NVIDIA GeForce GT 1010 vs NVIDIA GeForce GTX 980 Comparison

The NVIDIA GeForce GTX 980 and NVIDIA GeForce GT 1010 occupy opposite ends of the discrete GPU spectrum, and the recorded data makes the gap between them unmistakable. The GTX 980, a Maxwell 2.0 flagship from the GeForce 900 generation, sits at the 43rd percentile of all GPUs in the database, while the GT 1010, an entry-level Pascal card from the GeForce 10 generation, sits at the 38th percentile. A single direct head-to-head benchmark exists between them, and it is a rout: the GTX 980 wins Geekbench OpenCL by 417.7 percent. What follows is a full breakdown of where each card stands.

The Verdict

The data supports one clear recommendation: anyone choosing between these two cards for performance should take the GTX 980. Its average benchmark score of 8167 is well ahead of the GT 1010's 6698, and the only direct comparison in the database, Geekbench OpenCL, shows the GTX 980 at 34676 against 6698, a 417.7 percent margin. The GTX 980 also posts strong results across a wide suite of tests, including PassMark G3D at 11095, Geekbench Vulkan at 22543, Geekbench Metal at 15163, and 3DMark Steel Nomad DX12 at 474.

The GT 1010's case rests entirely on platform characteristics rather than speed. It draws just 30 W against the GTX 980's 165 W, needs no auxiliary power connectors where the GTX 980 requires 2x 6-pin, occupies a single slot versus dual-slot, and is only 147 mm long versus the GTX 980's 267 mm. Its suggested PSU is 200 W compared to the GTX 980's 450 W. In other words, the GT 1010 fits where the GTX 980 physically and electrically cannot. But as a performance proposition, the database shows no contest, with the GTX 980 winning the sole head-to-head benchmark and the GT 1010 winning none.

Architecture Differences

These cards bracket an entire architectural transition. The GTX 980 uses the GM204 chip built on Maxwell 2.0, fabricated at TSMC on a 28 nm process, with 5,200 million transistors packed into a 398 mm² die at a density of 13.1M per mm². The GT 1010 uses the GP108 chip on Pascal, fabricated at Samsung on a 14 nm process, with 1,800 million transistors in a 74 mm² die at 24.3M per mm². The Pascal part's density advantage is nearly double, but its die is roughly a fifth the size and its transistor budget is less than half, which explains most of the performance gap on its own.

The resource counts diverge sharply. The GTX 980 fields 2048 shading units, 128 TMUs, and 64 ROPs. The GT 1010 has 256 shading units, 16 TMUs, and 8 ROPs, exactly one-eighth the shading hardware and one-eighth the ROP count of the bigger card. Neither GPU has RT cores or tensor cores, so neither carries hardware-accelerated ray tracing or AI features.

Memory is another decisive difference. The GTX 980 pairs 4 GB of GDDR5 on a 256-bit bus running at 1753 MHz (7 Gbps effective), delivering 224.4 GB/s of bandwidth. The GT 1010 has 2 GB of GDDR5 on a narrow 64-bit bus at 1502 MHz (6 Gbps effective), yielding just 48.06 GB/s. That is less than a quarter of the GTX 980's bandwidth, a constraint that compounds the smaller frame buffer.

Clocks favor the newer, smaller chip: the GT 1010 boosts to 1468 MHz from a 1228 MHz base, while the GTX 980 boosts to 1216 MHz from 1127 MHz. But the clock advantage is nowhere near enough to offset the resource deficit. The theoretical output numbers tell the story: the GTX 980 delivers 77.82 GPixel/s of pixel rate, 155.6 GTexel/s of texture rate, and 4.981 TFLOPS of FP32 compute. The GT 1010 manages 11.74 GPixel/s, 23.49 GTexel/s, and 751.6 GFLOPS, each figure under a sixth of the Maxwell card's.

Feature support is essentially identical on paper. Both report DirectX 12 feature level 12_1, OpenGL 4.6, and Vulkan 1.4, and both use PCIe 3.0, though the GTX 980 gets a full x16 link while the GT 1010 is limited to x4. Both are end-of-life products now.

Head-to-Head Benchmarks

The database contains only one test run on both cards: Geekbench OpenCL. The GTX 980 scored 34676; the GT 1010 scored 6698. The delta is 417.7 percent in the GTX 980's favor, meaning the Maxwell card delivers more than five times the OpenCL compute throughput of the Pascal entry card. That single result aligns almost perfectly with the theoretical FP32 gap, 4.981 TFLOPS against 751.6 GFLOPS, so the OpenCL figure looks representative rather than anomalous.

The GTX 980's broader record reinforces the picture. Its PassMark G3D score of 11095 and PassMark GPU Compute score of 4753 sit alongside per-API PassMark results of 164 for DirectX 9, 83 for DirectX 11, 53 for DirectX 10, and 46 for DirectX 12, plus a PassMark G2D result of 792. Its Geekbench Metal score of 15163 and Geekbench Vulkan score of 22543 round out a full test suite. The GT 1010 has only the one OpenCL result in the database, which limits direct comparison but also underscores how little performance data the card has generated.

Context from each card's rival set sharpens the contrast. The GTX 980's average score of 8167 places it just 0.4 percent above the NVIDIA GeForce GTX 950M (8135), 0.5 percent above the AMD Radeon R9 M360 (8129), 0.8 percent above the NVIDIA GeForce 945M (8099), and 1.1 percent above the NVIDIA GRID K2 (8080). The GT 1010's 6698 average sits 1 percent below the AMD Radeon R7 M370 (6764), 1.3 percent above the AMD Radeon R7 M460 (6612), 1.8 percent above the AMD Radeon HD 7730M (6581), and 1.9 percent below the AMD FirePro M5100 (6830). Notably, the GTX 980's closest rivals are largely mobile parts, which speaks to how GPU performance has moved since its launch.

Specification Differences

• Chip and architecture: GM204 on Maxwell 2.0 (GeForce 900) versus GP108 on Pascal (GeForce 10).
• Process: 28 nm at TSMC versus 14 nm at Samsung; 398 mm² and 5,200 million transistors versus 74 mm² and 1,800 million.
• Shading resources: 2048 shading units, 128 TMUs, 64 ROPs versus 256, 16, and 8.
• Clocks: 1127 MHz base / 1216 MHz boost versus 1228 MHz base / 1468 MHz boost.
• Memory: 4 GB GDDR5, 256-bit, 224.4 GB/s versus 2 GB GDDR5, 64-bit, 48.06 GB/s.
• Output: 4.981 TFLOPS FP32, 77.82 GPixel/s, 155.6 GTexel/s versus 751.6 GFLOPS, 11.74 GPixel/s, 23.49 GTexel/s.
• Power: 165 W TDP with 2x 6-pin connectors and a 450 W suggested PSU versus 30 W TDP, no power connectors, and a 200 W suggested PSU.
• Interface: PCIe 3.0 x16 versus PCIe 3.0 x4.
• Outputs: 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2 versus 1x DVI, 1x mini-HDMI 2.0.
• Size: Dual-slot, 267 mm long, 111 mm tall, 40 mm wide versus single-slot, 147 mm long.
• Release: September 2014 with a launch MSRP of 549 USD versus January 2021 with no launch MSRP recorded.
• Lineage: the GTX 980 follows GeForce 700 and precedes GeForce 10; the GT 1010 follows GeForce 900 and precedes GeForce 20.

FAQ

Q: Which GPU is faster according to the benchmarks?

A: The GTX 980, decisively. It won the only direct head-to-head test, Geekbench OpenCL, 34676 to 6698, a 417.7 percent margin, and it leads on average benchmark score 8167 to 6698.

Q: How much more compute throughput does the GTX 980 have?

A: Its FP32 output is rated at 4.981 TFLOPS versus 751.6 GFLOPS for the GT 1010, roughly 6.6 times more raw shading compute.

Q: Does the GT 1010 have any hardware advantages?

A: Yes. It boosts higher (1468 MHz versus 1216 MHz), draws 30 W instead of 165 W, needs no power connectors, fits a single slot, and measures 147 mm versus 267 mm, making it viable in compact, power-limited builds.

Q: Do both cards support the same APIs?

A: Yes. Both report DirectX 12 (feature level 12_1), OpenGL 4.6, and Vulkan 1.4. Neither has RT cores or tensor cores.

Q: How do the two cards compare to other GPUs in the database?

A: The GTX 980 ranks at the 43rd percentile against all GPUs and the GT 1010 at the 38th percentile. The GTX 980's closest rival is the GTX 950M at just 0.4 percent behind, while the GT 1010 trades blows with the Radeon R7 M370, R7 M460, HD 7730M, and FirePro M5100.

Q: Are both cards still in production?

A: No. Both are listed as end-of-life. The GTX 980 launched in September 2014 with a launch MSRP of 549 USD; the GT 1010 launched in January 2021 with no launch MSRP recorded in the database.