StgSharp.Common Lock-free, fixed-capacity stack implemented on top of an array using atomic index operations. - No linked list; only array slots - Uses CAS on a shared top index to reserve/release slots - Per-slot state flag to safely publish/consume values under races Best-effort snapshot for enumeration; not strongly consistent under concurrent mutations. Determines whether the contains the specified key. The value to locate in the . true if the contains an element with the specified key; otherwise, false. Determines whether the contains the specified key. The value to locate in the . true if the contains an element with the specified key; otherwise, false. Cache hash result for obj with counts. Part of a collection. This contains a direct reference to _origin collection, so all change to _origin collection will be reflected in this object. Initializes a new instance of the RandomAccessStack class that is empty and has the default initial capacity. Initializes a new instance of the RandomAccessStack class that is empty and has the specified initial capacity. Gets or sets the element at the specified index in the stack (0 = bottom, Count-1 = top). Gets the number of elements contained in the stack. Removes all objects from the stack. Returns an enumerator that iterates through the stack from top to bottom. Throws if the stack is modified during iteration. Returns the object at the top of the stack without removing it. Removes and returns the object at the top of the stack. Adds an item to the top of the stack. CustomizeInit an instance of OpenGL program, This method should be called before any other World api. Version of current World platform. Main version represent in hundred's digit, and sub version represent in ten's digit. Write a log message to log file, and mark the _level of this message. A string contains information of how current evironment runs. The severity of current message. A class attempt to reduce cost of boxing, though still a bit slower than direct access to _origin value type instance. Type of instance to be boxed. A class contains a sets of methods to serialize several types An independent storage area, used as the only folder type in root folder. Contains several files reference as virtual disc. Data is actually stored in s in certain RAID mode. 当几何体调用了过多的CalcVec方法时，产生UnusedVertexException异常 Plain geometry only contains straight sides. Calculate movement to center CurrentPipeline time tick A square Get all uniform related to this . Shader program requires this camera. ContextName of all related uniforms. If you fallow form from A two dimension vector defined by two elements. Vec2 in World are default used as colum vector. Orthogonalize two s, let their length be 1 and perpendicular to each other. Length of the matrix in columns. Length of the matrix in rows. Reurns the value of ArcSin(x) The sine value of the angle Get the value of E ^ x; Returns the value of ln(x). calculate the value of x^y Returns the value of x^y returns a random float number between 0 and 1. Generates a random seed in int value. Interface of all kinds OpenGL BufferHandle objects Get the only handle to one of the Object instance Bind a BufferHandle instance to OpenGL Index of handle of the object in this instance to be bind Function handler to load an opengl function by searching its _label. the _label of the Opengl function An Intptr value representing the pointer to the function Hash code of time pixels updated last time. A collection of a sets of OpenGL textures. The collection can automatically select one of the texture with minimum costs to upgrade pixels to BufferHandle. Usually is used for loading, but not for target of . A collection of handle to Element BufferHandle Object in OpenglFunc. Each value indexed from instance of this class this is the only handle to one Element BufferHandle Object. Bind the frame BufferHandle object with _label frame BufferHandle to the frame BufferHandle target specified by target. will be set as default target. Index of object to be binded in handle set Bind the frame BufferHandle object with _label frame BufferHandle to the frame BufferHandle target specified by target. Target of this frame BufferHandle should be set manually. The frame BufferHandle target of the binding operation. Index of object to be binded in handle set Load an to certain texture. Set _data to current vertex BufferHandle object Type of bufferData Index to find certain VBO in this instance Data to write in How OpenGL use these _data, defined by Set _data to current vertex BufferHandle object Type of bufferData Index to find certain VBO in this instance Data to write in How OpenGL use these _data, defined by Set _data to current vertex BufferHandle object Type of bufferData Index to find certain VBO in this instance A of _data to write in How OpenGL use these _data, defined by Set data to current vertex BufferHandle object Type of bufferData Index to find certain VBO in this instance A of data to write in How OpenGL use these data, defined by Set _data to current vertex BufferHandle object Type of bufferData Index to find certain VBO in this instance A array of _data to write in How OpenGL use these _data, defined by Get the function pointer of given OpenGL api _label ContextName of the OpenGL api The funtion pointer of the api in stream of Load all supported OpenGL APIs and bind them to given . The handle to current OpenGL viewPortDisplay Create a set of . Amount of EBO to be created. Create a sets of Amount of textures to be created. Create a set of . Amount of VAO to be created. Create a set of . Amount of Object to be created. Initialize an opengl context and load all opengl functions. Initialize an OpenGl viewPortDisplay if current form has not been binded to any viewPortDisplay. and set the viewPortDisplay binded to current form as current OpenGL viewPortDisplay. Something goes wrong and World fails in creating a new OpenGL viewPortDisplay. Create an instance of program New instance of instance. OpenGL numeric type. An IEEE-754 floating-point value, clamped to the range [0,1]. OpenGL numeric type. An IEEE-754 double precision floating-point value, clamped to the range [0,1]. Attach one shader code to a shader program. Shader program to attach Index of shader code in current shader code set. CustomizeInit a program with no shader attached. Get a uniform form current shader, the value type of the uniform should be provided. ContextName of the Uniform in shader code Get a uniform form current shader, the value type of the uniform should be provided. ContextName of the Uniform in shader code Get a uniform form current shader, the value type of the uniform should be provided. ContextName of the Uniform in shader code Get a uniform form current shader, the value type of the uniform should be provided. ContextName of the Uniform in shader code Link the shader to current viewPortDisplay. Let the current viewPortDisplay use this shader. Set the Size limit of current form Minimum width of current form Minimum height of current form Maximum width of current form Maximum height of current form Get (TView is ) uniform from a shader. This uniform will be used as projection matrix of this camera. Additionally, for multi- shader rendering, each shader requires a uniform, so do not ignore return value of this method. Shader using this camera ContextName of projection camera named in shader (TView is ) representing projection matrix in shader program. Marking the start of current render operations, and set this renser context as current. Marking the end of current render operations, and release the binding of this renser context to current, making it available for another thread as current. Provide a lock for managing concurrent access to a buffer that can be expanded and use a common two-pase-copy expanding mechanism. An individual thread doing nothing, and its always returns false. A single-consumer bell/signal primitive. Producers can call multiple times to publish signals. The bell keeps a pending count of not-yet-consumed signals. The consumer uses or to consume signals. When a producer increments the pending count from 0 to 1, the internal event is set to wake the consumer. Consuming via or clears the entire pending count (aggregate consumption). If new signals arrive while resetting, the event is set again to avoid missed signals. Ownership: only the specified owner can call the consumer APIs. Initializes a new instance of . The unique consumer identity; subsequent calls to Wait/TryWait must pass the same reference. Spin count passed to the underlying . Thrown if is null. Disposes the bell, sets the event to wake any waiting consumer, and releases resources. Publishes a signal by incrementing the not-yet-consumed count and returns the current pending count. If the pending count transitions from 0 to 1, the event is set to wake the consumer. The current pending count including this call. Attempts to consume any pending signals without blocking. The consumer object; must be the same reference provided at construction. true if a signal was available (pending or event set); otherwise false . Thrown if is not the owner. Blocks until a signal is available, then consumes all pending signals. The consumer object; must be the same reference provided at construction. A token to cancel the wait. Thrown if is not the owner. A counting semaphore with adjustable maximum capacity at runtime. Implemented using one or more ManualResetEventSlim (MRES) gates and atomic counters to avoid kernel semaphore overhead while supporting high-frequency Release/Wait patterns. Notes: - This is a counting primitive (not a barrier). Permits accumulate up to . - Reducing does not revoke already issued permits; it only clamps future releases. - Striped MRES can reduce contention in highly parallel waits. Current number of available permits. Logical maximum number of permits. Can be changed at runtime. Release a single permit if under . Returns the previous count. Release multiple permits (capped by ). Returns the previous count. Try to take a permit without blocking. Returns true if acquired. Block until a permit is available, then consume one. Block until a permit is available or canceled, then consume one. Abstraction for a wide SIMD register or equivalent packed numeric storage. Provides raw reference access, broadcast initialization and element level access. Typical implementations map to 128/256/512-bit hardware registers or software emulated blocks. Gets a reference to the first element of the register reinterpreted as type . Unmanaged numeric element type implementing . Writable reference to element 0 (lowest lane / lowest bits). Intended for treating the register as a contiguous block of elements. The returned reference must remain valid for the lifetime of the register and be identical to with index 0. Caller must choose a type layout compatible with the actual register representation. Fills all elements (lanes) of the register with the provided scalar . Unmanaged numeric element type implementing . Scalar value to replicate to every element. Should map to a single broadcast/shuffle instruction where possible. After completion all lanes read back the same value. Gets a writable reference to the element at the specified . Unmanaged numeric element type implementing . Zero-based lane index (range defined by implementation). Writable reference to the lane. Enables scalar lane edits without unpacking the entire register. Implementations may omit bounds checks for performance. Caller is responsible for passing a valid index. Buffer for SSE targeted matrix panel, size of a panel is 4x4 elements, and no smaller than 16 bytes. Type of matrix element. Buffer for AVX targeted matrix panel, size of a panel is 4x4 elements, and no smaller than 16 bytes. Type of matrix element. Buffer for AVX512 targeted matrix panel, size of a panel is 4x4 elements, and no smaller than 16 bytes. Type of matrix element. Try to allocate a new 16MB block when all buckets are empty Output handle to the allocated Entry True if allocation succeeded Helper method for allocation - converts BucketNode to Entry BucketNode structure - specialized node for bucket linked lists Stored in main allocation area (user-overwritable region) Free allocated memory block and perform adjacent block merging Handle to the allocated memory block Determine segment index based on size and originLevel Check if two Entries are adjacent in memory Used in Free operations, efficiently remove from buckets and merge adjacent blocks. Current Entry will serve as the final merged block. The Merge will be in lazy strategy, any fail in acquiring spin lock will end a merging process. Remove node from position linked list Entry structure - streamlined version (only position and state information) Stored in external SLAB Creates a new SLAB allocator instance with the specified configuration. Initial number of objects that can be allocated. Memory layout strategy for the SLAB allocator. Whether to enable concurrent access support. A new SLAB allocator instance. Thrown when an invalid layout is specified. Thrown when requesting unsupported configurations. Important Threading Considerations: While SLAB /// allocators themselves may be designed for single-threaded use, allocated objects are /// frequently shared across multiple threads in real-world applications. Due to the /// special implementation of expandable concurrent sequential SLAB allocators, they rely /// on internal expansion locks to ensure that memory buffer expansions (which involve /// copying the entire buffer to a new location) do not occur while other threads are /// reading from allocated regions. Sequential Layout Limitations: ///
Sequential layout SLAB allocators cannot provide a single-threaded /// version because:They use contiguous memory /// layout that requires buffer expansion when capacity is exceeded ///Buffer expansion involves copying all /// existing data to a new memory location /// Without proper synchronization, concurrent access to allocated objects during /// expansion would cause memory corruptionThe /// expansion lock mechanism is essential even if allocation itself is single- /// threaded Chunked Layout: ///
Chunked layout allocators can optionally provide single-threaded /// versions since they don't require buffer expansion and copying. Each chunk is /// allocated independently, avoiding the memory relocation issues present in sequential /// layouts. CustomizeInit this form and render it on screen, and respond with operation Height of this form, represent in Width of this form, represent in Width of this form, represent in Set the Size limit of current form Minimum Size of current form. In form of a of (width,height) Maximum Size of current form. In form of a of (width,height) Set the Size limit of current form Minimum width of current form Minimum height of current form Maximum width of current form Maximum height of current form ViewBase is the base type for all view types, providing context binding and control element binding. A complete view should contain one complete , , and , which provide layout descriptions of control elements, user responses, and rendering implementations in the view. Data between instances of , , and is shared through . Get the status of a certain key on the keyboard. representing if a key is pressed or released. ViewDesigner provides descriptions of various controls in the view, binds the controls to specific user actions and behaviors, and describes the layout and interactions of each control. Reset pipeline has risk, this operation is a direct operation to port in current node, the port on the other node cannot get the change. Only the following occasions of calling are valid: 1. no bp is running or the only bp is quitting. 2. calling from the main thread or a thread the current bp is running on. A sets of complex logic designed for running in multi threads environment. Warning: do not run a instance in multi threads more than once. TimeSpanProvider: uses a bound TimeSequence and TimeSourceProviderBase to publish span ticks. Each OnSpanTick advances the sequence; when sequence ends provider unsubscribes. True if there is at least one unconsumed span tick (PendingSpanCount > 0). Number of span ticks not yet consumed by Wait ( > 0 means last tick not waited). Tick passed since binded time source. True if the span provider is working. False if time is over (the sequence ended). Efficient adaptive backoff strategy that dynamically selects optimal backoff methods based on system load and success rate Gets the currently used backoff method (for debugging/monitoring) Gets the globally considered optimal backoff method Creates a new adaptive backoff strategy instance Gets global statistics for the specified strategy (for monitoring) Called after CAS failure, performs backoff and may switch strategies Called after CAS success, resets state Evaluates and updates the global optimal strategy Fast pseudo-random number generator (LCG) Performs the actual backoff operation Checks whether global strategy evaluation should be performed Determines whether strategy should be switched Switches to the currently considered optimal strategy