NVIDIA GeForce GTX 1070 Ti

Benchmark Performance

The NVIDIA GeForce GTX 1070 Ti delivers an average benchmark score of 14,711 across the full test suite, placing it in the 56th percentile of all GPUs tracked by the database. This is a mid-pack position that reflects a card designed for solid 1440p-class performance rather than top-tier dominance. The aggregate score is built from a wide spread of API-specific results, with the most telling figures coming from the DirectX and compute workloads.

In 3DMark Steel Nomad DX12, the card scores 1,471 points, which is a modest result for a Pascal-generation part. The DX12 API path appears to be a relative weak point, as the PassMark DirectX 12 score of just 51 confirms. By contrast, legacy and compute-bound workloads show far more strength. The PassMark DirectX 11 score of 107 is more than double the DX12 result, while the DirectX 9 score of 207 is roughly four times higher. This pattern indicates that the GTX 1070 Ti was architected for an era when DX11 and DX9 were the primary gaming APIs, and its DX12 performance, while functional, does not lead the field.

Compute performance is a brighter spot. The Geekbench OpenCL score of 55,295 and the Vulkan score of 59,703 both sit well above the card's average percentile position. These numbers suggest that the GP104 chip's 2,432 shading units and 8.186 TFLOPS of FP32 throughput are put to good use in general-purpose workloads. The Geekbench Metal score of 22,180 is lower, which is unsurprising given that Metal is primarily an Apple-centric API and this is a Windows-oriented card.

The PassMark G3D score of 14,674 closely tracks the overall average, while the GPU compute score of 7,165 indicates that the card handles compute shaders reasonably well. The PassMark G2D score of 876 is a minor outlier, showing that 2D desktop workloads are not where this card's strengths lie. The DirectX 10 score of 90 is another legacy result that outperforms the DX12 figure, reinforcing the generational split in API performance.

Comparing to its nearest rivals, the GTX 1070 Ti leads the NVIDIA GeForce GTX 770 by 1.2% in average score. That is a slim margin, but the two cards are separated by several generations of architecture. The GTX 1070 Ti also edges out the NVIDIA Quadro M2000 by 1.2%, the AMD Radeon Pro 555X by 1.5%, and the AMD Radeon RX 5500 XT by 2.2%. These deltas are all within a narrow band, meaning the GTX 1070 Ti is not dramatically faster than any of these competitors in aggregate terms. However, the margin over the RX 5500 XT, a much newer card, is worth noting: a 2.2% lead in average score is respectable for a 2017 part facing a 2019 release.

Who Should Consider It

The benchmark data points to a card that is best suited for 1080p and 1440p gaming at high settings, with the caveat that DX12 titles will not extract maximum performance. The strong DX9 and DX11 scores suggest that players of older titles, esports games, and the many still-popular DX11 releases will see consistently high frame rates. The 8.186 TFLOPS of FP32 compute and the 107.7 GPixel/s pixel rate provide enough headroom for 1440p rendering without requiring the user to drop to medium presets in most games.

For 4K resolutions, the data is less encouraging. The modest DX12 score of 51 in PassMark and the 1,471-point Steel Nomad result indicate that the GTX 1070 Ti will struggle with the most demanding modern titles at 4K, particularly those that rely heavily on DX12 features. Users targeting 4K should look at higher-tier cards, as the 256-bit memory bus and 256.3 GB/s bandwidth, while adequate for 1440p, become a limiting factor at higher pixel counts.

The card's 180 W TDP and 450 W suggested PSU rating make it a practical choice for systems with mid-range power supplies. The dual-slot cooler and 267 mm length mean it will fit in most standard ATX cases, though smaller form-factor builds should check clearance. The 1x 8-pin power connector is a single, common connection that most modern PSUs provide without adapters.

Memory Subsystem

The GTX 1070 Ti comes equipped with 8 GB of GDDR5 memory on a 256-bit bus, yielding a total bandwidth of 256.3 GB/s. The memory clock runs at 2002 MHz, which translates to 8 Gbps effective. This configuration was standard for the high-end of the Pascal generation and remains adequate for 1440p gaming, where texture and geometry data typically fit within the 8 GB frame buffer.

At 4K resolutions, the memory subsystem becomes more of a constraint. The 256.3 GB/s bandwidth is sufficient for most current titles at high settings, but games with heavy texture streaming or large open worlds can push against this limit. The 8 GB capacity is the larger concern: while it handles today's games, future titles that require more than 8 GB for ultra textures will force the card to rely on slower system memory over PCIe, causing stutter and reduced frame pacing.

The 256-bit bus width is a balanced choice for this performance class. It provides enough bandwidth for the 8.186 TFLOPS of compute throughput without inflating the die cost. The pixel rate of 107.7 GPixel/s and texture rate of 255.8 GTexel/s are well matched to the memory bandwidth, indicating that the card does not have a severe bottleneck in either direction. For users who play at 1080p or 1440p, the memory subsystem will not be the limiting factor; for 4K enthusiasts, it is the first spec to scrutinize.

FAQ

Q: How does the GTX 1070 Ti compare to the GTX 770 in average benchmark score?

A: The GTX 1070 Ti averages 14,711 points, which is 1.2% higher than the GTX 770's average of 14,536.

Q: What is the card's best-performing API according to the data?

A: The PassMark DirectX 9 score of 207 is the highest among the DirectX tests, followed by DirectX 11 at 107. The DirectX 12 score of 51 is the lowest.

Q: Does the card support Vulkan?

A: Yes, the card supports Vulkan 1.4, and its Geekbench Vulkan score is 59,703, which is higher than its OpenCL score of 55,295.

Q: What is the memory configuration?

A: The GTX 1070 Ti has 8 GB of GDDR5 memory on a 256-bit bus, with 256.3 GB/s of bandwidth and an effective memory clock of 8 Gbps.

Q: What is the launch MSRP?

A: The launch MSRP is 399 USD.

Q: Is the card still in production?

A: No, the production status is listed as end-of-life, with a release date of 2017-11-01.

How It Compares

NVIDIA GeForce GTX 770 — The GTX 1070 Ti leads this older card by a narrow 1.2% in average score (14,711 vs. 14,536). That margin is surprisingly small given the architectural gap, but the GTX 770's strong DX9 and DX11 performance keeps it competitive in the aggregate. The GTX 1070 Ti's advantage becomes clearer in compute workloads, where its 8.186 TFLOPS and 7,165 PassMark compute score outclass the older part.

NVIDIA Quadro M2000 — The margin here is also 1.2%, with the GTX 1070 Ti averaging 14,711 against the Quadro's 14,532. The Quadro is a workstation card, so its driver optimizations favor professional applications rather than gaming. The GTX 1070 Ti's gaming-oriented feature set, including 2,432 shading units and higher pixel rate, makes it the more versatile choice for mixed-use systems.

AMD Radeon Pro 555X — The GTX 1070 Ti leads this mobile-class GPU by 1.5%, with the Pro 555X averaging 14,500. The Pro 555X is a lower-power part designed for laptops, so its performance per watt is different from the desktop GTX 1070 Ti. The 1.5% delta is small enough that real-world differences will be minimal in most workloads, but the GTX 1070 Ti's 8 GB of GDDR5 and 256-bit bus give it an edge in memory-heavy scenarios.

AMD Radeon RX 5500 XT — The largest gap among the nearest rivals is here: the GTX 1070 Ti leads by 2.2%, averaging 14,711 vs. 14,389. The RX 5500 XT is a newer card, but the GTX 1070 Ti's higher transistor count (7,200 million) and larger die size (314 mm²) provide more raw compute. The GTX 1070 Ti also has a higher memory bandwidth (256.3 GB/s) and a wider 256-bit bus, which helps at 1440p. This 2.2% lead is modest, but it shows that the older Pascal architecture still holds its own against a more recent competitor in aggregate benchmarks.