Write initial snapshot memory in place by ludfjig · Pull Request #1927 · hyperlight-dev/hyperlight

ludfjig · 2026-10-12T20:01:02Z

New sandboxes skip copying their initial memory. Snapshot::from_env writes the guest code, PEB, init data and page tables in place, and the host populates the rest, mostly the heap, on first guest touch. Sandbox creation gets 33 to 88% faster for small to large heaps.

Benchmarks

Compared to the PR below.

MSHV: Linux 6.6 mshv.
KVM: Linux 6.17.

Benchmark	MSHV	KVM
`sandboxes/create_initialized/small`	4.6 ms → 3.09 ms (−33%)	5.12 ms → 2.92 ms (−43%)
`sandboxes/create_initialized/medium`	11.8 ms → 5.93 ms (−50%)	9.81 ms → 3.11 ms (−68%)
`sandboxes/create_initialized/large`	38.5 ms → 11 ms (−71%)	28.8 ms → 3.59 ms (−88%)
`snapshot_files/cold_start_via_evolve/large`	38.8 ms → 10.7 ms (−72%)	29.1 ms → 4.24 ms (−85%)

No benchmark regresses.

Full results

All benchmarks on both hosts (click to expand)

Before → after (change). Negative is faster. Bold is a regression. ~ is no consistent change. A change needs agreement between a 0.5 s and a 3 s run, or a 3 s run alone.

snapshots

Benchmark	Size	MSHV	KVM
`call_and_create`	default	378 µs → 366 µs (−3%)	360 µs → 379 µs (~)
`call_and_create`	small	425 µs → 411 µs (~)	417 µs → 462 µs (~)
`call_and_create`	medium	789 µs → 777 µs (~)	769 µs → 769 µs (~)
`call_and_create`	large	1.84 ms → 1.82 ms (~)	1.85 ms → 1.87 ms (~)
`create`	default	249 µs → 234 µs (−6%)	262 µs → 254 µs (~)
`create`	small	291 µs → 282 µs (~)	324 µs → 313 µs (~)
`create`	medium	606 µs → 612 µs (~)	606 µs → 629 µs (~)
`create`	large	1.68 ms → 1.68 ms (~)	1.7 ms → 1.68 ms (~)
`restore`	default	69 µs → 71.4 µs (~)	22.2 µs → 24.7 µs (~)
`restore`	small	68.7 µs → 73.9 µs (~)	26.7 µs → 23.5 µs (~)
`restore`	medium	72.3 µs → 77.2 µs (~)	30.8 µs → 28 µs (~)
`restore`	large	113 µs → 110 µs (~)	85.2 µs → 61.7 µs (~)
`restore_alternate`	default	78.6 µs → 82.5 µs (~)	71.8 µs → 92.1 µs (~)
`restore_alternate`	small	82.9 µs → 82.2 µs (~)	93.4 µs → 94.4 µs (~)
`restore_alternate`	medium	86.8 µs → 85.6 µs (~)	97.9 µs → 98.8 µs (~)
`restore_alternate`	large	123 µs → 125 µs (~)	158 µs → 129 µs (~)

guest_calls

Benchmark	Size	MSHV	KVM
`call`	default	45.2 µs → 45.4 µs (~)	17.1 µs → 16.9 µs (~)
`call`	small	45.2 µs → 45.2 µs (~)	17.5 µs → 18.0 µs (~)
`call`	medium	45.4 µs → 45.0 µs (~)	17.5 µs → 17.5 µs (~)
`call`	large	44.7 µs → 45.2 µs (~)	17.6 µs → 17.5 µs (~)
`call_with_host_function`	default	75.2 µs → 73.5 µs (~)	34.5 µs → 31.9 µs (~)
`call_with_host_function`	small	74.1 µs → 74 µs (~)	34.7 µs → 33.4 µs (~)
`call_with_host_function`	medium	75.2 µs → 75.7 µs (~)	32.8 µs → 34.2 µs (~)
`call_with_host_function`	large	75.2 µs → 75.8 µs (~)	32.7 µs → 33.6 µs (~)
`call_with_restore`	default	133 µs → 148 µs (~)	62.9 µs → 62.5 µs (~)
`call_with_restore`	small	140 µs → 139 µs (~)	63.8 µs → 71.9 µs (~)
`call_with_restore`	medium	138 µs → 141 µs (~)	67.9 µs → 67.4 µs (~)
`call_with_restore`	large	166 µs → 166 µs (~)	89.1 µs → 89.8 µs (~)
`different_thread`	-	74 µs → 72.3 µs (~)	42.9 µs → 42.3 µs (~)
`interrupt_latency`	-	48.9 µs → 64.5 µs (~)	35.9 µs → 39.2 µs (~)

guest_functions_with_large_parameters

Benchmark	Size	MSHV	KVM
`guest_call_with_large_parameters`	-	163 ms → 172 ms (~)	916 ms → 890 ms (~)

sample_workloads

Benchmark	Size	MSHV	KVM
`24K_in_8K_out_c`	-	56.5 µs → 57.5 µs (~)	27.6 µs → 27.6 µs (~)
`24K_in_8K_out_rust`	-	56.1 µs → 55.3 µs (~)	27.9 µs → 27.8 µs (~)

sandboxes

Benchmark	Size	MSHV	KVM
`create_initialized`	default	1.72 ms → 1.49 ms (−13%)	2.8 ms → 2.83 ms (~)
`create_initialized`	small	4.6 ms → 3.09 ms (−33%)	5.12 ms → 2.92 ms (−43%)
`create_initialized`	medium	11.8 ms → 5.93 ms (−50%)	9.81 ms → 3.11 ms (−68%)
`create_initialized`	large	38.5 ms → 11 ms (−71%)	28.8 ms → 3.59 ms (−88%)
`create_initialized_and_drop`	default	21.4 ms → 24.3 ms (~)	14.1 ms → 11 ms (~)
`create_initialized_and_drop`	small	28.9 ms → 25.9 ms (~)	13.3 ms → 15.7 ms (~)
`create_initialized_and_drop`	medium	31.5 ms → 28.3 ms (−10%)	24.3 ms → 14.8 ms (−39%)
`create_initialized_and_drop`	large	57.9 ms → 30.7 ms (−47%)	39.6 ms → 12.4 ms (−69%)
`sandbox_from_snapshot`	default	665 µs → 631 µs (−5%)	971 µs → 947 µs (~)
`sandbox_from_snapshot`	small	812 µs → 805 µs (~)	978 µs → 971 µs (~)
`sandbox_from_snapshot`	medium	2.02 ms → 2.04 ms (~)	1.01 ms → 999 µs (~)
`sandbox_from_snapshot`	large	6.06 ms → 6.16 ms (~)	1.22 ms → 1.21 ms (~)

snapshot_files

Benchmark	Size	MSHV	KVM
`cold_start_via_evolve`	default	1.61 ms → 1.58 ms (~)	2.96 ms → 3.05 ms (~)
`cold_start_via_evolve`	small	4.55 ms → 3.22 ms (~)	5.76 ms → 3.14 ms (~)
`cold_start_via_evolve`	medium	12.3 ms → 5.31 ms (~)	11 ms → 3.52 ms (~)
`cold_start_via_evolve`	large	38.8 ms → 10.7 ms (~)	29.1 ms → 4.24 ms (~)
`cold_start_via_snapshot`	default	1.97 ms → 2.01 ms (~)	2.85 ms → 2.73 ms (~)
`cold_start_via_snapshot`	small	8.93 ms → 9.13 ms (~)	8.95 ms → 8.9 ms (~)
`cold_start_via_snapshot`	medium	53.8 ms → 54 ms (~)	50.9 ms → 50.5 ms (~)
`cold_start_via_snapshot`	large	205 ms → 205 ms (~)	194 ms → 196 ms (~)
`cold_start_via_snapshot_unverified`	default	1.4 ms → 1.43 ms (~)	2.2 ms → 1.95 ms (−11%)
`cold_start_via_snapshot_unverified`	small	2.08 ms → 2.21 ms (~)	2.17 ms → 2.3 ms (~)
`cold_start_via_snapshot_unverified`	medium	3.59 ms → 3.52 ms (~)	2.25 ms → 2.1 ms (~)
`cold_start_via_snapshot_unverified`	large	7.73 ms → 7.67 ms (~)	2.59 ms → 2.34 ms (~)
`load_snapshot`	default	701 µs → 703 µs (~)	701 µs → 692 µs (~)
`load_snapshot`	small	6.71 ms → 6.63 ms (~)	6.68 ms → 6.56 ms (~)
`load_snapshot`	medium	50.4 ms → 50.3 ms (~)	48.8 ms → 48.3 ms (~)
`load_snapshot`	large	198 ms → 197 ms (~)	192 ms → 192 ms (~)
`load_snapshot_unverified`	default	112 µs → 116 µs (~)	123 µs → 130 µs (~)
`load_snapshot_unverified`	small	113 µs → 112 µs (~)	119 µs → 116 µs (~)
`load_snapshot_unverified`	medium	114 µs → 123 µs (~)	120 µs → 123 µs (~)
`load_snapshot_unverified`	large	118 µs → 122 µs (~)	125 µs → 127 µs (~)
`save_snapshot`	default	941 µs → 943 µs (~)	43.4 ms → 39.5 ms (~)
`save_snapshot`	small	9.09 ms → 9.11 ms (~)	61.5 ms → 52.2 ms (~)
`save_snapshot`	medium	69.6 ms → 67.7 ms (~)	402 ms → 396 ms (~)
`save_snapshot`	large	271 ms → 266 ms (~)	1.75 s → 1.74 s (~)

Copilot

🟡 Changes recommended

Lazy no-reserve allocations need aggregate admission or commitment accounting to prevent deferred host-wide OOM.

1 open finding

Add memory admission control for lazy large-heap allocations · New

What changed in this PR

Optimizes sandbox creation by constructing initial snapshot memory directly in shared memory and relying on lazy page population.

Changes:

Writes guest initialization data in place.
Freezes shared memory without copying.
Documents the performance improvement.

File	Description
`src/hyperlight_host/src/sandbox/snapshot/mod.rs`	Builds snapshots directly in shared memory.
`src/hyperlight_host/src/mem/shared_mem.rs`	Restricts the copying helper to tests and HVF.
`CHANGELOG.md`	Documents lazy memory population.

🧠 Review effort: Balanced

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

hyperlight-gh-bot · 2026-10-21T21:03:57Z

Benchmark Results

Measured commit: 5631afe66fdd
Baseline commit: c7396235f55e

kvm / amd (Linux) (🌟 6.07x)

Top improvements

sandboxes/create_initialized_and_drop/medium — 🌟 6.07x faster

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`749.74 ns` (➖ 1.01x faster)
`vec_bytes`	`586.79 ns` (➖ 1.00x slower)
		`372.86 µs` (➖ 1.00x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`526.34 ns` (➖ 1.00x slower)
`65536`	`142.35 ns` (➖ 1.00x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`13.79 ms` (🌟 6.07x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`7.73 ns` (➖ 1.00x slower)	`7.72 ns` (➖ 1.00x faster)	`7.72 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`118.13 µs` (➖ 1.02x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.27 µs` (➖ 1.02x slower)	`6.98 µs` (➖ 1.04x faster)
`65536`	`2.05 µs` (➖ 1.05x slower)	`1.93 µs` (➖ 1.09x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`6.18 µs` (➖ 1.05x faster)	`6.22 µs` (➖ 1.07x faster)
`8192`	`1.05 µs` (➖ 1.05x faster)	`1.05 µs` (➖ 1.01x slower)
`262144`		`26.83 µs` (➖ 1.02x faster)

kvm / intel (Linux) (🌟 5.92x)

Top improvements

sandboxes/create_initialized_and_drop/medium — 🌟 5.92x faster

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`810.86 ns` (➖ 1.00x slower)
`vec_bytes`	`614.59 ns` (➖ 1.00x slower)
		`749.70 µs` (➖ 1.01x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`624.45 ns` (➖ 1.05x faster)
`65536`	`166.87 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`14.70 ms` (🌟 5.92x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`8.61 ns` (➖ 1.02x slower)	`8.57 ns` (➖ 1.00x slower)	`8.49 ns` (➖ 1.01x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`70.10 µs` (➖ 1.09x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.05 µs` (➖ 1.00x slower)	`9.03 µs` (➖ 1.02x faster)
`65536`	`2.58 µs` (➖ 1.01x slower)	`2.57 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.44 µs` (➖ 1.01x faster)	`8.44 µs` (➖ 1.02x faster)
`8192`	`880.07 ns` (➖ 1.03x faster)	`866.49 ns` (➖ 1.00x faster)
`262144`		`34.88 µs` (➖ 1.02x faster)

mshv3 / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`908.89 ns` (➖ 1.02x slower)
`vec_bytes`	`709.07 ns` (➖ 1.00x faster)
		`316.48 µs` (➖ 1.00x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`671.46 ns` (➖ 1.07x faster)
`65536`	`188.29 ns` (➖ 1.01x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`65.96 ms` (➖ 1.10x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.93 ns` (➖ 1.00x faster)	`9.93 ns` (➖ 1.01x faster)	`9.92 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`107.95 µs` (➖ 1.04x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.68 µs` (➖ 1.04x slower)	`9.12 µs` (➖ 1.00x faster)
`65536`	`2.63 µs` (➖ 1.02x faster)	`2.53 µs` (➖ 1.01x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.01 µs` (➖ 1.04x faster)	`7.97 µs` (➖ 1.01x faster)
`8192`	`1.36 µs` (➖ 1.03x slower)	`1.45 µs` (➖ 1.10x slower)
`262144`		`34.87 µs` (➖ 1.03x faster)

mshv3 / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`968.88 ns` (➖ 1.02x slower)
`vec_bytes`	`685.63 ns` (➖ 1.04x slower)
		`924.97 µs` (➖ 1.25x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`646.93 ns` (➖ 1.01x slower)
`65536`	`169.86 ns` (➖ 1.02x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`77.56 ms` (➖ 1.09x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.66 ns` (➖ 1.10x slower)	`8.82 ns` (➖ 1.00x faster)	`8.79 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`62.18 µs` (➖ 1.01x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`8.03 µs` (➖ 1.00x slower)	`7.99 µs` (➖ 1.02x faster)
`65536`	`2.31 µs` (➖ 1.01x slower)	`2.32 µs` (➖ 1.02x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`7.56 µs` (➖ 1.00x faster)	`7.50 µs` (➖ 1.02x faster)
`8192`	`932.01 ns` (➖ 1.01x faster)	`925.93 ns` (➖ 1.02x slower)
`262144`		`40.19 µs` (➖ 1.03x faster)

hyperv-ws2025 / amd (Windows) (🌟 6.72x)

Top improvements

sandboxes/create_initialized_and_drop/medium — 🌟 6.72x faster

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.19 µs` (➖ 1.00x faster)
`vec_bytes`	`800.31 ns` (➖ 1.00x faster)
		`1.84 ms` (➖ 1.05x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`794.77 ns` (➖ 1.00x faster)
`65536`	`231.35 ns` (➖ 1.00x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`11.81 ms` (🌟 6.72x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.85 ns` (➖ 1.00x faster)	`11.09 ns` (➖ 1.10x slower)	`10.07 ns` (➖ 1.01x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`1.70 ms` (➖ 2.26x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.13 µs` (➖ 1.03x faster)	`9.18 µs` (➖ 1.01x slower)
`65536`	`2.39 µs` (➖ 1.00x slower)	`2.38 µs` (➖ 1.01x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.93 µs` (➖ 1.01x slower)	`9.04 µs` (➖ 1.02x slower)
`8192`	`1.34 µs` (➖ 1.01x faster)	`1.34 µs` (➖ 1.02x slower)
`262144`		`39.36 µs` (➖ 1.18x faster)

hyperv-ws2025 / intel (Windows) (🌟 7.17x)

Top improvements

sandboxes/create_initialized_and_drop/medium — 🌟 7.17x faster

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.25 µs` (➖ 1.03x faster)
`vec_bytes`	`790.03 ns` (➖ 1.01x faster)
		`4.00 ms` (➖ 1.05x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`751.37 ns` (➖ 1.00x slower)
`65536`	`216.42 ns` (➖ 1.01x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`17.28 ms` (🌟 7.17x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`10.45 ns` (➖ 1.01x faster)	`10.10 ns` (➖ 1.00x faster)	`10.00 ns` (➖ 1.01x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`1.00 ms` (➖ 1.27x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`8.09 µs` (➖ 1.02x slower)	`8.16 µs` (➖ 1.03x slower)
`65536`	`2.37 µs` (➖ 1.02x faster)	`2.38 µs` (➖ 1.01x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.58 µs` (➖ 1.02x faster)	`8.91 µs` (➖ 1.01x faster)
`8192`	`1.16 µs` (➖ 1.10x faster)	`1.10 µs` (➖ 1.01x faster)
`262144`		`42.90 µs` (➖ 1.13x faster)

_{Reported by cargo ci bench-report --candidate run:39661853686 --config-file bench_report.toml.}

call_and_create times a guest call plus snapshot creation, so the call pays for memory remapping done by the previous snapshot. restore_alternate restores two snapshots in turn, so each restore targets memory that is not current. Signed-off-by: Ludvig Liljenberg <[email protected]>

Share immutable data blobs between related snapshots. Each blob holds guest data followed by the page tables of the snapshot that created it. Persist layered snapshots as OCI image layouts. Capture copies only pages that changed since the snapshot the sandbox was created from or last restored to. Signed-off-by: Ludvig Liljenberg <[email protected]>

Snapshot capture switches the running sandbox to the captured snapshot. The next capture reuses its layers and copies only pages changed after it. Scratch pages used before the capture become free again. They are zeroed so guest allocations return zeroed pages, as after restore. Capture removes map_region and map_file_cow regions. Pages of these regions that the guest has not mapped are lost. Signed-off-by: Ludvig Liljenberg <[email protected]>

walk_va_spaces merges mappings that have contiguous physical and virtual addresses and the same kind. Snapshot capture classifies and rebuilds each merged mapping once. Criterion snapshots/create: * MSHV: medium 1.18 ms to 950 us (-20%), large 3.56 ms to 2.85 ms (-20%) * KVM: medium 1.34 ms to 1.04 ms (-22%), large 3.59 ms to 2.91 ms (-19%) Signed-off-by: Ludvig Liljenberg <[email protected]>

A page-table walk reads every entry through GuestPhysicalMemoryView, which resolves the backing and copies 8 bytes through a locked volatile copy. The reader keeps a copy of the last table page it read, so each table costs one copy. Snapshot capture and crash dumps use the reader. Criterion snapshots/create: * MSHV: default 344 us to 248 us (-28%), medium 950 us to 594 us (-37%) * KVM: default 499 us to 379 us (-24%), medium 1.04 ms to 746 us (-28%) Signed-off-by: Ludvig Liljenberg <[email protected]>

By default, deleting a memslot flushes the guest mappings of every memslot. KVM keeps that flush as a workaround for VMs with assigned GPUs. Hyperlight assigns no devices and disables the quirk when the kernel supports it (Linux 6.12 and later). Deleting a memslot then flushes only that memslot's mappings. Snapshot capture deletes a memslot when a layer drops out of the snapshot. Restore to another snapshot and region unmapping also delete memslots. Criterion on KVM (Linux 6.17): * snapshots/call_and_create: 24% to 58% faster * snapshots/restore_alternate: 32% to 52% faster Signed-off-by: Ludvig Liljenberg <[email protected]>

* Iterate snapshot mappings in reverse and test membership with contains. * Rename GuestVirtualMemoryReader::read and the public SnapshotMemoryReader::read to read_exact. * Move the snapshot memory test helpers into the tests module. Signed-off-by: Ludvig Liljenberg <[email protected]>

Snapshot::from_env writes the guest code, PEB and init data straight into shared memory and freezes it. The host populates the remaining pages on first guest touch, so sandbox creation cost stays independent of heap size. Signed-off-by: Ludvig Liljenberg <[email protected]>

ludfjig added the kind/enhancement For PRs adding features, improving functionality, docs, tests, etc. label Oct 12, 2026

ludfjig requested review from andreiltd, danbugs, dblnz, devigned, jprendes, jsturtevant, simongdavies, squillace and syntactically as code owners October 12, 2026 20:01

ludfjig added this pull request to stack #1929 October 12, 2026 20:01

ludfjig mentioned this pull request Oct 12, 2026

Layered/incremental snapshots #1902

Closed

This comment has been minimized.

Sign in to view

Copilot AI balanced review requested due to automatic review settings October 13, 2026 18:56

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from 96fab64 to ab775e0 Compare October 13, 2026 18:56

Copilot started reviewing on behalf of ludfjig October 13, 2026 19:03

View session

Copilot AI reviewed Oct 13, 2026

View reviewed changes

Comment thread src/hyperlight_host/src/sandbox/snapshot/mod.rs

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from ab775e0 to 107a858 Compare October 13, 2026 21:01

This comment has been minimized.

Sign in to view

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from 107a858 to aa55340 Compare October 13, 2026 23:00

This comment has been minimized.

Sign in to view

andreiltd previously approved these changes Oct 14, 2026

View reviewed changes

ludfjig dismissed andreiltd’s stale review via 421a453 October 15, 2026 22:00

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from aa55340 to 421a453 Compare October 15, 2026 22:00

This comment has been minimized.

Sign in to view

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from 421a453 to 5631afe Compare October 21, 2026 19:27

ludfjig added 8 commits October 26, 2026 11:23

ludfjig force-pushed the ll/2-lazy-sandbox-memory branch from 5631afe to 5c04ce3 Compare October 26, 2026 18:44

Conversation

ludfjig commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmarks

Full results

snapshots

guest_calls

guest_functions_with_large_parameters

sample_workloads

sandboxes

snapshot_files

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

hyperlight-gh-bot Bot commented Oct 21, 2026

Benchmark Results

Top improvements

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Top improvements

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Top improvements

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Top improvements

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

ludfjig commented Oct 12, 2026 •

edited

Loading