Functions — github.com/anthropics/original_performance

hub / github.com/anthropics/original_performance_takehome / functions

Functions76 in github.com/anthropics/original_performance_takehome

⨍ Functions 76 ◇ Types & classes 17

↓ 9 callersMethodalloc_scratch

(self, name=None, length=1)

perf_takehome.py:61

↓ 8 callersFunctioncycles

()

tests/submission_tests.py:67

tests/frozen_problem.py:458

↓ 6 callersMethodscratch_const

(self, val, name=None)

↓ 4 callersMethodgenerate

(height: int)

problem.py:415

↓ 2 callersMethodbuild_kernel

Like reference_kernel2 but building actual instructions. Scalar implementation using only scalar ALU and load/store.

perf_takehome.py:88

↓ 2 callersFunctionbuild_mem_image

Build a flat memory image of the problem.

problem.py:487

↓ 2 callersMethoddebug_info

(self)

perf_takehome.py:48

↓ 2 callersFunctiondo_kernel_test

( forest_height: int, rounds: int, batch_size: int, seed: int = 123, trace: bool = False,

perf_takehome.py:178

↓ 2 callersFunctiondo_kernel_test

(forest_height: int, rounds: int, batch_size: int)

tests/submission_tests.py:30

↓ 2 callersMethodgenerate

(height: int)

tests/frozen_problem.py:415

↓ 2 callersFunctionreference_kernel2

Reference implementation of the kernel on a flat memory.

problem.py:535

↓ 2 callersMethodrewrite_slot

(self, slot)

problem.py:146

↓ 2 callersMethodrewrite_slot

(self, slot)

tests/frozen_problem.py:146

↓ 1 callersMethodalu

(self, core, op, dest, a1, a2)

problem.py:219

↓ 1 callersMethodalu

(self, core, op, dest, a1, a2)

tests/frozen_problem.py:219

↓ 1 callersMethodbuild

(self, slots: list[tuple[Engine, tuple]], vliw: bool = False)

perf_takehome.py:51

↓ 1 callersMethodbuild_hash

(self, val_hash_addr, tmp1, tmp2, round, i)

perf_takehome.py:77

↓ 1 callersFunctionbuild_mem_image

Build a flat memory image of the problem.

tests/frozen_problem.py:487

↓ 1 callersFunctioncdiv

(a, b)

problem.py:44

↓ 1 callersFunctioncdiv

(a, b)

tests/frozen_problem.py:44

↓ 1 callersFunctionkernel_builder

(forest_height: int, n_nodes: int, batch_size: int, rounds: int)

tests/submission_tests.py:24

↓ 1 callersFunctionmyhash

A simple 32-bit hash function

problem.py:449

↓ 1 callersFunctionmyhash

A simple 32-bit hash function

tests/frozen_problem.py:449

↓ 1 callersFunctionmyhash_traced

A simple 32-bit hash function

problem.py:516

↓ 1 callersFunctionmyhash_traced

A simple 32-bit hash function

tests/frozen_problem.py:516

↓ 1 callersMethodprint_step

(self, instr, core)

problem.py:134

↓ 1 callersMethodprint_step

(self, instr, core)

tests/frozen_problem.py:134

↓ 1 callersFunctionreference_kernel

Reference implementation of the kernel. A parallel tree traversal where at each node we set cur_inp_val = myhash(cur_inp_val ^ node_val)

problem.py:467

↓ 1 callersFunctionreference_kernel2

Reference implementation of the kernel on a flat memory.

tests/frozen_problem.py:535

↓ 1 callersMethodrewrite_instr

Rewrite an instruction to use scratch addresses instead of names

problem.py:123

↓ 1 callersMethodrewrite_instr

Rewrite an instruction to use scratch addresses instead of names

tests/frozen_problem.py:123

↓ 1 callersFunctionrun

(server_class=http.server.HTTPServer, handler_class=MyHandler)

tests/frozen_problem.py:197

↓ 1 callersMethodscratch_map

(self, core)

problem.py:140

↓ 1 callersMethodscratch_map

(self, core)

tests/frozen_problem.py:140

↓ 1 callersMethodsetup_trace

The simulator generates traces in Chrome's Trace Event Format for visualization in Perfetto (or chrome://tracing if you prefer it). S

problem.py:151

↓ 1 callersMethodsetup_trace

The simulator generates traces in Chrome's Trace Event Format for visualization in Perfetto (or chrome://tracing if you prefer it). S

tests/frozen_problem.py:151

↓ 1 callersMethodstep

Execute all the slots in each engine for a single instruction bundle

problem.py:352

↓ 1 callersMethodstep

Execute all the slots in each engine for a single instruction bundle

tests/frozen_problem.py:352

↓ 1 callersMethodtrace_post_step

(self, instr, core)

problem.py:337

↓ 1 callersMethodtrace_post_step

(self, instr, core)

tests/frozen_problem.py:337

↓ 1 callersMethodtrace_slot

(self, core, slot, name, i)

problem.py:347

↓ 1 callersMethodtrace_slot

(self, core, slot, name, i)

tests/frozen_problem.py:347

tests/frozen_problem.py:399

Method__init__

( self, mem_dump: list[int], program: list[Instruction], debug_info: DebugInfo

( self, mem_dump: list[int], program: list[Instruction], debug_info: DebugInfo

tests/frozen_problem.py:97

tests/frozen_problem.py:300

Methodgenerate

(forest: Tree, batch_size: int, rounds: int)

problem.py:433

Methodgenerate

(forest: Tree, batch_size: int, rounds: int)

tests/frozen_problem.py:433

tests/frozen_problem.py:269

Functionreference_kernel

Reference implementation of the kernel. A parallel tree traversal where at each node we set cur_inp_val = myhash(cur_inp_val ^ node_val)

tests/frozen_problem.py:467

tests/frozen_problem.py:288

Methodtest_kernel_correctness

(self)

tests/submission_tests.py:58

Methodtest_kernel_cycles

(self)

perf_takehome.py:257

Methodtest_kernel_speedup

(self)

tests/submission_tests.py:85

Methodtest_kernel_trace

(self)

perf_takehome.py:244

Methodtest_kernel_updated_starting_point

(self)

tests/submission_tests.py:88

Methodtest_opus45_11hr

(self)

tests/submission_tests.py:109

Methodtest_opus45_2hr

(self)

tests/submission_tests.py:101

Methodtest_opus45_casual

(self)

tests/submission_tests.py:96

Methodtest_opus45_improved_harness

(self)

tests/submission_tests.py:113

Methodtest_opus4_many_hours

(self)

tests/submission_tests.py:92

Methodtest_ref_kernels

Test the reference kernels against each other

perf_takehome.py:229

Methodtest_sonnet45_many_hours

(self)

tests/submission_tests.py:105

tests/frozen_problem.py:254