Trevor John Hastie (born 27 June 1953) is an American statistician and computer scientist. He is currently serving as the John A. Overdeck Professor of Mathematical Sciences and Professor of Statistics at Stanford University. Hastie is known for his contributions to applied statistics, especially in the field of machine learning, data mining, and bioinformatics. He has authored several popular books in statistical learning, including The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Hastie has been listed as an ISI Highly Cited Author in Mathematics by the ISI Web of Knowledge. He also contributed to the development of S. == Education and career == Hastie was born on 27 June 1953 in South Africa. He received his B.S. in statistics from the Rhodes University in 1976 and master's degree from University of Cape Town in 1979. Hastie joined the doctoral program at Stanford University in 1980 and received his Ph.D. in 1984 under the supervision of Werner Stuetzle. His dissertation was "Principal Curves and Surfaces". Hastie began his professional career in 1977 with the South African Medical Research Council. After receiving his master's degree in 1979, he spent a year interning at the London School of Hygiene & Tropical Medicine, the Johnson Space Center in Houston, and the Biomath department at Oxford University. After receiving his doctoral degree from Stanford, Hastie returned to South Africa to work with his former employer South African Medical Research Council. He returned to United States in 1986 and joined the AT&T Bell Laboratories in Murray Hill, New Jersey and remained there for nine years. Working with John Chambers, he co-directed the development of the S programming language. He joined Stanford University in 1994 as Associate Professor in Statistics and Biostatistics. He was promoted to full Professor in 1999. During the period 2006–2009, he was the chair of the Department of Statistics at Stanford University. In 2013 he was named the John A. Overdeck Professor of Mathematical Sciences. == Awards and honors == Hastie is a Fellow of the Royal Statistical Society since 1979. He is also an elected Fellow of several professional and scholarly societies, including the Institute of Mathematical Statistics, the American Statistical Association, and the South African Statistical Society. He is a recipient of 'Myrto Lefkopolou Distinguished Lectureship' award of Biostatistics Department at the Harvard School of Public Health. In 2018, he was elected a member of the National Academy of Sciences. In 2019 Hastie became a foreign member of the Royal Netherlands Academy of Arts and Sciences. Hastie was named for the C.R. and Bhargavi Rao Prize in 2025. Hastie and Hui Zou received the 2025 Founders of Statistics prize for their elastic net paper. == Publications == Hastie is a prolific author of scientific works on numerous topics in applied statistics, including statistical learning, data mining, statistical computing, and bioinformatics. He along with his collaborators has authored about 125 scientific articles. Many of Hastie's scientific articles were coauthored by his longtime collaborator, Robert Tibshirani. Hastie has been listed as an ISI Highly Cited Author in Mathematics by the ISI Web of Knowledge. He has coauthored the following books: T. Hastie and R. Tibshirani, Generalized Additive Models, Chapman and Hall, 1990. J. Chambers and T. Hastie, Statistical Models in S, Wadsworth/Brooks Cole, 1991. T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning: Prediction, Inference and Data Mining, Second Edition, Springer Verlag, 2009 (available for free from the author's website). G. James, D. Witten, T. Hastie, R. Tibshirani, An Introduction to Statistical Learning with Applications in R, Springer Verlag, 2013 (available for free from the co-author's website). T. Hastie, R. Tibshirani, M. Wainwright, Statistical Learning with Sparsity: the Lasso and Generalizations, CRC Press, 2015 (available for free from the author's website). Bradley Efron; Trevor Hastie (2016). Computer Age Statistical Inference. Cambridge University Press. ISBN 9781107149892.
Convolutional neural network
A convolutional neural network (CNN) is a type of feedforward neural network that learns features via filter (or kernel) optimization. This type of deep learning network has been applied to process and make predictions from many different types of data including text, images and audio. CNNs are the de-facto standard in deep learning-based approaches to computer vision and image processing, and have only recently been replaced—in some cases—by newer architectures such as the transformer. Vanishing gradients and exploding gradients, seen during backpropagation in earlier neural networks, are prevented by the regularization that comes from using shared weights over fewer connections. For example, for each neuron in the fully-connected layer, 10,000 weights would be required for processing an image sized 100 × 100 pixels. However, applying cascaded convolution (or cross-correlation) kernels, only 25 weights for each convolutional layer are required to process 5x5-sized tiles. Higher-layer features are extracted from wider context windows, compared to lower-layer features. Some applications of CNNs include: image and video recognition, recommender systems, image classification, image segmentation, medical image analysis, natural language processing, brain–computer interfaces, and financial time series. CNNs are also known as shift invariant or space invariant artificial neural networks, based on the shared-weight architecture of the convolution kernels or filters that slide along input features and provide translation-equivariant responses known as feature maps. Counter-intuitively, most convolutional neural networks are not invariant to translation, due to the downsampling operation they apply to the input. Feedforward neural networks are usually fully connected networks, that is, each neuron in one layer is connected to all neurons in the next layer. The "full connectivity" of these networks makes them prone to overfitting data. Typical ways of regularization, or preventing overfitting, include: penalizing parameters during training (such as weight decay) or trimming connectivity (skipped connections, dropout, etc.) Robust datasets also increase the probability that CNNs will learn the generalized principles that characterize a given dataset rather than the biases of a poorly-populated set. Convolutional networks were inspired by biological processes in that the connectivity pattern between neurons resembles the organization of the animal visual cortex. Individual cortical neurons respond to stimuli only in a restricted region of the visual field known as the receptive field. The receptive fields of different neurons partially overlap such that they cover the entire visual field. CNNs use relatively little pre-processing compared to other image classification algorithms. This means that the network learns to optimize the filters (or kernels) through automated learning, whereas in traditional algorithms these filters are hand-engineered. This simplifies and automates the process, enhancing efficiency and scalability overcoming human-intervention bottlenecks. == Architecture == A convolutional neural network consists of an input layer, hidden layers and an output layer. In a convolutional neural network, the hidden layers include one or more layers that perform convolutions. Typically this includes a layer that performs a dot product of the convolution kernel with the layer's input matrix. This product is usually the Frobenius inner product, and its activation function is commonly ReLU. As the convolution kernel slides along the input matrix for the layer, the convolution operation generates a feature map, which in turn contributes to the input of the next layer. This is followed by other layers such as pooling layers, fully connected layers, and normalization layers. Here it should be noted how close a convolutional neural network is to a matched filter. === Convolutional layers === In a CNN, the input is a tensor with shape: (number of inputs) × (input height) × (input width) × (input channels) After passing through a convolutional layer, the image becomes abstracted to a feature map, also called an activation map, with shape: (number of inputs) × (feature map height) × (feature map width) × (feature map channels). Convolutional layers convolve the input and pass its result to the next layer. This is similar to the response of a neuron in the visual cortex to a specific stimulus. Each convolutional neuron processes data only for its receptive field. Although fully connected feedforward neural networks can be used to learn features and classify data, this architecture is generally impractical for larger inputs (e.g., high-resolution images), which would require massive numbers of neurons because each pixel is a relevant input feature. A fully connected layer for an image of size 100 × 100 has 10,000 weights for each neuron in the second layer. Convolution reduces the number of free parameters, allowing the network to be deeper. For example, using a 5 × 5 tiling region, each with the same shared weights, requires only 25 neurons. Using shared weights means there are many fewer parameters, which helps avoid the vanishing gradients and exploding gradients problems seen during backpropagation in earlier neural networks. To speed processing, standard convolutional layers can be replaced by depthwise separable convolutional layers, which are based on a depthwise convolution followed by a pointwise convolution. The depthwise convolution is a spatial convolution applied independently over each channel of the input tensor, while the pointwise convolution is a standard convolution restricted to the use of 1 × 1 {\displaystyle 1\times 1} kernels. === Pooling layers === Convolutional networks may include local and/or global pooling layers along with traditional convolutional layers. Pooling layers reduce the dimensions of data by combining the outputs of neuron clusters at one layer into a single neuron in the next layer. Local pooling combines small clusters, tiling sizes such as 2 × 2 are commonly used. Global pooling acts on all the neurons of the feature map. There are two common types of pooling in popular use: max and average. Max pooling uses the maximum value of each local cluster of neurons in the feature map, while average pooling takes the average value. === Fully connected layers === Fully connected layers connect every neuron in one layer to every neuron in another layer. It is the same as a traditional multilayer perceptron neural network (MLP). Each neuron in the fully connected layer receives input from all the neurons in the previous layer. These inputs are weighted and summed with the corresponding biases, and then passed through an activation function to perform a nonlinear transformation, generating the output. The flattened matrix goes through a fully connected layer to classify the images. === Receptive field === In neural networks, each neuron receives input from some number of locations in the previous layer. In a convolutional layer, each neuron receives input from only a restricted area of the previous layer called the neuron's receptive field. Typically the area is a square (e.g. 5 by 5 neurons). Whereas, in a fully connected layer, the receptive field is the entire previous layer. Thus, in each convolutional layer, each neuron takes input from a larger area in the input than previous layers. This is due to applying the convolution over and over, which takes the value of a pixel into account, as well as its surrounding pixels. When using dilated layers, the number of pixels in the receptive field remains constant, but the field is more sparsely populated as its dimensions grow when combining the effect of several layers. To manipulate the receptive field size as desired, there are some alternatives to the standard convolutional layer. For example, atrous or dilated convolution expands the receptive field size without increasing the number of parameters by interleaving visible and blind regions. Moreover, a single dilated convolutional layer can comprise filters with multiple dilation ratios, thus having a variable receptive field size. === Weights === Each neuron in a neural network computes an output value by applying a specific function to the input values received from the receptive field in the previous layer. The function that is applied to the input values is determined by a vector of weights and a bias (typically real numbers). Learning consists of iteratively adjusting these biases and weights. The vectors of weights and biases are called filters and represent particular features of the input (e.g., a particular shape). A distinguishing feature of CNNs is that many neurons can share the same filter. This reduces the memory footprint because a single bias and a single vector of weights are used across all receptive fields that share that filter, as opposed to each receptive field having its own bias and vector
Ultra (cryptography)
Ultra was the designation adopted by British military intelligence in June 1941 for wartime signals intelligence obtained by breaking high-level encrypted enemy radio and teleprinter communications at the Government Code and Cypher School (GC&CS) at Bletchley Park. Ultra eventually became the standard designation among the western Allies for all such intelligence. The name arose because the intelligence obtained was considered more important than that designated by the highest British security classification then used (Most Secret) and so was regarded as being Ultra Secret. Several other cryptonyms had been used for such intelligence. The code name "Boniface" was used as a cover name for Ultra. In order to ensure that the successful code-breaking did not become apparent to the Germans, British intelligence created a fictional MI6 master spy, Boniface, who controlled a fictional series of agents throughout Germany. Information obtained through code-breaking was often attributed to the human intelligence from the Boniface network. The U.S. used the codename Magic for its decrypts from Japanese sources, including the "Purple" cipher. Much of the German cipher traffic was encrypted on the Enigma machine. Used properly, the German military Enigma would have been virtually unbreakable; in practice, shortcomings in operation allowed it to be broken. The term "Ultra" has often been used almost synonymously with "Enigma decrypts". However, Ultra also encompassed decrypts of the German Lorenz SZ 40/42 machines that were used by the German High Command, and the Hagelin machine. Many observers, at the time and later, regarded Ultra as immensely valuable to the Allies. Winston Churchill was reported to have told King George VI, when presenting to him Stewart Menzies (head of the Secret Intelligence Service and the person who controlled distribution of Ultra decrypts to the government): "It is thanks to the secret weapon of General Menzies, put into use on all the fronts, that we won the war!" F. W. Winterbotham quoted the western Supreme Allied Commander, Dwight D. Eisenhower, at war's end describing Ultra as having been "decisive" to Allied victory. Sir Harry Hinsley, Bletchley Park veteran and official historian of British Intelligence in World War II, made a similar assessment of Ultra, saying that while the Allies would have won the war without it, "the war would have been something like two years longer, perhaps three years longer, possibly four years longer than it was." However, Hinsley and others have emphasized the difficulties of counterfactual history in attempting such conclusions, and some historians, such as John Keegan, have said the shortening might have been as little as the three months it took the United States to deploy the atomic bomb. == Sources of intelligence == Most Ultra intelligence was derived from reading radio messages that had been encrypted with cipher machines, complemented by material from radio communications using traffic analysis and direction finding. In the early phases of the war, particularly during the eight-month Phoney War, the Germans could transmit most of their messages using land lines and so had no need to use radio. This meant that those at Bletchley Park had some time to build up experience of collecting and starting to decrypt messages on the various radio networks. German Enigma messages were the main source, with those of the German air force (the Luftwaffe) predominating, as they used radio more and their operators were particularly ill-disciplined. === German === ==== Enigma ==== "Enigma" refers to a family of electro-mechanical rotor cipher machines. These produced a polyalphabetic substitution cipher and were widely thought to be unbreakable in the 1920s, when a variant of the commercial Model D was first used by the Reichswehr. The German Army (Heer), Navy, Air Force, Nazi party, Gestapo and German diplomats used Enigma machines in several variants. Abwehr (German military intelligence) used a four-rotor machine without a plugboard and Naval Enigma used different key management from that of the army or air force, making its traffic far more difficult to cryptanalyse; each variant required different cryptanalytic treatment. The commercial versions were not as secure and Dilly Knox of GC&CS is said to have broken one before the war. German military Enigma was first broken in December 1932 by Marian Rejewski and the Polish Cipher Bureau, using a combination of brilliant mathematics, the services of a spy in the German office responsible for administering encrypted communications, and good luck. The Poles read Enigma to the outbreak of World War II and beyond, in France. At the turn of 1939, the Germans made the systems ten times more complex, which required a tenfold increase in Polish decryption equipment, which they could not meet. On 25 July 1939, the Polish Cipher Bureau handed reconstructed Enigma machines and their techniques for decrypting ciphers to the French and British. Gordon Welchman wrote, Ultra would never have got off the ground if we had not learned from the Poles, in the nick of time, the details both of the German military Enigma machine, and of the operating procedures that were in use. At Bletchley Park, some of the key people responsible for success against Enigma included mathematicians Alan Turing and Hugh Alexander and, at the British Tabulating Machine Company, chief engineer Harold Keen. After the war, interrogation of German cryptographic personnel led to the conclusion that German cryptanalysts understood that cryptanalytic attacks against Enigma were possible but were thought to require impracticable amounts of effort and investment. The Poles' early start at breaking Enigma and the continuity of their success gave the Allies an advantage when World War II began. ==== Lorenz cipher ==== In June 1941, the Germans started to introduce on-line stream cipher teleprinter systems for strategic point-to-point radio links, to which the British gave the code-name Fish. Several systems were used, principally the Lorenz SZ 40/42 (codenamed "Tunny" by the British) and Geheimfernschreiber ("Sturgeon"). These cipher systems were cryptanalysed, particularly Tunny, which the British thoroughly penetrated. It was eventually attacked using Colossus machines, which were the first digital programme-controlled electronic computers. In many respects the Tunny work was more difficult than for the Enigma, since the British codebreakers had no knowledge of the machine producing it and no head-start such as that the Poles had given them against Enigma. Although the volume of intelligence derived from this system was much smaller than that from Enigma, its importance was often far higher because it produced primarily high-level, strategic intelligence that was sent between Wehrmacht high command (Oberkommando der Wehrmacht, OKW). The eventual bulk decryption of Lorenz-enciphered messages contributed significantly, and perhaps decisively, to the defeat of Nazi Germany. Nevertheless, the Tunny story has become much less well known among the public than the Enigma one. At Bletchley Park, some of the key people responsible for success in the Tunny effort included mathematicians W. T. "Bill" Tutte and Max Newman and electrical engineer Tommy Flowers. === Italian === In June 1940, the Italians were using book codes for most of their military messages, except for the Italian Navy, which in early 1941 had started using a version of the Hagelin rotor-based cipher machine C-38. This was broken from June 1941 onwards by the Italian subsection of GC&CS at Bletchley Park. === Japanese === In the Pacific theatre, a Japanese cipher machine, called "Purple" by the Americans, was used for highest-level Japanese diplomatic traffic. It produced a polyalphabetic substitution cipher, but unlike Enigma, was not a rotor machine, being built around electrical stepping switches. It was broken by the US Army Signal Intelligence Service and disseminated as Magic. Detailed reports by the Japanese ambassador to Germany were encrypted on the Purple machine. His reports included reviews of German assessments of the military situation, reviews of strategy and intentions, reports on direct inspections by the ambassador (in one case, of Normandy beach defences), and reports of long interviews with Hitler. The Japanese are said to have obtained an Enigma machine in 1937, although it is debated whether they were given it by the Germans or bought a commercial version, which, apart from the plugboard and internal wiring, was the German Heer/Luftwaffe machine. Having developed a similar machine, the Japanese did not use the Enigma machine for their most secret communications. The chief fleet communications code system used by the Imperial Japanese Navy was called JN-25 by the Americans, and by early 1942 the US Navy had made considerable progress in decrypting Japanese naval messages. The US Army also made progress on the
Harvest now, decrypt later
Harvest now, decrypt later (HNDL) is a surveillance strategy that relies on the acquisition and long-term storage of currently unreadable encrypted data awaiting possible breakthroughs in decryption technology that would render it readable in the future—a hypothetical date referred to as Y2Q (a reference to Y2K), or Q-Day. The most common concern is the prospect of developments in quantum computing which would allow current strong encryption algorithms to be broken at some time in the future, making it possible to decrypt any stored material that had been encrypted using those algorithms. However, the improvement in decryption technology need not be due to a quantum-cryptographic advance; any other form of attack capable of enabling decryption would be sufficient. The existence of this strategy has led to concerns about the need to urgently deploy post-quantum cryptography; even though no practical quantum attacks yet exist, some data stored now may still remain sensitive even decades into the future. As of 2022, the U.S. federal government has proposed a roadmap for organizations to start migrating toward quantum-cryptography-resistant algorithms to mitigate these threats. This new version of Commercial National Security Algorithm Suite uses publicly-available algorithms and is allowed for government use up to the TOP SECRET level. == Terminology and scope == The term “harvest now, decrypt later” encompasses various surveillance or espionage operations in which ciphertext or encrypted communications are collected today with the view that they may one day be decrypted, given sufficient advances in computing power or cryptanalysis. The abbreviation HNDL is sometimes used in technical and policy documents. The “Y2Q” (or “Q-Day”) label draws an analogy to the Y2K date-change issue, emphasising a potential future point at which current cryptography may collapse. The strategy is particularly relevant for data with long confidentiality lifetimes, such as diplomatic communications, personal health records, critical infrastructure logs, or intellectual property. == Mitigation strategies == The primary defense against HNDL attacks is the transition to post-quantum cryptography (PQC), which utilizes algorithms believed to be secure against quantum computer attacks. However, because PQC protects the data payload digitally, rather than the transmission itself, the encrypted data can still be harvested and stored. A complementary approach involves physical layer security (also known as optical layer encryption or photonic shielding). Unlike algorithmic encryption, this method modifies the optical waveform itself—often by burying the signal within optical noise or using spectral phase encoding—to render the transmission unrecordable by standard receivers. By preventing the attacker from capturing a valid signal in the first place, this approach aims to eliminate the "harvest" phase of the threat. Commercial implementations of harvest-proof optical encryption have been developed by firms such as CyberRidge to secure long-haul fiber networks. Field trials have demonstrated 100 Gbps throughput over legacy DWDM networks using this method.
Manufacturing Automation Protocol
Manufacturing Automation Protocol (MAP) was a computer network standard released in 1982 for interconnection of devices from multiple manufacturers. It was developed by General Motors to combat the proliferation of incompatible communications standards used by suppliers of automation products such as programmable controllers. By 1985 demonstrations of interoperability were carried out and 21 vendors offered MAP products. In 1986 the Boeing corporation merged its Technical Office Protocol with the MAP standard, and the combined standard was referred to as "MAP/TOP". The standard was revised several times between the first issue in 1982 and MAP 3.0 in 1987, with significant technical changes that made interoperation between different revisions of the standard difficult. Although promoted and used by manufacturers such as General Motors, Boeing, and others, it lost market share to the contemporary Ethernet standard and was not widely adopted. Difficulties included changing protocol specifications, the expense of MAP interface links, and the speed penalty of a token-passing network. The token bus network protocol used by MAP became standardized as IEEE standard 802.4 but this committee disbanded in 2004 due to lack of industry attention.
Multiple buffering
In computer science, multiple buffering is the use of more than one buffer to hold a block of data, so that a "reader" will see a complete (though perhaps old) version of the data instead of a partially updated version of the data being created by a "writer". It is very commonly used for computer display images. It is also used to avoid the need to use dual-ported RAM (DPRAM) when the readers and writers are different devices. == Description == === Double buffering Petri net === The Petri net in the illustration shows double buffering. Transitions W1 and W2 represent writing to buffer 1 and 2 respectively while R1 and R2 represent reading from buffer 1 and 2 respectively. At the beginning, only the transition W1 is enabled. After W1 fires, R1 and W2 are both enabled and can proceed in parallel. When they finish, R2 and W1 proceed in parallel and so on. After the initial transient where W1 fires alone, this system is periodic and the transitions are enabled – always in pairs (R1 with W2 and R2 with W1 respectively). == Double buffering in computer graphics == In computer graphics, double buffering is a technique for drawing graphics that shows less stutter, tearing, and other artifacts. It is difficult for a program to draw a display so that pixels do not change more than once. For instance, when updating a page of text, it is much easier to clear the entire page and then draw the letters than to somehow erase only the pixels that are used in old letters but not in new ones. However, this intermediate image is seen by the user as flickering. In addition, computer monitors constantly redraw the visible video page (traditionally at around 60 times a second), so even a perfect update may be visible momentarily as a horizontal divider between the "new" image and the un-redrawn "old" image, known as tearing. === Software double buffering === A software implementation of double buffering has all drawing operations store their results in some region of system RAM; any such region is often called a "back buffer". When all drawing operations are considered complete, the whole region (or only the changed portion) is copied into the video RAM (the "front buffer"); this copying is usually synchronized with the monitor's raster beam in order to avoid tearing. Software implementations of double buffering necessarily require more memory and CPU time than single buffering because of the system memory allocated for the back buffer, the time for the copy operation, and the time waiting for synchronization. Compositing window managers often combine the "copying" operation with "compositing" used to position windows, transform them with scale or warping effects, and make portions transparent. Thus, the "front buffer" may contain only the composite image seen on the screen, while there is a different "back buffer" for every window containing the non-composited image of the entire window contents. === Page flipping === In the page-flip method, instead of copying the data, both buffers are capable of being displayed. At any one time, one buffer is actively being displayed by the monitor, while the other, background buffer is being drawn. When the background buffer is complete, the roles of the two are switched. The page-flip is typically accomplished by modifying a hardware register in the video display controller—the value of a pointer to the beginning of the display data in the video memory. The page-flip is much faster than copying the data and can guarantee that tearing will not be seen as long as the pages are switched over during the monitor's vertical blanking interval—the blank period when no video data is being drawn. The currently active and visible buffer is called the front buffer, while the background page is called the back buffer. == Triple buffering == In computer graphics, triple buffering is similar to double buffering but can provide improved performance. In double buffering, the program must wait until the finished drawing is copied or swapped before starting the next drawing. This waiting period could be several milliseconds during which neither buffer can be touched. In triple buffering, the program has two back buffers and can immediately start drawing in the one that is not involved in such copying. The third buffer, the front buffer, is read by the graphics card to display the image on the monitor. Once the image has been sent to the monitor, the front buffer is flipped with (or copied from) the back buffer holding the most recent complete image. Since one of the back buffers is always complete, the graphics card never has to wait for the software to complete. Consequently, the software and the graphics card are completely independent and can run at their own pace. Finally, the displayed image was started without waiting for synchronization and thus with minimum lag. Due to the software algorithm not polling the graphics hardware for monitor refresh events, the algorithm may continuously draw additional frames as fast as the hardware can render them. For frames that are completed much faster than interval between refreshes, it is possible to replace a back buffers' frames with newer iterations multiple times before copying. This means frames may be written to the back buffer that are never used at all before being overwritten by successive frames. Nvidia has implemented this method under the name "Fast Sync". An alternative method sometimes referred to as triple buffering is a swap chain three buffers long. After the program has drawn both back buffers, it waits until the first one is placed on the screen, before drawing another back buffer (i.e. it is a 3-long first in, first out queue). Most Windows games seem to refer to this method when enabling triple buffering. == Quad buffering == The term quad buffering is the use of double buffering for each of the left and right eye images in stereoscopic implementations, thus four buffers total (if triple buffering was used then there would be six buffers). The command to swap or copy the buffer typically applies to both pairs at once, so at no time does one eye see an older image than the other eye. Quad buffering requires special support in the graphics card drivers which is disabled for most consumer cards. AMD's Radeon HD 6000 Series and newer support it. 3D standards like OpenGL and Direct3D support quad buffering. == Double buffering for DMA == The term double buffering is used for copying data between two buffers for direct memory access (DMA) transfers, not for enhancing performance, but to meet specific addressing requirements of a device (particularly 32-bit devices on systems with wider addressing provided via Physical Address Extension). Windows device drivers are a place where the term "double buffering" is likely to be used. Linux and BSD source code calls these "bounce buffers". Some programmers try to avoid this kind of double buffering with zero-copy techniques. == Other uses == Double buffering is also used as a technique to facilitate interlacing or deinterlacing of video signals.
PGP word list
The PGP Word List ("Pretty Good Privacy word list", also called a biometric word list for reasons explained below) is a list of words for conveying data bytes in a clear unambiguous way via a voice channel. They are analogous in purpose to the NATO phonetic alphabet, except that a longer list of words is used, each word corresponding to one of the 256 distinct numeric byte values. == History and structure == The PGP Word List was designed in 1995 by Patrick Juola, a computational linguist, and Philip Zimmermann, creator of PGP. The words were carefully chosen for their phonetic distinctiveness, using genetic algorithms to select lists of words that had optimum separations in phoneme space. The candidate word lists were randomly drawn from Grady Ward's Moby Pronunciator list as raw material for the search, successively refined by the genetic algorithms. The automated search converged to an optimized solution in about 40 hours on a DEC Alpha, a particularly fast machine in that era. The Zimmermann–Juola list was originally designed to be used in PGPfone, a secure VoIP application, to allow the two parties to verbally compare a short authentication string to detect a man-in-the-middle attack (MiTM). It was called a biometric word list because the authentication depended on the two human users recognizing each other's distinct voices as they read and compared the words over the voice channel, binding the identity of the speaker with the words, which helped protect against the MiTM attack. The list can be used in many other situations where a biometric binding of identity is not needed, so calling it a biometric word list may be imprecise. Later, it was used in PGP to compare and verify PGP public key fingerprints over a voice channel. This is known in PGP applications as the "biometric" representation. When it was applied to PGP, the list of words was further refined, with contributions by Jon Callas. More recently, it has been used in Zfone and the ZRTP protocol, the successor to PGPfone. The list is actually composed of two lists, each containing 256 phonetically distinct words, in which each word represents a different byte value between 0 and 255. Two lists are used because reading aloud long random sequences of human words usually risks three kinds of errors: 1) transposition of two consecutive words, 2) duplicate words, or 3) omitted words. To detect all three kinds of errors, the two lists are used alternately for the even-offset bytes and the odd-offset bytes in the byte sequence. Each byte value is actually represented by two different words, depending on whether that byte appears at an odd or an even offset from the beginning of the byte sequence. The two lists are readily distinguished by the number of syllables; the odd list has words of three syllables, the even list has two. The two lists have a maximum word length of 11 and 9 letters, respectively. Using a two-list scheme was suggested by Zhahai Stewart. == Examples == Each byte in a bytestring is encoded as a single word. A sequence of bytes is rendered in network byte order, from left to right. For example, the leftmost (i.e. byte 0) is considered "even" and is encoded using the PGP Even Word table. The next byte to the right (i.e. byte 1) is considered "odd" and is encoded using the PGP Odd Word table. This process repeats until all bytes are encoded. Thus, "E582" produces "topmost Istanbul", whereas "82E5" produces "miser travesty". A PGP public key fingerprint that displayed in hexadecimal as E582 94F2 E9A2 2748 6E8B 061B 31CC 528F D7FA 3F19 would display in PGP Words (the "biometric" fingerprint) as topmost Istanbul Pluto vagabond treadmill Pacific brackish dictator goldfish Medusa afflict bravado chatter revolver Dupont midsummer stopwatch whimsical cowbell bottomless The order of bytes in a bytestring depends on endianness. == Other word lists for data == There are several other word lists for conveying data in a clear unambiguous way via a voice channel: the NATO phonetic alphabet maps individual letters and digits to individual words the S/KEY system maps 64 bit numbers to 6 short words of 1 to 4 characters each from a publicly accessible 2048-word dictionary. The same dictionary is used in RFC 1760 and RFC 2289. the Diceware system maps five base-6 random digits (almost 13 bits of entropy) to a word from a dictionary of 7,776 distinct words. the Electronic Frontier Foundation has published a set of improved word lists based on the same concept FIPS 181: Automated Password Generator converts random numbers into somewhat pronounceable "words". mnemonic encoding converts 32 bits of data into 3 words from a vocabulary of 1626 words. what3words encodes geographic coordinates in 3 dictionary words. the BIP39 standard permits encoding a cryptographic key of fixed size (128 or 256 bits, usually the unencrypted master key of a Cryptocurrency wallet) into a short sequence of readable words known as the seed phrase, for the purpose of storing the key offline. This is used in cryptocurrencies such as Bitcoin or Monero. Like the PGP word list, the Bytewords standard maps each possible byte to a word. There is only one list, rather than two. The words are uniformly four letters long and can be uniquely identified by their first and last letters