencies
 
 This package is meant to be run on a modern linux distro.
@@ -21,7 +31,8 @@ run:
 sudo apt-get install  -y wget build-essential xutils-dev bison zlib1g-dev flex \
       libglu1-mesa-dev git g++ libssl-dev libxml2-dev libboost-all-dev git g++ \
       libxml2-dev vim python-setuptools python-dev build-essential python-pip
-pip install pyyaml==5.1 plotly psutil
+
+pip3 install pyyaml plotly psutil
 wget http://developer.download.nvidia.com/compute/cuda/11.0.1/local_installers/cuda_11.0.1_450.36.06_linux.run
 sh cuda_11.0.1_450.36.06_linux.run --silent --toolkit
 rm cuda_11.0.1_450.36.06_linux.run
@@ -30,23 +41,25 @@ rm cuda_11.0.1_450.36.06_linux.run
 Note, that all the python scripts have more detailed options explanations when run with "--help"
 
 
-## Accel-Sim Repo Overview
+## Overview
 
-The code for the Accel-Sim framework is in this repo. Accel-Sim 1.0 uses the
-GPGPU-Sim 4.0 performance model, which was released as part of the original
+The code for the Accel-Sim and AccelWattch frameworks are in this repo. Accel-Sim 1.0 uses the
+[GPGPU-Sim 4.0](https://github.com/accel-sim/accel-sim-framework/blob/dev/gpu-simulator/gpgpu-sim4.md) performance model, which was released as part of the original
 Accel-Sim paper. Building the trace-based Accel-Sim will pull the right version of
-GPGPU-Sim 4.0 to use in Accel-Sim.
+GPGPU-Sim 4.0 and the AccelWattch power model to use in Accel-Sim. AccelWattch replaces the GPUWattch power model in GPGPU-Sim 4.0.
 
 There is an additional repo where we have collected a set of common GPU applications and a common infrastructure for building
 them with different versions of CUDA. If you use/extend this app framework, it makes Accel-Sim easily usable
 with a few simple command lines. The instructions in this README will take you through how to use Accel-Sim with
-the apps in from this collection as well as just on your own, with your own apps.
+the apps in from this collection as well as just on your own, with your own apps.  
 
 [GPU App Collection](https://github.com/accel-sim/gpu-app-collection)
 
+The 'release-accelwattch' branch of the [GPU App Collection](https://github.com/accel-sim/gpu-app-collection/tree/release-accelwattch) repository also includes the AccelWattch microbenchmarks and AccelWattch validation set benchmarks. For more information on these benchmarks, please look at our [MICRO 2021 paper](http://paragon.cs.northwestern.edu/papers/2021-MICRO-AccelWattch-Kandiah.pdf) and [AccelWattch MICRO'21 Artifact Manual](https://github.com/accel-sim/accel-sim-framework/blob/release-accelwattch/AccelWattch.md).
+
 ## Accel-Sim Components
 
-![Accel-Sim Overview](https://accel-sim.github.io/assets/img/accel-sim-crop.svg)
+![Accel-Sim Overview](https://github.com/VijayKandiah/accel-sim.github.io/blob/master/assets/img/accel-sim-crop.svg)
 
 1. **Accel-Sim Tracer**: An NVBit tool for generating SASS traces from CUDA applications. Code for the tool lives in ./util/tracer\_nvbit/. To make the tool:  
 
@@ -89,7 +102,7 @@ the apps in from this collection as well as just on your own, with your own apps
     ```  
     and follow the instructions.  
 
-2. **Accel-Sim SASS Frontend**: A simulator frontend that consumes SASS traces and feeds them into a performance model. The intial release of Accel-Sim coincides with the release of GPGPU-Sim 4.0, which acts as the detailed performance model. To build the Accel-Sim simulator that uses the traces, do the following:
+2. **Accel-Sim SASS Frontend and Simulation Engine**: A simulator frontend that consumes SASS traces and feeds them into a performance model. The intial release of Accel-Sim coincides with the release of [GPGPU-Sim 4.0](https://github.com/accel-sim/accel-sim-framework/blob/dev/gpu-simulator/gpgpu-sim4.md), which acts as the detailed performance model. To build the Accel-Sim simulator that uses the traces, do the following:
     ```bash
     source ./gpu-simulator/setup_environment.sh
     make -j -C ./gpu-simulator/
@@ -103,6 +116,10 @@ the apps in from this collection as well as just on your own, with your own apps
     ```bash
     ./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-SASS -T ./hw_run/traces/device-<device-num>/<cuda-version>/ -N myTest
     ```
+    The above command will run the workloads in Accel-Sim's SASS traces-driven mode. You can also run the workloads in PTX mode using:
+    ```bash
+    ./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-PTX -N myTest-PTX
+    ```
     You can monitor the tests using:
     ```bash
     ./util/job_launching/monitor_func_test.py -v -N myTest
@@ -119,6 +136,8 @@ the apps in from this collection as well as just on your own, with your own apps
     However, we encourage you to use our workload launch manager 'run_simulations' script as shown above, which will greatly simplify the simulation process and increase productivity.
 
     To understand what is going on and how to just run the simulator in isolation without the framework, read [this](https://github.com/accel-sim/accel-sim-framework/tree/dev/util/job_launching/README.md).  
+
+    To better undersatnd the Accel-Sim front-end and the interface with GPGPU-Sim, read [this](https://github.com/accel-sim/accel-sim-framework/blob/dev/gpu-simulator/README.md).
 
 3. **Accel-Sim Correlator**: A tool that matches, plots and correlates statistics from the performance model with real hardware statistics generated by profiling tools. To use the correlator, you must first generate hardware output and simulation statistics. To generate output from the GPU, use the scripts in [./util/hw_stats](./util/hw_stats).
 For example, to generate the profiler numbers for the short-running apps in our running example, do the following:
@@ -175,8 +194,23 @@ For a true validation, you should attempt correlating the fully-scaled set of ap
 ```
 
 
-4. **Accel-Sim Tuner**: An automated tuner that automates configuration file generation from a detailed microbenchmark suite. To be released soon!
-
+4. **Accel-Sim Tuner**: An automated tuner that automates configuration file generation from a detailed microbenchmark suite. You need to provide a C header file `hw_def` that contains minimal information about the hardware model. This file is used to configure and tune the microbenchmarks for the unduerline hardware. See an example of Ampere RTX 3060 card [here](https://github.com/accel-sim/accel-sim-framework/blob/dev/util/tuner/GPU_Microbenchmark/hw_def/ampere_RTX3070_hw_def.h). Then, compile and run the microbenchmarks and the tuner:
+
+  ```bash
+  # Make sure PATH includes nvcc  
+  # If your hardware has new compute capability, ensure to add it in the /GPU_Microbenchmark/common/common.mk
+  # Compile microbenchmarks
+  make -C ./util/tuner/GPU_Microbenchmark/
+  # Set the device id that you want to tune to 
+  # If you do not know the device id, run ./tuner/GPU_Microbenchmark/bin/list_devices
+  export CUDA_VISIBLE_DEVICES=0  
+  # Run the ubench and save output in stats.txt
+  ./util/tuner/GPU_Microbenchmark/run_all.sh | tee stats.txt
+  # Run the tuner with the stats.txt from the previous step
+  ./util/tuner/tuner.py -s stats.txt
+  ```  
+
+  The tuner.py script will parse the microbenchmarks output and generate a folder with the same device name (e.g. "RTX_3060"). The folder will contain the config files for GPGPU-Sim performance model and Accel-Sim trace-driven front-end that matche and model the underline hardware as much as possible. For more detilas about the Accel-Sim tuner and the microbemcakring suite, read [this](https://github.com/accel-sim/accel-sim-framework/tree/dev/util/tuner#readme).
 
 
 ### How do I quickly just run what Travis runs?
@@ -209,3 +243,30 @@ Just use a linux ditro with the packages detailed in dependencies, set
 CUDA\_INSTALL\_PATH, the run ./travis.sh.
 
 
+## AccelWattch Overview
+
+![AccelWattch Overview](https://github.com/VijayKandiah/accel-sim.github.io/blob/master/assets/img/accelwattch-flowchart.svg)
+
+1. **Running AccelWattch SASS SIM**: To run *the simple example from bullet 1* with AccelWattch power estimations enabled using the *AccelWattch SASS SIM* model,
+```bash
+./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-Accelwattch_SASS_SIM -T ./hw_run/traces/device-<device-num>/<cuda-version>/ -N myTest
+```
+This will use the *AccelWattch SASS SIM* xml configuration file for the power model. The configuration files for the AccelWattch power model presented in our [MICRO 2021 paper](http://paragon.cs.northwestern.edu/papers/2021-MICRO-AccelWattch-Kandiah.pdf) can be found [here](https://github.com/accel-sim/gpgpu-sim_distribution/tree/release-accelwattch/configs/tested-cfgs/SM7_QV100). Please look at `./util/job_launching/configs/define-standard-cfgs.yml` for a list of provided AccelWattch configurations. The *AccelWattch HYBRID* configuration provided there uses activity factors for L2 and NOC from Accel-Sim and the rest from hardware performance counters. You can create your own *AccelWattch HYBRID* configuration in this file with a different mix of AccelWattch activity factors from Accel-Sim and hardware execution. 
+Upon completion of simulations, AccelWattch power estimations are stored in a *accelwattch_power_report.log* in a per-kernel format in the run directory. 
+
+2. **Running AccelWattch HW or AccelWattch HYBRID:** To run *the simple example from bullet 1* with *AccelWattch HW* or *AccelWattch HYBRID* configurations, 
+```bash
+./util/job_launching/run_simulations.py -B rodinia_2.0-ft -a -C <QV100-Accelwattch_SASS_HW or QV100-Accelwattch_SASS_HYBRID> -T ./hw_run/traces/device-<device-num>/<cuda-version>/ -N myTest
+```
+Note that *AccelWattch HW* and *AccelWattch HYBRID* configurations require hardware performance counter information for the target application stored in a *hw_perf.csv* file in the run directory. A sample *hw_perf.csv* file with performance counter information collected from a QV100 card for validation suite benchmarks used in our [MICRO 2021 paper](http://paragon.cs.northwestern.edu/papers/2021-MICRO-AccelWattch-Kandiah.pdf) is copied over to the run directory by default with the above *run_simulations.py* command. The *-a* argument for *run_simulations.py* is used to feed the application name to AccelWattch. Please make sure that there is a hardware performance counter information entry with the same application name in *hw_perf.csv* for AccelWattch to obtain activity factors from. Please look at example entries in the provided `./util/accelwattch/accelwattch_hw_profiler/hw_perf.csv`. 
+
+3. **Running AccelWattch PTX SIM**: To run *the simple example from bullet 1* with AccelWattch power estimations enabled using the *AccelWattch PTX SIM* model,
+```bash
+./util/job_launching/run_simulations.py -B rodinia_2.0-ft -C QV100-Accelwattch_PTX_SIM -N myTest
+```
+
+4. **Hardware Power and Performance Profiler**: The AccelWattch hardware profiler scripts are located at `./util/accelwattch/accelwattch_hw_profiler/` in this repository. For more information on how to use them, please look at [this](https://github.com/accel-sim/accel-sim-framework/blob/release-accelwattch/AccelWattch.md#hardware-profiling-for-accelwattch-validation) section in our MICRO'21 Artifact Manual.
+
+5. **Microbenchmarks and Quadratic Optimization Solver**: The source code for the microbenchmarks used for AccelWattch dynamic power modeling are located [here](https://github.com/accel-sim/gpu-app-collection/tree/release-accelwattch/src/cuda/accelwattch-ubench) and can be compiled by following the README [here](https://github.com/accel-sim/gpu-app-collection/tree/release-accelwattch). The Quadratic Optimization Solver MATLAB script is located at `./util/accelwattch/quadprog_solver.m`.
+
+6. **SASS to Power Component Mapping**: The header file `gpu-simulator/ISA_Def/accelwattch_component_mapping.h` contains the Accel-Sim instruction opcode to AccelWattch power component mapping and can be extended to support new SASS instructions for future architectures. Please look at the *opcode.h* files for respective GPU Architectures in the same directory `gpu-simulator/ISA_Def/` for SASS instruction to Accel-Sim opcode mapping.

gpu-simulator/.gitignore

@@ -1,3 +1,3 @@
 bin/
 build/
-gpgpu-sim/
+gpgpu-sim/