NVIDIA GeForce GTX 680 vs NVIDIA GeForce RTX 2060 Comparison

The GeForce RTX 2060 and GeForce GTX 680 are separated by seven years of GPU design, and the benchmark data reflects a generational chasm. The RTX 2060 wins every head-to-head test in the database, but the GTX 680 remains a functional piece of history. This analysis breaks down the raw numbers, the architectural reasoning, and the practical implications for anyone considering these cards today.

Head-to-Head Benchmarks

The two cards share only two direct benchmark comparisons in this dataset, and the RTX 2060 dominates both. In Geekbench OpenCL, the RTX 2060 scores 65,014 against the GTX 680’s 16,906. That is a delta of 284.6% — the RTX 2060 delivers nearly four times the compute throughput. The gap is similarly stark in Geekbench Vulkan, where the RTX 2060 posts 65,846 versus 17,646, a 273.1% advantage. These are not marginal wins; they represent a complete tier shift in raw processing power.

Looking at the broader average benchmark scores, the RTX 2060 holds an average of 15,290 across all its tested workloads, while the GTX 680 averages 14,150. That overall gap is smaller than the head-to-head deltas suggest, largely because the RTX 2060’s dataset includes older DirectX tests where its lead narrows. For instance, in Passmark DirectX 9, the RTX 2060 scores 190, and in Passmark DirectX 10 it scores 98 — solid numbers, but not the 4x blowout seen in compute-heavy APIs. The GTX 680’s absence from those newer API tests (it only has Geekbench Metal, OpenCL, and Vulkan results) means its average is built from a thinner, older sample set.

The percentile rankings tell a similar story. The RTX 2060 sits at the 58th percentile of all GPUs, while the GTX 680 is at the 55th. That two-point difference might sound small, but it obscures the fact that the RTX 2060 is competing in a much larger and more modern field. The RTX 2060’s nearest rivals include the AMD Radeon RX 7600 (delta 0.8%) and the NVIDIA GeForce RTX 3050 OEM (delta 0.6%) — cards from the same era. The GTX 680’s nearest rivals are the NVIDIA Tesla K10 (delta 0.9%) and the Intel Iris Xe MAX Graphics (delta -1.2%), a mix of workstation and integrated-class parts.

FAQ

Q: Is the RTX 2060 faster than the GTX 680 in every benchmark?

A: Yes. The dataset shows the RTX 2060 winning both head-to-head tests (Geekbench OpenCL and Vulkan) with deltas of 284.6% and 273.1%, respectively. The GTX 680 has zero wins in the comparison.

Q: How much memory bandwidth does each card have?

A: The RTX 2060 has 336.0 GB/s of bandwidth from its 6 GB GDDR6 memory on a 192-bit bus. The GTX 680 has 192.3 GB/s from 2 GB GDDR5 on a 256-bit bus. The RTX 2060’s newer memory type more than compensates for the narrower bus.

Q: Which card has better compute performance for OpenCL workloads?

A: The RTX 2060 scores 65,014 in Geekbench OpenCL versus 16,906 for the GTX 680, a 284.6% advantage. The RTX 2060 also has a higher FP32 throughput at 6.451 TFLOPS compared to 3.250 TFLOPS for the GTX 680.

Q: What are the power requirements for each card?

A: The RTX 2060 has a TDP of 160 W and requires a single 8-pin power connector. The GTX 680 has a higher TDP of 195 W and needs two 6-pin connectors. Both cards recommend a 450 W power supply.

Q: Do both cards support modern graphics APIs?

A: The RTX 2060 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GTX 680 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RTX 2060 has a more feature-complete API set.

Q: What is the physical size difference between the two?

A: The RTX 2060 is 229 mm long, 113 mm tall, and 35 mm wide. The GTX 680 is longer at 256 mm, similar height at 111 mm, and slightly wider at 38 mm. Both are dual-slot cards.

The Verdict

The data is unambiguous: the RTX 2060 is the superior card by every measurable metric. It wins both head-to-head tests, has a higher average benchmark score (15,290 vs 14,150), and sits at a higher percentile ranking (58th vs 55th). The RTX 2060 offers more than double the FP32 compute (6.451 TFLOPS vs 3.250 TFLOPS), nearly double the memory bandwidth (336.0 GB/s vs 192.3 GB/s), and triple the memory capacity (6 GB vs 2 GB). It also does this while consuming less power (160 W vs 195 W) and occupying a shorter physical footprint.

For a builder choosing between these two today, the RTX 2060 is the only rational choice for any modern workload. The GTX 680’s 2 GB memory buffer is a hard limitation in current games and applications, and its Vulkan 1.2.175 support lacks the RTX 2060’s Vulkan 1.4 compatibility. The RTX 2060’s DirectX 12 Ultimate support and ray-tracing cores make it viable for contemporary titles, while the GTX 680 is strictly a legacy card. The GTX 680 does have a wider memory bus (256-bit vs 192-bit), but that advantage is negated by the slower GDDR5 memory and lower bandwidth.

Specification Differences

The two cards differ in nearly every core specification. The RTX 2060 uses the TU106 chip on a 12 nm process with 10,800 million transistors, while the GTX 680 uses the GK104 chip on a 28 nm process with 3,540 million transistors. The RTX 2060 has a larger die (445 mm² vs 294 mm²) and a much higher transistor density (24.3M / mm² vs 12.0M / mm²). Clock speeds favor the RTX 2060 with a base of 1365 MHz and boost of 1680 MHz, compared to the GTX 680’s 1006 MHz base and 1058 MHz boost. Memory configurations are entirely different: the RTX 2060 uses 6 GB GDDR6 at 14 Gbps effective, while the GTX 680 uses 2 GB GDDR5 at 6 Gbps effective. The RTX 2060 has more shading units (1920 vs 1536), fewer TMUs (120 vs 128), and more ROPs (48 vs 32). The RTX 2060 also brings 30 RT cores and 240 tensor cores, which the GTX 680 lacks entirely.

Architecture Differences

The architectural gap is the root of the performance divide. The RTX 2060 is built on Turing, NVIDIA’s architecture that introduced dedicated ray-tracing hardware and tensor cores for AI acceleration. It is fabricated on TSMC’s 12 nm process, enabling a transistor density of 24.3M per mm². The GTX 680 uses the older Kepler architecture on a 28 nm process, with a transistor density of just 12.0M per mm². Kepler was designed for a different era, without any consideration for ray tracing or tensor workloads. The RTX 2060’s FP16 throughput of 12.90 TFLOPS (2:1) is a notable feature absent from the GTX 680, which has no FP16 data in the pack. The RTX 2060 also supports USB Type-C output alongside its DVI, HDMI 2.0, and dual DisplayPort 1.4a outputs; the GTX 680 offers dual DVI, HDMI 1.4a, and a single DisplayPort 1.2. The RTX 2060’s display outputs are modern, while the GTX 680’s are dated.

Where Each One Wins

The RTX 2060 wins in every scenario where performance matters. For compute-heavy tasks like OpenCL and Vulkan workloads, it is in a different league — the 284.6% and 273.1% deltas are not close calls. Its 6 GB memory buffer handles modern textures and datasets where the GTX 680’s 2 GB would hit a wall. The RTX 2060 is also the only card here with ray-tracing cores and tensor cores, making it suitable for games that use those features. Its lower TDP and single 8-pin connector make it easier to power in a modern build.

The GTX 680’s only theoretical advantages are its wider 256-bit memory bus and higher TMU count (128 vs 120), but these do not translate into any benchmark wins. It could be considered for a retro build or a collector’s piece, given its Kepler architecture and historical significance. The GTX 680 does have a higher launch MSRP of 499 USD compared to the RTX 2060’s 349 USD, but that is a historical fact, not a current buying signal. In practical terms, the GTX 680 is a card for showcasing older technology, not for performing modern tasks. The RTX 2060 is the card to pick for any real workload.