NVIDIA GeForce GTX 1060 3 GB vs NVIDIA GeForce RTX 5090 D V2 Comparison

Head-to-Head Benchmarks

The benchmark database contains no direct head-to-head tests between the GTX 1060 3 GB and the RTX 5090 D V2. Instead, the comparison must rely on aggregate scores and percentile placements. The GTX 1060 3 GB records an average benchmark score of 19,334 across three tests: Geekbench Metal at 13,527, Geekbench OpenCL at 27,001, and Geekbench Vulkan at 17,475. The RTX 5090 D V2 has a single recorded 3DMark Steel Nomad DX12 score of 16,504, which serves as its average.

The difference in test methodologies complicates direct comparison. The GTX 1060 3 GB score derives from compute-oriented Geekbench workloads, while the RTX 5090 D V2 score comes from a modern DX12 gaming benchmark. Still, the percentile data offers context. The GTX 1060 3 GB sits at the 64th percentile among all GPUs in the database, while the RTX 5090 D V2 sits at the 59th percentile. A higher percentile for the older card does not imply superior performance; it reflects the distribution of scores across different test suites and the fact that the RTX 5090 D V2 has only one recorded benchmark.

Looking at nearest rivals for each card reveals interesting patterns. The GTX 1060 3 GB sits within 1.3% of the AMD Radeon RX 7900 XTX (delta -0.4%), NVIDIA TITAN Xp (delta 0.8%), GeForce GTX 780 (delta 0.9%), and Tesla K20m (delta 1.3%). This clustering suggests the Geekbench suite produces compressed scores across very different hardware generations. The RTX 5090 D V2, by contrast, sits within 0.9% of the NVIDIA T400 (delta 0%), AMD Radeon PRO W7500 (delta 0.5%), RTX PRO 6000 Blackwell (delta 0.6%), and AMD Radeon RX 5700 XT (delta 0.9%) on the Steel Nomad test.

The raw specification gap is enormous. The GTX 1060 3 GB delivers 3.935 TFLOPS FP32, while the RTX 5090 D V2 delivers 104.8 TFLOPS, approximately 26.6 times higher. Texture rate jumps from 123.0 GTexel/s to 1,636.8 GTexel/s, over 13 times higher. Pixel rate increases from 81.98 GPixel/s to 423.6 GPixel/s, roughly 5.2 times higher. Memory bandwidth scales from 192.2 GB/s to 1.34 TB/s, a factor of about 7. These are not incremental improvements; they represent multiple generational leaps.

Where Each One Wins

The GTX 1060 3 GB wins in the Geekbench suite by raw score. Its OpenCL result of 27,001 is the highest single benchmark score recorded for either card. Its Vulkan score of 17,475 also exceeds the RTX 5090 D V2's Steel Nomad result of 16,504. However, this does not indicate real-world superiority. The GTX 1060 3 GB benefits from mature driver optimization in compute APIs and from the fact that Geekbench workloads scale differently than modern gaming workloads.

The RTX 5090 D V2 wins in every architectural metric that matters for contemporary workloads. It has 21,760 shading units versus 1,152, an 18.9 times increase. It features 170 ray tracing cores and 680 tensor cores, capabilities entirely absent from the GTX 1060 3 GB. Its FP16 performance is 104.8 TFLOPS at a 1:1 ratio with FP32, while the GTX 1060 3 GB delivers only 61.49 GFLOPS FP16 at a 1:64 ratio. For any workload using half-precision arithmetic, the RTX 5090 D V2 is in a different universe.

The memory subsystem favors the newer card decisively. The RTX 5090 D V2 has 24 GB of GDDR7 across a 384-bit bus, while the GTX 1060 3 GB has 3 GB of GDDR5 across a 192-bit bus. Eight times the capacity and roughly seven times the bandwidth. Modern games with high-resolution textures will exhaust the 3 GB frame buffer quickly; the 24 GB capacity provides headroom for 4K and beyond.

The GTX 1060 3 GB wins on power efficiency per frame in legacy titles, but the database records no such measurement. Its 120 W TDP versus 575 W TDP shows a 4.8 times difference in power draw, yet the performance gap is far larger in theoretical throughput. The RTX 5090 D V2 delivers roughly 26.6 times the FP32 throughput at 4.8 times the power, making it substantially more efficient per FLOP.

Architecture Differences

The GTX 1060 3 GB uses the GP106 chip built on Pascal architecture, fabricated on a 16 nm process at TSMC. The RTX 5090 D V2 uses the GB202 chip built on Blackwell 2.0 architecture, fabricated on a 5 nm process, also at TSMC. The process node shrink from 16 nm to 5 nm allows for a dramatic increase in transistor count: 4,400 million versus 92,200 million, a 21 times increase. Die size grows from 200 mm² to 750 mm², a 3.75 times increase. Transistor density improves from 22.0M per mm² to 122.9M per mm², a 5.6 times improvement.

Pascal was a monolithic design without dedicated ray tracing or tensor cores. The GTX 1060 3 GB relies entirely on traditional shader pipelines. Blackwell 2.0 introduces 170 RT cores and 680 tensor cores, enabling hardware-accelerated ray tracing and AI-driven features like DLSS. The API support reflects this: the GTX 1060 3 GB supports DirectX 12 (12_1), while the RTX 5090 D V2 supports DirectX 12 Ultimate (12_2), which includes features like mesh shaders and variable rate shading.

Memory technology differs fundamentally. The GTX 1060 3 GB uses GDDR5 at 2002 MHz (8 Gbps effective), while the RTX 5090 D V2 uses GDDR7 at 1750 MHz (28 Gbps effective). The newer memory standard achieves higher effective data rates despite lower physical clock speeds. The bus width also differs: 192-bit versus 384-bit. Combined, these yield 192.2 GB/s versus 1.34 TB/s.

The FP16 implementation reveals architectural philosophy changes. The GTX 1060 3 GB has FP16 at 1:64 the rate of FP32, meaning it is effectively a compute afterthought. The RTX 5090 D V2 runs FP16 at 1:1 with FP32, making it a first-class citizen. This matters for AI inference, scientific computing, and any workload that can exploit half precision.

Specification Differences

The two cards differ across nearly every specification. Clock speeds: the GTX 1060 3 GB has a base clock of 1506 MHz and boost of 1708 MHz, while the RTX 5090 D V2 has a base of 2017 MHz and boost of 2407 MHz. The boost clock difference is 699 MHz, a 40.9% increase.

Shading units: 1,152 versus 21,760. Texture mapping units: 72 versus 680. Raster operation units: 48 versus 176. The RTX 5090 D V2 has 9.4 times more TMUs and 3.7 times more ROPs.

Memory: 3 GB GDDR5 versus 24 GB GDDR7. Bus width: 192-bit versus 384-bit. Bandwidth: 192.2 GB/s versus 1.34 TB/s.

Power: 120 W TDP versus 575 W TDP. Power connectors: 1x 6-pin versus 1x 16-pin. Suggested PSU: 300 W versus 950 W.

Bus interface: PCIe 3.0 x16 versus PCIe 5.0 x16. The newer card has four times the per-lane bandwidth of the older generation.

Display outputs: the GTX 1060 3 GB has 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. The RTX 5090 D V2 has 1x HDMI 2.1b and 3x DisplayPort 2.1b. The newer card drops DVI entirely and upgrades both connector standards.

Physical dimensions: the GTX 1060 3 GB is 250 mm (9.8 inches) long, while the RTX 5090 D V2 is 304 mm (12 inches) long, 137 mm (5.4 inches) high, and 48 mm (1.9 inches) wide. Both are dual-slot cards.

Release dates: August 17, 2016 versus August 14, 2025, a gap of nine years. The GTX 1060 3 GB is end-of-life; the RTX 5090 D V2 is active. The launch MSRP for the GTX 1060 3 GB was 199 USD; for the RTX 5090 D V2 it was 2,299 USD.

FAQ

Q: Why does the GTX 1060 3 GB have a higher percentile rank than the RTX 5090 D V2?

A: The percentile reflects the distribution of scores in the database. The GTX 1060 3 GB has three Geekbench scores averaging 19,334, placing it at the 64th percentile. The RTX 5090 D V2 has one Steel Nomad score of 16,504, placing it at the 59th percentile. Different test suites and different score distributions explain the apparent inversion.

Q: Can the RTX 5090 D V2 run DirectX 12 Ultimate features that the GTX 1060 3 GB cannot?

A: Yes. The GTX 1060 3 GB supports DirectX 12 (12_1), while the RTX 5090 D V2 supports DirectX 12 Ultimate (12_2). The Ultimate tier includes hardware features like mesh shaders and variable rate shading that are absent from the older card.

Q: How much faster is the RTX 5090 D V2 in FP32 compute?

A: The RTX 5090 D V2 delivers 104.8 TFLOPS FP32, while the GTX 1060 3 GB delivers 3.935 TFLOPS. This is a 26.6 times increase in raw shader throughput.

Q: What is the memory capacity difference?

A: The GTX 1060 3 GB has 3 GB of GDDR5, while the RTX 5090 D V2 has 24 GB of GDDR7. This is an eight times increase in capacity, along with a bandwidth increase from 192.2 GB/s to 1.34 TB/s.

Q: Does the RTX 5090 D V2 support ray tracing?

A: Yes. It has 170 dedicated ray tracing cores. The GTX 1060 3 GB has no ray tracing cores; its Pascal architecture predates hardware-accelerated ray tracing.

Q: What is the power draw difference?

A: The GTX 1060 3 GB has a 120 W TDP, while the RTX 5090 D V2 has a 575 W TDP. The newer card also requires a 16-pin power connector and a 950 W suggested PSU, compared to a 6-pin connector and 300 W PSU for the older card.

The Verdict

The data paints a clear picture: these cards serve entirely different eras and use cases. The GTX 1060 3 GB is a product of 2016, built for 1080p gaming at a time when 3 GB of VRAM was sufficient for most titles. Its Pascal architecture lacks ray tracing, tensor cores, and meaningful FP16 support. Its 192.2 GB/s bandwidth and 3.935 TFLOPS FP32 were competitive then but are severely limited now.

The RTX 5090 D V2 is the opposite: a 2025 flagship with 24 GB of GDDR7, 104.8 TFLOPS FP32, 170 RT cores, and 680 tensor cores. It supports PCIe 5.0, DisplayPort 2.1b, and DirectX 12 Ultimate. Its 1.34 TB/s bandwidth and 1,636.8 GTexel/s texture rate place it in a performance class that the GTX 1060 3 GB cannot approach.

For legacy 1080p gaming on older titles, the GTX 1060 3 GB remains functional, and its lower power draw of 120 W makes it easier to accommodate in older systems. For any modern workload, whether gaming at high resolutions, ray tracing, AI inference, or compute tasks, the RTX 5090 D V2 is the only viable choice from this pairing. The specification gap is not marginal; it spans multiple architectural generations and represents a fundamental shift in GPU capability.

Users with the GTX 1060 3 GB should consider upgrading if their workloads involve modern games, creative applications, or any compute beyond basic graphics. Users with the RTX 5090 D V2 have no reason to consider the older card. The benchmark scores, while not directly comparable due to different test suites, reinforce the qualitative assessment: the newer card operates in a performance tier that the older card cannot reach, despite the older card's higher percentile placement in the database.