Performance measurements › arm_1 › kem-33 · system: x86_1 · arm_1

kem-33 QUBE — performance on AArch64 (system arm_1)

Independent measurement following the structure of the NICCS ARM self-assessment guide, §3.5 (1)–(7). Not a submitter self-assessment.

1. Basic information

Category: public-key algorithm; function: key encapsulation
Algorithm: QUBE
Implementation versions measured: reference
Parameter sets: qube-128, qube-192, qube-256, qube-384, qube-512
Security evaluation: kem-33 report
Measurement method: arm_1 method
Submission: NICCS page

2. Assessment environment

item	value
processor	Qualcomm Oryon (CPU 2, one core)
machine	ASUS Vivobook S 15
clock	fixed 2.71 GHz (governor performance, minimum = maximum), boost off, SMT none
memory	30562 MiB
OS / kernel	Ubuntu 26.04.1 LTS / 7.0.0-34-generic
compiler / build tool	gcc (Ubuntu 15.2.0-16ubuntu1) 15.2.0 / cmake version 4.2.3
campaign start / end (UTC)	2026-09-28T15:05:46 / 2026-09-29T08:43:52

3. Functional testing (KAT)

Each library was checked against the submitted KAT vectors (SHA-256 manifest kem-33/kat.sha256 in the harness) before timing.

instance	build flags	KAT
`qube-128`	harness-default	MISMATCH [1]
`qube-192`	harness-default	NOKAT [1]
`qube-256`	harness-default	MISMATCH [1]
`qube-384`	harness-default	MISMATCH [1]
`qube-512`	harness-default	MISMATCH [1]

[1] The reference sources reproduce their own in-tree KAT files byte for byte, but the top-level Test_Vectors carry the optimized implementation's key and ciphertext sizes at four levels, and no top-level qube-192 vector was submitted (kem-33/README.md, pseudocode.md). These instances are timed anyway; their output is not validated.

4. Performance

instance	operation	mean cycles	mean time	ops/s	median trial	n (trials × iters)
`qube-128`	keygen	551.6 k	205 µs	4.89e+03	205 µs	14535 (5 × 2907)
`qube-128`	enc	600.8 k	223 µs	4.49e+03	223 µs	13560 (5 × 2712)
`qube-128`	dec	1.03 M	381 µs	2.62e+03	379 µs	8255 (5 × 1651)
`qube-192`	keygen	1.25 M	466 µs	2.15e+03	462 µs	6410 (5 × 1282)
`qube-192`	enc	1.32 M	489 µs	2.04e+03	489 µs	6340 (5 × 1268)
`qube-192`	dec	2.36 M	877 µs	1.14e+03	876 µs	3585 (5 × 717)
`qube-256`	keygen	2.55 M	947 µs	1.06e+03	947 µs	3250 (5 × 650)
`qube-256`	enc	2.60 M	965 µs	1.04e+03	962 µs	3340 (5 × 668)
`qube-256`	dec	4.78 M	1.77 ms	563	1.78 ms	1785 (5 × 357)
`qube-384`	keygen	6.15 M	2.28 ms	438	2.28 ms	1355 (5 × 271)
`qube-384`	enc	6.26 M	2.32 ms	430	2.31 ms	1340 (5 × 268)
`qube-384`	dec	11.10 M	4.12 ms	243	4.11 ms	760 (5 × 152)
`qube-512`	keygen	16.78 M	6.23 ms	161	6.22 ms	490 (5 × 98)
`qube-512`	enc	16.76 M	6.22 ms	161	6.21 ms	510 (5 × 102)
`qube-512`	dec	25.27 M	9.38 ms	107	9.37 ms	340 (5 × 68)

5. Resource consumption

Static memory is approximated by the library's loadable ELF segments; peak memory is the process high-water mark (VmHWM) including the benchmark driver and libc, so both are upper-bound proxies rather than isolated algorithm memory.

instance	operation	static: ELF image (bytes)	baseline RSS	peak RSS
`qube-128`	keygen	37348	1436 KiB	3552 KiB
`qube-128`	enc	37348	1460 KiB	1536 KiB
`qube-128`	dec	37348	1468 KiB	1548 KiB
`qube-192`	keygen	38996	1448 KiB	1572 KiB
`qube-192`	enc	38996	3508 KiB	3572 KiB
`qube-192`	dec	38996	1516 KiB	1604 KiB
`qube-256`	keygen	39132	1460 KiB	1604 KiB
`qube-256`	enc	39132	1552 KiB	1636 KiB
`qube-256`	dec	39132	1572 KiB	1648 KiB
`qube-384`	keygen	39020	1492 KiB	1712 KiB
`qube-384`	enc	39020	3716 KiB	3780 KiB
`qube-384`	dec	39020	1740 KiB	1824 KiB
`qube-512`	keygen	39420	3468 KiB	3680 KiB
`qube-512`	enc	39420	1872 KiB	1940 KiB
`qube-512`	dec	39420	3848 KiB	3912 KiB

6. Transmission and storage overhead

External public-key, ciphertext and signature sizes follow the curated performance/external_sizes.csv catalog; secret-key and shared-secret lengths remain API figures. See the size audit for disagreements.

instance	public key	secret key	ciphertext	shared secret
`qube-128`	3294	3326	3287	16
`qube-192`	6364	6412	6362	24
`qube-256`	10446	10510	10423	32
`qube-384`	20738	20834	20633	48
`qube-512`	34028	34156	33846	64

Symmetric primitives

Survey verdict: ICCS-only

Share of each operation spent in the ICCS placeholder hash functions and in the ICCS DRNG (reference build, measured in the same run with link-time wrappers):

instance	operation	pseudohash/XOF/sm3	DRNG	calls per operation
`qube-128`	keygen	57%	1.6%	drng 2, pseudoXOF 7.07, sm3hash 1
`qube-128`	enc	62%	1.4%	drng 2, pseudoXOF 7, sm3hash 2
`qube-128`	dec	52%	0.0%	pseudoXOF 11, sm3hash 4
`qube-192`	keygen	50%	0.7%	drng 2, pseudoXOF 8, pseudohash 1
`qube-192`	enc	57%	0.7%	drng 2, pseudoXOF 8, pseudohash 1, sm3hash 1
`qube-192`	dec	45%	0.0%	pseudoXOF 13, pseudohash 1, sm3hash 3
`qube-256`	keygen	49%	0.3%	drng 2, pseudoXOF 9, pseudohash 1
`qube-256`	enc	55%	0.3%	drng 2, pseudoXOF 8.57, pseudohash 1, sm3hash 1
`qube-256`	dec	40%	0.0%	pseudoXOF 14, pseudohash 1, sm3hash 3
`qube-384`	keygen	41%	0.2%	drng 2, pseudoXOF 10, pseudohash 1
`qube-384`	enc	51%	0.2%	drng 2, pseudoXOF 9, pseudohash 2
`qube-384`	dec	44%	0.0%	pseudoXOF 15, pseudohash 4
`qube-512`	keygen	54%	0.1%	drng 2, pseudoXOF 10, pseudohash 1
`qube-512`	enc	63%	0.1%	drng 2, pseudoXOF 10, pseudohash 2
`qube-512`	dec	53%	0.0%	pseudoXOF 17, pseudohash 4

7. Raw evidence index

Paths are relative to performance/data/arm_1/ in the harness; each JSON file records its commands, environment and trials.

instance	item	file
`qube-128`	KAT log (sha256 `430d6f37988fdac0…`)	`kat/kem-33/qube-128.log`
`qube-128`	timing dec	`records/kem-33/qube-128__dec.json`
`qube-128`	timing enc	`records/kem-33/qube-128__enc.json`
`qube-128`	timing keygen	`records/kem-33/qube-128__keygen.json`
`qube-128`	hash profile dec	`profile/kem-33/qube-128__dec.json`
`qube-128`	hash profile enc	`profile/kem-33/qube-128__enc.json`
`qube-128`	hash profile keygen	`profile/kem-33/qube-128__keygen.json`
`qube-192`	KAT log (sha256 `8e39d7098e70e062…`)	`kat/kem-33/qube-192.log`
`qube-192`	timing dec	`records/kem-33/qube-192__dec.json`
`qube-192`	timing enc	`records/kem-33/qube-192__enc.json`
`qube-192`	timing keygen	`records/kem-33/qube-192__keygen.json`
`qube-192`	hash profile dec	`profile/kem-33/qube-192__dec.json`
`qube-192`	hash profile enc	`profile/kem-33/qube-192__enc.json`
`qube-192`	hash profile keygen	`profile/kem-33/qube-192__keygen.json`
`qube-256`	KAT log (sha256 `8572024e2a5eee13…`)	`kat/kem-33/qube-256.log`
`qube-256`	timing dec	`records/kem-33/qube-256__dec.json`
`qube-256`	timing enc	`records/kem-33/qube-256__enc.json`
`qube-256`	timing keygen	`records/kem-33/qube-256__keygen.json`
`qube-256`	hash profile dec	`profile/kem-33/qube-256__dec.json`
`qube-256`	hash profile enc	`profile/kem-33/qube-256__enc.json`
`qube-256`	hash profile keygen	`profile/kem-33/qube-256__keygen.json`
`qube-384`	KAT log (sha256 `f55f27e3a733b7ae…`)	`kat/kem-33/qube-384.log`
`qube-384`	timing dec	`records/kem-33/qube-384__dec.json`
`qube-384`	timing enc	`records/kem-33/qube-384__enc.json`
`qube-384`	timing keygen	`records/kem-33/qube-384__keygen.json`
`qube-384`	hash profile dec	`profile/kem-33/qube-384__dec.json`
`qube-384`	hash profile enc	`profile/kem-33/qube-384__enc.json`
`qube-384`	hash profile keygen	`profile/kem-33/qube-384__keygen.json`
`qube-512`	KAT log (sha256 `8447864d076f8758…`)	`kat/kem-33/qube-512.log`
`qube-512`	timing dec	`records/kem-33/qube-512__dec.json`
`qube-512`	timing enc	`records/kem-33/qube-512__enc.json`
`qube-512`	timing keygen	`records/kem-33/qube-512__keygen.json`
`qube-512`	hash profile dec	`profile/kem-33/qube-512__dec.json`
`qube-512`	hash profile enc	`profile/kem-33/qube-512__enc.json`
`qube-512`	hash profile keygen	`profile/kem-33/qube-512__keygen.json`

Scripts: performance/campaign.py, performance/ngcc_perf.c, performance/hashprof/ in the harness.