NVIDIA GeForce GTX 590 vs NVIDIA GeForce GTX 680 Comparison

The NVIDIA GeForce GTX 680 and the NVIDIA GeForce GTX 590 represent two distinct approaches to high-end graphics from the same manufacturer, separated by a single generation. The GTX 680, built on the Kepler architecture, arrived in March 2012 as the flagship of the GeForce 600 series, while the GTX 590, based on the Fermi 2.0 architecture, launched in March 2011 as the top-end dual-GPU card of the GeForce 500 series. The recorded data shows only one direct benchmark comparison between the two, but that single result, combined with their respective specification lists and rival comparisons, provides a clear picture of where each card stands.

Head-to-Head Benchmarks

The database contains one head-to-head benchmark result: Geekbench OpenCL. In this test, the GTX 680 scores 16,906 points, while the GTX 590 scores 12,830 points. The delta percentage is 31.8% in favor of the GTX 680. This is a substantial margin, indicating that the GTX 680 delivers significantly higher OpenCL compute performance than the older dual-GPU card. The GTX 680 wins the sole head-to-head comparison, giving it one win and the GTX 590 zero wins.

Interpreting this score in context, the GTX 680’s OpenCL result of 16,906 places it well above its own average benchmark score of 14,150 points, suggesting that OpenCL is a particularly strong workload for this card. The GTX 590’s OpenCL score of 12,830, by contrast, is exactly equal to its average benchmark score, meaning this workload is representative of its overall performance profile. The 31.8% gap is not a marginal difference; it reflects a fundamental architectural and clock-speed advantage for the GTX 680 in this compute task.

Looking at the nearest rivals for each card further contextualizes these numbers. The GTX 680’s average benchmark score of 14,150 places it at the 55th percentile of all GPUs in the database. Its closest competitors include the NVIDIA Tesla K10 (average score 14,029, delta 0.9% ahead of the GTX 680), the NVIDIA GeForce GTX 1070 Ti (average score 14,277, delta -0.9% behind the GTX 680), the Intel Iris Xe MAX Graphics (average score 14,315, delta -1.2%), and the AMD Radeon Vega 11 (average score 14,352, delta -1.4%). This shows the GTX 680 sits in a tight cluster of mid-range and older high-end cards, within roughly 1.5% of its nearest rivals. Its performance is competitive with much newer hardware, but not dominant.

The GTX 590’s average benchmark score of 12,830 places it at the 52nd percentile of all GPUs. Its nearest rivals are the AMD Radeon Pro 455 (average score 12,831, delta 0%), the AMD FirePro W5100 (average score 12,847, delta -0.1%), the AMD Radeon 740M (average score 12,870, delta -0.3%), and the NVIDIA GeForce GTX 670 (average score 12,773, delta 0.4%). The GTX 590 is effectively tied with the Radeon Pro 455, and its performance sits within a narrow band of about 0.7% of its closest competitors. This suggests the GTX 590’s overall compute capability is comparable to a mid-range mobile workstation GPU, despite its dual-chip design.

In terms of raw compute throughput, the GTX 680 has a clear theoretical edge. It offers 3.250 TFLOPS of FP32 performance, while the GTX 590 delivers 1,244.2 GFLOPS (or 1.244 TFLOPS). The GTX 680 more than doubles the FP32 throughput of the GTX 590. This aligns with the observed OpenCL benchmark result, where the GTX 680’s higher compute capacity translates into a 31.8% performance advantage. The GTX 680 also has a higher pixel rate at 33.86 GPixel/s versus the GTX 590’s 19.46 GPixel/s, and a higher texture rate at 135.4 GTexel/s versus 38.91 GTexel/s. These numbers indicate the GTX 680 is faster across both pixel and texture processing workloads.

The Verdict

Based strictly on the recorded data, the NVIDIA GeForce GTX 680 is the superior card in the single head-to-head benchmark available. It wins the Geekbench OpenCL test by 31.8%, and its average benchmark score of 14,150 is 10.3% higher than the GTX 590’s average of 12,830. The GTX 680 also occupies a higher percentile position (55th versus 52nd), indicating it outperforms a larger fraction of the GPU population. For users seeking maximum compute performance in OpenCL applications, the data clearly favors the GTX 680.

However, the GTX 590 is not without merit, but its advantages are limited to specific areas. The GTX 590 has a wider memory bus (384-bit versus 256-bit) and more ROPs (48 versus 32), which could theoretically benefit certain memory-intensive or rasterization-heavy workloads. Yet, the recorded benchmark data shows no direct test where the GTX 590 wins; the GTX 680 wins the only comparison. The GTX 590’s lower power draw of 365 W versus 195 W is not a point in its favor, as it consumes more power while delivering less compute performance.

For the user who values compute performance, the GTX 680 is the clear choice. Its higher FP32 throughput, faster pixel and texture rates, and superior OpenCL score make it the better all-around performer. The GTX 590, with its dual-GPU design and older architecture, appears to be a less efficient solution, offering lower performance despite higher power consumption and a larger physical footprint. The data suggests that the GTX 680 is the more capable card for modern workloads, and the GTX 590 is best suited for users who specifically need the wider memory bus or the higher ROP count, though no benchmark evidence exists to validate these advantages in practice.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 680 has an average benchmark score of 14,150, while the NVIDIA GeForce GTX 590 has an average score of 12,830. The GTX 680 is higher by approximately 10.3%.

Q: How much faster is the GTX 680 in the OpenCL head-to-head test?

A: The GTX 680 scores 16,906 in Geekbench OpenCL, compared to the GTX 590’s 12,830. This gives the GTX 680 a 31.8% performance advantage.

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

A: The GTX 680 has 2 GB of GDDR5 memory on a 256-bit bus with 192.3 GB/s bandwidth. The GTX 590 has 1536 MB (1.5 GB) of GDDR5 memory on a 384-bit bus with 164.0 GB/s bandwidth.

Q: Which card has a higher transistor density?

A: The GTX 680 has a transistor density of 12.0 million transistors per square millimeter, while the GTX 590 has 5.8 million per square millimeter. The GTX 680’s density is more than double that of the GTX 590.

Q: What is the launch MSRP for each card?

A: The GTX 680 has a launch MSRP of 499 USD, and the GTX 590 has a launch MSRP of 699 USD.

Q: Which card supports Vulkan?

A: The GTX 680 supports Vulkan version 1.2.175, while the GTX 590 has no Vulkan support listed in the database.

Specification Differences

The two cards differ across nearly every major specification. The GTX 680 uses the GK104 chip fabricated on a 28 nm process at TSMC, with 3,540 million transistors on a 294 mm² die. The GTX 590 uses the GF110 chip on a 40 nm process, also at TSMC, with 3,000 million transistors on a 520 mm² die. The GTX 680’s smaller process node allows for higher transistor density (12.0M per mm² versus 5.8M per mm²).

Clock speeds vary significantly. The GTX 680 has a base clock of 1006 MHz and a boost clock of 1058 MHz, while the GTX 590 has no base or boost clock listed. The memory clock for the GTX 680 is 1502 MHz (6 Gbps effective), whereas the GTX 590 runs its memory at 854 MHz (3.4 Gbps effective).

Memory configurations differ: the GTX 680 has 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s bandwidth; the GTX 590 has 1536 MB on a 384-bit bus, yielding 164.0 GB/s. The shading unit count is 1536 for the GTX 680 versus 512 for the GTX 590. Texture mapping units (TMUs) are 128 versus 64, and ROPs are 32 versus 48.

Power requirements diverge sharply. The GTX 680 has a TDP of 195 W and uses two 6-pin power connectors with a suggested PSU of 450 W. The GTX 590 has a TDP of 365 W, uses two 8-pin connectors, and requires a 750 W PSU. The bus interface differs: the GTX 680 uses PCIe 3.0 x16, while the GTX 590 uses PCIe 2.0 x16.

Physical dimensions also change: the GTX 680 is 256 mm long, 111 mm high, and 38 mm wide; the GTX 590 is 279 mm long, 111 mm high, and 40 mm wide. Display outputs are 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2 for the GTX 680, versus 3x DVI and 1x mini-DisplayPort for the GTX 590. The GTX 680 was released on 2012-03-21, and the GTX 590 on 2011-03-23.

Architecture Differences

The GTX 680 is built on the Kepler architecture, a design that emphasizes efficiency and higher clock speeds per watt. Its chip, GK104, is a single-GPU configuration with 1536 shading units. The GTX 590, in contrast, uses the Fermi 2.0 architecture with the GF110 chip, which is a dual-GPU design combining two GF110 dies. This explains the GTX 590’s larger die size of 520 mm² and its higher transistor count of 3,000 million, despite being on an older 40 nm process.

The process node difference is significant: 28 nm for the GTX 680 versus 40 nm for the GTX 590. This allows the GTX 680 to pack more transistors into a smaller area, achieving a density of 12.0M per mm² compared to 5.8M per mm². The GTX 680’s smaller transistors and newer architecture enable it to deliver higher FP32 throughput (3.250 TFLOPS) while consuming less power (195 W versus 365 W).

The GTX 590’s dual-chip design means it has two separate GF110 GPUs, each contributing 256 shading units for a total of 512, but the overall compute performance is limited by the older architecture’s inefficiencies. The GTX 680’s single GK104 chip with 1536 shading units achieves better raw compute despite having fewer total transistors per chip. The memory subsystem also differs: the GTX 590 uses a 384-bit bus for its 1536 MB, while the GTX 680 uses a 256-bit bus for 2 GB, but the GTX 680’s higher memory clock (1502 MHz versus 854 MHz) results in higher bandwidth (192.3 GB/s versus 164.0 GB/s).

In terms of API support, both cards support DirectX 12 (11_0) and OpenGL 4.6, but the GTX 680 adds Vulkan 1.2.175, while the GTX 590 lists no Vulkan support. This is a notable feature gap for modern applications that rely on Vulkan. The GTX 680 also has a boost clock mechanism (1058 MHz) that the GTX 590 lacks, allowing dynamic frequency scaling under load. The GTX 590’s lack of a base clock suggests it may run at a fixed frequency, but the database does not list that value.

The GTX 680’s successor is the GeForce 700 series, and its predecessor is the GeForce 500 series, which includes the GTX 590. This places the two cards in a direct generational transition: the GTX 590 represents the end of the Fermi era, while the GTX 680 marks the beginning of Kepler. The data indicates that Kepler’s architectural improvements, combined with the 28 nm process, yield a card that is both faster and more power-efficient than the dual-GPU Fermi flagship.