Skip to content
Shop Bitcoin Australia Buy a Bitaxe →
Mining · 10 min read

Overclocking Your Bitaxe Gamma: A Stepped Approach (and When to Stop)

Stock is 490 MHz at 1.20 V. A stable overclock nets 1.35 TH/s on good silicon. A reckless overclock bricks the chip or burns a VR. Here's the dial-in methodology we've run on 60+ Aussie customer units — plus the summer-thermal cap you need to respect.

A Bitaxe ASIC miner in the centre with overlay frequency and voltage tuning dials, a hashrate curve climbing with a red thermal-cap marker.

A stock Bitaxe Gamma ships at 490 MHz and 1.20 V. That’s 1.20 TH/s at ~17 W — 14.2 J/TH. These are conservative numbers. They’re designed to land every chip the factory ships inside a safe envelope, with headroom for the worst silicon and the worst thermal conditions.

Your specific chip is probably better than that.

How much better depends on the silicon lottery — the wafer-level variance in dopant concentration, transistor switching speed, and thermal characteristics that makes every ASIC subtly different. Some Gammas run flawlessly at 560 MHz / 1.26 V. Some give up at 515 MHz. You won’t know until you measure.

This piece is the methodology we use to dial in customer units before shipping (yes, we do a pre-tune on request). It’s conservative. It won’t give you the highest number — it’ll give you the highest stable number, which is what you actually want if you’re leaving the miner on 24/7.

What you’re tuning

Two dials:

  1. Frequency (MHz) — how fast the ASIC clocks. More MHz = more hash attempts per second. Costs power proportionally.
  2. Voltage (mV) — how much juice the core runs on. Higher voltage lets the chip clock higher stably; lower voltage is more efficient but caps your achievable frequency.

The goal is to find the highest frequency that’s stable at the lowest voltage. Efficiency (J/TH) drops when you push voltage up; hashrate goes up when you push frequency up. The Pareto frontier — more hashrate without burning efficiency — is narrow and you find it by walking the two dials in sequence.

Before you start: baseline your chip

  1. Run the Bitaxe at stock (490 MHz / 1.20 V) for at least 48 hours, 24 hours minimum. Let the chip’s behaviour stabilise.
  2. Note the baseline numbers on your dashboard:
    • Measured hashrate
    • Hashrate error %
    • ASIC temperature
    • VR temperature
    • Efficiency (J/TH)
  3. Confirm cooling is clean: fan at 100% PWM produces 5000+ RPM, no dust in the heatsink fins, ASIC temp under 62°C at stock.
  4. Confirm your PSU is stable: input voltage reads 4.98–5.02 V under load on the dashboard. If it sags, fix this before overclocking — an overclock on a marginal PSU is a fool’s errand.

If the baseline looks healthy, proceed. If it doesn’t — fix the baseline first. Overclocking is about pushing silicon past its rated envelope; doing that on a borderline setup amplifies problems rather than revealing them.

The three-phase method

Phase 1: Push frequency at stock voltage

Goal: find the highest frequency your chip can sustain at 1.20 V.

  • Bump frequency by +10 MHz via AxeOS settings
  • Save, wait 10 minutes
  • Check hashrate error % and measured hashrate
  • Exit condition: hashrate error exceeds 3%, OR measured hashrate stops climbing despite higher expected hashrate

Work in 10 MHz steps: 490 → 500 → 510 → 520 → 530 → 540 → …

A typical Gamma walking this path:

FrequencyExpectedMeasuredError %J/TH
490 MHz1.200 TH/s1.185 TH/s1.3%14.3
510 MHz1.249 TH/s1.238 TH/s0.9%14.0
525 MHz1.286 TH/s1.275 TH/s0.9%13.8
540 MHz1.322 TH/s1.303 TH/s1.4%13.7
550 MHz1.347 TH/s1.308 TH/s2.9%13.9
560 MHz1.371 TH/s1.301 TH/s5.1%14.6

At 560 MHz the chip starts missing cycles (hashrate error > 3%) and measured hashrate actually drops vs 550 MHz. This is the classic overclock wall. Back off to 550 MHz, which gave you 2.9% error (acceptable) and peak measured hashrate.

Record your wall. This Gamma’s frequency ceiling at stock voltage is ~550 MHz. Every Gamma is different — some will cap at 520, some will go to 580+.

Phase 2: Dial voltage up to extend the frequency ceiling

Now try the same push at slightly higher voltage. Voltage lets silicon run reliably at higher frequencies but costs efficiency.

From 550 MHz / 1.20 V:

  • Bump voltage +25 mV → 1.225 V
  • Leave frequency at 550 MHz, wait 10 minutes
  • Check hashrate error — should drop
  • If error is now under 1.5%, push frequency another 10 MHz
  • Repeat: +10 MHz, then if still stable, +25 mV, alternating

Example continuation:

FreqVoltageError %MeasuredJ/TH
550 MHz1.20 V2.9%1.308 TH/s13.9
550 MHz1.225 V0.8%1.336 TH/s14.1
560 MHz1.225 V1.6%1.349 TH/s14.3
570 MHz1.225 V3.2%1.355 TH/s14.6
570 MHz1.25 V1.1%1.383 TH/s14.9

The efficiency cost is visible. At 570 MHz / 1.25 V you’re making 1.38 TH/s but at 14.9 J/TH vs 13.7 J/TH at 540 MHz / 1.20 V. You bought ~6% more hashrate at the cost of ~9% worse efficiency.

Whether that’s worthwhile depends on your power tariff. On solar or off-peak, push voltage without hesitation. On peak retail (AU$0.28+/kWh), stop at the highest frequency you can hold at stock voltage.

Phase 3: Thermal sanity check

Once you’ve found your stable-at-voltage point, leave the miner running for 24 hours under load. Watch:

  • ASIC temperature — must stay under 70°C even during hot-weather spikes
  • VR temperature — must stay under 82°C
  • Hashrate error % — must stay under 3% sustained (brief spikes to 4-5% during thermal events are OK)
  • Input voltage — must stay at 4.95-5.05 V; any sag means the PSU can’t sustain the overclock load

If any metric drifts, back off the overclock 10-25 MHz. Stability over peak performance.

The three stopping rules

It’s tempting to keep pushing. Don’t past these:

Rule 1: Hardware limits

Absolute caps for Bitaxe Gamma (BM1368):

  • Frequency: 600 MHz (physical chip limit; running higher wastes clock cycles on unclockable logic)
  • Core voltage: 1.30 V (above this, chip degradation accelerates noticeably)
  • ASIC temperature: 75°C (throttling begins, chip life shortens)

These are not opinion. These are the silicon physics.

Rule 2: Aussie summer thermal cap

Your overclock must be valid at your worst ambient temperature, not your winter ambient.

A Gamma running at 70°C ASIC in July Melbourne (8°C ambient) will run at 95°C in February Brisbane (36°C ambient) for roughly the same voltage/frequency setting. The chip’s cooling capacity is ambient-relative, not absolute.

Practical rule: set your summer overclock to keep ASIC at 55°C in winter. That gives you ~15°C of thermal headroom for summer, which matches the ~15°C ambient swing between winter and summer most Aussie cities see.

Two overclock profiles, easily toggled:

AxeOS v2.7+ lets you save multiple configuration profiles. Create:

  • Winter profile: aggressive — 570 MHz / 1.25 V, targeting 62°C ASIC at 15°C ambient
  • Summer profile: conservative — 530 MHz / 1.21 V, targeting 55°C ASIC at 15°C ambient (which becomes ~70°C at 35°C ambient)

Flip them April and October if you’re running somewhere that actually has seasons.

Rule 3: Diminishing returns

At some point you’re burning efficiency for a 2% hashrate gain. Do the maths.

Say you push from 540 MHz / 1.20 V (1.30 TH/s at 17.8 W) to 570 MHz / 1.25 V (1.38 TH/s at 20.6 W). That’s +6% hashrate at +16% power. In economics terms:

  • At AU$0.08/kWh (solar), you gain ~AU$0.007/day of sats, cost ~AU$0.005/day of electricity. Net: +AU$0.73/year.
  • At AU$0.30/kWh (peak retail), you cost ~AU$0.020/day of electricity, gain ~AU$0.007/day of sats. Net: -AU$4.75/year.

The aggressive overclock makes money on solar, loses money on peak retail. Know your tariff before chasing MHz.

Special case: underclocking for efficiency

The opposite of this exercise is also useful. If you’re running on marginal power, or if silence matters more than hashrate, you can underclock:

  • 460 MHz / 1.15 V typically delivers ~1.12 TH/s at 14 W (12.5 J/TH)
  • 420 MHz / 1.10 V typically delivers ~1.02 TH/s at 10 W (9.8 J/TH)

Yes, 9.8 J/TH — better efficiency than an Antminer S21 Pro, on a Bitaxe Gamma. The silicon is capable; the frequency tradeoff is in your hands.

Underclock scenarios that make sense:

  • Running on a very small solar array. Maximise efficiency per watt of solar capacity.
  • Bedroom miner. Lower power = lower fan speed = quieter.
  • Older Bitaxe Gamma with cumulative thermal damage. Gentler operation extends life.

Common overclock failures and what they mean

SymptomLikely causeFix
Hashrate error jumps to 10%+ immediately after bumpChip can’t clock that highBack off 20 MHz
Hashrate error climbs over hoursThermal — chip heating upImprove cooling or back off
Measured hashrate drops below pre-overclockOverclocked past the chip’s useful limitBack off
Miner resets spontaneouslyPSU sag under overclock loadBetter PSU, or reduce overclock
VR temperature climbs above 85°CVoltage too high for coolingReduce voltage or improve airflow
Fan hits 100% and noise becomes unacceptableOverclock generating too much heatBack off or improve case airflow

The Aussie-specific notes

Summer cap: as above, plan overclock around your worst ambient, not your best. We see more dead Gammas from summer-overclocked units than from any other cause.

PSU sourcing: a 5V/6A PSU from a reputable brand (Mean Well, the genuine Bitaxe-spec unit we ship) handles 22 W of mining load without sag. Cheap USB-C bricks from Bunnings, Officeworks, or AliExpress do not. If you’re overclocking, validate your PSU first.

Power board loading: four overclocked Bitaxes on one 10A GPO pulling ~100 W combined is fine electrically, but shared with a kettle or heater on the same circuit causes the RCD to trip during appliance inrush. Dedicated circuit is cheap insurance.

Dust: particularly in dry-interior Australia (WA goldfields, Central Qld), expect dust accumulation on the heatsink within weeks. An overclocked Gamma that’s fine at shipping gets thermally throttled three months later because its heatsink is clogged. Quarterly maintenance — compressed air, nothing more.

TL;DR

  1. Phase 1: push frequency at stock voltage until hashrate error exceeds 3% or hashrate stops climbing.
  2. Phase 2: if you want more, bump voltage +25 mV and push frequency further. Watch efficiency — you’re burning sats for hashrate.
  3. Phase 3: thermal-validate over 24 hours. Back off if ASIC exceeds 70°C or VR exceeds 82°C.
  4. Hard caps: 600 MHz, 1.30 V, 75°C ASIC, 82°C VR. Don’t cross them.
  5. Summer profile matters in AU. Build one for hot months and flip it in October.
  6. Underclocking is a legitimate strategy — 9.8 J/TH on a Gamma at 420 MHz / 1.10 V is better than any industrial miner shipping.
  7. Economics decide how aggressively to push: solar and off-peak reward overclocks, peak retail punishes them.

Every Bitaxe Gamma is a slightly different chip. Use the framework; trust your measurements over someone else’s screenshot.


Methodology refined across 60+ Aussie customer units pre-tuned on our workbench before shipping. If you want us to do it for you, DM us — pre-tune is a free service for authenticated Bitaxes bought through the shop.

Frequently Asked Questions

What's a realistic overclock on a Bitaxe Gamma?

On good silicon, 540–560 MHz at 1.25 V is stable and nets ~1.32–1.38 TH/s at 20–22 W. On marginal silicon, you'll cap at 510–525 MHz before hashrate error climbs. Every chip is different — don't trust someone else's numbers, tune yours in steps.

Will overclocking kill my chip?

Sustained over-voltage (>1.30 V without upgraded cooling) and over-temperature (>75°C ASIC) will degrade a chip over months to years. A conservative overclock (frequency up, voltage held low) is safe indefinitely. An aggressive overclock (frequency AND voltage pushed) shortens silicon life.

Should I overclock in summer?

In Australian summer — particularly Queensland, WA, and inland NSW/Vic — ambient can hit 38°C. Your overclock needs to be capped so the ASIC stays under 70°C at peak ambient. Many Aussies run a winter overclock and a summer profile. AxeOS v2.7+ makes this two-button easy.

What's the single most important thing to watch while overclocking?

Hashrate error percentage. It's the hardware's way of telling you the chip can't actually hit the clock you requested. Keep it under 3% and you're fine. Over 5% means you've pushed too far — back off.

Published 21 April 2026
By Shop Bitcoin Australia