#include "clearstate_gfx11.h"
#include "v11_structs.h"
#include "gfx_v11_0.h"
+#include "gfx_v11_0_cleaner_shader.h"
#include "gfx_v11_0_3.h"
#include "nbio_v4_3.h"
#include "mes_v11_0.h"
}
switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+ case IP_VERSION(11, 0, 3):
+ adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+ adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+ if (adev->gfx.me_fw_version >= 2280 &&
+ adev->gfx.pfp_fw_version >= 2370 &&
+ adev->gfx.mec_fw_version >= 2450 &&
+ adev->mes.fw_version[0] >= 99) {
+ adev->gfx.enable_cleaner_shader = true;
+ r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+ if (r) {
+ adev->gfx.enable_cleaner_shader = false;
+ dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+ }
+ }
+ break;
default:
adev->gfx.enable_cleaner_shader = false;
+ break;
}
/* Enable CG flag in one VF mode for enabling RLC safe mode enter/exit */
--- /dev/null
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is first 64 Dwords or 256 bytes of 192 Dwords cleaner shader.
+//To turn this shader program on for complitaion change this to main and lower shader main to main_1
+
+// Navi3 : Clear SGPRs, VGPRs and LDS
+// Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot
+// Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD
+// Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS)
+// It takes 2 workgroups to use all of LDS: one on each CU of the WGP
+// Each wave clears SGPRs 0 - 107
+// Each wave clears VGPRs 0 - 63
+// The first wave of the workgroup clears its 64KB of LDS
+// The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+// before any wave in the workgroup could end. Without this, it is possible not all SGPRs get cleared.
+
+shader main
+ asic(GFX11)
+ type(CS)
+ wave_size(32)
+// Note: original source code from SQ team
+
+// Takes about 2500 clocks to run.
+// (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks)
+//
+ S_BARRIER
+
+ //
+ // CLEAR VGPRs
+ //
+ s_mov_b32 m0, 0x00000058 // Loop 96/8=12 times (loop unrolled for performance)
+
+label_0005:
+ v_movreld_b32 v0, 0
+ v_movreld_b32 v1, 0
+ v_movreld_b32 v2, 0
+ v_movreld_b32 v3, 0
+ v_movreld_b32 v4, 0
+ v_movreld_b32 v5, 0
+ v_movreld_b32 v6, 0
+ v_movreld_b32 v7, 0
+ s_sub_u32 m0, m0, 8
+ s_cbranch_scc0 label_0005
+ //
+ //
+
+ s_mov_b32 s2, 0x80000000 // Bit31 is first_wave
+ s_and_b32 s2, s2, s0 // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+ s_cbranch_scc0 label_0023 // Clean LDS if its first wave of ThreadGroup/WorkGroup
+ // CLEAR LDS
+ //
+ s_mov_b32 exec_lo, 0xffffffff
+ s_mov_b32 exec_hi, 0xffffffff
+ v_mbcnt_lo_u32_b32 v1, exec_hi, 0 // Set V1 to thread-ID (0..63)
+ v_mbcnt_hi_u32_b32 v1, exec_lo, v1 // Set V1 to thread-ID (0..63)
+ v_mul_u32_u24 v1, 0x00000008, v1 // * 8, so each thread is a double-dword address (8byte)
+ s_mov_b32 s2, 0x00000003f // 64 loop iterations
+ s_mov_b32 m0, 0xffffffff
+ // Clear all of LDS space
+ // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+ ds_write2_b64 v1, v[2:3], v[2:3] offset1:32
+ ds_write2_b64 v1, v[4:5], v[4:5] offset0:64 offset1:96
+ v_add_co_u32 v1, vcc, 0x00000400, v1
+ s_sub_u32 s2, s2, 1
+ s_cbranch_scc0 label_001F
+ //
+ // CLEAR SGPRs
+ //
+label_0023:
+ s_mov_b32 m0, 0x00000068 // Loop 108/4=27 times (loop unrolled for performance)
+label_sgpr_loop:
+ s_movreld_b32 s0, 0
+ s_movreld_b32 s1, 0
+ s_movreld_b32 s2, 0
+ s_movreld_b32 s3, 0
+ s_sub_u32 m0, m0, 4
+ s_cbranch_scc0 label_sgpr_loop
+
+ //clear vcc
+ s_mov_b64 vcc, 0 //clear vcc
+ s_mov_b32 flat_scratch_lo, 0 //clear flat scratch lo SGPR
+ s_mov_b32 flat_scratch_hi, 0 //clear flat scratch hi SGPR
+ s_mov_b64 ttmp0, 0 //Clear ttmp0 and ttmp1
+ s_mov_b64 ttmp2, 0 //Clear ttmp2 and ttmp3
+ s_mov_b64 ttmp4, 0 //Clear ttmp4 and ttmp5
+ s_mov_b64 ttmp6, 0 //Clear ttmp6 and ttmp7
+ s_mov_b64 ttmp8, 0 //Clear ttmp8 and ttmp9
+ s_mov_b64 ttmp10, 0 //Clear ttmp10 and ttmp11
+ s_mov_b64 ttmp12, 0 //Clear ttmp12 and ttmp13
+ s_mov_b64 ttmp14, 0 //Clear ttmp14 and ttmp15
+
+ s_endpgm
+
+end
+
--- /dev/null
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* Define the cleaner shader gfx_11_0_3 */
+static const u32 gfx_11_0_3_cleaner_shader_hex[] = {
+ 0xb0804006, 0xbe8200ff,
+ 0x00000058, 0xbefd0080,
+ 0x7e008480, 0x7e028480,
+ 0x7e048480, 0x7e068480,
+ 0x7e088480, 0x7e0a8480,
+ 0x7e0c8480, 0x7e0e8480,
+ 0xbefd0002, 0x80828802,
+ 0xbfa1fff5, 0xbe8200ff,
+ 0x80000000, 0x8b020002,
+ 0xbfa10012, 0xbefe00c1,
+ 0xbeff00c1, 0xd71f0001,
+ 0x0001007f, 0xd7200001,
+ 0x0002027e, 0x16020288,
+ 0xbe8200bf, 0xbefd00c1,
+ 0xd9382000, 0x00020201,
+ 0xd9386040, 0x00040401,
+ 0xd7006a01, 0x000202ff,
+ 0x00000400, 0x80828102,
+ 0xbfa1fff7, 0xbefd00ff,
+ 0x00000068, 0xbe804280,
+ 0xbe814280, 0xbe824280,
+ 0xbe834280, 0x80fd847d,
+ 0xbfa1fffa, 0xbeea0180,
+ 0xbeec0180, 0xbeee0180,
+ 0xbef00180, 0xbef20180,
+ 0xbef40180, 0xbef60180,
+ 0xbef80180, 0xbefa0180,
+ 0xbfb00000, 0xbf9f0000,
+ 0xbf9f0000, 0xbf9f0000,
+ 0xbf9f0000, 0xbf9f0000,
+};
projects / users / jedix / linux-maple.git / commitdiff