NVIDIA GeForce GTX 570 vs NVIDIA GeForce GTX 680 Comparison

The NVIDIA GeForce GTX 680 and NVIDIA GeForce GTX 570 represent two distinct generations of NVIDIA’s high-end graphics lineup, separated by a significant architectural shift. The GTX 680, built on the Kepler architecture, arrives as a successor to the Fermi-based GTX 570. Benchmark data indicates a clear performance hierarchy, with the GTX 680 holding a substantial lead in the available OpenCL test, yet both cards occupy similar positions in the overall GPU percentile rankings. This analysis examines their architectural differences, benchmark results, and the specific scenarios where each card may still hold relevance.

FAQ

Q: How much faster is the GTX 680 than the GTX 570 in the head-to-head benchmark?

A: In the Geekbench OpenCL test, the GTX 680 scores 16906 compared to the GTX 570’s 13515, resulting in a 25.1% higher score for the GTX 680.

Q: Which card has a higher average benchmark score?

A: The GTX 680 has an average benchmark score of 14150 across all its tested workloads, while the GTX 570’s average is 13515, based solely on its single OpenCL result.

Q: How do their overall GPU percentile rankings compare?

A: The GTX 680 sits at the 55th percentile among all GPUs, while the GTX 570 is at the 54th percentile, placing them nearly adjacent in overall performance distribution.

Q: What is the memory configuration difference between the two cards?

A: The GTX 680 features 2 GB of GDDR5 memory on a 256-bit bus with 192.3 GB/s bandwidth, whereas the GTX 570 has 1280 MB of GDDR5 on a 320-bit bus with 152.0 GB/s bandwidth.

Q: Which card has a higher transistor density despite having a smaller die?

A: The GTX 680, with a 294 mm² die and 3,540 million transistors, achieves a density of 12.0M / mm². The GTX 570’s larger 520 mm² die holds 3,000 million transistors, yielding a lower density of 5.8M / mm².

Q: Does the GTX 570 support Vulkan like the GTX 680 does?

A: No, the GTX 680 lists Vulkan 1.2.175 support, while the GTX 570’s API list shows no Vulkan version. Both cards support DirectX 12 (11_0) and OpenGL 4.6.

Architecture Differences

The GTX 680 and GTX 570 are built on fundamentally different architectures from NVIDIA. The GTX 680 uses the GK104 chip based on the Kepler architecture, manufactured on a 28 nm process at TSMC. In contrast, the GTX 570 uses the GF110 chip based on the Fermi 2.0 architecture, also produced by TSMC but on a larger 40 nm process node. This process shrink allows the GTX 680 to pack 3,540 million transistors into a 294 mm² die, achieving a transistor density of 12.0M / mm². The GTX 570 holds 3,000 million transistors on a much larger 520 mm² die, with a significantly lower density of 5.8M / mm².

The compute configurations diverge sharply. The GTX 680 features 1536 shading units, 128 texture mapping units (TMUs), and 32 render output units (ROPs). The GTX 570 has 480 shading units, 60 TMUs, and 40 ROPs. This means the GTX 680 has over three times the shading units and more than double the TMUs, though the GTX 570 retains a higher ROP count. Consequently, the GTX 680 achieves a pixel rate of 33.86 GPixel/s and a texture rate of 135.4 GTexel/s, while the GTX 570 manages 21.96 GPixel/s and 43.92 GTexel/s respectively. The FP32 compute throughput is also vastly different: 3.250 TFLOPS for the GTX 680 versus 1,405.4 GFLOPS for the GTX 570.

Memory architecture also differs. The GTX 680 uses a 256-bit memory bus with 2 GB of GDDR5, running at 1502 MHz (6 Gbps effective) to deliver 192.3 GB/s bandwidth. The GTX 570 relies on a wider 320-bit bus but has only 1280 MB of GDDR5 memory, clocked at 950 MHz (3.8 Gbps effective), providing 152.0 GB/s bandwidth. Both cards are dual-slot designs with identical 2x 6-pin power connectors, though the GTX 570 has a higher TDP of 219 W compared to the GTX 680’s 195 W. The GTX 680 also uses a newer PCIe 3.0 x16 interface, while the GTX 570 is limited to PCIe 2.0 x16.

Where Each One Wins

The benchmark data is one-sided, with the GTX 680 winning the only head-to-head test available. In the Geekbench OpenCL workload, the GTX 680’s 25.1% lead over the GTX 570 is decisive. This performance advantage is consistent with the GTX 680’s superior compute resources, including more than triple the shading units and a much higher FP32 throughput. For any OpenCL-based application, the GTX 680 is clearly the stronger option.

The GTX 570, however, is not without its merits. Its higher ROP count of 40 versus the GTX 680’s 32 suggests it may have an advantage in certain pixel-bound operations, though no direct benchmark confirms this. Its wider 320-bit memory bus could theoretically provide better efficiency in specific memory-access patterns, despite lower overall bandwidth. Additionally, the GTX 570 has a lower launch MSRP of 349 USD compared to the GTX 680’s 499 USD, which may have made it a more accessible option at release, though pricing is not a current performance factor.

Both cards occupy nearly identical percentile positions (55th for GTX 680, 54th for GTX 570), indicating that in the broader GPU landscape, they are considered comparable in overall capability despite the significant generational gap. The GTX 570’s single benchmark score of 13515 places it within 0.2% of the AMD Radeon HD 7770M and RX 9070 XT, while the GTX 680’s 14150 average is within 1.4% of the Intel Iris Xe MAX Graphics. This suggests that in mixed workloads, the real-world difference may be less pronounced than raw specs imply.

Specification Differences

The two cards differ across nearly every major specification category. The GTX 680 uses the GK104 chip on a 28 nm process, while the GTX 570 uses the GF110 chip on a 40 nm process. Transistor count is 3,540 million versus 3,000 million, with die sizes of 294 mm² and 520 mm² respectively. The GTX 680 has a base clock of 1006 MHz and a boost clock of 1058 MHz; the GTX 570 has no listed base or boost clocks. Memory clocks are 1502 MHz (6 Gbps effective) for the GTX 680 and 950 MHz (3.8 Gbps effective) for the GTX 570.

Memory configuration differs: 2 GB on a 256-bit bus versus 1280 MB on a 320-bit bus. Bandwidth is 192.3 GB/s for the GTX 680 and 152.0 GB/s for the GTX 570. Shading units are 1536 versus 480, TMUs are 128 versus 60, and ROPs are 32 versus 40. Pixel rates are 33.86 GPixel/s and 21.96 GPixel/s, while texture rates are 135.4 GTexel/s and 43.92 GTexel/s. FP32 compute is 3.250 TFLOPS versus 1,405.4 GFLOPS. TDP is 195 W for the GTX 680 and 219 W for the GTX 570, with suggested PSUs of 450 W and 550 W respectively.

Bus interfaces are PCIe 3.0 x16 for the GTX 680 and PCIe 2.0 x16 for the GTX 570. Display outputs are 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2 for the GTX 680, while the GTX 570 offers 2x DVI and 1x mini-HDMI 1.3a. The GTX 680 supports Vulkan 1.2.175; the GTX 570 has no Vulkan support. Physical dimensions are 256 mm length for the GTX 680 versus 267 mm for the GTX 570, with only the GTX 680 listing height and width. Release dates are March 2012 for the GTX 680 and December 2010 for the GTX 570, with launch MSRPs of 499 USD and 349 USD respectively.

Head-to-Head Benchmarks

The single head-to-head benchmark is Geekbench OpenCL, where the GTX 680 scores 16906 against the GTX 570’s 13515. This represents a delta of 25.1% in favor of the GTX 680. The magnitude of this difference is substantial and reflects the architectural advantages of Kepler over Fermi. The GTX 680’s 3.250 TFLOPS FP32 throughput is more than double the GTX 570’s 1,405.4 GFLOPS, and its 1536 shading units provide three times the parallel processing capacity. The memory bandwidth advantage of 192.3 GB/s versus 152.0 GB/s also contributes, though the GTX 570’s wider 320-bit bus partially compensates.

In the broader benchmark context, the GTX 680’s average score of 14150 is 4.7% higher than the GTX 570’s 13515. However, the GTX 680’s average is pulled down by its Geekbench Metal score of 7897 and Vulkan score of 17646, which are lower than its OpenCL result. The GTX 570 has no Metal or Vulkan scores, making its average equal to its OpenCL score. When comparing against nearest rivals, the GTX 680 is 0.9% ahead of the NVIDIA Tesla K10 (14029) and 0.9% behind the GTX 1070 Ti (14277). The GTX 570 is 0.2% ahead of the AMD Radeon HD 7770M (13536) and 0.8% behind the AMD Radeon Pro 555 (13407).

The wins ledger shows GTX 680 with 1 win and GTX 570 with 0 wins in head-to-head tests. This outcome is expected given the generational gap and the GTX 680’s superior specifications across nearly all compute metrics. The GTX 570’s higher ROP count and wider memory bus do not manifest as a benchmark victory in the available data. For users prioritizing OpenCL performance, the GTX 680’s 25.1% advantage is the definitive metric. The percentile rankings of 55th and 54th, however, indicate that in the full GPU landscape, these cards are closer in overall standing than their direct comparison suggests, likely due to the GTX 570’s efficiency in certain legacy workloads not captured by the single test.