paged attention
vllm, memory
PagedAttention, introduced in vLLM, revolutionizes KV cache management by treating it like operating system virtual memory with fixed-size pages. Traditional implementations…
29,259 technical terms and definitions
vllm, memory
PagedAttention, introduced in vLLM, revolutionizes KV cache management by treating it like operating system virtual memory with fixed-size pages. Traditional implementations…
optimization
Paged Attention is a memory-management approach that stores KV cache blocks in pageable non-contiguous segments - It is a core method in modern semiconductor AI serving…
inference optimization
PagedAttention is a memory-management technique for LLM inference that applies operating-system-style virtual-memory paging to the attention key-value (KV) cache. Introduced…
virtual memory kv cache, paged memory management, kv cache blocks, memory efficient serving
PagedAttention is the attention mechanism that manages KV cache using virtual memory techniques with fixed-size blocks (pages) — eliminating memory fragmentation and enabling…
graph algorithms
PageRank is the seminal graph centrality algorithm originally designed for Google Search that ranks nodes by recursive importance — a node is important if it is pointed to…
chemistry ai
PaiNN (Polarizable Atom Interaction Neural Network) is an E(3)-equivariant message passing neural network that maintains both scalar (invariant) and vector (equivariant) features…
graph neural networks
PaiNN is an equivariant atomistic graph model that couples scalar and vector features for molecular interactions - It captures directional physics by jointly propagating magnitude…
quality & reliability
A paired t-test compares two measurements that belong together, such as a before-and-after measurement on the same wafer. Instead of treating the two columns as independent…
evaluation
Pairwise comparison is an evaluation method where two model outputs are placed side by side and a judge (human or LLM) determines which response is better. It is the most common…
training techniques
Pairwise Comparison is an evaluation method where two model outputs are judged against each other for preference or quality - It is a core method in modern LLM training and safety…
Pairwise Learning for Recommendations Introduction & Motivation Pairwise Learning for Recommendations addresses a central problem in recsys systems and (target) optimization: how…
machine learning
Pairwise ranking learns from item comparisons — training models to predict which of two items should rank higher, directly learning relative preferences rather than absolute…
recommendation systems
Pairwise Ranking is ranking optimization that learns preferences between item pairs for a given user or query - It improves ordering sensitivity by directly modeling which item…
programmable array logic, AND-OR array, fusible link, glue logic, simple PLD
A PAL device, from programmable array logic, is an early and highly successful type of simple programmable logic device, built around a programmable array of AND and OR logic that…
Paleontology: Fossils, Biostratigraphy, Evolution, and Extinction Paleontology reconstructs the history of life from incomplete physical evidence. Fossils record organisms, traces…
palm, pathways language model, foundation model
PaLM (Pathways Language Model) is Google's large-scale language model that demonstrated breakthrough capabilities through massive scaling, achieving state-of-the-art results on…
stanford, stanford university, hp, hewlett packard
Palo Alto is location-and-institution intent linking Palo Alto with Stanford and adjacent technology heritage context - It is a core method in modern semiconductor AI…
dataframe, pandas series, pandas groupby, pandas merge, pandas read csv, pandas categorical, pandas copy on write, pandas apply, pandas parquet, pandas performance, pandas memory, pandas index
Pandas is a Python library built on NumPy that introduces two primary data structures—Series (a labeled 1-D array) and DataFrame (a labeled 2-D table of columns)—and a query…
Pandera Statistical Schema Validation for DataFrames Introduction & Motivation Pandera Statistical Schema Validation for DataFrames addresses a central problem in (subject)…
packaging, large-scale, processing, throughput, cost, RDL, singulation
Panel-level packaging processes multiple chip packages simultaneously on a large rectangular panel, rather than on a single circular wafer the way wafer-level packaging does…
panel-level fan-out rdl registration prediction, panel-level fan-out rdl, fan-out rdl registration, rdl registration prediction, panel-level fan-out, fan-out rdl
Panel-Level Fan-Out RDL Registration Prediction with Machine Learning Introduction & Motivation Panel-Level Fan-Out RDL Registration Prediction with Machine Learning addresses…
panel-level packaging warpage prediction, panel-level packaging warpage, packaging warpage prediction, panel-level packaging, packaging warpage, warpage prediction
Panel-Level Packaging Warpage Prediction with Machine Learning Introduction & Motivation Panel-Level Packaging Warpage Prediction with Machine Learning addresses a central problem…
panel level packaging warpage prediction with machine learning, panel level packaging warpage prediction machine learning, panel level, packaging warpage prediction, cte rdl
Panel-Level Packaging Warpage Prediction with Machine Learning Introduction & Motivation Panel-Level Packaging Warpage Prediction with Machine Learning addresses a central problem…
Panoptic Quality Metric Introduction & Motivation Panoptic Quality Metric addresses a central problem in vision processing and stuff-thing-balance prediction: how to Build vision…
Universal Panoptic Segmentation with Mask2Former Introduction & Motivation Universal Panoptic Segmentation with Mask2Former addresses a central problem in vision processing…
panoptic segmentation unified framework, panoptic segmentation, segmentation unified, ade20k coco fn fp gt id, panoptic segmentation unified, segmentation unified framework, unified framework
Panoptic Segmentation - Unified Framework Introduction & Motivation Panoptic segmentation: unified semantic + instance segmentation. Predict class labels and instance IDs.…
panoptic segmentation unified, panoptic segmentation, segmentation unified
Panoptic Segmentation Unified Introduction & Motivation Panoptic Segmentation Unified addresses a central problem in vision process control and semantic-instance-joint-prediction…
panoptic segmentation unified prediction, panoptic segmentation unified, segmentation unified prediction, panoptic segmentation, segmentation unified, unified prediction
Panoptic Segmentation Unified Prediction with Machine Learning Introduction & Motivation Panoptic Segmentation Unified Prediction with Machine Learning addresses a central problem…
panoptic segmentation, unified stuff things, 3d coco crf fn fp passed, panoptic segmentation unified stuff & things segmentation, panoptic segmentation unified, segmentation unified
Panoptic Segmentation: Unified Stuff & Things Segmentation Introduction & Motivation Panoptic segmentation unifies semantic (stuff) and instance (things) segmentation. Stuff…
generative models
Panorama generation is the image synthesis process for producing wide-aspect or 360-degree scenes with coherent global perspective - it extends diffusion pipelines to cinematic…
panorama stitching mosaicing, panorama stitching, stitching mosaicing
Panorama Stitching Mosaicing Introduction & Motivation Panorama Stitching Mosaicing addresses a central problem in vision process control and image-alignment-homography…
panther lake, intel panther lake, what is panther lake, panther lake processor, panther lake architecture
Panther Lake is Intel’s codename for the mobile PC processor platform sold as Intel Core Ultra Series 3. Intel launched the family at CES in January 2026 after an architecture…
gradient, ml
Paperspace Gradient is a cloud ML platform that provides managed GPU-powered Jupyter notebooks, scalable training, and one-click model deployment — offering free-tier GPU access…
reduction, algorithms, tree, Kogge-Stone, cascade
Parallel Reduction Algorithms is strategies for combining values using an associative operator (sum, max, product, etc.) across distributed processes or threads, minimizing steps…
scan, prefix, sum, algorithm, Blelloch, work-efficient
Parallel Scan Prefix Sum Algorithm is a technique computing for each position the cumulative result of an associative operation (like addition) from element 0 to current position…
graph, algorithms, BFS, DFS, depth, breadth, first, search
Parallel Graph Algorithms BFS DFS is strategies for traversing or searching graph structures using distributed or shared-memory parallelism, balancing workload distribution…
stencil, computation, finite, difference, halo, exchange
Parallel Stencil Computation is numerical computation applying local patterns (stencils) to grid points independently, combining neighboring values according to kernel…
matrix, factorization, LU, QR, Cholesky, dense, decomposition
Parallel Matrix Factorization LU QR is decomposition of matrices into products of structured forms (triangular, unitary) enabling efficient solution of linear systems…
sorting, distributed, merge, quicksort, bitonic, hypercube
Parallel Sorting Distributed is algorithms ordering distributed data across multiple processors efficiently, minimizing communication while maintaining balanced computation…
programming, memory, consistency, sequential, release, acquire, models
Parallel Programming Memory Consistency Models is a formal specification of guarantees about memory access ordering across threads/processes, defining what memory values threads…
debugging, correctness, tools, verification
Parallel Debugging Correctness Tools is a specialized toolset for identifying and correcting bugs in parallel programs including race conditions, deadlocks, and correctness…
graph traversal parallel, parallel bfs gpu, graph processing parallel, vertex edge parallel
Parallel Breadth-First Search (BFS) is the foundational graph traversal algorithm that explores vertices level by level from a source vertex — where parallelizing BFS requires…
lz4 parallel, zstd parallel, data compression gpu, parallel decompression, compression throughput
Parallel Compression and Decompression is the high-throughput implementation of data compression algorithms (LZ4, Zstandard, Snappy, gzip) that exploits multi-core CPUs, SIMD…
parallel gzip lz4, gpu compression, data compression parallel, parallel decompression
Parallel Data Compression is the application of parallel computing to the inherently sequential problem of lossless data compression — where standard algorithms like DEFLATE…
parallel computing basics, parallel processing, parallel programming, parallel computation, concurrent computing
Parallel computing is the simultaneous use of multiple compute resources — cores, processors, accelerators, or nodes — to solve a problem faster or at larger scale. Every layer…
hpc carpentry tutorial, cuda udacity course, parallel programming textbook, programming massively parallel processors
HPC Education and Training: Pathways to Parallel Computing — textbooks, courses, and workshops for skill development High-performance computing education spans textbooks, online…
timing attack hpc, gpu side channel attack, spectre meltdown vulnerability, secure parallel computation
Security in Parallel Computing is the emerging discipline addressing the unique attack surfaces introduced by shared parallel hardware — where multiple tenants sharing GPU…
data
Parallel corpora is paired datasets that contain source and target sentences aligned at sentence or segment level across two languages - Alignment links each source segment to its…
race detector, thread sanitizer, parallel bug, concurrency debug
Parallel Debugging is the discipline of detecting, diagnosing, and fixing concurrency bugs (race conditions, deadlocks, livelocks, ordering violations) in multi-threaded…
race condition detection, thread sanitizer, helgrind, data race debugging, parallel bug detection
Parallel Debugging and Race Condition Detection is the specialized discipline of finding and fixing bugs unique to concurrent programs — race conditions, deadlocks, data races…