Name EXT_timer_query Name Strings GL_EXT_timer_query Contact James Jones, NVIDIA Corporation (jajones 'at' nvidia.com) Contributors Axel Mamode, Sony Brian Paul, Tungsten Graphics Pat Brown, NVIDIA Remi Arnaud, Sony Status Shipping (version 1.0) Supported by NVIDIA Release 80 drivers. Version Last Modified Date: 2013/06/25 Revision: 3 Number 319 Dependencies Written based on the wording of the OpenGL 2.0 specification. OpenGL 1.5 is required. This extension modifies ARB_occlusion_query and NV_occlusion_query. Overview Applications can benefit from accurate timing information in a number of different ways. During application development, timing information can help identify application or driver bottlenecks. At run time, applications can use timing information to dynamically adjust the amount of detail in a scene to achieve constant frame rates. OpenGL implementations have historically provided little to no useful timing information. Applications can get some idea of timing by reading timers on the CPU, but these timers are not synchronized with the graphics rendering pipeline. Reading a CPU timer does not guarantee the completion of a potentially large amount of graphics work accumulated before the timer is read, and will thus produce wildly inaccurate results. glFinish() can be used to determine when previous rendering commands have been completed, but will idle the graphics pipeline and adversely affect application performance. This extension provides a query mechanism that can be used to determine the amount of time it takes to fully complete a set of GL commands, and without stalling the rendering pipeline. It uses the query object mechanisms first introduced in the occlusion query extension, which allow time intervals to be polled asynchronously by the application. Issues What time interval is being measured? RESOLVED: The timer starts when all commands prior to BeginQuery() have been fully executed. At that point, everything that should be drawn by those commands has been written to the framebuffer. The timer stops when all commands prior to EndQuery() have been fully executed. What unit of time will time intervals be returned in? RESOLVED: Nanoseconds (10^-9 seconds). This unit of measurement allows for reasonably accurate timing of even small blocks of rendering commands. The granularity of the timer is implementation-dependent. A 32-bit query counter can express intervals of up to approximately 4 seconds. What should be the minimum number of counter bits for timer queries? RESOLVED: 30 bits, which will allow timing sections that take up to 1 second to render. How are counter results of more than 32 bits returned? RESOLVED: Using the int64 and uint64 datatypes (introduced in OpenGL 3.2 and used by this extension as well), and corresponding GetQueryObject entry points. These types hold integer values and have a minimum bit width of 64. Should the extension measure total time elapsed between the full completion of the BeginQuery and EndQuery commands, or just time spent in the graphics library? RESOLVED: This extension will measure the total time elapsed between the full completion of these commands. Future extensions may implement a query to determine time elapsed at different stages of the graphics pipeline. This extension introduces a second query type supported by BeginQuery/EndQuery. Can multiple query types be active simultaneously? RESOLVED: Yes; an application may perform an occlusion query and a timer query simultaneously. An application can not perform multiple occlusion queries or multiple timer queries simultaneously. An application also can not use the same query object for an occlusion query and a timer query simultaneously. Do query objects have a query type permanently associated with them? RESOLVED: No. A single query object can be used to perform different types of queries, but not at the same time. Having a fixed type for each query object simplifies some aspects of the implementation -- not having to deal with queries with different result sizes, for example. It would also mean that BeginQuery() with a query object of the "wrong" type would result in an INVALID_OPERATION error. How predictable/repeatable are the results returned by the timer query? RESOLVED: In general, the amount of time needed to render the same primitives should be fairly constant. But there may be many other system issues (e.g., context switching on the CPU and GPU, virtual memory page faults, memory cache behavior on the CPU and GPU) that can cause times to vary wildly. Note that modern GPUs are generally highly pipelined, and may be processing different primitives in different pipeline stages simultaneously. In this extension, the timers start and stop when the BeginQuery/EndQuery commands reach the bottom of the rendering pipeline. What that means is that by the time the timer starts, the GL driver on the CPU may have started work on GL commands issued after BeginQuery, and the higher pipeline stages (e.g., vertex transformation) may have started as well. What should the new 64 bit integer type be called? RESOLVED: The new types will be called GLint64/GLuint64 The new command suffixes will be i64 and ui64. These names clearly convey the minimum size of the types. These types are similar to the C99 standard type int_least64_t, but we use names similar to the C99 optional type int64_t for simplicity. (note: previous versions of EXT_timer_query used GLint64EXT and and GLuint64EXT. These types were later promoted to core in OpenGL 3.2, and this extension was changed to use the core datatypes for compatibility with changes to the OpenGL ES EXT_disjoint_timer_query extension, which introduces the same query entry points, and would have otherwise have had different function signatures). New Procedures and Functions void GetQueryObjecti64vEXT(uint id, enum pname, int64 *params); void GetQueryObjectui64vEXT(uint id, enum pname, uint64 *params); New Tokens Accepted by the parameter of BeginQuery, EndQuery, and GetQueryiv: TIME_ELAPSED_EXT 0x88BF Additions to Chapter 2 of the OpenGL 2.0 Specification (OpenGL Operation) (Modify table 2.1, Correspondence of command suffix letters to GL argument types, p. 8) Add two new types and suffixes: Letter Corresponding GL Type ------ --------------------- i64 int64 ui64 uint64 (Modify table 2.2, GL data types, p. 9) Add two new types: Minimum GL Type Bit Width Description --------- --------- ------------------------------------ int64 64 signed 2's complement binary integer uint64 64 unsigned binary integer Additions to Chapter 3 of the OpenGL 2.0 Specification (Rasterization) None. Additions to Chapter 4 of the OpenGL 2.0 Specification (Per-Fragment Operations and the Framebuffer) (Replace Section 4.1.7, Occlusion Queries, p.204) Section 4.1.7, Asynchronous Queries Asynchronous queries provide a mechanism to return information about the processing of a sequence of GL commands. There are two query types supported by the GL. Occlusion queries (section 4.1.7.1) count the number of fragments or samples that pass the depth test. Timer queries (section 4.1.12) record the amount of time needed to fully process these commands. The results of asynchronous queries are not returned by the GL immediately after the completion of the last command in the set; subsequent commands can be processed while the query results are not complete. When available, the query results are stored in an associated query object. The commands described in section 6.1.12 provide mechanisms to determine when query results are available and return the actual results of the query. The name space for query objects is the unsigned integers, with zero reserved by the GL. Each type of query supported by the GL has an active query object name. If the active query object name for a query type is non-zero, the GL is currently tracking the information corresponding to that query type and the query results will be written into the corresponding query object. If the active query object for a query type name is zero, no such information is being tracked. A query object is created by calling void BeginQuery(enum target, uint id); with an unused name . indicates the type of query to be performed; valid values of are defined in subsequent sections. When a query object is created, the name is marked as used and associated with a new query object. BeginQuery sets the active query object name for the query type given by to . If BeginQuery is called with an of zero, if the active query object name for is non-zero, or if is the active query object name for any query type, the error INVALID_OPERATION is generated. The command void EndQuery(enum target); marks the end of the sequence of commands to be tracked for the query type given by . The active query object for is updated to indicate that query results are not available, and the active query object name for is reset to zero. When the commands issued prior to EndQuery have completed and a final query result is available, the query object active when EndQuery is called is updated by the GL. The query object is updated to indicate that the query results are available and to contain the query result. If the active query object name for is zero when EndQuery is called, the error INVALID_OPERATION is generated. The command void GenQueries(sizei n, uint *ids); returns previously unused query object names in . These names are marked as used, but no object is associated with them until the first time they are used by BeginQuery. Query objects are deleted by calling void DeleteQueries(sizei n, const uint *ids); contains names of query objects to be deleted. After a query object is deleted, its name is again unused. Unused names in are silently ignored. Calling either GenQueries or DeleteQueries while any query of any target is active causes an INVALID_OPERATION error to be generated. Query objects contain two pieces of state: a single bit indicating whether a query result is available, and an integer containing the query result value. The number of bits used to represent the query result is implementation-dependent. In the initial state of a query object, the result is available and its value is zero. The necessary state for each query type is an unsigned integer holding the active query object name (zero if no query object is active), and any state necessary to keep the current results of an asynchronous query in progress. Section 4.1.7.1, Occlusion Queries Occlusion queries use query objects to track the number of fragments or samples that pass the depth test. An occlusion query can be started and finished by calling BeginQuery and EndQuery, respectively, with a of SAMPLES_PASSED. When an occlusion query starts, the samples-passed count maintained by the GL is set to zero. When an occlusion query is active, the samples-passed count is incremented for each fragment that passes the depth test. If the value of SAMPLE BUFFERS is 0, then the samples-passed count is incremented by 1 for each fragment. If the value of SAMPLE BUFFERS is 1, then the samples-passed count is incremented by the number of samples whose coverage bit is set. However, implementations, at their discretion, may instead increase the samples-passed count by the value of SAMPLES if any sample in the fragment is covered. When an occlusion query finishes and all fragments generated by the commands issued prior to EndQuery have been generated, the samples-passed count is written to the corresponding query object as the query result value, and the query result for that object is marked as available. If the samples-passed count overflows, (i.e., exceeds the value 2^n - 1, where n is the number of bits in the samples-passed count), its value becomes undefined. It is recommended, but not required, that implementations handle this overflow case by saturating at 2^n - 1 and incrementing no further. (Add new Section 4.1.12, Timer Queries, p.212) Timer queries use query objects (section 4.1.7) to track the amount of time needed to fully complete a set of GL commands. A timer query can be started and finished by calling BeginQuery and EndQuery, respectively, with a of TIME_ELAPSED_EXT. When BeginQuery and EndQuery are called with a of TIME_ELAPSED_EXT, the GL prepares to start and stop the timer used for timer queries. The timer is started or stopped when the effects from all previous commands on the GL client and server state and the framebuffer have been fully realized. The BeginQuery and EndQuery commands may return before the timer is actually started or stopped. When the timer query timer is finally stopped, the elapsed time (in nanoseconds) is written to the corresponding query object as the query result value, and the query result for that object is marked as available. If the elapsed time overflows the number of bits, , available to hold elapsed time, its value becomes undefined. It is recommended, but not required, that implementations handle this overflow case by saturating at 2^n - 1. Additions to Chapter 5 of the OpenGL 2.0 Specification (Special Functions) None. Additions to Chapter 6 of the OpenGL 2.0 Specification (State and State Requests) (Replace Section 6.1.12, Occlusion Queries, p. 254) Section 6.1.12, Asynchronous Queries The command boolean IsQuery(uint id); returns TRUE if is the name of a query object. If is zero, or if is a non-zero value that is not the name of a query object, IsQuery returns FALSE. Information about a query target can be queried with the command void GetQueryiv(enum target, enum pname, int *params); identifies the query target and can be SAMPLES_PASSED for occlusion queries or TIME_ELAPSED_EXT for timer queries. If is CURRENT_QUERY, the name of the currently active query for , or zero if no query is active, will be placed in . If is QUERY_COUNTER_BITS, the implementation-dependent number of bits used to hold the query result for will be placed in params. The number of query counter bits may be zero, in which case the counter contains no useful information. For occlusion queries (SAMPLES_PASSED), if the number of bits is non-zero, the minimum number of bits allowed is a function of the implementation's maximum viewport dimensions (MAX_VIEWPORT_DIMS). The counter must be able to represent at least two overdraws for every pixel in the viewport. The formula to compute the allowable minimum value (where n is the minimum number of bits) is: n = min(32, ceil(log_2(maxViewportWidth * maxViewportHeight * 2))). For timer queries (TIME_ELAPSED_EXT), if the minimum number if bits is non-zero, it must be at least 30. The state of a query object can be queried with the commands void GetQueryObjectiv(uint id, enum pname, int *params); void GetQueryObjectuiv(uint id, enum pname, uint *params); void GetQueryObjecti64vEXT(uint id, enum pname, int64 *params); void GetQueryObjectui64vEXT(uint id, enum pname, uint64 *params); If is not the name of a query object, or if the query object named by is currently active, then an INVALID_OPERATION error is generated. If is QUERY_RESULT, then the query object's result value is returned as a single integer in . If the value is so large in magnitude that it cannot be represented with the requested type, then the nearest value representable using the requested type is returned. If the number of query counter bits for any is zero, then the result is returned as a single integer with a value of 0. There may be an indeterminate delay before the above query returns. If is QUERY_RESULT_AVAILABLE, FALSE is returned if such a delay would be required, TRUE is returned otherwise. It must always be true that if any query object returns a result available of TRUE, all queries of the same type issued prior to that query must also return TRUE. Querying the state for any given query object forces the corresponding query to complete within a finite amount of time. If multiple queries are issued using the same object name prior to calling GetQueryObject[u]iv, the result and availability information returned will always be from the last query issued. The results from any queries before the last one will be lost if they are not retrieved before starting a new query on the same and . GLX Protocol (Modification to the GLX 1.3 Protocol Encoding Specification) Add to Section 1.4 (p.2), Common Types INT64 A 64-bit signed integer value. CARD64 A 64-bit unsigned integer value. Two new non-rendering GL commands are added. These commands are sent seperately (i.e., not as part of a glXRender or glXRenderLarge request), using the glXVendorPrivateWithReply request: GetQueryObjecti64vEXT 1 CARD8 opcode (X assigned) 1 1328 GLX opcode (glXVendorPrivateWithReply) 2 4 request length 4 GLX_CONTEXT_TAG context tag 4 CARD32 id 4 ENUM pname => 1 1 reply 1 unused 2 CARD16 sequence number 4 m reply length, m=(n==1?0:n) 4 unused 4 CARD32 n if (n=1) this follows: 8 INT64 params 8 unused otherwise this follows: 16 unused n*8 LISTofINT64 params GetQueryObjectui64vEXT 1 CARD8 opcode (X assigned) 1 1329 GLX opcode (glXVendorPrivateWithReply) 2 4 request length 4 GLX_CONTEXT_TAG context tag 4 CARD32 id 4 ENUM pname => 1 1 reply 1 unused 2 CARD16 sequence number 4 m reply length, m=(n==1?0:n) 4 unused 4 CARD32 n if (n=1) this follows: 8 CARD64 params 8 unused otherwise this follows: 16 unused n*8 CARD64 params Errors All existing errors for query objects apply unchanged from the ARB_occlusion_query spec, except the modification below: The error INVALID_ENUM is generated if BeginQueryARB, EndQueryARB, or GetQueryivARB is called where is not SAMPLES_PASSED or TIME_ELAPSED_EXT. The error INVALID_OPERATION is generated if GetQueryObjecti64vEXT or GetQueryObjectui64vEXT is called where is not the name of a query object. The error INVALID_OPERATION is generated if GetQueryObjecti64vEXT or GetQueryObjectui64vEXT is called where is the name of a currently active query object. The error INVALID_ENUM is generated if GetQueryObjecti64vEXT or GetQueryObjectui64vEXT is called where is not QUERY_RESULT or QUERY_RESULT_AVAILABLE. New State (table 6.37, p 298) Update the occlusion query / query object state to cover timer queries: Get Value Type Get Command Init. Value Description Sec Attribute ---------------------- ---- ---------------- ----------- ------------------------- ----- --------- CURRENT_QUERY 2xZ+ GetQueryiv 0 Active query object name 4.1.7 - (occlusion and timer) QUERY_RESULT 2xZ+ GetQueryObjectiv 0 Query object result 4.1.7 - (samples passed or time elapsed) QUERY_RESULT_AVAILABLE 2xB GetQueryObjectiv TRUE Query object result 4.1.7 - available? New Implementation Dependent State (table 6.34, p. 295) Update the occlusion query / query object state to cover timer queries: Get Value Type Get Command Minimum Value Description Sec Attribute -------------------- ---- ----------- ------------- -------------------------- ------ --------- QUERY_COUNTER_BITS 2xZ+ GetQueryiv see 6.1.12 Asynchronous query counter 6.1.12 - bits (occlusion and timer queries) Dependencies on ARB_occlusion_query and NV_occlusion_query If ARB_occlusion_query or NV_occlusion_query is supported, the previous spec edits are considered to apply to the nearly identical language in these extension specifications. Note that the functionality provided by these extensions is included in OpenGL versions 1.5 and greater. Usage Examples Here is some rough sample code that demonstrates the intended usage of this extension. GLint queries[N]; GLint available = 0; // timer queries can contain more than 32 bits of data, so always // query them using the 64 bit types to avoid overflow GLuint64 timeElapsed = 0; // Create a query object. glGenQueries(N, queries); // Start query 1 glBeginQuery(GL_TIME_ELAPSED_EXT, queries[0]); // Draw object 1 .... // End query 1 glEndQuery(GL_TIME_ELAPSED_EXT); ... // Start query N glBeginQuery(GL_TIME_ELAPSED_EXT, queries[N-1]); // Draw object N .... // End query N glEndQuery(GL_TIME_ELAPSED_EXT); // Wait for all results to become available while (!available) { glGetQueryObjectiv(queries[N-1], GL_QUERY_RESULT_AVAILABLE, &available); } for (i = 0; i < N; i++) { // See how much time the rendering of object i took in nanoseconds. glGetQueryObjectui64vEXT(queries[i], GL_QUERY_RESULT, &timeElapsed); // Do something useful with the time. Note that care should be // taken to use all significant bits of the result, not just the // least significant 32 bits. AdjustObjectLODBasedOnDrawTime(i, timeElapsed); } This example is sub-optimal in that it stalls at the end of every frame to wait for query results. Ideally, the collection of results would be delayed one frame to minimize the amount of time spent waiting for the GPU to finish rendering. Revision History Version 3, June 25, 2013 (Jon Leech) - replace int64EXT / uint64EXT with core int64/uint64 types, for compatibility with EXT_disjoint_timer_query (Bug 10449).