Skip to main content

Single-Slot RTX 3060 Mod Runs on PCIe Slot Power Alone

·1842 words·9 mins
NVIDIA GeForce RTX 3060 GPU Modding Shunt Mod PCIe Graphics Cards SFF PCs GPU Cooling Hardware
Table of Contents

Single-Slot RTX 3060 Mod Runs on PCIe Slot Power Alone

A heavily modified GeForce RTX 3060 has attracted attention for doing something the standard desktop card was never designed to do: operate without an external PCIe power connector.

Featured in a recently surfaced Bilibili video, the custom card combines a single-slot, low-profile PCB design with a compact blower-style cooler and draws its power entirely from the motherboard’s PCIe slot.

The result is an RTX 3060 that is barely larger than a smartphone.

But the extreme reduction in size and power budget comes with a major trade-off.

In 3DMark Time Spy, the modified card reportedly achieved a graphics score of only 4,821 points, roughly half the performance expected from a conventional RTX 3060.

The project is therefore less about creating a practical high-performance RTX 3060 and more about demonstrating how far the GPU can be pushed toward an ultra-low-power, ultra-compact form factor.

🧩 An RTX 3060 Compressed Into a Single Slot
#

The custom card uses a deep-blue PCB and an unusually compact physical layout.

Its defining characteristics include:

  • Single-slot form factor
  • Low-profile PCB
  • No external PCIe power connector
  • PCIe-slot-only power delivery
  • Compact blower-style cooling
  • One HDMI output
  • One DisplayPort output

Physically, the card is approximately smartphone-sized, making it dramatically smaller than most conventional RTX 3060 implementations.

A typical RTX 3060 is designed around a substantially larger cooling solution and additional board-level power delivery.

This modified design removes both luxuries.

Standard RTX 3060
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Large heatsink + fan          β”‚
β”‚                               β”‚
β”‚ GPU + VRAM + VRM              β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
       β”‚
       └── External PCIe Power

Modified RTX 3060
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Compact GPU   β”‚
β”‚ + VRAM + VRM  β”‚
β”‚ + Blower      β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
       β”‚
       └── PCIe Slot Only

The second configuration is considerably more restrictive from both electrical and thermal perspectives.

⚑ The 75W PCIe Slot Power Problem
#

The biggest engineering challenge is power.

A conventional desktop RTX 3060 has a default 170W Total Board Power (TBP), while the official configurable lower limit is substantially below that but still above what a motherboard PCIe slot is designed to provide by itself.

A standard PCIe x16 slot is generally specified to provide up to 75W of power to an add-in card.

That creates an immediate mismatch:

Typical RTX 3060
β‰ˆ 170W board power

PCIe slot alone
β‰ˆ 75W available power

The modified card therefore cannot simply operate like a normal RTX 3060 while drawing all of its power from the slot.

Something in the board’s power-management system has to enforce a radically lower operating point.

πŸ”§ Shunt Modding May Be Behind the Power Reduction
#

The exact modification used by the builder is not publicly confirmed in the available description, but the video reportedly suggests changes around the card’s power-sensing circuitry.

One possibility is a technique commonly known as a shunt mod.

Shunt resistors are used by GPU power-monitoring circuits to measure current.

By changing the resistance or modifying the feedback relationship seen by the power controller, a board can potentially alter how the GPU interprets its own power consumption.

Conceptually:

Normal Power Monitoring

Actual Current
      β”‚
      β–Ό
Shunt Resistor
      β”‚
      β–Ό
Power Controller
      β”‚
      β–Ό
GPU Power Management

A modified configuration can alter the feedback relationship:

Modified Feedback

Actual Current
      β”‚
      β–Ό
Altered Sense Circuit
      β”‚
      β–Ό
Modified Controller Feedback
      β”‚
      β–Ό
Lower Effective Power Target

However, this should not be interpreted as physically creating additional power capacity.

A shunt modification does not turn a 75W PCIe slot into a 170W power source.

Instead, it can change how the GPU’s power-management system interprets or limits its operating conditions.

That distinction is critical.

πŸ“‰ Why GPU-Z Power Readings May Be Misleading
#

The reported experiment also highlights an important consequence of modifying power-sensing circuitry.

Software utilities such as GPU-Z depend on hardware telemetry.

If the underlying current-sensing circuitry has been altered, the reported values may no longer correspond directly to actual electrical consumption.

The modified card reportedly showed readings around:

  • 145W board power
  • 176W GPU chip power

Those figures appear incompatible with a card supposedly operating exclusively from a 75W PCIe slot.

That does not necessarily mean the motherboard was delivering those power levels through the slot.

Instead, the readings may reflect the modified board’s altered power-monitoring calibration.

The basic principle is:

Normal Hardware
Actual Power
     β”‚
     β–Ό
Sensor
     β”‚
     β–Ό
Software Reading

After modifying the sensor:

Actual Power
     β”‚
     β–Ό
Modified Sensor
     β”‚
     β–Ό
Incorrect / Non-Standard Reading

Consequently, software-reported power values from a heavily modified GPU should be interpreted cautiously.

Actual electrical measurements would require appropriate hardware instrumentation.

🚦 Clock Speeds Collapse Under the Power Limit
#

The most visible consequence of the extreme power restriction is reduced GPU frequency.

The RTX 3060 normally has a boost clock around 1,777 MHz, depending on the specific implementation.

The modified card reportedly struggled to sustain frequencies above approximately 900 MHz.

That represents a dramatic reduction in operating frequency.

The relationship can be visualized as:

Standard RTX 3060
Boost target
β‰ˆ 1777 MHz
       β”‚
       β–Ό
Higher power budget
       β”‚
       β–Ό
Normal GPU performance


Modified RTX 3060
β‰ˆ 900 MHz sustained
       β”‚
       β–Ό
Severe power constraint
       β”‚
       β–Ό
Heavy performance reduction

The GPU’s architecture remains the same, but it cannot operate near its normal performance envelope when its available power is heavily restricted.

πŸ“Š Time Spy Performance Falls to 4,821
#

The benchmark results illustrate the cost of the modification.

The custom RTX 3060 reportedly scored:

4,821 points in 3DMark Time Spy Graphics.

A conventional RTX 3060 commonly scores above 9,000 points in the same benchmark under typical configurations.

That means the modified card delivers approximately half the performance of a standard RTX 3060.

The reported result also places it below several older mainstream GPUs in raw Time Spy graphics performance.

Standard RTX 3060
β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ  β‰ˆ 9,000+

Modified RTX 3060
β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ            4,821

The benchmark demonstrates that the limiting factor is not necessarily the GPU architecture itself.

Instead, the card is being constrained by the electrical and thermal envelope imposed by its miniature design.

🌑️ Cooling Becomes the Second Major Bottleneck
#

Power reduction solves one problem but does not eliminate thermal limitations.

The card reportedly uses a very small single-slot blower cooler.

Such a cooler has considerably less heatsink volume and airflow capacity than the dual-slot or triple-slot solutions commonly found on desktop RTX 3060 cards.

During testing, the GPU hotspot reportedly exceeded 90Β°C, triggering additional thermal throttling.

The performance ceiling is therefore determined by two interacting constraints:

Limited PCIe Power
       β”‚
       β–Ό
Lower GPU Frequency
       β”‚
       β–Ό
Reduced Performance
       
        +

Limited Cooler
       β”‚
       β–Ό
High GPU Hotspot
       β”‚
       β–Ό
Thermal Throttling

The result is a GPU constrained from both directions.

Even if the power-management problem were further optimized, the tiny cooler would still impose a significant limit on sustained performance.

🏎️ Why the Card Cannot Simply Run at 1,777 MHz
#

GPU boost clocks are not guaranteed operating frequencies.

Modern GPUs dynamically adjust clocks according to several constraints, including:

  • Power consumption
  • Temperature
  • Voltage
  • Workload
  • Current limits
  • Reliability constraints

A card operating under a severe power ceiling cannot simply command the GPU to remain at its advertised boost frequency.

If maintaining a higher frequency would exceed the available electrical budget, the GPU reduces voltage and clock speed.

This produces the characteristic behavior observed in the modified card:

Higher Clock
    β”‚
    β–Ό
Higher Voltage / Power Demand
    β”‚
    β–Ό
Power Limit Reached
    β”‚
    β–Ό
Clock Reduction
    β”‚
    β–Ό
Lower Performance

This is why reducing the physical size of the card has such a dramatic impact on benchmark results.

πŸ–₯️ A Niche Solution for Extreme SFF Systems
#

Despite its poor benchmark performance, the concept has a potential niche.

Some small-form-factor systems have extremely limited internal space and may lack conventional external PCIe power connectors.

In those environments, physical dimensions can matter more than absolute GPU performance.

Potential applications include:

  • Ultra-compact SFF PCs
  • Specialized embedded systems
  • Prebuilt systems with limited expansion power
  • Industrial compact workstations
  • Custom small-form-factor builds
  • Experimental GPU projects

A conventional low-profile RTX-class GPU with a larger power budget would generally be more practical when space and power constraints allow it.

The modified RTX 3060 instead represents an extreme point on the design spectrum:

Maximum Performance
        β”‚
        β–Ό
Larger GPU
More Power
Larger Cooler
        β”‚
        β”‚
        β–Ό
────────────────────────
        β”‚
        β”‚
        β–Ό
Smaller GPU
Lower Power
Smaller Cooler
        β”‚
        β–Ό
Maximum Compactness

The closer the design moves toward minimum physical size, the more performance it sacrifices.

⚠️ Why This Is More Engineering Experiment Than Product
#

The biggest limitation is the balance between power, cooling, and performance.

A 170W-class GPU was not designed to deliver its normal performance envelope through a 75W slot alone.

The custom modification therefore changes the fundamental operating conditions of the RTX 3060.

That makes the project interesting from an engineering perspective, but difficult to justify as a mainstream product.

There are also important risks associated with modifying GPU power-sensing and power-delivery circuitry.

Changing feedback components can affect:

  • Current monitoring
  • Voltage regulation
  • Protection mechanisms
  • GPU stability
  • Thermal behavior
  • Component reliability

Incorrect modifications could potentially damage the graphics card, motherboard, or power-delivery components.

Therefore, the reported configuration should be viewed as an experimental hardware modification rather than a recommended way to build an RTX 3060 system.

🧠 The Interesting Part Is the Power-Delivery Engineering
#

The most compelling aspect of this project is not the 4,821-point benchmark score.

It is the demonstration of how much engineering work is required to fit a relatively powerful desktop GPU into an extremely restrictive electrical and mechanical envelope.

Three constraints dominate the design:

             RTX 3060 GPU
                  β”‚
        β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”
        β–Ό         β–Ό         β–Ό
     Power     Thermal    Physical
      75W       >90Β°C     Single Slot
        β”‚         β”‚         β”‚
        β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                  β–Ό
          ~900 MHz Operation
                  β”‚
                  β–Ό
          4,821 Time Spy

Removing the external power connector does not make the GPU more efficient.

It simply forces the entire system to operate within a much smaller power envelope.

That creates cascading consequences for clock speed, heat dissipation, and benchmark performance.

🏁 Conclusion
#

The custom single-slot RTX 3060 featured in the Bilibili video demonstrates just how far a desktop GPU can be pushed toward ultra-compact, slot-powered operation.

By eliminating the external PCIe power connector and relying entirely on motherboard slot power, the modified card achieves a remarkable physical footprint.

But the engineering compromises are substantial.

The reported ~900 MHz operating range, hotspot temperatures above 90Β°C, and 4,821 Time Spy Graphics score show that the card is operating far outside the performance envelope of a conventional RTX 3060.

The suspected shunt-style modification is particularly interesting because it highlights an important principle of GPU power management: changing the power-sensing feedback can alter how the card regulates itself, but it cannot eliminate the physical limits imposed by the motherboard, VRM, and cooling system.

Ultimately, this RTX 3060 is best understood as a hardware engineering experiment rather than a practical replacement for a standard graphics card.

It demonstrates the trade-off at the heart of ultra-compact GPU design:

Smaller size and lower power can be achievedβ€”but when a 170W-class GPU is forced into a 75W slot-powered envelope, performance inevitably becomes the price.

Related

NVIDIA RTX 50 Hot Spot Temperatures Hidden: Tests Reveal Up to 39Β°C Higher Junction Temps
·990 words·5 mins
NVIDIA GeForce RTX 50 GPU Cooling Graphics Cards Hardware Thermal Management PC Hardware GPU Diagnostics
RTX 5070 Ti SUPER Leak: 24GB GDDR7, 350W TBP, and AI-Ready VRAM
·1134 words·6 mins
NVIDIA GeForce RTX 5070 Ti SUPER Blackwell GDDR7 Graphics Cards Gaming AI GPU Hardware
NVIDIA Vera CPU Specifications Revealed: 88 Cores, 176 Threads, and 1.5TB LPDDR5X
·789 words·4 mins
NVIDIA Vera CPU Server Processors AI Infrastructure High-Performance Computing Data Center LPDDR5X Hardware