Abstract We present a novel deep neural network (DNN) training scheme and resistive RAM (RRAM) inmemory computing (IMC) hardware evaluation towards achieving high accuracy against RRAM device/array variations and enhanced robustness against adversarial input attacks. We present improved IMC inference accuracy results evaluated on stateoftheart DNNs including ResNet18, AlexNet, and VGG with binary, 2bit, and 4bit activation/weight precision for the CIFAR10 dataset. These DNNs are evaluated with measured noise data obtained from three different RRAMbased IMC prototype chips. Across these various DNNs and IMC chip measurements, we show that our proposed hardware noiseaware DNN training consistently improves DNN inference accuracy for actual IMC hardware, up to 8% accuracy improvement for the CIFAR10 dataset. We also analyze the impact of our proposed noise injection scheme on the adversarial robustness of ResNet18 DNNs with 1bit, 2bit, and 4bit activation/weight precision. Our results show up to 6% improvement in the robustness to blackbox adversarial input attacks.
more »
« less
This content will become publicly available on September 1, 2024
FPIMC: A 28nm AllDigital Configurable FloatingPoint InMemory Computing Macro
Inmemory computing (IMC) provides energy efficient solutions to deep neural networks (DNN). Most IMC de signs for DNNs employ fixedpoint precisions. However, floating point precision is still required for DNN training and complex inference models to maintain high accuracy. There have not been floatpoint precision based IMC works in the literature where the floatpoint computation is immersed into the weight memory storage. In this work, we propose a novel floatingpoint precision IMC macro with a configurable architecture that supports both normal 8bit floating point (FP8) and 8bit block floating point (BF8) with a shared exponent. The proposed FPIMC macro implemented in 28nm CMOS demonstrates 12.1 TOPS/W for FP8 precision and 66.6 TOPS/W for BF8 precision, improving energyefficiency beyond the stateoftheart FP IMC macros.
more »
« less
 NSFPAR ID:
 10462009
 Date Published:
 Journal Name:
 Proceedings of ESSCIRC
 Format(s):
 Medium: X
 Sponsoring Org:
 National Science Foundation
More Like this


Mainstream math libraries for floating point (FP) do not produce correctly rounded results for all inputs. In contrast, CRLIBM and RLIBM provide correctly rounded implementations for a specific FP representation with one rounding mode. Using such libraries for a representation with a new rounding mode or with different precision will result in wrong results due to double rounding. This paper proposes a novel method to generate a single polynomial approximation that produces correctly rounded results for all inputs for multiple rounding modes and multiple precision configurations. To generate a correctly rounded library for n bits, our key idea is to generate a polynomial approximation for a representation with n +2bits using the roundtoodd mode. We prove that the resulting polynomial approximation will produce correctly rounded results for all five rounding modes in the standard and for multiple representations with k bits such that  E  +1 < k ≤ n , where  E  is the number of exponent bits in the representation. Similar to our prior work in the RLIBM project, we approximate the correctly rounded result when we generate the library with n +2bits using the roundtoodd mode. We also generate polynomial approximations by structuring it as a linear programming problem but propose enhancements to polynomial generation to handle the roundtoodd mode. Our prototype is the first 32bit float library that produces correctly rounded results with all rounding modes in the IEEE standard for all inputs with a single polynomial approximation. It also produces correctly rounded results for any FP configuration ranging from 10bits to 32bits while also being faster than mainstream libraries.more » « less

Inmemorycomputing (IMC) SRAM architecture has gained significant attention as it achieves high energy efficiency for computing a convolutional neural network (CNN) model [1]. Recent works investigated the use of analogmixedsignal (AMS) hardware for high area and energy efficiency [2], [3]. However, AMS hardware output is well known to be susceptible to process, voltage, and temperature (PVT) variations, limiting the computing precision and ultimately the inference accuracy of a CNN. We reconfirmed, through the simulation of a capacitorbased IMC SRAM macro that computes a 256D binary dot product, that the AMS computing hardware has a significant rootmeansquare error (RMSE) of 22.5% across the worstcase voltage, temperature (Fig. 16.1.1 top left) and 3sigma process variations (Fig. 16.1.1 top right). On the other hand, we can implement an IMC SRAM macro using robust digital logic [4], which can virtually eliminate the variability issue (Fig. 16.1.1 top). However, digital circuits require more devices than AMS counterparts (e.g., 28 transistors for a mirror full adder [FA]). As a result, a recent digital IMC SRAM shows a lower area efficiency of 6368F2/b (22nm, 4b/4b weight/activation) [5] than the AMS counterpart (1170F2/b, 65nm, 1b/1b) [3]. In light of this, we aim to adopt approximate arithmetic hardware to improve area and power efficiency and present two digital IMC macros (DIMC) with different levels of approximation (Fig. 16.1.1 bottom left). Also, we propose an approximationaware training algorithm and a number format to minimize inference accuracy degradation induced by approximate hardware (Fig. 16.1.1 bottom right). We prototyped a 28nm test chip: for a 1b/1b CNN model for CIFAR10 and across 0.5to1.1V supply, the DIMC with doubleapproximate hardware (DIMCD) achieves 2569F2/b, 9322219TOPS/W, 47520032GOPS, and 86.96% accuracy, while for a 4b/1b CNN model, the DIMC with the singleapproximate hardware (DIMCS) achieves 3814F2/b, 458990TOPS/Wmore » « less

Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)The Neuronix highperformance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from lowend consumer grade devices such as the Nvidia GTX 970 to higherend devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely computeintensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3] as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floatingpoint calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should produce identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly wellknown issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely timeconsuming for algorithm research in which a single run often taxes a computing environment to its limits. Wellknown techniques such as crossvalidation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when rerunning the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Crossvalidation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Largescale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPUaccelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are nondeterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multithreading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32bit since Python defaults to 64bit. We try to avoid using 64bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [1113]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues.more » « less

Vectormatrix multiplication (VMM) is a core operation in many signal and data processing algorithms. Previous work showed that analog multipliers based on nonvolatile memories have superior energy efficiency as compared to digital counterparts at lowtomedium computing precision. In this paper, we propose extremely energy efficient analog mode VMM circuit with digital input/output interface and configurable precision. Similar to some previous work, the computation is performed by gatecoupled circuit utilizing embedded floating gate (FG) memories. The main novelty of our approach is an ultralow power sensing circuitry, which is designed based on translinear Gilbert cell in topological combination with a floating resistor and a lowgain amplifier. Additionally, the digitaltoanalog input conversion is merged with VMM, while currentmode algorithmic analogtodigital circuit is employed at the circuit backend. Such implementations of conversion and sensing allow for circuit operation entirely in a current domain, resulting in high performance and energy efficiency. For example, postlayout simulation results for 400×400 5bit VMM circuit designed in 55 nm process with embedded NOR flash memory, show up to 400 MHz operation, 1.68 POps/J energy efficiency, and 39.45 TOps/mm2 computing throughput. Moreover, the circuit is robust against processvoltagetemperature variations, in part due to inclusion of additional FG cells that are utilized for offset compensation.more » « less