NVIDIA GeForce GTX 580 vs NVIDIA GeForce GTX 950 Comparison

Head-to-Head Benchmarks

The benchmark data gives the NVIDIA GeForce GTX 950 a clear, if narrow, victory over the GeForce GTX 580 in the one directly comparable compute test. In Geekbench OpenCL, the GTX 950 scores 15,662 points, while the GTX 580 scores 15,283 points. That is a 2.4% advantage for the GTX 950, a modest but decisive margin. For a card released roughly five years later, the gain is smaller than the generational gap might suggest, but it still flips the result in favor of the newer architecture.

The GTX 580’s single benchmark entry, Geekbench OpenCL at 15,283, places it at the 58th percentile of all GPUs in the database. Its nearest rival is the NVIDIA GeForce RTX 2060, which averages 15,290 points, a delta of 0%. The GTX 580 essentially ties the RTX 2060 in this specific workload, which is remarkable given the RTX 2060’s much later release. The AMD Radeon 680M sits just 0.1% behind at 15,270 points, and the RTX 3050 OEM trails by 0.6% with 15,199 points. The AMD Radeon RX 7600 is 0.7% behind at 15,171 points. These deltas are tiny, meaning the GTX 580’s raw compute throughput in OpenCL lands right in the middle of a cluster of modern and semi-modern GPUs.

The GTX 950, by contrast, has three benchmark results recorded: Geekbench OpenCL at 15,662, Geekbench Vulkan at 16,734, and Geekbench Metal at 7,172. Its average benchmark score across all recorded tests is 13,189, which puts it at the 53rd percentile of all GPUs. The OpenCL score alone is higher than the GTX 580’s, but the average is dragged down by the Metal result, which is less than half of the OpenCL score. The GTX 950’s nearest rivals in the database include the NVIDIA GeForce GTX 480 with an average score of 13,300 (0.8% higher), the Tesla M2090 at 13,075 (0.9% lower), the AMD Radeon Pro 555X at 13,321 (1% higher), and the AMD FirePro M6100 at 13,354 (1.2% higher). The GTX 950’s average sits between these, slightly behind the GTX 480 but ahead of the Tesla M2090.

Looking strictly at the head-to-head test, the GTX 950 wins the only direct comparison. That is the entire story of the direct matchup: one test, one winner. But the broader benchmark picture adds context. The GTX 580’s single score is consistent across all recorded data, while the GTX 950 shows wider variance depending on the API. In OpenCL, the GTX 950 is faster. In Vulkan, the GTX 950 scores even higher at 16,734, which is 9.5% above its own OpenCL result. The GTX 580 has no Vulkan score recorded, so no direct comparison exists there, but the GTX 950’s Vulkan figure indicates the Maxwell architecture handles modern low-level APIs well. The Metal score of 7,172 is the outlier, likely reflecting API-specific driver behavior rather than raw hardware capability.

Where Each One Wins

The GTX 950 wins in the only directly comparable benchmark: Geekbench OpenCL. It also demonstrates strength in Vulkan, where it scores 16,734, the highest number recorded for either card in any test. This suggests the GTX 950 is the better choice for workloads that leverage Vulkan, such as recent games or compute applications built around that API. Its OpenCL score of 15,662 also edges out the GTX 580, so for general compute tasks using OpenCL, the GTX 950 holds the advantage.

The GTX 580, however, shows a narrower but still competitive profile. Its OpenCL score of 15,283 is within 2.4% of the GTX 950’s, so in pure compute throughput the two are close. The GTX 580’s percentile rank of 58 is higher than the GTX 950’s 53, which reflects the fact that the GTX 580’s single score is more consistent and does not suffer from a low Metal result. If the comparison is limited to OpenCL only, the GTX 580 is nearly on par with the GTX 950, and it ties the RTX 2060 in the database, which is an impressive result for a 2010 card.

For gaming, the GTX 950 has the edge due to its Vulkan support and higher OpenCL score. The GTX 580 has no Vulkan support listed, which limits its modern API compatibility. The GTX 950 also offers a higher texture rate (57.02 GTexel/s versus 49.41 GTexel/s) and a higher pixel rate (38.02 GPixel/s versus 24.70 GPixel/s), which directly benefit rasterization-heavy workloads. The GTX 950’s shading unit count is 768 versus 512 for the GTX 580, and its FP32 throughput is 1.825 TFLOPS versus 1.581 TFLOPS. These numbers favor the GTX 950 in shader-bound scenarios.

The GTX 580, however, has a much wider memory bus: 384 bit versus 128 bit, and higher memory bandwidth at 192.4 GB/s versus 105.8 GB/s. That bandwidth advantage can matter in memory-intensive compute tasks, even if the GTX 950 has more memory capacity (2 GB versus 1536 MB). For tasks that are bandwidth-limited, the GTX 580 could pull ahead, though no direct benchmark in the database tests that scenario.

FAQ

Q: Which card wins the Geekbench OpenCL test?

A: The NVIDIA GeForce GTX 950 wins with a score of 15,662, while the GTX 580 scores 15,283. The delta is 2.4% in favor of the GTX 950.

Q: How does the GTX 580 compare to its nearest rivals in the database?

A: The GTX 580’s average score of 15,283 is essentially tied with the NVIDIA GeForce RTX 2060 (15,290, 0% delta) and slightly ahead of the AMD Radeon 680M (15,270, 0.1% delta), the RTX 3050 OEM (15,199, 0.6% delta), and the AMD Radeon RX 7600 (15,171, 0.7% delta).

Q: Does the GTX 950 support Vulkan?

A: Yes, the GTX 950 lists Vulkan 1.4 as a supported API, and its Geekbench Vulkan score is 16,734. The GTX 580 does not list any Vulkan support in the database.

Q: What is the average benchmark score for each card?

A: The GTX 580 has an average benchmark score of 15,283 based on one recorded test. The GTX 950 has an average of 13,189 across three tests (OpenCL, Vulkan, and Metal).

Q: Which card has higher memory bandwidth?

A: The GTX 580 has a memory bandwidth of 192.4 GB/s, while the GTX 950 has 105.8 GB/s. The GTX 580’s memory bus is 384 bit, compared to the GTX 950’s 128 bit.

Q: How do their percentile ranks compare?

A: The GTX 580 sits at the 58th percentile of all GPUs, while the GTX 950 sits at the 53rd percentile.

Specification Differences

The two cards differ in nearly every core specification. The GTX 580 uses a 40 nm process node, while the GTX 950 uses a 28 nm node. The GTX 580 has 3,000 million transistors on a 520 mm² die, giving a transistor density of 5.8 million per mm². The GTX 950 has 2,940 million transistors on a 228 mm² die, for a density of 12.9 million per mm². That is a significantly denser design for the newer card.

Clock speeds differ: the GTX 580 lists a memory clock of 1002 MHz (4 Gbps effective) but no base or boost clock. The GTX 950 has a base clock of 1024 MHz and a boost clock of 1188 MHz, with memory at 1653 MHz (6.6 Gbps effective). Memory configuration also diverges: the GTX 580 has 1536 MB of GDDR5 on a 384 bit bus, while the GTX 950 has 2 GB of GDDR5 on a 128 bit bus. Bandwidth is 192.4 GB/s for the GTX 580 versus 105.8 GB/s for the GTX 950.

Shading units favor the GTX 950: 768 versus 512. The GTX 580 has 64 TMUs and 48 ROPs, while the GTX 950 has 48 TMUs and 32 ROPs. Pixel rate is 24.70 GPixel/s for the GTX 580 and 38.02 GPixel/s for the GTX 950. Texture rate is 49.41 GTexel/s for the GTX 580 and 57.02 GTexel/s for the GTX 950. FP32 compute is 1.581 TFLOPS for the GTX 580 and 1.825 TFLOPS for the GTX 950.

Power requirements are starkly different: the GTX 580 has a TDP of 244 W, requires a 550 W power supply, and uses 1x 6-pin plus 1x 8-pin connectors. The GTX 950 has a TDP of 90 W, requires a 250 W power supply, and uses a single 6-pin connector. Physical dimensions also differ: the GTX 580 is 267 mm (10.5 inches) long and 111 mm (4.4 inches) high, while the GTX 950 is 202 mm (8 inches) long with no height listed. The bus interface is PCIe 2.0 x16 for the GTX 580 and PCIe 3.0 x16 for the GTX 950.

Display outputs differ: the GTX 580 offers 2x DVI and 1x mini-HDMI 1.3a, while the GTX 950 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2. API support also differs: the GTX 580 supports DirectX 12 (11_0) and OpenGL 4.6, with no Vulkan listed. The GTX 950 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4.

Architecture Differences

The GTX 580 is built on the Fermi 2.0 architecture, using the GF110 chip, and belongs to the GeForce 500 generation. The GTX 950 is built on Maxwell 2.0, using the GM206 chip, and belongs to the GeForce 900 generation. Both are manufactured by TSMC, but at different nodes: 40 nm for Fermi and 28 nm for Maxwell.

The transistor counts are nearly identical (3,000 million versus 2,940 million), but the die size is dramatically different: 520 mm² for the GTX 580 versus 228 mm² for the GTX 950. This means the GTX 950 packs roughly the same number of transistors into less than half the area, a direct result of the smaller process node. The GTX 950’s higher transistor density (12.9M / mm² versus 5.8M / mm²) reflects the architectural and manufacturing improvements.

Core configuration differs significantly. The GTX 950 has 768 shading units, 48 TMUs, and 32 ROPs, while the GTX 580 has 512 shading units, 64 TMUs, and 48 ROPs. Despite fewer TMUs and ROPs, the GTX 950 achieves higher pixel and texture rates due to its higher clocks. The GTX 950’s boost clock of 1188 MHz is substantially higher than anything recorded for the GTX 580, which has no base or boost clock listed.

Memory architecture is a key difference. The GTX 580 uses a 384 bit bus with 1536 MB, while the GTX 950 uses a 128 bit bus with 2 GB. The GTX 580’s wider bus gives it almost double the bandwidth, but the GTX 950 compensates with faster effective memory speed (6.6 Gbps versus 4 Gbps) and higher capacity. The GTX 950 also supports PCIe 3.0, while the GTX 580 is limited to PCIe 2.0, which affects data transfer to the host system.

Feature support also diverges. The GTX 950 supports Vulkan 1.4 and DirectX 12 (12_1), while the GTX 580 only reaches DirectX 12 (11_0) and has no Vulkan support. Display output capabilities are more modern on the GTX 950, with HDMI 2.0 and DisplayPort 1.2, versus the GTX 580’s older mini-HDMI 1.3a. Power efficiency is a major architectural win for the GTX 950: 90 W TDP versus 244 W, and a suggested PSU of 250 W versus 550 W.

The Verdict

The data points to the GTX 950 as the stronger card for modern workloads. It wins the only direct head-to-head benchmark (Geekbench OpenCL, 15,662 versus 15,283, a 2.4% margin), offers Vulkan support that the GTX 580 lacks, delivers higher FP32 throughput (1.825 TFLOPS versus 1.581 TFLOPS), and does so at a fraction of the power draw (90 W versus 244 W). For anyone choosing between these two for general compute, gaming, or API-compatible tasks, the GTX 950 is the safer pick.

The GTX 580, however, is not without merit. Its OpenCL score is within striking distance of the GTX 950, and it ties the RTX 2060 in the database, a signal of its raw compute capability. Its 384 bit memory bus and 192.4 GB/s bandwidth are superior to the GTX 950’s 128 bit bus and 105.8 GB/s, which could matter for bandwidth-sensitive applications. Its percentile rank of 58 is also higher than the GTX 950’s 53, though that is partly due to the GTX 950’s low Metal score dragging its average down.

Who should pick which? If the priority is modern API support, efficiency, and higher raw shader throughput, the GTX 950 is the clear choice. It wins the direct test, supports Vulkan, and uses far less power. If the priority is memory bandwidth and a higher percentile standing in the database, the GTX 580 offers a competitive OpenCL result and a wider memory bus, but it lacks Vulkan and draws 244 W. The recorded data favors the GTX 950 in the only direct comparison, and that is the deciding factor.