Geekerwan Benchmark
EN
English简体中文繁體中文

Geekerwan Benchmark: Phone CPU, GPU & Efficiency Rankings

Geekerwan benchmark data for phones: CPU, GPU and efficiency rankings of 150+ chips, plus Geekbench, GFXBench and 3DMark results. Filter, sort and search.

View all →
8 benchmarks150+ phone chips
REAL TEST DATA / 2026-11
GEEKERWAN
BENCHMARK
GB6 single-core #14,019
GB6 multi-core #114,905
chips indexed150+
01 / ABOUT

About Geekerwan and the Geekerwan Benchmark

Geekerwan is a Chinese tech media team known for hardcore, in-depth hardware testing and technical analysis. With more than five million followers on Bilibili (as of 2026) and multiple "Top 100 Uploader" awards, they cover phones, PCs, graphics cards and the principles behind chip design. Their best-known videos include large-scale battery life comparisons, flagship phone chip shootouts and investigations into gaming performance "cheating".

Beyond standard tools such as Geekbench, 3DMark and GFXBench, Geekerwan developed its own testing methodology over years of reviews and launched distinctive features such as efficiency curves and battery tests. The benchmark, battery and efficiency figures gathered along the way were compiled into widely shared charts — collectively known as the Geekerwan Benchmark.

Geekerwan on BilibiliSOCPK Chip DatabaseGeekerwan on YouTube
02 / RANKINGS

Geekerwan Benchmark Overview

{N} benchmark charts covering overall, CPU / GPU performance and efficiency, plus Geekbench 5/6, GFXBench 5 and 3DMark WLE scores.

Phone Chip Performance Benchmark

158 chips ranked by overall index: CPU 70%, GPU 30%, Snapdragon 865 = 100.

Phone CPU Benchmark

158 chips ranked by CPU index: single-core 30%, multi-core 70%.

Phone GPU Benchmark

158 chips ranked by GPU index: peak 50%, sustained 50%, throttling included.

Phone CPU Efficiency Benchmark

51 chips ranked by multi-core performance per watt in the 3W-5W range.

Geekbench 6 Phone CPU Benchmark

82 chips by raw Geekbench 6 scores: single-core for daily speed, multi-core for heavy loads.

Geekbench 5 Phone CPU Benchmark

147 chips across generations, flagship to entry level: look up older phones and see the gains.

GFXBench 5 Phone GPU Benchmark

143 chips by graphics fps: Metal on Apple, best of OpenGL ES / Vulkan on Android.

3DMark Wild Life Extreme GPU Benchmark

82 chips on a heavy, cross-platform GPU stress test of peak and sustained graphics.

03 / METHODOLOGY

How Geekerwan Tests

Geekerwan’s testing methodology has been explained across many videos. Taking efficiency curves and battery life tests as examples, the core idea is to remove vendor "tuning / cheating" from the equation through controlled variables and automated simulation. The key steps are summarised below:

1. Chip efficiency curve testing

  • Hardware sourcing

    No vendor-supplied "press-tuned" review units: every phone is bought at retail, and chip development boards are used to cross-check baseline data.

  • Controlled variables

    All tests run at a constant 25°C, with memory, storage and other peripheral specs kept as comparable as possible to reduce their impact on chip power draw.

  • Sustained load, not peak scores

    Brief AnTuTu or Geekbench peak runs are ignored because their power draw cannot be sustained in daily use. SPEC 2026, Geekbench 6 and 3DMark map the sustained-load efficiency curve across power bands and clock speeds, focusing on the sweet spot where low clocks are highly efficient and high clocks fall off a cliff.

2. Battery life testing (the Battery Test 5.0 standard)

  • AI Agent 5.0 automation

    A locally deployed dual vision-language model drives the phone autonomously instead of a human tapping, producing realistic, highly varied and reproducible everyday usage.

  • Strict, unified test conditions
    • Every phone is set to balanced performance mode.
    • Screen brightness locked at 400 nits.
    • Location services and haptics stay on throughout, a weak 5G network is simulated, and audio plays through a Bluetooth speaker.
  • Looped everyday scenario script

    One full loop takes one hour and repeats until the phone shuts down. Each loop randomises the order of high-frequency daily tasks:

    • Messaging, shopping and scrolling social feeds.
    • Background WeChat voice/video calls alongside music playback.
    • Online video, popular mobile games, navigation and rapid app open/close cycles, with at least one WeChat message per minute to mirror real mixed heavy-to-moderate usage.
  • Anti-cheating and cell monitoring

    Before each run, vendor-side scheduler locks aimed at battery tests are identified and excluded. Cell-level capacity-lock checks and charging energy curves are recorded, so the headline battery capacity is never the only number considered.

04 / GLOSSARY

Benchmark Glossary

To read the Geekerwan Benchmark properly, it helps to know these terms first — they make it much easier to understand what the benchmark rankings actually tell you.

TermWhat it isHow to read it
Geekbench single/multi-coreThe most common cross-platform CPU benchmarkSingle-core for daily speed, multi-core for heavy jobs
SPEC CPU 2026Industry-grade CPU benchmark with unified compiler flagsTougher than phone benchmarks; the core of efficiency curves
GFXBench / 3DMark WLEFixed-scene GPU benchmarks for rendering ceilingHigh score = strong peak graphics; sustained performance is a separate chart
Efficiency curvePerformance vs. power (W), measured point by pointHigher line at the same wattage = more efficient
Board powerChip power measured with a dummy battery, display excludedChip-vs-chip comparison without screen noise
Peak vs. sustainedPeak is a single run; sustained is long-session outputGames ride on sustained; throttling is already counted
Snapdragon 865 = 100Index charts are normalized to Snapdragon 865200 means twice the performance — cross-gen at a glance
Battery Test 5.0Standardized full-run battery methodology, on videoOnly same-methodology results compare; check battery Wh too
Benchmark scoreRaw results from Geekbench, GFXBench, 3DMark and similar toolsOnly comparable within the same tool and version
IndexRelative score normalized to Snapdragon 865 = 100 (overall, CPU, GPU, efficiency)Comparable across generations; not the same as a benchmark score
3DMark WLEShort for 3DMark Wild Life Extreme, a heavy cross-platform GPU stress testHigh score = strong peak graphics; sustained performance is separate
Frame rate (fps)Frames rendered per secondOnly comparable at the same resolution and scene
Why are Raspberry Pi and Switch chips listed?Geekbench and GFXBench are cross-platform; the source lists every SoC it has testedA few tablet, handheld and dev-board chips appear, which does not break comparison
Efficiency (performance per watt)Performance at the same power, or power for the same performanceHere measured as multi-core performance in the 3W-5W band
05 / HOW TO USE

How to Use These Benchmarks to Pick a Phone

Start from what you care about, then pick the matching chart. Three most common scenarios:

01Daily fluidity

Geekbench single-core + CPU efficiency benchmark

02Heavy gaming

GPU performance benchmark (peak + sustained)

03Battery anxiety

CPU efficiency benchmark + battery capacity

STEP 1

Orient: overall benchmark first

Use the overall performance benchmark to place the chip’s ballpark.

View chart →
STEP 2

Zoom in: per-dimension benchmarks

Check the CPU / GPU benchmarks to see whether it leans burst or graphics.

View chart →
STEP 3

Verify: efficiency over time

A fast chip does not equal a cool phone — check the CPU efficiency benchmark at the same power.

View chart →
06 / TRUSTED

Why the Geekerwan Benchmark Is Widely Cited

Credibility comes from method, not traffic.

01

Retail units, self-purchased

All test devices are bought at retail — no vendor samples, no commercial influence.

02

Transparent methodology

Compiler flags, power methodology and weights are published with every chart.

03

Board-level measurement

Dummy-battery rigs isolate the chip’s own power draw from the display.

04

Video records available

Battery tests are recorded end-to-end and open for anyone to review.

07 / VALUE

What These Benchmarks Are Good For

Scores are not the point. Turning scattered test data into comparable benchmarks makes every purchase and discussion evidence-based.

→

Evidence for buyers

No more fuzzy “flagship / mid-range” impressions — verifiable relative positions.

→

Visible progress

147 chips across years in the Geekbench 5 archive show generational gains under one methodology.

→

Tech, explained

Benchmark scores, efficiency curves and battery test conditions become concepts you can actually understand, not just numbers to memorise.

08 / VIDEOS

Geekerwan Videos

Videos by Geekerwan on Bilibili (in Chinese). Click to play.

Geekerwan 2026 Flagship Smartphone Chipset Showdown: Apple vs. Qualcomm vs. MediaTek vs. Xring – Who Is the Best?

Snapdragon 8 Elite Extreme Gen 6 First Look: Qualcomm's New Flagship Mobile Chip Is Here!

iPhone 18 Pro Performance Analysis: The Biggest Chip Upgrade in Five Years!

Dimensity 9600 Pro First Look: How Efficient Is This 2nm Flagship Mobile Chip?

09 / FAQ

Benchmark FAQ

Questions people often ask.

Is the data accurate?

＋

It is compiled from SOCPK’s public test results, with data version and entry counts shown per page. Values are rounded for display — the original site is authoritative.

How often is it updated?

＋

Updates follow the original site’s data versions; version and date are shown at the top of each chart.

Why is my chip not on the list?

＋

Coverage follows the original tests — only tested devices make the list; older chips may live in the Geekbench 5 archive.

Does a higher score mean a faster phone?

＋

Not necessarily. Cooling, scheduling and OS optimization shape real experience; use these benchmarks for ceilings and ballpark positions, alongside efficiency and battery charts.

START EXPLORING

Start exploring now
Geekerwan Benchmark

Performance, efficiency, battery — make a clearer choice with real test data.

View all benchmarks →Geekerwan SOCPK (original site) ↗