AMD Radeon HD 8870M vs NVIDIA GeForce GTX 1070 Comparison

FAQ

Q: How does the NVIDIA GeForce GTX 1070 compare to the AMD Radeon HD 8870M in average benchmark score?

A: The GTX 1070 records an average benchmark score of 9780, while the HD 8870M scores 8462. This places the GTX 1070 about 15.6% higher in overall performance, based on the recorded data.

Q: What is the percentile ranking of each GPU in the database?

A: The GTX 1070 sits at the 47th percentile among all GPUs, while the HD 8870M is at the 43rd percentile. This indicates both are near the middle of the performance distribution, though the GTX 1070 holds a slight edge.

Q: In the head-to-head benchmark, how large is the performance gap?

A: In the only shared test, Geekbench OpenCL, the GTX 1070 scores 44700 versus 8462 for the HD 8870M. That is a delta of 428.2%, meaning the GTX 1070 delivers more than five times the OpenCL performance.

Q: Which GPU has a higher transistor count?

A: The GTX 1070 uses 7,200 million transistors on a 314 mm² die, while the HD 8870M uses 1,500 million transistors on a 123 mm² die. The GTX 1070 has roughly 4.8 times the transistor count.

Q: What are the memory specifications for each card?

A: The GTX 1070 has 8 GB of GDDR5 memory on a 256-bit bus with 256.3 GB/s bandwidth. The HD 8870M has 2 GB of GDDR5 on a 128-bit bus with 72.00 GB/s bandwidth.

Q: Which GPU supports a newer DirectX version?

A: The GTX 1070 supports DirectX 12 (12_1), while the HD 8870M supports DirectX 12 (11_1). Both support OpenGL 4.6, but the GTX 1070 has a newer Vulkan version (1.4 versus 1.2.170).

Architecture Differences

The two GPUs represent vastly different eras and design philosophies. The NVIDIA GeForce GTX 1070 is built on the Pascal architecture using the GP104 chip, fabricated on a 16 nm process at TSMC. The AMD Radeon HD 8870M uses the GCN 1.0 architecture with the Venus chip, also from TSMC but on a larger 28 nm node. This process gap alone explains much of the efficiency and density difference.

The GTX 1070 packs 7,200 million transistors into a 314 mm² die, yielding a transistor density of 22.9M per mm². The HD 8870M has 1,500 million transistors on a 123 mm² die, a density of 12.2M per mm². The newer process allows the GTX 1070 to nearly double the density, which directly contributes to its higher clock speeds and larger compute resources.

Shader and texture hardware differ dramatically. The GTX 1070 has 1920 shading units, 120 texture mapping units (TMUs), and 64 render output units (ROPs). The HD 8870M has 640 shading units, 40 TMUs, and 16 ROPs. That is three times the shading units, three times the TMUs, and four times the ROPs on the NVIDIA part.

Memory architecture also sets them apart. The GTX 1070 uses a 256-bit bus with 8 GB of GDDR5, while the HD 8870M uses a 128-bit bus with 2 GB. The resulting bandwidth is 256.3 GB/s versus 72.00 GB/s, a 3.5x advantage for the GTX 1070. The GTX 1070 also runs its memory at 2002 MHz (8 Gbps effective) versus 1125 MHz (4.5 Gbps effective) on the HD 8870M.

Feature support shows the generational gap. The GTX 1070 supports DirectX 12 (12_1), Vulkan 1.4, and OpenGL 4.6. The HD 8870M supports DirectX 12 (11_1), Vulkan 1.2.170, and OpenGL 4.6. The Pascal card also lists a full set of display outputs (1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a), while the HD 8870M has no recorded display output information.

The GTX 1070 is a desktop dual-slot card with a 150 W TDP and a 1x 8-pin power connector. The HD 8870M is a mobile chip with no recorded TDP, slot width, or power connector data. The GTX 1070 measures 267 mm in length, 112 mm in height, and 40 mm in width. The HD 8870M has no recorded dimensions.

The Verdict

The data makes the choice straightforward for most workloads. The NVIDIA GeForce GTX 1070 wins the only head-to-head benchmark by 428.2% in Geekbench OpenCL, and its average benchmark score of 9780 compared to 8462 confirms a substantial overall lead. If the task involves compute, gaming, or any GPU-accelerated workload, the GTX 1070 is the clear pick.

However, context matters. The HD 8870M is a mobile part from 2013, while the GTX 1070 is a desktop card from 2016. The HD 8870M has no recorded TDP, making it likely a lower-power option for notebooks, but the database provides no power figures to confirm this. For anyone constrained to a laptop form factor, the HD 8870M might be the only option that physically fits, but the performance trade-off is severe.

For desktop users, the GTX 1070 is the only sensible choice based on the recorded metrics. It has more memory, more bandwidth, more shading units, and a newer architecture. The HD 8870M's 43rd percentile ranking places it just below the GTX 1070's 47th percentile, but the actual score gap is much larger than that percentile difference suggests.

Specification Differences

The two cards differ in nearly every measurable specification. The GTX 1070 uses a 16 nm process, while the HD 8870M uses 28 nm. Transistor count is 7,200 million versus 1,500 million. Die size is 314 mm² versus 123 mm². Transistor density is 22.9M per mm² versus 12.2M per mm².

Clock speeds favor the GTX 1070 significantly. Base clock is 1506 MHz versus 725 MHz. Boost clock is 1683 MHz versus 775 MHz. Memory clock is 2002 MHz (8 Gbps effective) versus 1125 MHz (4.5 Gbps effective).

Memory capacity is 8 GB versus 2 GB. Bus width is 256 bit versus 128 bit. Bandwidth is 256.3 GB/s versus 72.00 GB/s.

Compute resources differ by multiples. Shading units are 1920 versus 640. TMUs are 120 versus 40. ROPs are 64 versus 16. Pixel rate is 107.7 GPixel/s versus 12.40 GPixel/s. Texture rate is 202.0 GTexel/s versus 31.00 GTexel/s. FP32 performance is 6.463 TFLOPS versus 992.0 GFLOPS. The GTX 1070 also has an FP16 rating of 101.0 GFLOPS (1:64), while the HD 8870M has no recorded FP16 value.

The GTX 1070 has a 150 W TDP, is dual-slot, uses a 1x 8-pin power connector, and suggests a 450 W PSU. The HD 8870M has none of these recorded. The GTX 1070 lists dimensions of 267 mm by 112 mm by 40 mm, while the HD 8870M has none.

Release dates differ: the GTX 1070 launched on 2016-06-09, the HD 8870M on 2013-03-31. The GTX 1070 has a launch MSRP of 379 USD, while the HD 8870M has no recorded launch MSRP.

Head-to-Head Benchmarks

The database includes only one directly comparable benchmark between these two GPUs: Geekbench OpenCL. The results are lopsided. The GTX 1070 scores 44700, while the HD 8870M scores 8462. The delta is 428.2%, meaning the GTX 1070 is more than five times faster in this OpenCL compute test.

This single result dominates the comparison. The GTX 1070 wins 1 head-to-head benchmark, while the HD 8870M wins none. The margin is so large that it suggests the HD 8870M is not in the same performance class for compute-heavy tasks.

Looking at the broader benchmark data, the GTX 1070 shows a wide range of scores across different tests. It records 1082 in 3DMark Steel Nomad DX12, 18801 in Geekbench Metal, 22121 in Geekbench Vulkan, and 13498 in Passmark G3D. The HD 8870M has no data for those specific tests, so direct comparisons are limited to OpenCL.

The GTX 1070's nearest rivals in the database are the AMD FirePro W5000 (avg score 9803, delta -0.2%), NVIDIA Quadro M2000M (9832, -0.5%), NVIDIA Tesla M10 (9724, +0.6%), and NVIDIA Tesla C2070 (9716, +0.7%). These deltas are all within 1%, indicating the GTX 1070 sits in a tightly packed performance cluster.

The HD 8870M's nearest rivals are the NVIDIA GeForce MX330 (8458, +0.1%), AMD Radeon 880M (8436, +0.3%), NVIDIA GeForce GTX 675MX (8427, +0.4%), and Intel Arc A380 (8558, -1.1%). Again, the deltas are small, but the absolute scores are roughly 1300 points lower than the GTX 1070's cluster.

Where Each One Wins

The GTX 1070 wins in every recorded category. It has higher raw compute, more memory, faster memory, and a newer architecture. The only benchmark where both have data, Geekbench OpenCL, shows a 428.2% advantage for the GTX 1070. That is the definitive win condition.

For gaming and DirectX workloads, the GTX 1070's DirectX 12 (12_1) support and higher pixel rate (107.7 GPixel/s versus 12.40 GPixel/s) give it a clear edge. The texture rate of 202.0 GTexel/s versus 31.00 GTexel/s means the GTX 1070 can handle far more detailed scenes. The 8 GB memory capacity also allows higher resolution textures without running out of VRAM.

For compute tasks, the FP32 performance of 6.463 TFLOPS versus 992.0 GFLOPS is a 6.5x difference. The GTX 1070 also offers FP16 support at 101.0 GFLOPS, which the HD 8870M lacks entirely.

The HD 8870M's only potential advantage is its form factor. It is a mobile chip with no recorded TDP, meaning it could fit in laptops where the GTX 1070's dual-slot, 267 mm length, and 150 W TDP would be impossible. The HD 8870M also has an earlier release date, which might matter for legacy system compatibility, but the database does not record any performance advantage that would justify choosing it on merit.

In practical terms, the GTX 1070 is the winner for any desktop use case. The HD 8870M is only relevant for mobile systems where the GTX 1070 cannot physically be installed. Even then, the performance gap is so large that any workload requiring GPU acceleration would be severely limited on the HD 8870M. The data suggests the GTX 1070 is in a different performance tier entirely.