NVIDIA GeForce GTX 780 vs NVIDIA GeForce RTX 5090 D V2 Comparison

The Verdict

The data presents a stark generational contrast. The GTX 780, a Kepler-era card from 2013, shows its age in raw specifications but holds a surprisingly strong position in legacy compute benchmarks. The RTX 5090 D V2, built on Blackwell 2.0, obliterates the older card in every architectural metric, yet its single recorded benchmark score places it in a lower percentile ranking. The GTX 780 sits in the 64th percentile of all GPUs, while the RTX 5090 D V2 sits in the 59th. This is a curious inversion, driven by the fact that the two cards are tested under completely different benchmark suites. For a user seeking modern gaming features, ray tracing, or massive memory capacity, the RTX 5090 D V2 is the only logical choice. For legacy compute tasks using OpenCL or Vulkan, the GTX 780 still delivers competitive results, sitting within 0.7% of an AMD Radeon RX 6600 in average score. The RTX 5090 D V2, on the other hand, is effectively tied with the NVIDIA T400 in its own benchmark, a delta of 0%. This suggests that the newer card's single Steel Nomad result does not reflect its theoretical dominance; it is a specialized DirectX 12 test, not a general-purpose compute measurement.

Architecture Differences

The two GPUs are separated by more than a decade of silicon evolution. The GTX 780 uses the GK110 chip on a 28 nm TSMC process, packing 7,080 million transistors into a 561 mm² die. The RTX 5090 D V2 uses the GB202 chip on a 5 nm TSMC process, with 92,200 million transistors on a 750 mm² die. Transistor density tells the story: the older card manages 12.6 million transistors per square millimeter, while the newer card achieves 122.9 million, a near tenfold increase in integration. The GTX 780 is built on the Kepler architecture, supporting DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RTX 5090 D V2 is built on Blackwell 2.0, supporting DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The newer card also introduces 170 ray tracing cores and 680 tensor cores, features entirely absent from the GTX 780. Clock speeds jump from a base of 863 MHz and boost of 902 MHz on the older card to a base of 2017 MHz and boost of 2407 MHz on the newer one. The memory subsystem is equally divergent: the GTX 780 uses 3 GB of GDDR5 on a 384-bit bus at 6 Gbps effective, yielding 288.4 GB/s, while the RTX 5090 D V2 uses 24 GB of GDDR7 on the same 384-bit bus at 28 Gbps effective, yielding 1.34 TB/s. The RTX 5090 D V2 also moves to PCIe 5.0 x16, doubling the bus bandwidth of the GTX 780's PCIe 3.0 x16. Power delivery is another clear divide: the older card draws up to 250 W with a 600 W suggested PSU and two connectors, while the newer card consumes up to 575 W, requires a 950 W PSU, and uses a single 16-pin connector.

Where Each One Wins

The GTX 780 wins in legacy compute workloads. Its recorded benchmarks include Geekbench Metal (10,114), Geekbench OpenCL (22,863), and Geekbench Vulkan (24,514). Its average benchmark score is 19,164, which places it in the 64th percentile. The RTX 5090 D V2 has only one recorded benchmark: 3DMark Steel Nomad DX12, scoring 16,504. This places it in the 59th percentile, below the older card's rank. The data suggests that if a workload relies on OpenCL or Vulkan, the GTX 780 is a proven performer, essentially matching the NVIDIA TITAN Xp (within 0.1%) and slightly beating the NVIDIA Tesla K20m (0.4% faster). Conversely, the RTX 5090 D V2 wins on every architectural front. It has 21,760 shading units versus 2,304, 680 texture mapping units versus 192, and 176 ROPs versus 48. Its pixel rate is 423.6 GPixel/s versus 43.30 GPixel/s, and its texture rate is 1,636.8 GTexel/s versus 173.2 GTexel/s. FP32 compute jumps from 4.156 TFLOPS to 104.8 TFLOPS. Memory capacity jumps from 3 GB to 24 GB, and bandwidth from 288.4 GB/s to 1.34 TB/s. The RTX 5090 D V2 is also the only card with ray tracing, tensor cores, and DirectX 12 Ultimate support. For modern gaming, high-resolution texturing, or any modern gaming workload that uses the latest API standards, the RTX 5090 D V2 is the clear winner. The GTX 780 is end-of-life, while the RTX 5090 D V2 is still active in production. The RTX 5090 D V2 also supports Vulkan 1.4, a newer version of the API, and has a newer display output set: 1x HDMI 2.1b and 3x DisplayPort 2.1b, versus the GTX 780's 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

FAQ

Q: The RTX 5090 D V2 has a lower percentile rank than the GTX 780. Does that mean it is a slower overall GPU?

A: The percentile ranks come from different test suites. The GTX 780's 64th percentile is based on Metal, OpenCL, and Vulkan results from older GPUs. The RTX 5090 D V2's 59th percentile is based on a single 3DMark Steel Nomad DX12 score. The newer card has a recorded average benchmark score of 16,504, lower than the GTX 780's average of 19,164, but the tests are not directly comparable.

Q: Which card is more suitable for modern gaming with ray tracing and high refresh rates

A: The RTX 5090 D V2 is the only card with 170 ray tracing cores and 170 tensor cores. It also supports DirectX 12 Ultimate (12_2) with Vulkan 1.4 support. The GTX 780 supports DirectX 12 (11_0) and has no ray tracing or tensor cores. The RTX 5090 D V2 also has 24 GB of memory and a 1.34 TB/s bandwidth, which is suitable for high-resolution textures.

Q: How do the two compare in raw compute throughput, specifically FP32 and pixel rate

A: The RTX 5090 D V2 has a FP32 compute of 104.8 TFLOPS and FP16 compute of 104.8 TFLOPS (1:1 ratio). The GTX 780 has FP32 of 4.156 TFLOPS. The newer card also has a pixel rate of 423.6 GPixel/s, while the GTX 780 has a pixel rate of 43.30 GPixel/s. Texture rate is 1,636.8 GTexel/s for the RTX 5090 D V2, versus 173.2 GTexel/s for the GTX 780.

Q: What are the memory differences in terms of capacity bandwidth and type

A: The GTX 780 comes with 3 GB of GDDR5 memory on a 384-bit bus at 1502 MHz with 6 Gbps effective speed. The RTX 5090 D V2 comes with 24 GB of GDDR7 memory on a 384-bit bus at 1750 MHz with 28 Gbps effective speed. The bandwidth is 288.4 GB/s for the older card, while the newer one is 1.34 TB/s, a difference of more than four times the bandwidth.

Q: Which card has been on the market longer and what is its production status

**A: The GTX 780 was released on 2013-05-22 and has a production status of end-of-life, with a launch MSRP of 649 USD. The RTX 5090 D V2, on the other hand, was released on 2025-08-14 and is active in production, with a launch MSRP of 2,299 USD, and is now in the GeForce 50-series generation.

Head-to-Head Benchmark Walkthrough

The only head-to-head benchmark data field is empty. The database records zero wins for the GTX 780 and zero wins for the RTX 5090 D V2. The GTX 780, however, has three recorded benchmarks that can be interpreted individually. In Geekbench Metal, it scores 10,114. In Geekbench OpenCL, it scores 22,863, a solid result that contributes to its average score of 19,164. In Geekbench Vulkan, it scores 24,514. These results place it in the 64th percentile. The GTX 780's nearest rivals include the NVIDIA TITAN Xp with an average score of 19,177 and a delta of -0.1% relative to it. It is 0.4% faster than the NVIDIA Tesla K20m, which has an average score of 19,089. The NVIDIA GeForce RTX 4050 Mobile has an average score of 19,049, and the GTX 780 is 0.6% faster, and the AMD Radeon RX 6600 has an average score of 19,036, with the GTX 780 being 0.7% faster. The RTX 5090 D V2, in its single recorded benchmark, 3DMark Steel Nomad DX12, scores 16,504. This gives it an average score of 16,504 and a 59th percentile ranking. Its closest rival is the NVIDIA T400, which scores 16,508, a delta of 0%. The AMD Radeon PRO W7500 scores 16,415, with the RTX 5090 D V2 being 0.5% faster. The NVIDIA RTX PRO 6000 Blackwell scores 16,408, and the RTX 5090 D V2 is 0.6% faster. The AMD Radeon RX 5700 XT scores 16,361, and the RTX 5090 D V2 is 0.9% faster. The data implies that the GTX 780 is a mature, well-tested GPU with a stable position relative to other legacy cards, while the RTX 5090 D V2 has only one data point, which is a much newer and more demanding test.

Specification Differences

The two cards differ in nearly every field. The GTX 780 has a chip of GK110 with Kepler architecture, while the RTX 5090 D V2 has a chip of GB202 with Blackwell 2.0 architecture. The process node is 28 nm for the GTX 780 and 5 nm for the RTX 5090 D V2. Transistor count is 7,080 million versus 92,200 million. Die size is 561 mm² versus 750 mm². Transistor density is 12.6M / mm² versus 122.9M / mm². Base clock is 863 MHz versus 2017 MHz. Boost clock is 902 MHz versus 2407 MHz. Memory clock is 1502 MHz (6 Gbps effective) versus 1750 MHz (28 Gbps effective). Memory size is 3 GB versus 24 GB. Memory type is GDDR5 versus GDDR7. Bus width is identical at 384 bit. Bandwidth is 288.4 GB/s versus 1.34 TB/s. Shading units are 2304 versus 21760. TMUs are 192 versus 680. ROPs are 48 versus 176. The RTX 5090 D V2 has 170 ray tracing cores and 680 tensor cores, while the GTX 780 has none. Pixel rate is 43.30 GPixel/s versus 423.6 GPixel/s. Texture rate is 173.2 GTexel/s versus 1,636.8 GTexel/s. FP32 is 4.156 TFLOPS versus 104.8 TFLOPS. The RTX 5090 D V2 has FP16 at 104.8 TFLOPS (1:1), while the GTX 780 has no FP16 data. TDP is 250 W versus 575 W. Power connectors are 1x 6-pin + 1x 8-pin versus 1x 16-pin. Suggested PSU is 600 W versus 950 W. Bus interface is PCIe 3.0 x16 versus PCIe 5.0 x16. Display outputs are 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 versus 1x HDMI 2.1b, 3x DisplayPort 2.1b. DirectX support is 12 (11_0) versus 12 Ultimate (12_2). Vulkan support is 1.2.175 versus 1.4. OpenGL is the same at 4.6 for both. Dimensions are 267 mm length, 111 mm height, 38 mm width versus 304 mm length, 137 mm height, 48 mm width. Production status is end-of-life versus active. Release date is 2013-05-22 versus 2025-08-14. The GTX 780 belongs to the GeForce 700 generation, while the RTX 5090 D V2 belongs to the GeForce 50 generation. The GTX 780 has a launch MSRP of 649 USD, while the RTX 5090 D V2 has a launch MSRP of 2,299 USD. The GTX 780's predecessor is GeForce 600 and successor is GeForce 900. The RTX 5090 D V2's predecessor is GeForce 40 and successor is GeForce 60.