Quick Summary (AI Overview):
While both GPUs belong to NVIDIA’s Turing architecture generation (PG150 vs PG180 reference boards), the RTX 2080 Ti PCB and RTX 2080 PCB are fundamentally different at the hardware layer. The RTX 2080 Ti uses the massive TU102 GPU die with a 352-bit memory bus (11 GDDR6 chips) and a heavy-duty 13+3 or 16+3 phase VRM power delivery system fed by dual 8-pin connectors. In contrast, the RTX 2080 uses the smaller TU104 die with a 256-bit bus (8 GDDR6 chips) and an 8+2 or 10+2 phase VRM fed by an 8-pin + 6-pin combo. Due to different mounting hole patterns, die sizes, and VRAM layouts, water blocks, custom cooling plates, and BIOS binaries are not interchangeable between the two cards.
When NVIDIA launched the Turing microarchitecture, hardware enthusiasts, modders, and custom water-cooling builders were eager to see how the flagship cards were built on the inside. At first glance, the Founder’s Edition shrouds for both the RTX 2080 and the RTX 2080 Ti look nearly identical. However, stripping off the heatsinks reveals two entirely different PCB (Printed Circuit Board) designs under the hood.
Whether you are repairing a damaged card, planning a custom loop water block installation, shunt-modding for overclocking records, or simply curious about GPU engineering, understanding the component-level differences between the RTX 2080 Ti PCB (NVIDIA PG150 reference design) and the RTX 2080 PCB (NVIDIA PG180 reference design) is crucial. In this deep dive, we compare GPU dies, power delivery systems, VRAM configurations, PCB layer stackups, and water block compatibility.
Visual Architecture: Structural PCB Layout Differences
To grasp why these two GPUs require different water blocks and power delivery designs, examine the layout comparison diagram below:
RTX 2080 Ti vs RTX 2080 Reference PCB Topology
RTX 2080 Ti PCB (PG150)
- GPU Silicon: TU102-300A / TU102-300 (754 mm²)
- Memory Bus Layout: 352-bit (11 GDDR6 BGA Pads)
- Core VRM: 13 to 16 Power Phases (uP9512P Controller)
- Memory VRM: 3 Power Phases
- PCIe Power Inputs: 8-pin + 8-pin (300W+ Board Limit)
- Physical Dimensions: Wider PCB, 70.5 x 70.5 mm Cooler Mounting
RTX 2080 PCB (PG180)
- GPU Silicon: TU104-400A / TU104-400 (545 mm²)
- Memory Bus Layout: 256-bit (8 GDDR6 BGA Pads)
- Core VRM: 8 to 10 Power Phases (uP9512Q Controller)
- Memory VRM: 2 Power Phases
- PCIe Power Inputs: 8-pin + 6-pin (225W Board Limit)
- Physical Dimensions: Standard PCB, 58.4 x 58.4 mm Cooler Mounting
Figure 1: High-level architectural and component breakdown comparing PG150 (2080 Ti) and PG180 (2080) board layouts.
Technical Specification Matrix: PG150 vs. PG180
The table below presents a direct, side-by-side component comparison between the reference Founder’s Edition PCBs of both graphics cards:
| Hardware Feature | NVIDIA RTX 2080 Ti PCB (PG150) | NVIDIA RTX 2080 PCB (PG180) |
|---|---|---|
| GPU Core Code Name | TU102-300 / TU102-300A | TU104-400 / TU104-400A |
| Die Size / Transistor Count | 754 mm² / 18.6 Billion Transistors | 545 mm² / 13.6 Billion Transistors |
| Memory Interface & Bus Width | 352-bit (11x 1GB GDDR6 Modules) | 256-bit (8x 1GB GDDR6 Modules) |
| GPU Core Power Phases | 13 Dynamic Phases (Expandable to 16 on AIBs) | 8 Dynamic Phases (Expandable to 10 on AIBs) |
| Memory Power Phases | 3 Dedicated Phases | 2 Dedicated Phases |
| Smart Power Stages (DrMOS) | FDMF3170 (Fairchild/ON Semi 70A rated) | FDMF3170 / NCP302155 (55A/70A rated) |
| PWM Controller IC | uPI Semiconductor uP9512P (8-channel) | uPI Semiconductor uP9512Q (8-channel) |
| Power Input Connectors | Dual 8-pin PCIe (300W Reference TDP) | 8-pin + 6-pin PCIe (225W Reference TDP) |
| NVLink Edge Connector | Dual-Channel x8 NVLink Finger | Single-Channel x8 NVLink Finger |
| Cooler Mounting Hole Pitch | 70.5 mm x 70.5 mm Square | 58.4 mm x 58.4 mm Square |
Deep Component Analysis: 5 Key PCB Differences
1. Silicon Package Footprint & Die Substrate
The core difference between these two PCBs stems from the underlying GPU processor. The RTX 2080 Ti utilizes the TU102 silicon, a giant 754 mm² processor package containing 18.6 billion transistors. In comparison, the RTX 2080 uses the smaller TU104 die, measuring 545 mm² with 13.6 billion transistors.
Because the physical silicon and substrate sizes differ drastically, the surrounding PCB trace layouts are fundamentally non-interchangeable:
- Mounting Hole Spacing: The 2080 Ti uses a wider 70.5 x 70.5 mm heatsink mounting hole pattern, whereas the 2080 uses the traditional 58.4 x 58.4 mm spacing found on older GTX 1080 cards.
- Keep-Out Zones: Surface-mount capacitors and inductors around the GPU die are spaced further apart on the 2080 Ti to accommodate the wider TU102 die package.
2. Memory Bus Routing & GDDR6 Chip Placement
NVIDIA’s Turing architecture paired each 32-bit memory controller with a single GDDR6 VRAM chip. The 2080 Ti features a 352-bit memory bus, which requires 11 physical GDDR6 memory pads positioned around the die, leaving an empty space for a 12th unpopulated pad (reserved for full 384-bit Titan RTX variants).
By contrast, the RTX 2080 operates on a 256-bit memory bus, requiring only 8 GDDR6 memory modules. The traces running from the TU104 die to the VRAM chips on the 2080 PCB are noticeably shorter and arranged differently, drastically altering signal timing and trace impedance constraints on the board.
3. Voltage Regulator Module (VRM) & Power Delivery
High power consumption under load requires heavy-duty onboard power regulation. The power delivery setups between the two cards show significant differences in phase count and component rating:
- RTX 2080 Ti (13+3 Phase VRM): The reference PG150 board employs 13 phases dedicated to supplying the VCORE (GPU core) voltage and 3 phases dedicated to the GDDR6 VRAM rail. NVIDIA used high-efficiency Fairchild/ON Semiconductor FDMF3170 Smart Power Stages (DrMOS) rated for 70A per phase, managed dynamically by a uPI uP9512P digital PWM controller.
- RTX 2080 (8+2 Phase VRM): The reference PG180 board utilizes an 8-phase VCORE VRM and a 2-phase memory VRM controlled by a uPI uP9512Q IC. While custom AIB models (like the ASUS ROG Strix or EVGA FTW3) added more phases, reference boards strictly adhered to this lighter configuration.
4. Shunt Resistors and Power Balancing Topology
NVIDIA uses low-resistance SMD shunt resistors (typically marked R005 for 5 mΩ) to monitor current draw across the PCIe slot and external power connectors. The RTX 2080 Ti PCB features a complex current-balancing system with three shunt resistors monitoring the dual 8-pin inputs and the PCIe slot line.
If current draw across any individual rail exceeds BIOS power limits, the uP9512 controller throttles clock speeds. Extreme overclockers frequently perform “shunt mods” by soldering lower resistance (e.g., 2 mΩ) resistors in parallel across these pads. Due to differing trace configurations and rail assignments on the 2080 Ti vs. 2080 PCB, shunt resistor locations and power balancing loops are completely distinct.
5. NVLink Edge Connectors
Both GPUs feature NVIDIA’s high-speed NVLink interconnect for multi-GPU configurations, but their physical PCB traces differ:
- The 2080 Ti PCB has a dual-channel x8 NVLink interface capable of up to 100 GB/s bidirectional bandwidth.
- The 2080 PCB features a scaled-down single-channel x8 NVLink interface delivering up to 50 GB/s bidirectional bandwidth. The physical gold finger width on the top left edge of the PCB reflects this trace difference.
Cross-Compatibility & Water Block Guidelines
If you are planning a custom water-cooling loop or buying aftermarket GPU coolers, keep these strict compatibility rules in mind:
CRITICAL COMPATIBILITY WARNING:
Never attempt to install an RTX 2080 water block on an RTX 2080 Ti, or vice versa!
The cooler mounting hole dimensions (70.5mm vs 58.4mm), VRM component heights, and GDDR6 chip positions are physically incompatible. Forcing an incorrect block onto either PCB will cause uneven contact, severe thermal throttling, physical board strain, or immediate catastrophic VRM short circuits.
AIB Custom PCBs vs. NVIDIA Reference Designs
When selecting aftermarket components, note that custom board designs differ from standard reference layouts:
- Reference Layouts (PG150 / PG180): Used on Founder’s Edition models and baseline custom cards (e.g., EVGA XC, MSI Ventus, Zotac Twin Fan). Water blocks built for “Reference 2080 Ti” or “Reference 2080” fit these specific boards.
- Custom AIB Layouts: Higher-tier cards (e.g., MSI Lightning Z, ASUS ROG Strix, EVGA FTW3) use custom, wider PCBs with extra VRM phases and dual/triple 8-pin connectors. These require custom water blocks designed specifically for that board model.
Frequently Asked Questions (FAQs)
Q1: Can I flash an RTX 2080 Ti BIOS onto an RTX 2080 GPU?
Answer: Absolutely not. Flashing a 2080 Ti BIOS onto a standard 2080 will brick the card instantly. The two boards use different GPU silicon (TU102 vs TU104), memory controllers (352-bit vs 256-bit), and VRM controller configurations that will fail POST.
Q2: Are 2080 Ti and 2080 power connectors physically interchangeable?
Answer: No. The RTX 2080 Ti reference PCB requires dual 8-pin (8+8) PCIe power inputs to supply its higher power limit (300W+). The RTX 2080 reference PCB uses an 8-pin + 6-pin (8+6) power configuration designed for a 225W power envelope.
Q3: Why does the 2080 Ti PCB have an unpopulated VRAM pad?
Answer: The PG150 reference PCB is shared with the workstation-grade Titan RTX. The Titan RTX utilizes the full 384-bit bus with 12 GDDR6 chips (24GB total). The RTX 2080 Ti uses a cut-down 352-bit bus, leaving one GDDR6 pad empty.
Q4: Are the air coolers from a 2080 Ti FE and 2080 FE interchangeable?
Answer: While the outer metal shrouds look identical, the underlying heatsink cold plates are not interchangeable. The GPU cold plate screw patterns (70.5mm vs 58.4mm) and VRM contact pad heights differ between the two models.
Q5: What is the board power limit difference between the two PCBs?
Answer: Reference RTX 2080 PCBs have a stock power target of 215W to 225W (scalable up to ~250W via software). Reference RTX 2080 Ti PCBs feature a baseline target of 250W to 260W (scalable up to 320W–380W depending on VBIOS power limits).
Q6: Can you solder extra VRAM onto an RTX 2080 PCB to make it a 2080 Ti?
Answer: No. Memory capacity is physically tied to the GPU memory controllers embedded inside the silicon die. The TU104 processor on the 2080 lacks the physical silicon traces and memory controllers needed to support 11GB of GDDR6 memory on a 352-bit bus.
Q7: Which PCB is easier to repair and trace for hardware faults?
Answer: The RTX 2080 PCB is generally simpler to diagnose because it has fewer power phases, a smaller trace density around the 256-bit bus, and fewer power rail balance loops than the denser RTX 2080 Ti PCB.
Q8: Do both PCBs support the USB-C VirtualLink port?
Answer: Yes. Both the reference PG150 and PG180 PCBs feature a dedicated Cypress CCG5 USB-C PD Controller chip and onboard power rails to drive the VirtualLink port on Founder’s Edition cards.
Q9: What happens if you use a 2080 water block on a 2080 Ti?
Answer: The block will fail to line up with the mounting holes. Forcing the block can crack the TU102 glass die edge, crush surface-mount capacitors, or leave VRM power stages uncooled, leading to immediate thermal failure.
Q10: What is the main PCB difference between ‘A’ and ‘Non-A’ Turing dies?
Answer: Early Turing GPUs were divided into factory-overclocked dies (e.g., TU102-300A / TU104-400A) and non-overclocked reference dies (TU102-300 / TU104-400). While the physical PCB trace layout was identical, ‘Non-A’ dies were restricted by NVIDIA from using higher power limit factory VBIOS files.
Final Thoughts
Although the RTX 2080 Ti and RTX 2080 share the same Turing architecture generation, their underlying board designs serve two distinct performance tiers. The RTX 2080 Ti (PG150 PCB) is a heavy-duty flagship platform featuring a 352-bit bus layout, 13+3 phase VRM, and dual 8-pin power inputs designed to drive the massive TU102 die.
The RTX 2080 (PG180 PCB) provides a streamlined design built around the smaller TU104 die, utilizing a 256-bit bus layout, 8+2 phase VRM, and 8-pin + 6-pin power setup. Understanding these component-level differences ensures smooth hardware repairs, successful custom water-block installations, and safe GPU modding.
