[Kernel] Migrate scattered quantization kernels to sglang.kernels (RFC #29630, Phase 2.5, 1/7) (#30784)

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

test/registered/quant/test_awq_dequant.py

@@ -12,7 +12,7 @@ import unittest
 
 import torch
 
-from sglang.srt.layers.quantization.awq.awq_triton import (
+from sglang.kernels.ops.quantization.awq_triton import (
     AWQ_TRITON_SUPPORTED_GROUP_SIZES,
     awq_dequantize_triton,
     awq_gemm_triton,

test/registered/quant/test_fp8_blockwise_row_padding.py

@@ -11,7 +11,7 @@ import unittest
 
 import torch
 
-from sglang.srt.layers.quantization.fp8_kernel import (
+from sglang.kernels.ops.quantization.fp8_kernel import (
     fp8_dtype,
     per_token_group_quant_fp8,
     sglang_per_token_group_quant_fp8_row_padded,

test/registered/quant/test_fp8_kernel.py

@@ -2,7 +2,7 @@ import unittest
 
 import torch
 
-from sglang.srt.layers.quantization.fp8_kernel import (
+from sglang.kernels.ops.quantization.fp8_kernel import (
     per_token_group_quant_fp8,
     w8a8_block_fp8_matmul,
 )

test/registered/quant/test_int8_kernel.py

@@ -3,10 +3,10 @@ import unittest
 
 import torch
 
+from sglang.kernels.ops.quantization.int8_kernel import per_token_quant_int8
 from sglang.srt.layers.activation import SiluAndMul
 from sglang.srt.layers.moe.moe_runner.triton_utils.fused_moe import fused_moe
 from sglang.srt.layers.moe.topk import TopKConfig, select_experts
-from sglang.srt.layers.quantization.int8_kernel import per_token_quant_int8
 from sglang.srt.server_args import ServerArgs, set_global_server_args_for_scheduler
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase

test/registered/quant/test_triton_scaled_mm.py

@@ -4,7 +4,7 @@ from typing import Optional
 import torch
 import torch.testing
 
-from sglang.srt.layers.quantization.fp8_kernel import triton_scaled_mm
+from sglang.kernels.ops.quantization.fp8_kernel import triton_scaled_mm
 from sglang.srt.utils.common import get_device
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase