Mercurial > hg > graal-compiler
comparison src/gpu/ptx/vm/ptxKernelArguments.cpp @ 11596:91e5f927af63
Initial implementation of PTXRuntime (RegisterConfig, PTX description etc); guarded with new flag UseGPU. Specify -XX:+UseGPU to exercise this new implementation.
author | bharadwaj |
---|---|
date | Tue, 10 Sep 2013 22:39:50 -0400 |
parents | src/gpu/ptx/vm/kernelArguments.cpp@49bb1bc983c6 |
children | d8659ad83fcc |
comparison
equal
deleted
inserted
replaced
11595:003be97acdda | 11596:91e5f927af63 |
---|---|
1 /* | |
2 * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved. | |
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. | |
4 * | |
5 * This code is free software; you can redistribute it and/or modify it | |
6 * under the terms of the GNU General Public License version 2 only, as | |
7 * published by the Free Software Foundation. | |
8 * | |
9 * This code is distributed in the hope that it will be useful, but WITHOUT | |
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
12 * version 2 for more details (a copy is included in the LICENSE file that | |
13 * accompanied this code). | |
14 * | |
15 * You should have received a copy of the GNU General Public License version | |
16 * 2 along with this work; if not, write to the Free Software Foundation, | |
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. | |
18 * | |
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA | |
20 * or visit www.oracle.com if you need additional information or have any | |
21 * questions. | |
22 * | |
23 */ | |
24 | |
25 #include "precompiled.hpp" | |
26 #include "ptxKernelArguments.hpp" | |
27 #include "runtime/javaCalls.hpp" | |
28 | |
29 gpu::Ptx::cuda_cu_memalloc_func_t gpu::Ptx::_cuda_cu_memalloc; | |
30 gpu::Ptx::cuda_cu_memcpy_htod_func_t gpu::Ptx::_cuda_cu_memcpy_htod; | |
31 | |
32 // Get next java argument | |
33 oop PTXKernelArguments::next_arg(BasicType expectedType) { | |
34 assert(_index < _args->length(), "out of bounds"); | |
35 oop arg=((objArrayOop) (_args))->obj_at(_index++); | |
36 assert(expectedType == T_OBJECT || java_lang_boxing_object::is_instance(arg, expectedType), "arg type mismatch"); | |
37 return arg; | |
38 } | |
39 | |
40 void PTXKernelArguments::do_int() { | |
41 // If the parameter is a return value, | |
42 if (is_return_type()) { | |
43 // Allocate device memory for T_INT return value pointer on device. Size in bytes | |
44 int status = gpu::Ptx::_cuda_cu_memalloc(&_return_value_ptr, T_INT_BYTE_SIZE); | |
45 if (status != GRAAL_CUDA_SUCCESS) { | |
46 tty->print_cr("[CUDA] *** Error (%d) Failed to allocate memory for return value pointer on device", status); | |
47 _success = false; | |
48 return; | |
49 } | |
50 // Push _return_value_ptr to _kernelBuffer | |
51 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = _return_value_ptr; | |
52 _bufferOffset += sizeof(_return_value_ptr); | |
53 } | |
54 else { | |
55 // Get the next java argument and its value which should be a T_INT | |
56 oop arg = next_arg(T_INT); | |
57 // Copy the java argument value to kernelArgBuffer | |
58 jvalue intval; | |
59 if (java_lang_boxing_object::get_value(arg, &intval) != T_INT) { | |
60 tty->print_cr("[CUDA] *** Error: Unexpected argument type; expecting T_INT"); | |
61 _success = false; | |
62 return; | |
63 } | |
64 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = intval.i; | |
65 _bufferOffset += sizeof(intval.i); | |
66 } | |
67 return; | |
68 } | |
69 | |
70 void PTXKernelArguments::do_long() { | |
71 // If the parameter is a return value, | |
72 if (is_return_type()) { | |
73 // Allocate device memory for T_LONG return value pointer on device. Size in bytes | |
74 int status = gpu::Ptx::_cuda_cu_memalloc(&_return_value_ptr, T_LONG_BYTE_SIZE); | |
75 if (status != GRAAL_CUDA_SUCCESS) { | |
76 tty->print_cr("[CUDA] *** Error (%d) Failed to allocate memory for return value pointer on device", status); | |
77 _success = false; | |
78 return; | |
79 } | |
80 // Push _return_value_ptr to _kernelBuffer | |
81 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = _return_value_ptr; | |
82 _bufferOffset += sizeof(_return_value_ptr); | |
83 } | |
84 else { | |
85 // Get the next java argument and its value which should be a T_LONG | |
86 oop arg = next_arg(T_LONG); | |
87 // Copy the java argument value to kernelArgBuffer | |
88 jvalue val; | |
89 if (java_lang_boxing_object::get_value(arg, &val) != T_LONG) { | |
90 tty->print_cr("[CUDA] *** Error: Unexpected argument type; expecting T_LONG"); | |
91 _success = false; | |
92 return; | |
93 } | |
94 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = val.j; | |
95 _bufferOffset += sizeof(val.j); | |
96 } | |
97 return; | |
98 } | |
99 | |
100 void PTXKernelArguments::do_byte() { | |
101 // If the parameter is a return value, | |
102 if (is_return_type()) { | |
103 // Allocate device memory for T_BYTE return value pointer on device. Size in bytes | |
104 int status = gpu::Ptx::_cuda_cu_memalloc(&_return_value_ptr, T_BYTE_SIZE); | |
105 if (status != GRAAL_CUDA_SUCCESS) { | |
106 tty->print_cr("[CUDA] *** Error (%d) Failed to allocate memory for return value pointer on device", status); | |
107 _success = false; | |
108 return; | |
109 } | |
110 // Push _return_value_ptr to _kernelBuffer | |
111 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = _return_value_ptr; | |
112 _bufferOffset += sizeof(_return_value_ptr); | |
113 } | |
114 else { | |
115 // Get the next java argument and its value which should be a T_BYTE | |
116 oop arg = next_arg(T_BYTE); | |
117 // Copy the java argument value to kernelArgBuffer | |
118 jvalue val; | |
119 if (java_lang_boxing_object::get_value(arg, &val) != T_BYTE) { | |
120 tty->print_cr("[CUDA] *** Error: Unexpected argument type; expecting T_BYTE"); | |
121 _success = false; | |
122 return; | |
123 } | |
124 *((gpu::Ptx::CUdeviceptr*) &_kernelArgBuffer[_bufferOffset]) = val.b; | |
125 _bufferOffset += sizeof(val.b); | |
126 } | |
127 return; | |
128 } | |
129 | |
130 // TODO implement other do_* |