Computer Architecture analysis has an extended historical past of growing simulators and instruments to guage and form the design of computer programs. For instance, the SimpleScalar simulator was launched within the late Nineteen Nineties and allowed researchers to discover varied microarchitectural concepts. Computer architecture simulators and instruments, reminiscent of gem5, DRAMSys, and lots of extra have performed a big position in advancing computer architecture analysis. Since then, these shared sources and infrastructure have benefited trade and academia and have enabled researchers to systematically construct on one another’s work, resulting in important advances within the subject.
Nonetheless, computer architecture analysis is evolving, with trade and academia turning in direction of machine learning (ML) optimization to satisfy stringent domain-specific necessities, reminiscent of ML for computer architecture, ML for TinyML acceleration, DNN accelerator datapath optimization, reminiscence controllers, energy consumption, safety, and privateness. Although prior work has demonstrated the advantages of ML in design optimization, the shortage of robust, reproducible baselines hinders truthful and goal comparability throughout completely different strategies and poses a number of challenges to their deployment. To guarantee regular progress, it’s crucial to know and sort out these challenges collectively.
To alleviate these challenges, in “ArchGym: An Open-Source Gymnasium for Machine Learning Assisted Architecture Design”, accepted at ISCA 2023, we launched ArchGym, which incorporates quite a lot of computer architecture simulators and ML algorithms. Enabled by ArchGym, our outcomes point out that with a sufficiently giant variety of samples, any of a various assortment of ML algorithms are able to find the optimum set of architecture design parameters for every goal drawback; nobody resolution is essentially higher than one other. These outcomes additional point out that deciding on the optimum hyperparameters for a given ML algorithm is important for discovering the optimum architecture design, however selecting them is non-trivial. We launch the code and dataset throughout a number of computer architecture simulations and ML algorithms.
Challenges in ML-assisted architecture analysis
ML-assisted architecture analysis poses a number of challenges, together with:
- For a particular ML-assisted computer architecture drawback (e.g., discovering an optimum resolution for a DRAM controller) there isn’t any systematic technique to determine optimum ML algorithms or hyperparameters (e.g., learning charge, warm-up steps, and so forth.). There is a wider vary of ML and heuristic strategies, from random stroll to reinforcement learning (RL), that may be employed for design area exploration (DSE). While these strategies have proven noticeable efficiency enchancment over their selection of baselines, it isn’t evident whether or not the enhancements are due to the selection of optimization algorithms or hyperparameters.
Thus, to make sure reproducibility and facilitate widespread adoption of ML-aided architecture DSE, it’s mandatory to stipulate a scientific benchmarking methodology. - While computer architecture simulators have been the spine of architectural improvements, there’s an rising want to deal with the trade-offs between accuracy, pace, and value in architecture exploration. The accuracy and pace of efficiency estimation broadly varies from one simulator to a different, relying on the underlying modeling particulars (e.g., cycle-accurate vs. ML-based proxy fashions). While analytical or ML-based proxy fashions are nimble by advantage of discarding low-level particulars, they often undergo from excessive prediction error. Also, as a result of industrial licensing, there will be strict limits on the variety of runs collected from a simulator. Overall, these constraints exhibit distinct efficiency vs. pattern effectivity trade-offs, affecting the selection of optimization algorithm for architecture exploration.
It is difficult to delineate systematically examine the effectiveness of varied ML algorithms below these constraints. - Finally, the panorama of ML algorithms is quickly evolving and a few ML algorithms want information to be helpful. Additionally, rendering the result of DSE into significant artifacts reminiscent of datasets is essential for drawing insights concerning the design area.
In this quickly evolving ecosystem, it’s consequential to make sure amortize the overhead of search algorithms for architecture exploration. It isn’t obvious, nor systematically studied leverage exploration information whereas being agnostic to the underlying search algorithm.
ArchGym design
ArchGym addresses these challenges by offering a unified framework for evaluating completely different ML-based search algorithms pretty. It contains two most important elements: 1) the ArchGym atmosphere and a pair of) the ArchGym agent. The atmosphere is an encapsulation of the architecture price mannequin — which incorporates latency, throughput, space, power, and so forth., to find out the computational price of working the workload, given a set of architectural parameters — paired with the goal workload(s). The agent is an encapsulation of the ML algorithm used for the search and consists of hyperparameters and a guiding coverage. The hyperparameters are intrinsic to the algorithm for which the mannequin is to be optimized and might considerably affect efficiency. The coverage, alternatively, determines how the agent selects a parameter iteratively to optimize the goal goal.
Notably, ArchGym additionally features a standardized interface that connects these two elements, whereas additionally saving the exploration information because the ArchGym Dataset. At its core, the interface entails three most important alerts: {hardware} state, {hardware} parameters, and metrics. These alerts are the naked minimal to determine a significant communication channel between the atmosphere and the agent. Using these alerts, the agent observes the state of the {hardware} and suggests a set of {hardware} parameters to iteratively optimize a (user-defined) reward. The reward is a perform of {hardware} efficiency metrics, reminiscent of efficiency, power consumption, and so forth.
 |
| ArchGym contains two most important elements: the ArchGym atmosphere and the ArchGym agent. The ArchGym atmosphere encapsulates the fee mannequin and the agent is an abstraction of a coverage and hyperparameters. With a standardized interface that connects these two elements, ArchGym offers a unified framework for evaluating completely different ML-based search algorithms pretty whereas additionally saving the exploration information because the ArchGym Dataset. |
ML algorithms may very well be equally favorable to satisfy user-defined goal specs
Using ArchGym, we empirically show that throughout completely different optimization aims and DSE issues, a minimum of one set of hyperparameters exists that ends in the identical {hardware} efficiency as different ML algorithms. A poorly chosen (random choice) hyperparameter for the ML algorithm or its baseline can result in a deceptive conclusion {that a} specific household of ML algorithms is best than one other. We present that with enough hyperparameter tuning, completely different search algorithms, even random stroll (RW), are capable of determine the very best reward. However, word that discovering the best set of hyperparameters might require exhaustive search and even luck to make it aggressive.
 |
| With a enough variety of samples, there exists a minimum of one set of hyperparameters that ends in the identical efficiency throughout a variety of search algorithms. Here the dashed line represents the utmost normalized reward. Cloud-1, cloud-2, stream, and random point out 4 completely different reminiscence traces for DRAMSys (DRAM subsystem design area exploration framework). |
Dataset building and high-fidelity proxy mannequin coaching
Creating a unified interface utilizing ArchGym additionally allows the creation of datasets that can be utilized to design higher data-driven ML-based proxy architecture price fashions to enhance the pace of architecture simulation. To consider the advantages of datasets in constructing an ML mannequin to approximate architecture price, we leverage ArchGym’s capacity to log the information from every run from DRAMSys to create 4 dataset variants, every with a special variety of information factors. For every variant, we create two classes: (a) Diverse Dataset, which represents the information collected from completely different brokers (ACO, GA, RW, and BO), and (b) ACO solely, which exhibits the information collected completely from the ACO agent, each of that are launched together with ArchGym. We prepare a proxy mannequin on every dataset utilizing random forest regression with the target to foretell the latency of designs for a DRAM simulator. Our outcomes present that:
- As we improve the dataset measurement, the common normalized root imply squared error (RMSE) barely decreases.
- However, as we introduce range within the dataset (e.g., gathering information from completely different brokers), we observe 9× to 42× decrease RMSE throughout completely different dataset sizes.
 |
| Diverse dataset assortment throughout completely different brokers utilizing ArchGym interface. |
 |
| The impression of a various dataset and dataset measurement on the normalized RMSE. |
The want for a community-driven ecosystem for ML-assisted architecture analysis
While, ArchGym is an preliminary effort in direction of creating an open-source ecosystem that (1) connects a broad vary of search algorithms to computer architecture simulators in an unified and easy-to-extend method, (2) facilitates analysis in ML-assisted computer architecture, and (3) types the scaffold to develop reproducible baselines, there are plenty of open challenges that want community-wide help. Below we define a number of the open challenges in ML-assisted architecture design. Addressing these challenges requires a properly coordinated effort and a group pushed ecosystem.
 |
| Key challenges in ML-assisted architecture design. |
We name this ecosystem Architecture 2.0. We define the important thing challenges and a imaginative and prescient for constructing an inclusive ecosystem of interdisciplinary researchers to sort out the long-standing open issues in making use of ML for computer architecture analysis. If you have an interest in serving to form this ecosystem, please fill out the curiosity survey.
Conclusion
ArchGym is an open supply gymnasium for ML architecture DSE and allows an standardized interface that may be readily prolonged to go well with completely different use instances. Additionally, ArchGym allows truthful and reproducible comparability between completely different ML algorithms and helps to determine stronger baselines for computer architecture analysis issues.
We invite the computer architecture group in addition to the ML group to actively take part within the improvement of ArchGym. We imagine that the creation of a gymnasium-type atmosphere for computer architecture analysis can be a big step ahead within the subject and supply a platform for researchers to make use of ML to speed up analysis and result in new and progressive designs.
Acknowledgements
This blogpost relies on joint work with a number of co-authors at Google and Harvard University. We want to acknowledge and spotlight Srivatsan Krishnan (Harvard) who contributed a number of concepts to this venture in collaboration with Shvetank Prakash (Harvard), Jason Jabbour (Harvard), Ikechukwu Uchendu (Harvard), Susobhan Ghosh (Harvard), Behzad Boroujerdian (Harvard), Daniel Richins (Harvard), Devashree Tripathy (Harvard), and Thierry Thambe (Harvard). In addition, we’d additionally prefer to thank James Laudon, Douglas Eck, Cliff Young, and Aleksandra Faust for their help, suggestions, and motivation for this work. We would additionally prefer to thank John Guilyard for the animated determine used on this put up. Amir Yazdanbakhsh is now a Research Scientist at Google DeepMind and Vijay Janapa Reddi is an Associate Professor at Harvard.