NVIDIA GeForce GTX 1060 6 GB

The NVIDIA GeForce GTX 1060 6 GB is a GeForce 10-series GPU from NVIDIA, built around the GP106 die on TSMC's 16 nm process. It integrates 4,400 million transistors on a 200 mm² die, resulting in 22.0 million transistors per mm². It launched on 2016-07-18 at a launch MSRP of 299 USD and is now marked end-of-life. The database record places it at the 50th percentile of all GPUs, but the benchmarks array is empty and the average benchmark score is 0.

Who Should Consider It

The strongest audience signal in the fact pack is structural, not measured: 6 GB GDDR5, 192.2 GB/s bandwidth, 4.375 TFLOPS FP32, 82.03 GPixel/s pixel rate, and 136.7 GTexel/s texture rate. Because the benchmark list contains no scores, the database cannot validate specific resolution or settings combinations. The card sits at percentileVsAllGpus 50, meaning it is the midpoint of the database's GPU population. That positional data supports neither a high-end nor a low-end classification; it is a middle-tier part by percentile.

The physical and electrical requirements are concrete. The card is dual-slot, 250 mm long, and 111 mm tall. It is rated at 120 W TDP, uses one 6-pin PCIe power connector, and the suggested PSU is 300 W. A user with a 300 W-capable supply and clearance for a 250 mm card can consider it without exceeding the stated power envelope. The display output set, 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a, allows multiple monitors to be attached directly. It connects through a PCIe 3.0 x16 bus interface.

However, anyone seeking settings guidance for a particular game will not find it in this record, because no benchmark scores are present. The data supports a general suitability based on 6 GB capacity, moderate power draw, and display connectivity, but it does not support a claim that the card can drive a given resolution at a given quality level. In short, the card is a 50th-percentile GPU whose suitability comes from its memory capacity, power requirements, and interface flexibility rather than from measured performance data.

Memory Subsystem

The memory subsystem consists of 6 GB of GDDR5 on a 192-bit bus. The memory clock is 2002 MHz, which the fact pack also expresses as 8 Gbps effective, and the resulting bandwidth is 192.2 GB/s. Those three numbers, capacity, bus width, and bandwidth, define how much data can be moved during rendering. At high resolutions, a larger frame buffer and higher bandwidth are generally advantageous. The card provides 6 GB and 192.2 GB/s without a stated resolution ceiling in the fact pack.

The 192-bit interface is the data path width, and the 8 Gbps effective transfer rate is the per-pin speed. Together they produce the 192.2 GB/s figure. The 48 ROPs consume pixel fill from this memory pipe; pixel rate is 82.03 GPixel/s. The fact pack does not include a memory temperature or cooler specification, so thermal behavior of the VRAM is not assessed here. What the data shows is a complete memory picture: capacity, type, bus, clock, and bandwidth are all specified. The capacity is 6 GB, the type is GDDR5, the bus width is 192 bit, and the bandwidth limit is exactly 192.2 GB/s. No further memory-related constraints are listed.

Power and Cooling

The card is specified at a 120 W TDP, with a suggested PSU of 300 W. Power is delivered through one 6-pin PCIe connector. The card occupies two slots. Its length is 250 mm, or 9.8 inches, and its height is 111 mm, or 4.4 inches, so case and PSU planning require at least this physical envelope. The fact pack does not name a cooler model. The cooling description is limited to dual-slot occupancy.

The 300 W PSU suggestion and single 6-pin requirement are the only power-delivery facts in the record. There is no wattage comparison to other graphics cards in the fact pack, so no statement about relative power draw can be made beyond the 120 W figure. The 120 W TDP is the card's own power target, while the 300 W suggested PSU is the system-level recommendation. A single 6-pin PCIe power connector is the only external power connection listed. The 250 mm length and 111 mm height are the dimensions to check before installation, and the dual-slot designation means adjacent slot clearance may also matter. The PCIe 3.0 x16 bus interface is the host connection.

How It Compares

The nearestRivals list is empty. There are no rival names, no rival scores, and no deltaPct values in the fact pack. Consequently, a rival-by-rival comparison section cannot be constructed from the data. The only quantitative position in the database is percentileVsAllGpus: 50. That places the card at the midpoint of all GPUs tracked in this database.

In the GeForce lineup, this model belongs to the 10-series generation, and the fact pack lists GeForce 900 as its predecessor and GeForce 20 as its successor. The architecture is Pascal, and the chip is GP106. These product-line facts provide context, but they are not performance comparisons. The card's generation placement is clear: it replaced the GeForce 900 line and was itself replaced by GeForce 20. No exact percentage advantage or disadvantage over any specific GPU can be stated, because none is present in the source data. The only comparative statement supported by the fact pack is the 50th percentile position across all GPUs.

Ray Tracing and Feature Set

The architecture is Pascal, implemented by the GP106 chip. The fact pack fields for RT cores and tensor cores are null; no counts are listed. Therefore, the data contains no hardware ray tracing acceleration and no tensor core hardware. On the API side, the card supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. This API set is available despite the lack of RT and tensor core data.

The card also has 1280 shading units, 80 TMUs, and 48 ROPs. The FP32 throughput is 4.375 TFLOPS, while FP16 throughput is 68.36 GFLOPS at a 1:64 ratio. That ratio shows FP16 execution is a small fraction of FP32. The display output stage includes 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. The bus interface is PCIe 3.0 x16. The fact pack lists no additional feature names beyond these APIs, core counts, and display outputs, so any feature analysis must remain within those listed items.

FAQ

Q: What is the die and architecture?

A: The GPU is the GP106 chip on the Pascal architecture, manufactured by TSMC on a 16 nm process with 4,400 million transistors on a 200 mm² die.

Q: What are the memory specifications?

A: 6 GB GDDR5 with a 192-bit bus, 2002 MHz memory clock (8 Gbps effective), and 192.2 GB/s bandwidth.

Q: What are the core counts?

A: 1280 shading units, 80 TMUs, and 48 ROPs.

Q: What power supply and connectors are required?

A: The TDP is 120 W, the suggested PSU is 300 W, and the card uses one 6-pin PCIe power connector.

Q: Does the card support modern APIs?

A: It supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The fact pack lists no RT cores or tensor cores.

Q: What is the production status and product generation?

A: It is end-of-life. It was released on 2016-07-18, is part of the GeForce 10-series, succeeded GeForce 900, and was succeeded by GeForce 20.

Benchmark Performance

The benchmarks array is empty and the average benchmark score is 0. This means the database has no measured score for the card, so there are no exact deltas to compute against rivals. The nearestRivals array is also empty; no rival names, scores, or deltaPct values are available. The one benchmark-adjacent number is percentileVsAllGpus: 50, which puts this card exactly at the midpoint of the all-GPU distribution.

Clock behavior is specified as a 1506 MHz base and 1709 MHz boost. Derived throughput rates are 82.03 GPixel/s pixel fill, 136.7 GTexel/s texture fill, 4.375 TFLOPS FP32, and 68.36 GFLOPS FP16 at a 1:64 ratio. These rates describe the hardware's specified limits, but they are not benchmark results. The memory clock is 2002 MHz with 8 Gbps effective data transfer, feeding 192.2 GB/s of bandwidth. Without benchmark scores, the 50th percentile is the only aggregate performance signal in the record. Any percentage comparison to a named rival would require data that is not present in the fact pack, so no such comparison can be provided.