Neural graphics primitives, parameterized by fully connected neural networks, can be costly to train and evaluate. We reduce this cost with a versatile new input encoding that permits the use of a ...smaller network without sacrificing quality, thus significantly reducing the number of floating point and memory access operations: a small neural network is augmented by a multiresolution hash table of trainable feature vectors whose values are optimized through stochastic gradient descent. The multiresolution structure allows the network to disambiguate hash collisions, making for a simple architecture that is trivial to parallelize on modern GPUs. We leverage this parallelism by implementing the whole system using fully-fused CUDA kernels with a focus on minimizing wasted bandwidth and compute operations. We achieve a combined speedup of several orders of magnitude, enabling training of high-quality neural graphics primitives in a matter of seconds, and rendering in tens of milliseconds at a resolution of 1920×1080.
Explainable clustering provides human-understandable reasons for decisions in black-box learning models. In a previous work, a decision tree built on the set of dimensions was used to define ranges ...of values for k-means clusters. For explainable graph clustering, we use expander graphs instead of dense subgraphs since powering an expander graph is guaranteed to result in a clique after at most a logarithmic number of steps.
Consider a set of multi-dimensional points labeled with k labels. We introduce the heat map sorting problem as reordering the rows and columns of an input matrix (each point is a column and each row is a dimension) such that the labels of the entries of the matrix form connected components (clusters). A cluster is preserved if it remains connected, i.e., if it is not split into several clusters and no two clusters are merged. In the massively parallel computation model (MPC), each machine has a sublinear memory and the total memory of the machines is linear.
We prove the problem is NP-hard. We give a fixed-parameter algorithm in MPC and an approximation algorithm based on expander decomposition. We empirically compare our algorithm with explainable k-means on several graphs of email and computer networks.
•A general method for explainable clustering of high-dimensional data.•A fixed-parameter algorithms for explainable graph clustering.•A Massively Parallel Computation (MPC) algorithm for explainable clustering.•An approximation algorithm for graph clustering on expander graphs.
The
Massively Parallel Computation
(MPC) model serves as a common abstraction of many modern large-scale data processing frameworks, and has been receiving increasingly more attention over the past ...few years, especially in the context of classical graph problems. So far, the only way to argue lower bounds for this model is to condition on conjectures about the hardness of some specific problems, such as graph connectivity on promise graphs that are either one cycle or two cycles, usually called the
one cycle versus two cycles
problem. This is unlike the traditional arguments based on conjectures about complexity classes (e.g.,
P
≠
NP
), which are often more robust in the sense that refuting them would lead to groundbreaking algorithms for a whole bunch of problems. In this paper we present connections between problems and classes of problems that allow the latter type of arguments. These connections concern the class of problems solvable in a sublogarithmic amount of rounds in the MPC model, denoted by
MPC
(
o
(
log
N
)
)
, and the standard space complexity classes
L
and
NL
, and suggest conjectures that are robust in the sense that refuting them would lead to many surprisingly fast new algorithms in the MPC model. We also obtain new conditional lower bounds, and prove new reductions and equivalences between problems in the MPC model. Specifically, our main results are as follows.
Lower bounds conditioned on the one cycle versus two cycles conjecture can be instead argued under the
L
⊈
MPC
(
o
(
log
N
)
)
conjecture: these two assumptions are equivalent, and refuting either of them would lead to
o
(
log
N
)
-round MPC algorithms for a large number of challenging problems, including list ranking, minimum cut, and planarity testing. In fact, we show that these problems and many others require asymptotically the same number of rounds as the seemingly much easier problem of distinguishing between a graph being one cycle or two cycles.
Many lower bounds previously argued under the one cycle versus two cycles conjecture can be argued under an even more robust (thus harder to refute) conjecture, namely
NL
⊈
MPC
(
o
(
log
N
)
)
. Refuting this conjecture would lead to
o
(
log
N
)
-round MPC algorithms for an even larger set of problems, including all-pairs shortest paths, betweenness centrality, and all aforementioned ones. Lower bounds under this conjecture hold for problems such as perfect matching and network flow.
Polynomial sharings on two secrets: Buy one, get one free Arnold, Paula; Berndt, Sebastian; Eisenbarth, Thomas ...
IACR transactions on cryptographic hardware and embedded systems,
07/2024, Letnik:
2024, Številka:
3
Journal Article
Recenzirano
Odprti dostop
While passive side-channel attacks and active fault attacks have been studied intensively in the last few decades, strong attackers combining these attacks have only been studied relatively recently. ...Due to its simplicity, most countermeasures against passive attacks are based on additive sharing. Unfortunately, extending these countermeasures against faults often leads to quite a significant performance penalty, either due to the use of expensive cryptographic operations or a large number of shares due to massive duplication. Just recently, Berndt, Eisenbarth, Gourjon, Faust, Orlt, and Seker thus proposed to use polynomial sharing against combined attackers (CRYPTO 2023). While they construct gadgets secure against combined attackers using only a linear number of shares, the overhead introduced might still be too large for practical scenarios.In this work, we show how the overhead of nearly all known constructions using polynomial sharing can be reduced by nearly half by embedding two secrets in the coefficients of one polynomial at the expense of increasing the degree of the polynomial by one. We present a very general framework that allows adapting these constructions to this new sharing scheme and prove the security of this approach against purely passive side-channel attacks, purely active fault attacks, and combined attacks. Furthermore, we present new gadgets allowing us to operate upon the different secrets in a number of useful ways.
A feedback node set (FNS) of a graph is a subset of the nodes of the graph whose deletion makes the residual graph acyclic. By finding an FNS in an interconnection network, we can set a check point ...at each node in it to avoid a livelock configuration. Hence, to find an FNS is a critical issue to enhance the dependability of a parallel computing system. In this paper, we propose a method to find FNS's in n-pancake graphs and n-burnt pancake graphs. By analyzing the types of cycles proposed in our method, we also give the number of the nodes in the FNS in an n-pancake graph, (n-2.875)(n-1)!+1.5(n-3)!, and that in an n-burnt pancake graph, 2n-1(n-1)!(n-3.5).
As one of the arbitrary Lagrangian–Eulerian methods, the material point method (MPM) owns intrinsic advantages in simulation of large deformation problems by combining the merits of the Lagrangian ...and Eulerian approaches. Significant computational intensity is involved in the calculations of the MPM due to its very fine mesh needed to achieve a sufficiently high accuracy. A new multiple‐GPU parallel strategy is developed based on a single‐root complex architecture of the computer purely within a CUDA environment. Peer‐to‐Peer (P2P) communication between the GPUs is performed to exchange the information of the crossing particles and ghost element nodes, which is faster than the heavy send/receive operations between different computers through the infiniBand network. Domain decomposition is performed to split the whole computational task over the GPUs with a number of subdomains. The computations within each subdomain are allocated on a corresponding GPU using an enhanced “Particle‐List” scheme to tackle the data race during the interpolation from associated particles to common nodes. The acceleration effect of the parallelization is evaluated with two benchmarks cases, mini‐slump test after a dam break and cone penetration test in clay, where the maximum speedups with 1 and 8 GPUs are 88 and 604, respectively.
The Petaflops supercomputer “Zhores” recently launched in the “Center for Computational and Data-Intensive Science and Engineering” (CDISE) of Skolkovo Institute of Science and Technology (Skoltech) ...opens up new exciting opportunities for scientific discoveries in the institute especially in the areas of data-driven modeling, machine learning and artificial intelligence. This supercomputer utilizes the latest generation of Intel and NVidia processors to provide resources for the most compute intensive tasks of the Skoltech scientists working in digital pharma, predictive analytics, photonics, material science, image processing, plasma physics and many more. Currently it places 7
in the Russian and CIS TOP-50 (2019) supercomputer list. In this article we summarize the cluster properties and discuss the measured performance and usage modes of this new scientific instrument in Skoltech.
This work discusses the implementation, capabilities, and validation of Shift, a massively parallel Monte Carlo radiation transport package authored at Oak Ridge National Laboratory. Shift has been ...developed to scale well from laptops to small computing clusters to advanced supercomputers and includes features such as support for multiple geometry and physics engines, hybrid capabilities for variance reduction methods such as the Consistent Adjoint-Driven Importance Sampling methodology, advanced parallel decompositions, and tally methods optimized for scalability on supercomputing architectures. The scaling studies presented in this paper demonstrate good weak and strong scaling behavior for the implemented algorithms. Shift has also been validated and verified against various reactor physics benchmarks, including the Consortium for Advanced Simulation of Light Water Reactors' Virtual Environment for Reactor Analysis criticality test suite and several Westinghouse AP1000® problems presented in this paper. These benchmark results compare well to those from other contemporary Monte Carlo codes such as MCNP5 and KENO.