Skip to content

feat: add support for GPU in abstraction + extend benchmarks with iterations #10

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ttrenty merged 2 commits into from
Jun 29, 2025
Merged

feat: add support for GPU in abstraction + extend benchmarks with iterations #10

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
529e2d4
feat: add support for GPU in abstraction + extend benchmarks with ite…
ttrenty Jun 29, 2025
ab216e0
fix: add has_accelerator() before DeviceContext() call
ttrenty Jun 29, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 6 additions & 4 deletions benchmarks/bench_main.mojo
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,13 +16,14 @@ def main():
# bench_qubit_wise_multiply()
bench_qubit_wise_multiply_inplace[
min_number_qubits=1,
max_number_qubits=20,
max_number_qubits=25,
number_qubits_step_size=1,
min_number_layers=5,
max_number_layers=3500,
number_layers_step_size=400,
fixed_number_qubits=13,
fixed_number_layers=5,
fixed_number_layers=15,
fixed_number_iterations=1,
]()

@parameter
Expand All @@ -31,13 +32,14 @@ def main():
else:
bench_qubit_wise_multiply_inplace_gpu[
min_number_qubits=1,
max_number_qubits=26, # 29 is OOM for my 3070 Ti Laptop GPU
max_number_qubits=30, # 29 is OOM for my 3070 Ti Laptop GPU
number_qubits_step_size=1,
min_number_layers=5,
max_number_layers=7000,
number_layers_step_size=400,
fixed_number_qubits=13,
fixed_number_layers=5,
fixed_number_layers=15,
fixed_number_iterations=1,
]()

# bench_qubit_wise_multiply_extended()
Expand Down
161 changes: 84 additions & 77 deletions benchmarks/bench_qubit_wise_multiply.mojo
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,7 +90,7 @@ fn benchmark_qubit_wise_multiply[
@parameter
@always_inline
fn benchmark_qubit_wise_multiply_inplace[
num_qubits: Int, number_layers: Int
num_qubits: Int, number_layers: Int, number_iterations: Int = 1
](mut b: Bencher) raises:
gates_list: List[Gate] = [Hadamard, PauliX, PauliY, PauliZ]

Expand All @@ -99,7 +99,7 @@ fn benchmark_qubit_wise_multiply_inplace[
)
random.seed() # Seed on current time
random.randint(
indexes, number_layers * 2 * num_qubits, 0, len(gates_list) - 1
indexes, number_layers * 1 * num_qubits, 0, len(gates_list) - 1
)

@parameter
Expand All @@ -125,80 +125,86 @@ fn benchmark_qubit_wise_multiply_inplace[
quantum_state_1 = StateVector.from_bitstring("0" * num_qubits)

current_state = 0
for layer in range(number_layers):
for i in range(num_qubits):
# NOTE Works but is slow with the dictionary
# qubit_wise_multiply_inplace(
# gates_list[Int(indexes[layer * num_qubits + i])].matrix,
# i,
# quantum_states[current_state],
# quantum_states[1 - current_state],
# )
# NOTE: Fast buty doesn't actually use the next state for new operations
# qubit_wise_multiply_inplace(
# gates_list[Int(indexes[layer * num_qubits + i])].matrix,
# i,
# quantum_state_0,
# quantum_state_1,
# )
if current_state == 0:
qubit_wise_multiply_inplace(
gates_list[Int(indexes[layer * num_qubits + i])].matrix,
i,
quantum_state_0,
quantum_state_1,
)
current_state = 1
else:
qubit_wise_multiply_inplace(
gates_list[Int(indexes[layer * num_qubits + i])].matrix,
i,
quantum_state_1,
quantum_state_0,
)
current_state = 0
for i in range(num_qubits - 1):
# qubit_wise_multiply_inplace(
# gates_list[
# Int(indexes[layer * num_qubits + num_qubits + i])
# ].matrix,
# i,
# quantum_states[current_state],
# quantum_states[1 - current_state],
# [[(i + 1) % num_qubits, 1]],
# )
# current_state = 1 - current_state
# qubit_wise_multiply_inplace(
# gates_list[
# Int(indexes[layer * num_qubits + num_qubits + i])
# ].matrix,
# i,
# quantum_state_0,
# quantum_state_1,
# [[(i + 1) % num_qubits, 1]],
# )
if current_state == 0:
qubit_wise_multiply_inplace(
gates_list[
Int(indexes[layer * num_qubits + num_qubits + i])
].matrix,
i,
quantum_state_0,
quantum_state_1,
[[(i + 1) % num_qubits, 1]],
)
current_state = 1
else:
qubit_wise_multiply_inplace(
gates_list[
Int(indexes[layer * num_qubits + num_qubits + i])
].matrix,
i,
quantum_state_1,
quantum_state_0,
[[(i + 1) % num_qubits, 1]],
)
current_state = 0
for iteration in range(number_iterations):
for layer in range(number_layers):
for i in range(num_qubits):
# NOTE Works but is slow with the dictionary
# qubit_wise_multiply_inplace(
# gates_list[Int(indexes[layer * num_qubits + i])].matrix,
# i,
# quantum_states[current_state],
# quantum_states[1 - current_state],
# )
# NOTE: Fast but doesn't actually use the next state for new operations
# qubit_wise_multiply_inplace(
# gates_list[Int(indexes[layer * num_qubits + i])].matrix,
# i,
# quantum_state_0,
# quantum_state_1,
# )
if current_state == 0:
qubit_wise_multiply_inplace(
gates_list[
Int(indexes[layer * num_qubits + i])
].matrix,
i,
quantum_state_0,
quantum_state_1,
)
current_state = 1
else:
qubit_wise_multiply_inplace(
gates_list[
Int(indexes[layer * num_qubits + i])
].matrix,
i,
quantum_state_1,
quantum_state_0,
)
current_state = 0
# Compare with GPU which currently doesn't support controls qubits yet
# for i in range(num_qubits - 1):
# # qubit_wise_multiply_inplace(
# # gates_list[
# # Int(indexes[layer * num_qubits + num_qubits + i])
# # ].matrix,
# # i,
# # quantum_states[current_state],
# # quantum_states[1 - current_state],
# # [[(i + 1) % num_qubits, 1]],
# # )
# # current_state = 1 - current_state
# # qubit_wise_multiply_inplace(
# # gates_list[
# # Int(indexes[layer * num_qubits + num_qubits + i])
# # ].matrix,
# # i,
# # quantum_state_0,
# # quantum_state_1,
# # [[(i + 1) % num_qubits, 1]],
# # )
# if current_state == 0:
# qubit_wise_multiply_inplace(
# gates_list[
# Int(indexes[layer * num_qubits + num_qubits + i])
# ].matrix,
# i,
# quantum_state_0,
# quantum_state_1,
# [[(i + 1) % num_qubits, 1]],
# )
# current_state = 1
# else:
# qubit_wise_multiply_inplace(
# gates_list[
# Int(indexes[layer * num_qubits + num_qubits + i])
# ].matrix,
# i,
# quantum_state_1,
# quantum_state_0,
# [[(i + 1) % num_qubits, 1]],
# )
# current_state = 0

bench_ctx = DeviceContext()
b.iter_custom[qubit_wise_multiply_inplace_workflow](bench_ctx)
Expand Down Expand Up @@ -369,6 +375,7 @@ def bench_qubit_wise_multiply_inplace[
number_layers_step_size: Int = 200,
fixed_number_qubits: Int = 5,
fixed_number_layers: Int = 2,
fixed_number_iterations: Int = 1,
]():
print("Running qubit_wise_multiply_inplace() Benchmarks...")
print("-" * 80)
Expand All @@ -382,7 +389,7 @@ def bench_qubit_wise_multiply_inplace[
):
bench.bench_function[
benchmark_qubit_wise_multiply_inplace[
number_qubits, fixed_number_layers
number_qubits, fixed_number_layers, fixed_number_iterations
]
](
BenchId(
Expand Down
Loading