blob: 1fdc5d6fcb68940f9f29a6144c56d75e9e7a1f16 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
|
Name
NV_device_cuda
Name Strings
EGL_NV_device_cuda
Contributors
Michael Chock
James Jones
Contact
Michael Chock (mchock 'at' nvidia.com)
Status
Complete
Version
Version 1, June 21, 2014
Number
EGL Extension #74
Extension Type
EGL device extension
Dependencies
This extension is written against the language of EGL 1.5 as
modified by EGL_EXT_device_base.
EGL_EXT_device_base is required.
Overview
EGL and CUDA both have the capability to drive multiple devices,
such as GPUs, within a single system. To interoperate with one
another, both APIs must have compatible notions of such devices.
This extension defines a mapping from an EGL device to a CUDA device
enumerant.
IP Status
No known claims.
New Types
None.
New Procedures and Functions
None.
New Tokens
Accepted as a queried <attribute> in eglQueryDeviceAttribEXT:
EGL_CUDA_DEVICE_NV 0x323A
Add a new section 2.1.3 (CUDA Devices) after 2.1.2 (Devices)
"Somewhat analogous to an EGL device, a CUDA device establishes a
namespace for CUDA operations. In the CUDA API, such a device is
represented by a C int. For more details, see the CUDA
documentation."
Changes to section 3.2 (Device Enumeration)
Replace the paragraph immediately following the prototype for
eglQueryDeviceAttribEXT:
"The only valid value of <attribute> is EGL_CUDA_DEVICE_NV. On
success, EGL_TRUE is returned, and a valid CUDA device handle
corresponding to the EGL device is returned in <value>. This handle
is compatible with CUDA API functions."
Issues
None
Revision History
Version 1, 2014/06/24 (Michael Chock)
- initial version.
|
