PC (12th Gen Intel Core i7-12700KF) vs Xiaomi 17T

9,194Math4,591

2,251 ops/s

Throughput

1,124 ops/s

15,434JSON & XML6,688

1,664 ops/s

Throughput

721 ops/s

13,207Compression & crypto5,711

1,277 ops/s

Throughput

552 ops/s

28,951Multi-core4,585

24,140 ops/s

Throughput

3,823 ops/s

12,142RAM5,028

87.2 GB/s	Read	41.4 GB/s
52.3 GB/s	Write	16.7 GB/s
41.0 GB/s	Copy	16.7 GB/s
89.0 ns	Latency	187 ns

3,562Storage7,595

1,192 MB/s	Sequential read	2,982 MB/s
327 MB/s	Sequential write	1,667 MB/s
5,548 IOPS	Random read (4 KB)	39,603 IOPS
50,833 IOPS	Random write (4 KB)	11,531 IOPS
direct I/O	Reads	direct I/O

18,490Database8,686

15,536 tx/s	Insert	16,142 tx/s
121,302 tx/s	Update	82,898 tx/s
110,075 /s	Query	16,072 /s

26,234Lake Valley2,537

567 fps	Average	54.8 fps
215 fps	1% low	41.7 fps

35,345Neon City2,978

155 fps	Average	13.1 fps
86.6 fps	1% low	7.7 fps

68,719Neon City RT5,470

153 fps	Average	12.2 fps
106 fps	1% low	9.2 fps

16,639Web browsing4,123

6,832	Layout	3,673
4,818	Canvas	1,720
4,000	CSS	340
10,579	SVG	1,470
5,680 ops/s	JavaScript	2,338 ops/s

76,987Web scrolling4,997

854,555 px/s	Speed held	55,465 px/s
0.2 %	Frames dropped	3.7 %

17,329Photo gallery5,200

367 MP/s	JPEG	113 MP/s
263 MP/s	WebP	87.2 MP/s
310 MP/s	AVIF	81.7 MP/s

26,473Photo editing3,261

1,752 MP/s

Editing chain

216 MP/s

6,910Video playback2,707

203 fps · hardware	4K60 H.264	174 fps · hardware
438 fps · hardware	4K60 HEVC	195 fps · hardware
335 fps · hardware	4K60 VP9	184 fps · hardware
365 fps · hardware	4K60 AV1	164 fps · hardware
95.4 fps · hardware	8K30 HEVC	not supported
88.2 fps · hardware	8K30 AV1	not supported

1,442Video export2,849

81.7 fps · 25.6 dB · hardware · quality too low, not counted	4K60 H.264	93.3 fps · 39.3 dB · hardware
96.1 fps · 27.0 dB · hardware	4K60 HEVC	96.7 fps · 42.0 dB · hardware
no encoder	4K60 AV1	no encoder
31.2 fps · 25.7 dB · hardware · quality too low, not counted	8K30 HEVC	no encoder

10,392QR scanning5,170

974 scans/s	ZXing	485 scans/s
n/a	ML Kit	66.8 scans/s

15,654CPU4,838

330 inferences/s	Classify	243 inferences/s
9.97 inferences/s	Detect	13.5 inferences/s
5.44 inferences/s	Segment	4.44 inferences/s
28.8 inferences/s	Upscale	2.51 inferences/s
54.5 inferences/s	Depth	30.6 inferences/s
2,288 inferences/s	Sound	2,137 inferences/s

19,831GPU1,471

1,062 inferences/s	Classify	99.9 inferences/s
376 inferences/s	Detect	23.5 inferences/s
189 inferences/s	Segment	10.8 inferences/s
355 inferences/s	Upscale	27.8 inferences/s
519 inferences/s	Depth	53.2 inferences/s
2,097 inferences/s	Sound	184 inferences/s

0NPU0

none ran	Runtime	none ran
—	Classify	—
—	Detect	—
—	Segment	—
—	Upscale	—
—	Depth	—
—	Sound	—

31,013Prompt (prefill)294

70.2 tok/s	CPU	17.8 tok/s
1,875 tok/s	GPU	5.0 tok/s

48,582Reply (decode)3,956

42.4 tok/s	CPU	11.4 tok/s
140 tok/s	GPU	6.9 tok/s

Run

NVIDIA GeForce RTX 3080 Ti	GPU used	Mali-G720 MC8
24.6 GB	Free memory at start	5.3 GB