Difference between revisions of "Buffer Object"

From OpenGL.org
Jump to: navigation, search
(Immutable Storage: Immutable storage and usage. Still need to add a section on persistent and coherent.)
(Buffer Corruption: stub section for persistent mapping.)
Line 213: Line 213:
How often does this happen? On Microsoft Windows 5.1 (XP) and below, video memory could get trashed anytime an application didn't have input focus. This is why alt-tabbing away from games takes a long time to recover from; the application/OpenGL has to reload all of this data back to video memory. Fortunately, on Windows 6.0 (Vista) and above, this is fixed; Windows itself manages video memory and will ensure that all video memory is retained. Thus, at least theoretically, this should never be a problem on Vista or above machines.
How often does this happen? On Microsoft Windows 5.1 (XP) and below, video memory could get trashed anytime an application didn't have input focus. This is why alt-tabbing away from games takes a long time to recover from; the application/OpenGL has to reload all of this data back to video memory. Fortunately, on Windows 6.0 (Vista) and above, this is fixed; Windows itself manages video memory and will ensure that all video memory is retained. Thus, at least theoretically, this should never be a problem on Vista or above machines.
==== Persistent mapping ====
{{infobox feature
| name = Immutable Storage
| core = 4.4
| core_extension = {{extref|buffer_storage}}
==== Performance Concerns ====
==== Performance Concerns ====

Revision as of 00:40, 28 July 2013

Buffer Objects are OpenGL Objects that store an array of unformatted memory allocated by the OpenGL context (aka: the GPU). These can be used to store vertex data, pixel data retrieved from images or the framebuffer, and a variety of other things.


Buffer Objects are OpenGL Objects; they therefore follow all of the rules of regular OpenGL objects. To create a buffer object, you call glGenBuffers. Deleting them uses glDeleteBuffers. These use the standard Gen/Delete paradigm as most OpenGL objects.

As with the standard OpenGL object paradigm, this only creates the object's name, the reference to the object. To allocate storage, you must bind it to the context. You do this using the following API:

 void glBindBuffer(enum target, uint bufferName)

The target defines how you intend to use this binding of the buffer object. When you're just creating and/or filling the buffer object with data, the target you use doesn't technically matter. It matters more when you intend to tell OpenGL to use the data in the buffer in some way.

Performance Note: In the technical sense, the target a buffer is bound to does not matter for the purposes of creating the memory storage for it. However, OpenGL implementations are allowed to make judgments about your intended use of the buffer object based on the first target you bind it to. So if you intend for your buffer object to be used as a vertex array buffer, you should bind that buffer to GL_ARRAY_BUFFER first. You may later use it as a GL_TRANSFORM_FEEDBACK buffer for readback or whatever, but binding it to GL_ARRAY_BUFFER gives the implementation important information about how you plan to use it overall.

Buffer objects hold a linear array of memory of arbitrary size. This memory must be allocated before it can be uploaded to or used. There are two ways to allocate storage for buffer objects: mutable or immutable. Allocating immutable storage for a buffer changes the nature of how you can interact with the buffer object.

Immutable Storage

Immutable Storage
Core in version 4.5
Core since version 4.4
Core ARB extension ARB_buffer_storage

Much like immutable storage textures, the storage for buffer objects can be allocated immutably. When this is done, you will be unable to reallocate that storage. You may still invalidate it with an explicit invalidation command or through mapping the buffer. But you cannot do the glBufferData(..., NULL)​ trick to invalidate it if the storage is immutable.

To allocate immutable storage for buffer objects, you call this function:

void glBufferStorage(GLenum target​, GLsizeiptr size​, const GLvoid * data​, GLbitfield flags​);

The target​ parameter is just like the one for glBindBuffer; it says which bound buffer to modify. size​ represents how many bytes you want to allocate in this buffer object.

data​ is a pointer to an array of bytes of size​ in length. OpenGL will copy that data into the buffer object upon initialization. You may pass NULL for this parameter; if you do, the initial contents of the buffer will be undefined. You can clear the buffer after allocation (if it is modifiable from the client, see below) if you wish to update it.

The flags​ field sets up a contract between you and OpenGL, describing how you may and may not access the contents of the buffer.

Immutable access methods

Immutable storage buffer objects allow you to tell OpenGL that you will only access the buffer object in a certain way. The flags​ bitfield is a series of bits that describes what restrictions you will operate under.

No client modification: If you set flags​ to 0, then the only processes that can modify the data within the buffer are OpenGL's internal processes. You are forbidden from uploading data to the buffer after the initial creation (you can give the buffer initial values with the data​ parameter). You can still write to it in the following ways:

Basically, you're forbidden to use any operation that transfer bulk data directly from the client (ie: you) to OpenGL.

You can still read from the buffer, but not by mapping it. You can only use glGetBufferSubData to read from it.

This is useful for buffers who's contents are pure in-OpenGL processes. For example, if you're doing indirect rendering, that's generally because the rendering parameters are computed by a Compute Shader or some other in-GPU process. Such a buffer's contents will be generated by the shader or GPU process; all the user needs to do is allocate the storage and hand it off to the process that generates it. This is also useful for storing the results of Transform Feedback operations, to later be used for vertex data.

However, this is also useful for operations when you only need to update the storage once (remember: glBufferStorage's data​ parameter will still be copied in). Static vertex data is one such application.

No client modifications, with read mapping: You can set the GL_MAP_READ_BIT in flags​. This will allow you to map the buffer, but only for reading data from it. Attempting to map it for writing will still fail.

This can be useful for buffer objects that are used for downloading rendered images from OpenGL. You would map them to make it easier or faster to read from them.

Client update, without mapping: The flags field can take the GL_DYNAMIC_STORAGE_BIT. This allows you to directly upload data to the buffer after creation. However, you are still forbidden from mapping the buffer in any way.

This is probably not the most useful combination of flags. Especially since mapping can sometimes give you a big performance win over non-mapped buffer updates. Particularly when streaming data.

Client update, with mapping: You can set the GL_MAP_WRITE_BIT and the GL_DYNAMIC_STORAGE_BIT bits in flags​. This will allow you to map the buffer for writing, in addition to being able to update it with the usual functions. However, if you set the map write bit, you must always specify GL_DYNAMIC_STORAGE_BIT. So you can't cheat the meaning of not allowing user modification by saying that you want to map the buffer for writing.

This combination is probably the most flexible (potentially in tandem with GL_MAP_READ_BIT). But you should only ask for functionality you intend to use. If you truly are going to update the buffer only once, it's best to collate that data ASAP, then update the buffer with it at creation time. This is as opposed to creating the buffer, then uploading pieces of your data to it, and then never uploading to it again. If your buffer really will be static after a point, then wait until you have all the data before you create the buffer.

Note: flags​ is a bitfield, and many of the bits are orthogonal to one another. So you can set both GL_MAP_READ_BIT and GL_MAP_WRITE_BIT just fine. Of course, you'll also need GL_DYNAMIC_STORAGE_BIT because of the needs of the write bit. The map read bit doesn't care if you can also update it from the client, and vice-versa.

Mapping while in use: Normally, buffers cannot be used by other OpenGL processes while they are mapped. Attempting to do so will either result in an OpenGL error or undefined behavior.

However, if the buffer uses immutable storage (and only if it does so), you may specify the GL_PERSISTENT_BIT, which allows the buffer to be used while it is mapped. You can also add the GL_COHERENT_BIT to make it a bit more user-friendly (though almost certainly slower). The specific meaning of these will be described in the section on buffer mapping.

If you set the GL_PERSISTENT_BIT bit, you must set one of the mapping bits. And if you set the GL_COHERENT_BIT, you must also set the GL_PERSISTENT_BIT bit.

That other thing: You can also set the GL_CLIENT_STORAGE_BIT in flags​. This has no actual meaning for how you can use the buffer object. It's just a hint that OpenGL should allocate memory that is closer to the client, if such memory pools exist.

Mutable Storage

To create mutable storage for a buffer object, you use this API:

 void glBufferData(enum target, sizeiptr size, const void *data, enum usage)

The target parameter is just like the one for glBindBuffer; it says which bound buffer to modify. size represents how many bytes you want to allocate in this buffer object.

The data parameter is a pointer to user memory that will be copied into the buffer object's data store. If this value is NULL, then no copying will occur, and the buffer object's data will be undefined.

The usage parameter can be very confusing.

Buffer Object Usage

Buffer objects are general purpose memory storage blocks allocated by OpenGL. They are intended to be used in a great many ways. To give the implementation great flexibility in exactly what a particular buffer object's data store will be, so as to better optimize performance, the user is required to give usage hints. These provide a general description as to how exactly the user will be using the buffer object.

There are two independent parts to the usage pattern: how the user will be reading/writing from/to the buffer, and how often the user will be changing it relative to the use of the data.

There are two ways for data to cause a change in the data contents of a buffer object. One way is for the user to explicitly upload some binary data. The other way is for the user to issue GL commands that cause the buffer to be filled in. For example, if you want to have a buffer store the results of a vertex shader computation through the use of transform feedback, the user is not directly changing the buffer information. So this is the latter kind of change.

Similarly, the user can read a buffer's data, using a variety of commands. Or, the user can execute an OpenGL command that causes the GL to read the contents of the buffer and do something based on it. Buffers storing vertex data are read by the GL when rendering.

There are three hints that the user can specify the data. They are all based on what the user will be doing with the buffer. That is, whether the user will be directly reading or writing the buffer's data.

  • DRAW: The user will be writing data to the buffer, but the user will not read it.
  • READ: The user will not be writing data, but the user will be reading it back.
  • COPY: The user will be neither writing nor reading the data.

DRAW is useful for, as the name suggests, drawing. The user is uploading data, but only the GL is reading it. Buffer objects holding vertex data are generally specified as DRAW, though there can be exceptions.

READ is used when a buffer object is used as the destination for OpenGL commands. This could be rendering to a Buffer Texture, using arbitrary writes to buffer textures, doing a pixel transfer into a buffer object, using Transform Feedback, or any other OpenGL operation that writes to buffer objects.

COPY is used when a buffer object is used to pass data from one place in OpenGL to another. For example, you can read image data into a buffer, then use that image data as vertex data in a draw call. Your code never actually sends data to the buffer directly, nor does it read data back. You can also use Transform Feedback to achieve the same thing in a more direct way. You have the feedback data go to a buffer object, then use that buffer object as vertex data. And while the user is causing the buffer to be updated via rendering commands, at no time is the user directly either reading from or writing to the buffer's storage.

There are three hints for how frequently the user will be changing the buffer's data.

  • STATIC: The user will set the data once.
  • DYNAMIC: The user will set the data occasionally.
  • STREAM: The user will be changing the data after every use. Or almost every use.

STREAM is pretty easy to understand: the buffer object's storage will be updated after almost every use. STATIC is pretty easy to understand too. The buffer object's contents will be updated once and never changed.

What is unclear is when DYNAMIC becomes STREAM or STATIC. These are only hints, after all. It is perfectly legal OpenGL code to modify a STATIC buffer after it has been created, or to never modify a STREAM buffer.

Is it better to use STATIC for buffers that are updated very infrequently? Is it better to use DYNAMIC for buffers that get updated frequently, but not at STREAM speed? Is it better to use DYNAMIC for buffers that get partially updated? These are questions that can only be answered with careful profiling. And even then, the answer will only be accurate for that particular driver version from that particular hardware vendor.

In any case, STREAM, STATIC, and DYNAMIC can be matched with READ, DRAW, and COPY in any combination. STREAM_COPY means that you will be doing transform feedback writes (or other kinds of GL-based writes) into the buffer after almost every use; it will not be updated with glBufferSubData or similar functions. STATIC_READ means that you will fill the buffer up from the GL, but you will only do this once.

Data Specification

We have seen that glBufferData can be used to update the data in a buffer object. However, this also reallocates the buffer object's storage. This is not usually what one wants, as recreating the buffer can often be a heavyweight operation.

Instead, one can use the following API:

 void glBufferSubData(enum target, intptr offset, sizeiptr size, const void *data)

The offset parameter is an integer offset into the buffer object where we should begin updating. The size parameter is the number of bytes we should copy out of data. For obvious reasons, data cannot be NULL.


Buffer Object
Core in version 4.5
Core since version 4.3
Core ARB extension ARB_clear_buffer_object

A buffer object's storage can be cleared, in part or in full, to a specific value. These functions work in a similar fashion to Pixel Transfer operations, though with some significant differences:

void glClearBufferData(GLenum internalformat​, GLenum format​, GLenum type​, const void * data​);
void glClearBufferSubData(GLenum internalformat​, GLintptr offset​, GLsizeiptr size​, GLenum format​, GLenum type​, const void * data​);

glClearBufferData works as glClearBufferSubData, except it operates on the entire buffer's contents. It does not respecify the storage for the buffer.

internalformat​ must be a sized Image Format, but only of the kind that can be used for buffer textures. This defines how OpenGL will store the data in the buffer object. format​ and type​ operate as normal for Pixel Transfer operations.

data​ is a pointer to a single pixel's worth of data, rather than the rows that are used in actual pixel transfers. So if format​ is GL_RG and type​ is GL_UNSIGNED_BYTE, then data​ should be a pointer to an array of two GLubyte​s.

This function will copy the given data repeatedly throughout the specified range of the buffer. offset​ must be a multiple of the byte size defined by the internalformat​, as must size​.


Data can be copied from one buffer object to another. To do this, first bind the source and destination buffers to different target​s. These could be any target, but GL_COPY_READ_BUFFER​ and GL_COPY_WRITE_BUFFER have no special semantics, so they make useful targets for this purpose.

Once both are bound, use this function:

void glCopyBufferSubData(GLenum readtarget​, GLenum writetarget​, GLintptr readoffset​, GLintptr writeoffset​, GLsizeiptr size​);

readtarget​ is the buffer you bound the source buffer to. So this is where you get the data to copy from. writetarget​ is the buffer you bound the destination buffer to. readoffset​ is the byte offset from the beginning of the source buffer to start reading. writeoffset​ is the byte offset from the beginning of the destination buffer to start writing to. size​ is the number of bytes to read.

Errors will be given if the offset/sizes would cause reading to or writing from locations outside of the respective buffer objects' storage.


glBufferSubData is a nice way to present data to a buffer object. But it can be wasteful in performance, depending on your use patterns.

For example, if you have an algorithm that generates data that you want to store in the buffer object, you must first allocate some temporary memory to store that data in. Then you can use glBufferSubData to transfer it to OpenGL's memory. Similarly, if you want to read data back, glGetBufferSubData is perhaps not what you need, though this is less likely. It would be really nice if you could just get a pointer to the buffer object's storage and write directly to it.

You can. To do this, you must map the buffer. This gives you a pointer to memory that you can write to or read from, theoretically, just like any other. When you unmap the buffer, this invalidates the pointer (don't use it again), and the buffer object will be updated with the changes you made to it.

While a buffer is mapped, you can freely unbind the buffer. However, you cannot call any function that would cause OpenGL to read, modify, or write to that buffer while it is mapped. Thus, calling glBufferData is out, as is using any function that would cause OpenGL to read from it (rendering with a VAO that uses it, etc).

To map a buffer, you call glMapBufferRange. The signature of this function is this:

 void *glMapBufferRange(GLenum target​, GLintptr offset​, GLsizeiptr length​, GLbitfield access​);

The return value is the pointer to the buffer object's data. The parameters offset​ and length​ allow you to specify a particular range within the buffer to map; you do not have to map the entire buffer. The target​ parameter refers to the particular target that you have the buffer you want to map bound to.

The access​ parameter is a bit complicated.

You can tell OpenGL what you intend to do with the pointer. If you're just adding new data to the buffer, then the pointer that gets returned isn't something you need to read from. Similarly, it could be your intention to read from this pointer and not change the buffer; OpenGL could do a simple copy from the buffer's memory to scratch memory in the client address space to make reading the buffer faster (buffer memory may not be optimized for reading).

You specify this by setting bitflags in the access​ parameter. This parameter must have either the GL_MAP_READ_BIT or GL_MAP_WRITE_BIT set; it can have both (ie: the memory will be fit for reading and writing) but it can't have neither. There are a number of other flags that can be set.

Unmapping the buffer is done when you are finished with the pointer and want to let OpenGL know that the buffer is free to be used. This is done with the function glUnmapBuffer. This function takes only the target that the buffer in question is bound to. After calling this function, you should not use the pointer returned in the map call again.


Map Buffer Alignment
Core in version 4.5
Core since version 4.2
Core ARB extension ARB_map_buffer_alignment

The alignment of the pointer returned by mapping functions is important when dealing with types that need highly restricted alignment. For example, SSE intrinsics need 16-byte alignment, while AVX intrinsics need 32-byte alignment.

The base alignment of a buffer object is defined as the alignment of the pointer retrieved by calling glMapBuffer or glMapBufferRange with a zero offset​. The base alignment of a buffer is implementation defined, but it must be no less than GL_MIN_MAP_BUFFER_ALIGNMENT. The smallest this value will be is 64, thus allowing this to work for any strictly aligned type in existence and some that aren't.

The alignment of the pointer returned by glMapBufferRange with a non-zero offset​ is the modulus of the offset​ by the base alignment.

Buffer Corruption

There is one major gotcha when mapping a buffer.

During normal OpenGL operations, the OpenGL specification requires that all data stored in OpenGL objects be preserved. Thus, if something in the operating system or other such things causes video memory to be trashed, the OpenGL implementation must ensure that this data is restored properly.

Mapping is not a normal operation. Because of its low-level nature, these protections have to be relaxed. Therefore, it is possible that, during the time a buffer is mapped, some kind of corruption happens. If this occurs, calling glUnmapBuffer will return GL_FALSE. At that point, the contents of the buffer in question are considered undefined. It may have your data, or it may have random garbage.

How often does this happen? On Microsoft Windows 5.1 (XP) and below, video memory could get trashed anytime an application didn't have input focus. This is why alt-tabbing away from games takes a long time to recover from; the application/OpenGL has to reload all of this data back to video memory. Fortunately, on Windows 6.0 (Vista) and above, this is fixed; Windows itself manages video memory and will ensure that all video memory is retained. Thus, at least theoretically, this should never be a problem on Vista or above machines.

Persistent mapping

Immutable Storage
Core in version 4.5
Core since version 4.4
Core ARB extension ARB_buffer_storage

Performance Concerns

One thing to remember about buffer mapping is this: the implementation is not obligated in any way to give you an actual pointer to the buffer object's memory. It is perfectly capable of giving you a pointer to some memory that OpenGL allocated just for the purpose of mapping, then it will do the copy on its own time.

The other thing to remember is that you should not care. OpenGL likes to give implementations flexibility to make performance optimizations. If mapping a certain buffer gives you some other pointer, and the implementation will do the copy on its own time, you should assume that this is probably the fastest way to work. It's still potentially faster than glBufferSubData, since the copy in glBufferSubData must happen before the function call returns, whereas the copy in the mapping case can happen in a thread the GL spawns. The worst case is that it's no slower than glBufferSubData.

However, you should not use the pointer you are given like any other pointer you might have. If this pointer is a pointer to non-standard memory (uncached or video memory), then writing to it haphazardly can be problematic. If you are attempting to stream data to the buffer, you should always map the buffer only for writing and you should write sequentially. You do not need to write every byte, but you should avoid going backwards or skipping around in the memory.

The purpose in writing sequentially is to be able to use write-combine memory, a feature of some processors (most x86's). It allows sequential writes to uncached memory to flow fairly quickly, compared to random writes to uncached memory.


Buffer invalidation
Core in version 4.5
Core since version 4.3
Core ARB extension ARB_invalidate_subdata

glMapBufferRange can invalidate part of a buffer or the entire thing using GL_MAP_INVALIDATE_RANGE_BIT and GL_MAP_INVALIDATE_BUFFER_BIT, respectively. You can force invalidation of a buffer without mapping it by using the following functions:

void glInvalidateBufferData(GLuint buffer​);
void glInvalidateBufferSubData(GLuint buffer​, GLintptr offset​, GLsizeiptr length​);

glInvalidateBufferData is equivalent to calling glInvalidateBufferSubData with offset​ as 0 and length​ as the size of the buffer's storage.

When a buffer or region thereof is invalidated, it means that the contents of that buffer are now undefined. How OpenGL handles invalidation is up to it. Note that any pending operations on the buffer will still complete. Any pending reads from OpenGL will still get their original values, and any pending writes will still write their values.

The idea is that, by invalidating a buffer or range thereof, the implementation will simply grab a new piece of memory to use. Thus, while previously executed GL commands can still read the buffer's data, you can fill the invalidated buffer with new values (via mapping or glBufferSubData) without causing synchronization.


Streaming is the process of frequently uploading data to a buffer object and then using that buffer object in some OpenGL process. Making this as efficient as possible is a delicate operation. Buffer objects provide a number of possible usage patterns for streaming, and which ones work best is not entirely clear. Testing should be done with the hardware of interest to make sure that you get optimal streaming performance.

The key to streaming is parallelism. The OpenGL specification permits an implementation to delay the execution of drawing commands. This allows you to draw a lot of stuff, and then let OpenGL handle things on its own time. Because of this, it is entirely possible that well after you called the rendering function with a buffer object, you might start trying to stream vertex data into that buffer. If this happens, the OpenGL specification requires that the thread halt until all drawing commands that could be affected by your update of the buffer object complete. This obviously misses the whole point of streaming.

The key to effective streaming is avoiding this synchronization. Or rather, avoiding it at all costs.

General use

Most of the uses of buffer objects involve binding them to a certain target, then calling a function that behaves differently based on having a buffer object in that target. Usually, these functions take a pointer as one of their parameters. When a buffer object is bound to certain targets, it causes some functions that take a pointer parameter to treat that parameter as an offset into the buffer object. Thus, rather than pulling data from client memory pointers, they pull it from the bound buffer object.

Here are the various targets for buffer objects and their associated uses:

The buffer will be used as a source for vertex data, but only when glVertexAttribPointer is called. The pointer field of this function is taken as a byte offset from the beginning of whatever buffer is currently bound to this target.
All rendering functions of the form gl*Draw*Elements*​ will use the pointer field as a byte offset from the beginning of the buffer object bound to this target. The indices used for indexed rendering will be taken from the buffer object. Note that this binding target is part of a Vertex Array Objects state, so a VAO must be bound before binding a buffer here.
These have no particular semantics. Because they have no actual meaning, they are useful targets for copying buffer object data with glCopyBufferSubData. You do not have to use these targets when copying, but by using them, you avoid disturbing buffer targets that have actual semantics.
These are for performing asynchronous pixel transfer operations. If a buffer is bound to GL_PIXEL_UNPACK_BUFFER, glTexImage*​, glTexSubImage*​, glCompressedTexImage*​, and glCompressedTexSubImage*​ are all affected. These functions will read their data from the bound buffer object instead of where a client pointer points. Similarly, if a buffer is bound to GL_PIXEL_PACK_BUFFER, glGetTexImage, and glReadPixels will store their data to the bound buffer object instead of where a client pointer points.
This target has no special semantics, but if you intend to use a buffer object for Buffer Textures, it is a good idea to bind it here when you first create it.
An indexed buffer binding for buffers used in Transform Feedback operations.
An indexed buffer binding for buffers used as storage for uniform blocks.
The buffer bound to this target will be used as the source for the indirect data when performing indirect rendering. This is only available in core OpenGL 4.0 or with ARB_draw_indirect.
An indexed buffer binding for buffers used as storage for atomic counters. This requires OpenGL 4.2 or ARB_shader_atomic_counters
The buffer bound to this target will be used as the source for indirect compute dispatch operations, via glDispatchComputeIndirect. This requires OpenGL 4.3 or ARB_compute_shader.
An indexed buffer binding for buffers used as storage for shader storage blocks. This requires OpenGL 4.3 or ARB_shader_storage_buffer_object.

Binding indexed targets

Some buffer targets are indexed, as noted above. This is used for binding a number of buffers that do similar things. For example, a GLSL program can use a number of different uniform buffers.

To bind a buffer object to an indexed location, you may use this function:

 void glBindBufferRange(GLenum target​, GLuint index​, GLuint buffer​, GLintptr offset​, GLsizeiptr size​ );

This causes the buffer​ to be bound to the indexed location target​ at the index​ location. The only valid values for target​ are indexed targets (see below).

The valid values for index​ depend on the kind of target​ being bound. The valid target​ values, and their associated index limits, are:


The offset​ is the byte offset into buffer​ that the should be used for the binding, and the size​ is how many bytes after this are valid for this use of the buffer object. This allows you to bind subsections of a buffer object. If buffer​ is zero, then this index is unbound.

Do note that this does not replace standard buffer binding with glBindBuffer. It does in fact bind the buffer to the target​ parameter, thus unbinding whatever was bound to that target. But usually when you use glBindBufferRange, you are serious about wanting to use the buffer rather than just modify it.

Think of glBindBufferRange as binding the buffer to two places: the particular index and the target​. glBindBuffer only binds to the target​, not the index.

There is a more limited form of this function, glBindBufferBase, that binds the entire buffer to an index. It simply omits the offset​ and size​ fields.