create engine folder and move code (#3948)

* retry

* older tf

* that

test/external/external_benchmark_multitensor_allreduce.py

@@ -3,8 +3,8 @@ from tinygrad import Tensor, Device
 from tinygrad.lazy import LazyBuffer
 from tinygrad.ops import ReduceOps, GlobalCounters
 from tinygrad.features.multi import MultiLazyBuffer, all_reduce
-from tinygrad.features.jit import TinyJit
-from tinygrad.realize import create_schedule, run_schedule
+from tinygrad.engine.jit import TinyJit
+from tinygrad.engine.realize import create_schedule, run_schedule
 from tinygrad.helpers import getenv, Context, RING
 from typing import List, Union
 

test/external/external_model_benchmark.py

@@ -66,7 +66,7 @@ def benchmark_model(m, devices, validate_outs=False):
     tinygrad_model = get_run_onnx(onnx_model)
     benchmark(m, f"tinygrad_{device.lower()}_jitless", lambda: {k:v.numpy() for k,v in tinygrad_model(inputs).items()})
 
-    from tinygrad.features.jit import TinyJit
+    from tinygrad.engine.jit import TinyJit
     tinygrad_jitted_model = TinyJit(lambda **kwargs: {k:v.realize() for k,v in tinygrad_model(kwargs).items()})
     for _ in range(3): {k:v.numpy() for k,v in tinygrad_jitted_model(**inputs).items()}
     benchmark(m, f"tinygrad_{device.lower()}_jit", lambda: {k:v.numpy() for k,v in tinygrad_jitted_model(**inputs).items()}) # noqa: F821

test/external/external_test_hip_compile.py

@@ -2,7 +2,7 @@ import time, unittest
 from tinygrad.runtime.driver.hip_comgr import compile_hip
 from tinygrad import Tensor
 from tinygrad.device import Device
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.codegen.linearizer import Linearizer
 
 class TestHIPCompileSpeed(unittest.TestCase):

test/external/external_test_hsa_driver.py

@@ -4,7 +4,7 @@ from tinygrad.device import Device, Buffer, BufferXfer
 from tinygrad.dtype import dtypes
 from tinygrad.runtime.driver.hsa import AQLQueue
 from tinygrad.runtime.graph.hsa import VirtAQLQueue, HSAGraph
-from tinygrad.features.jit import JitItem
+from tinygrad.engine.jit import JitItem
 
 def get_hsa_inc_prog(dev, inc=1):
   prg = f"""

test/external/external_test_jit_on_models.py

@@ -2,7 +2,7 @@
 import unittest
 import numpy as np
 from tinygrad import Tensor, dtypes
-from tinygrad.features.jit import TinyJit
+from tinygrad.engine.jit import TinyJit
 from tinygrad.helpers import CI
 from test.helpers import derandomize_model
 

test/external/external_test_opt.py

@@ -17,7 +17,7 @@ from tinygrad.helpers import getenv
 from tinygrad.nn import optim
 #from tinygrad.lazy import PUSH_PERMUTES
 PUSH_PERMUTES = False
-from tinygrad.features.jit import CacheCollector
+from tinygrad.engine.jit import CacheCollector
 
 class CLCache:
   def __init__(self, allowed=None, strict=False, preclear=True, var_vals=None):

test/external/external_test_uops_graphing.py

@@ -4,7 +4,7 @@ from tinygrad.tensor import Tensor
 from tinygrad.codegen.linearizer import Linearizer
 from tinygrad.renderer.cstyle import OpenCLRenderer
 from tinygrad.features.graph import graph_uops
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.nn import Conv2d
 
 class TestUopsGraph(unittest.TestCase):

test/models/test_real_world.py

@@ -2,7 +2,7 @@ import unittest, time, gc
 import numpy as np
 from tinygrad.nn import optim
 from tinygrad.nn.state import get_parameters
-from tinygrad.features.jit import TinyJit
+from tinygrad.engine.jit import TinyJit
 from tinygrad import Tensor, Device, GlobalCounters, dtypes
 from tinygrad.helpers import CI, getenv
 from tinygrad.shape.symbolic import Variable

test/test_conv_shapetracker.py

@@ -3,7 +3,7 @@ import unittest
 from tinygrad.tensor import Tensor
 from tinygrad.ops import LoadOps
 from tinygrad.nn import Conv2d
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 
 class TestConvShapetracker(unittest.TestCase):
   def test_conv_3x3_one_view(self):

test/test_custom_function.py

@@ -87,7 +87,7 @@ class TestCustomFunction(unittest.TestCase):
   @unittest.skipIf(Device.DEFAULT in ["CPU"], "atan2_cpu not jittable")
   def test_atan2_jit(self):
     # custom ops even work in the JIT!
-    from tinygrad.features.jit import TinyJit
+    from tinygrad.engine.jit import TinyJit
 
     @TinyJit
     def jitted_atan2(a:Tensor, b:Tensor) -> Tensor:

test/test_dtype_alu.py

@@ -6,7 +6,7 @@ import numpy as np
 from hypothesis import given, strategies as strat, settings
 from tinygrad.dtype import DType
 from tinygrad.helpers import CI, getenv
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.ops import UnaryOps, get_lazyop_info
 from test.helpers import is_dtype_supported
 

test/test_fusion_op.py

@@ -2,7 +2,7 @@ import unittest
 import time
 import numpy as np
 from tinygrad import Tensor, dtypes
-from tinygrad.realize import run_schedule, create_schedule, lower_schedule_item
+from tinygrad.engine.realize import run_schedule, create_schedule, lower_schedule_item
 
 class TestFusionOp(unittest.TestCase):
   def test_contiguous_add(self):

test/test_jit.py

@@ -4,7 +4,7 @@ import numpy as np
 
 from test.helpers import assert_jit_cache_len
 from tinygrad.tensor import Tensor
-from tinygrad.features.jit import TinyJit
+from tinygrad.engine.jit import TinyJit
 from tinygrad.device import Device
 from tinygrad.helpers import CI
 

test/test_lazybuffer.py

@@ -3,7 +3,7 @@ import numpy as np
 import unittest
 from tinygrad import Tensor, Device, dtypes
 from tinygrad.lazy import LazyBuffer, ReduceOps
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 
 class TestLazyBuffer(unittest.TestCase):
   def test_fromcpu_shape_tracker(self):

test/test_lazyop.py

@@ -1,6 +1,6 @@
 import unittest
 from tinygrad.tensor import Tensor
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 
 # stuff needed to unpack a kernel
 # ruff: noqa: F401

test/test_linearizer.py

@@ -9,8 +9,8 @@ from tinygrad.shape.shapetracker import ShapeTracker
 from tinygrad.shape.view import View
 from tinygrad.shape.symbolic import MulNode, Variable, NumNode, Node
 from tinygrad.tensor import Tensor
-from tinygrad.features.jit import CacheCollector
-from tinygrad.realize import create_schedule, run_schedule
+from tinygrad.engine.jit import CacheCollector
+from tinygrad.engine.realize import create_schedule, run_schedule
 from tinygrad.helpers import prod, Context
 from tinygrad.dtype import DType, dtypes
 from tinygrad.codegen.uops import UOpGraph

test/test_multitensor.py

@@ -5,7 +5,7 @@ from tinygrad.device import BufferCopy
 from tinygrad.ops import LoadOps, ReduceOps
 from tinygrad.helpers import CI, prod, Context
 from tinygrad.nn.state import get_parameters, get_state_dict
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.features.multi import all_reduce, MultiLazyBuffer
 from random import randint
 import numpy as np

test/test_schedule.py

@@ -9,7 +9,7 @@ from tinygrad.ops import LoadOps
 from tinygrad.helpers import DEBUG, GRAPH
 from tinygrad.codegen.linearizer import Linearizer
 from tinygrad.features.graph import print_tree, realized_lazybuffer
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad import nn, dtypes
 
 def check_schedule(t:Tensor, allowed:int, to_prerealize:Optional[List[Tensor]]=None, filter_loadops=True):

test/test_search.py

@@ -1,7 +1,7 @@
 import unittest
 
 from tinygrad.codegen.linearizer import Linearizer
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.features.search import time_linearizer, bufs_from_lin
 from tinygrad.device import Device, Buffer
 from tinygrad.ops import LoadOps

test/test_speed_v_torch.py

@@ -13,7 +13,7 @@ from tinygrad import Device, GlobalCounters
 from tinygrad.tensor import Tensor
 from tinygrad.nn import Conv2d
 from tinygrad.helpers import colored, getenv, CI
-from tinygrad.features.jit import TinyJit
+from tinygrad.engine.jit import TinyJit
 
 IN_CHANS = [int(x) for x in getenv("IN_CHANS", "4,16,64").split(",")]
 

test/test_symbolic_jit.py

@@ -1,7 +1,7 @@
 import unittest
 
 from test.helpers import assert_jit_cache_len
-from tinygrad.features.jit import TinyJit
+from tinygrad.engine.jit import TinyJit
 from tinygrad.shape.symbolic import Variable
 from tinygrad.tensor import Tensor
 import numpy as np

test/test_uops.py

@@ -5,7 +5,7 @@ from tinygrad.tensor import Tensor
 from tinygrad.dtype import dtypes, DType, PtrDType
 from tinygrad.device import Buffer, Device, CompiledASTRunner
 from tinygrad.ops import UnaryOps, BinaryOps, TernaryOps
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 from tinygrad.codegen.linearizer import UOps, UOp
 from tinygrad.codegen.uops import exec_alu, UOpGraph
 from test.helpers import is_dtype_supported

test/test_uops_stats.py

@@ -1,6 +1,6 @@
 import unittest
 from tinygrad import Tensor
-from tinygrad.realize import create_schedule, lower_schedule_item
+from tinygrad.engine.realize import create_schedule, lower_schedule_item
 
 # TODO: can copy this in here when we remove it
 #from tinygrad.ops import get_lazyop_info

test/test_winograd.py

@@ -3,7 +3,7 @@ from tinygrad import Tensor, GlobalCounters
 from tinygrad.helpers import Timing, CI, Profiling, WINO, DEBUG
 from tinygrad.ops import LoadOps
 from tinygrad.codegen.linearizer import Linearizer
-from tinygrad.realize import create_schedule
+from tinygrad.engine.realize import create_schedule
 
 class TestWinograd(unittest.TestCase):
   def setUp(self):