Computer graphics processing and selective visual display system – Computer graphic processing system – Plural graphics processors
Reexamination Certificate
1999-01-29
2002-03-26
Chauhan, Ulka J. (Department: 2671)
Computer graphics processing and selective visual display system
Computer graphic processing system
Plural graphics processors
C345S552000, C345S536000, C345S567000, C345S572000
Reexamination Certificate
active
06362824
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to computer graphics. More particularly, the invention relates to techniques for accessing texture data stored in a computer for use in texture mapping operations by a graphics display subsystem.
BACKGROUND
In the field of computer graphics, texture mapping is a known technique used to create the appearance of complexity on the surface of rendered objects without actually having to model every detail of the object's surface. Typically, the technique involves mapping a two-dimensional function or array onto an object in three-dimensional object space and then projecting the resultant image back to two-dimensional screen space for display. The term “texture map” refers to the function or array that is used in the texture mapping process. A common array-based texture map might consist of a repeatable pattern for representing a material, such as wood or marble for example. Such texture maps are made up of a plurality of numerical values called texels. A texel's numerical value usually corresponds to an RGB color value and perhaps also to an alpha transparency value. (Although other parameters may be included in texture maps in addition to, or in lieu of, RGB and alpha values.) A texel's location within a texture map may be designated using s,t coordinates.
A technique known as mipmapping is also used in texture mapping. The acronym “mip” stands for multum in parvo—multiple things in one place. In mipmapping, numerous versions of the same texture map are created, each version representing the texture at a different resolution. After this has been done, the process of texture mapping may be performed using only those versions of the texture map that are appropriate given the resolution required for the object being drawn. To create the different versions of the texture map, a base texture map is down-sampled numerous times to develop a series of smaller texture maps, each of which represents the base map at a predetermined lower level of resolution. A more detailed explanation of mipmapping may be found, for example, in Lance Williams, “Pyramidal Parametrics,” Computer Graphics Vol. 17, No. 3, pp.1-10 (July, 1983). Hereinafter, the different resolution versions of a texture map that in the aggregate constitute a mipmap will be referred to as “pages,” or “mipmap pages.”
While mipmaps yield important efficiencies for rendering complex images, they can become burdensome in terms of the amount of memory that is required to store them. Indeed, the size of the mipmaps used to render an image can in some cases be larger than the rendered image itself. This is particularly problematic given that, in the past, texture data was stored exclusively within the graphics subsystem (such as in the frame buffer memory or in a dedicated texture memory). One technique now being used to work around the texture data storage problem is to store mipmaps in the system memory of the host computer as well as in memory located within the graphics subsystem. When this technique is employed, the graphics subsystem uses direct memory access to retrieve texture data from system memory as needed. This new technique is beneficial to the extent that it reduces the need for expensive memory in the graphics subsystem. (Ample and relatively inexpensive storage space is often available in system memory.) Unfortunately, the technique also creates new problems:
First, there is overhead associated with retrieving texture data from system memory into graphics subsystem memory. For example, each time a texel address is generated by graphics rendering hardware, the graphics subsystem must determine whether to initiate a direct memory access request to system memory for the required texel data, or whether simply to retrieve the texel data from a memory located locally within the graphics subsystem. In prior art machines, a texture address comparator is provided for this purpose. Reference addresses are stored to indicate the boundaries of system memory and graphics subsystem memory. Every time a texel address is generated, it is compared with the reference addresses to determine whether a direct memory access will be required. Texel addresses are generated with enormous frequency, and each address comparison takes time. Thus, storing texture data in system memory can reduce the performance of a graphics system even when texture data is not being moved from system memory to graphics subsystem memory. The extra hardware required to perform the address comparison function has a cost associated with it as well. It is therefore an object of the invention to provide a mechanism and technique for very quickly determining whether it is necessary to access the system memory of the host computer for the purpose of retrieving a particular unit of texture data.
Another problem that arises with storage of texture data in system memory has to do with fixed relative addressing. In prior art systems that use fixed relative addressing, all of the pages of a mipmap must be located within the same memory in order to be used. In other words, the texels in a mipmap structure can be accessed individually only as long as all of the pages of the mipmap are located either in system memory, frame buffer memory or texture memory. It is not permitted in such systems to locate one of the pages of a mipmap in frame buffer memory and the remaining pages in system memory, for example. This is because, in fixed relative addressing, the location of each texel in a mipmap is defined relative to the beginning address of the base page of the overall mipmap structure.
For example, referring now to
FIG. 1
, the pages of a mipmap n are used to represent one texture at varying levels of detail. The largest page of mipmap n, page 0, is used to represent the texture at the highest level of detail. Pages 1, 2 and 3 are used to represent the same texture at successively lesser levels of detail. Each of pages 1, 2 and 3 is a down-sampled version of the previous page. If page 0 were 256×256 texels in size, then pages 1, 2 and 3 would typically be 128×128, 64×64 and 32×32 texels in size, respectively. Each of the pages in mipmap n has a base address associated with it, indicated in the drawing at 100-106. In a fixed relative addressing scheme, the pages of a mipmap n are stored in memory in such a manner that only the base address of page 0 need be known in order to access texels within any of the pages of the mipmap n. In short, the base address location for a given page is “fixed” relative to the base addresses for the other pages in the same mipmap. The base address for page 0 of the mipmap is used as the primary point of reference. For example, the pages may be stored contiguously so that the base address for page 1 will always be equal to [the base address for page 0]+65,536*[size of a texel]. The base address for page 2 will be equal to [the base address for page 1]+16,384*[size of a texel], and so on. Alternatively, the pages of the mipmaps may be stored so that they are not contiguous with one another, so long as the base addresses for each of the down-sampled pages of a given mipmap may calculated solely from the base address of page 0 of that mipmap.
Although fixed relative addressing provides simplicity, it also involves constraints. For instance, in some applications it would be desirable to have some of the frequently-accessed pages of a mipmap reside in (fast) graphics subsystem memory and to have some of the less-frequently-accessed pages of the mipmap reside in (slower) system memory. In a system that uses fixed relative addressing, such a scheme is not possible. It is therefore a further object of the invention to provide the flexibility to store the pages of a mipmap at random locations either within the system memory of the host computer, in a frame buffer memory or in a dedicated texture memory.
SUMMARY OF THE INVENTION
The invention includes a number of unique aspects, each of which contributes to the achievement of th
Chauhan Ulka J.
Hart Kevin M.
Hewlett--Packard Company
LandOfFree
System-wide texture offset addressing with page residence... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System-wide texture offset addressing with page residence..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System-wide texture offset addressing with page residence... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2880699