Mercurial > hg > graal-compiler
view src/gpu/hsail/vm/gpu_hsail.hpp @ 15921:6d8c901814eb
Support for compressed constants in HSAIL backend.
author | Roland Schatz <roland.schatz@oracle.com> |
---|---|
date | Tue, 27 May 2014 12:08:11 +0200 |
parents | 66d31e70bd79 |
children | 06eedda53e14 |
line wrap: on
line source
/* * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #ifndef GPU_HSAIL_HPP #define GPU_HSAIL_HPP #include "utilities/exceptions.hpp" #include "graal/graalEnv.hpp" #include "gpu_hsail_Frame.hpp" class Hsail : public Gpu { public: class HSAILKernelDeoptimization VALUE_OBJ_CLASS_SPEC { friend class VMStructs; private: // TODO: separate workitemid and actionAndReason out // since they are there only once even if there are multiple frames // for now, though we only ever have one hsail frame jint _workitemid; jint _actionAndReason; // the first (innermost) "hsail frame" starts after the above fields public: inline jint workitem() { return _workitemid; } inline jint reason() { return _actionAndReason; } inline jint pc_offset() { return first_frame()->pc_offset(); } inline HSAILFrame *first_frame() { // starts after the "header" fields return (HSAILFrame *) (((jbyte *) this) + sizeof(*this)); } }; // 8 compute units * 40 waves per cu * wavesize 64 // TODO: query the device to get this number #define MAX_DEOPT_SLOTS (8 * 40 * 64) class HSAILDeoptimizationInfo : public CHeapObj<mtInternal> { friend class VMStructs; private: jint* _notice_safepoints; jint _deopt_occurred; jint _deopt_next_index; JavaThread** _donor_threads; jint _num_slots; jint _deopt_span; char _ignore; // keep a pointer last so save area following it is word aligned jboolean * _never_ran_array; public: HSAILKernelDeoptimization _deopt_save_states[1]; // number and size of these can vary per kernel static inline size_t hdr_size() { return sizeof(HSAILDeoptimizationInfo); } inline jbyte * save_area_start() { return (jbyte*) (this) + hdr_size(); } inline HSAILDeoptimizationInfo(int numSlots, int bytesPerSaveArea, int dimX, JavaThread** donorThreads) { _notice_safepoints = &Hsail::_notice_safepoints; _deopt_occurred = 0; _deopt_next_index = 0; _num_slots = numSlots; _never_ran_array = NEW_C_HEAP_ARRAY(jboolean, dimX, mtInternal); memset(_never_ran_array, 0, dimX * sizeof(jboolean)); _donor_threads = donorThreads; _deopt_span = sizeof(HSAILKernelDeoptimization) + sizeof(HSAILFrame) + bytesPerSaveArea; if (TraceGPUInteraction) { tty->print_cr("HSAILDeoptimizationInfo allocated, %d slots of size %d, total size = 0x%lx bytes", _num_slots, _deopt_span, (_num_slots * _deopt_span + sizeof(HSAILDeoptimizationInfo))); } } inline ~HSAILDeoptimizationInfo() { FREE_C_HEAP_ARRAY(jboolean, _never_ran_array, mtInternal); } inline jint deopt_occurred() { return _deopt_occurred; } inline jint num_deopts() { return _deopt_next_index; } inline jboolean *never_ran_array() { return _never_ran_array; } inline jint num_slots() {return _num_slots;} inline HSAILKernelDeoptimization * get_deopt_save_state(int slot) { // use _deopt_span to index into _deopt_states return (HSAILKernelDeoptimization *) (save_area_start() + _deopt_span * slot); } void * operator new (size_t hdrSize, int numSlots, int bytesPerSaveArea) { assert(hdrSize <= hdr_size(), ""); size_t totalSizeBytes = hdr_size() + numSlots * (sizeof(HSAILKernelDeoptimization) + sizeof(HSAILFrame) + bytesPerSaveArea); return NEW_C_HEAP_ARRAY(char, totalSizeBytes, mtInternal); } void operator delete (void *ptr) { FREE_C_HEAP_ARRAY(char, ptr, mtInternal); } }; private: static JNINativeMethod HSAIL_methods[]; // static native boolean initialize(); JNIEXPORT static jboolean initialize(JNIEnv *env, jclass); // static native long generateKernel(byte[] targetCode, String name); JNIEXPORT static jlong generate_kernel(JNIEnv *env, jclass, jbyteArray code_handle, jstring name_handle); // static native boolean executeKernel0(HotSpotInstalledCode kernel, int jobSize, Object[] args); JNIEXPORT static jboolean execute_kernel_void_1d(JNIEnv *env, jclass, jobject hotspotInstalledCode, jint dimX, jobject args, jobject oopsSave, jobject donorThreads, int allocBytesPerWorkitem, jobject oop_map_array); // static native void getThreadPointers(Object[] donorThreads, long[] threadPointersOut); JNIEXPORT static void get_thread_pointers(JNIEnv *env, jclass, jobject donor_threads_handle, jobject thread_ptrs_handle); static void getNewTlabForDonorThread(ThreadLocalAllocBuffer* tlab, size_t tlabMinHsail); static jboolean execute_kernel_void_1d_internal(address kernel, int dimX, jobject args, methodHandle& mh, nmethod *nm, jobject oopsSave, jobject donorThreads, int allocBytesPerWorkitem, jobject oop_map_array, TRAPS); static void register_heap(); static GraalEnv::CodeInstallResult install_code(Handle& compiled_code, CodeBlob*& cb, Handle installed_code, Handle triggered_deoptimizations); public: // Registers the implementations for the native methods in HSAILHotSpotBackend static bool register_natives(JNIEnv* env); virtual const char* name() { return "HSAIL"; } virtual void notice_safepoints(); virtual void ignore_safepoints(); #if defined(__x86_64) || defined(AMD64) || defined(_M_AMD64) typedef unsigned long long CUdeviceptr; #else typedef unsigned int CUdeviceptr; #endif private: typedef void* (*okra_create_context_func_t)(); typedef void* (*okra_create_kernel_func_t)(void*, unsigned char *, const char *); typedef bool (*okra_push_object_func_t)(void*, void*); typedef bool (*okra_push_boolean_func_t)(void*, jboolean); typedef bool (*okra_push_byte_func_t)(void*, jbyte); typedef bool (*okra_push_double_func_t)(void*, jdouble); typedef bool (*okra_push_float_func_t)(void*, jfloat); typedef bool (*okra_push_int_func_t)(void*, jint); typedef bool (*okra_push_long_func_t)(void*, jlong); typedef bool (*okra_execute_with_range_func_t)(void*, jint); typedef bool (*okra_clearargs_func_t)(void*); typedef bool (*okra_register_heap_func_t)(void*, size_t); public: static okra_create_context_func_t _okra_create_context; static okra_create_kernel_func_t _okra_create_kernel; static okra_push_object_func_t _okra_push_object; static okra_push_boolean_func_t _okra_push_boolean; static okra_push_byte_func_t _okra_push_byte; static okra_push_double_func_t _okra_push_double; static okra_push_float_func_t _okra_push_float; static okra_push_int_func_t _okra_push_int; static okra_push_long_func_t _okra_push_long; static okra_execute_with_range_func_t _okra_execute_with_range; static okra_clearargs_func_t _okra_clearargs; static okra_register_heap_func_t _okra_register_heap; protected: static void* _device_context; // true if safepoints are activated static jint _notice_safepoints; }; #endif // GPU_HSAIL_HPP