AMD Radeon HD 8870M vs NVIDIA GeForce GTX 650 Ti Boost Comparison

# Head-to-Head Benchmarks

The only direct benchmark comparison available is Geekbench OpenCL, and the data is unambiguous: the NVIDIA GeForce GTX 650 Ti Boost wins decisively. It scores 9,302 points against the AMD Radeon HD 8870M’s 8,462, a 9% advantage. This is not a marginal victory — it places the GTX 650 Ti Boost nearly a full tier above its rival in raw compute throughput. In practical terms, the NVIDIA card delivers roughly 10% more OpenCL performance, which translates into measurable gains in GPU-accelerated workloads such as OpenCL-based rendering, physics simulation, and compute shaders.

The delta is consistent with the underlying hardware. The GTX 650 Ti Boost carries 768 shading units compared to 640 on the HD 8870M, and its 1.585 TFLOPS FP32 throughput eclipses the AMD part’s 992.0 GFLOPS by roughly 60%. The benchmark gap is narrower than the theoretical compute gap suggests, which indicates that the HD 8870M’s GCN 1.0 architecture extracts efficiency from its 1,500 million transistors better than the Kepler design does from its 2,540 million. But the verdict remains: in the one head-to-head test available, the GTX 650 Ti Boost wins outright, with no benchmark category favoring the AMD card.

# Architecture Differences

The two GPUs come from fundamentally different design philosophies. The AMD Radeon HD 8870M uses the Venus chip built on GCN 1.0 architecture, part of the Solar System generation (HD 8800M series). The NVIDIA GeForce GTX 650 Ti Boost uses the GK106 chip on the older Kepler architecture, belonging to the GeForce 600 generation. Both are fabricated on TSMC’s 28 nm process node, but the similarities end there.

Die size and transistor counts diverge sharply. NVIDIA’s GK106 packs 2,540 million transistors onto a 221 mm² die, while AMD’s Venus fits 1,500 million onto a much smaller 123 mm² die. This gives the AMD chip a higher transistor density at 12.2M per mm² versus NVIDIA’s 11.5M per mm². The smaller AMD die suggests lower manufacturing cost potential and better thermal efficiency per area, though the GTX 650 Ti Boost compensates with sheer silicon scale.

Memory subsystems differ substantially. The HD 8870M uses a 128-bit memory bus with 2 GB of GDDR5 running at 4.5 Gbps effective, yielding 72.00 GB/s bandwidth. The GTX 650 Ti Boost widens the bus to 192-bit, also with 2 GB GDDR5 but at 6 Gbps effective, nearly doubling bandwidth to 144.2 GB/s. This 2x bandwidth advantage is critical for texture-heavy workloads and high-resolution framebuffer operations.

Compute resources also favor NVIDIA numerically. The GTX 650 Ti Boost has 768 shading units, 64 texture mapping units, and 24 ROPs, versus 640 shaders, 40 TMUs, and 16 ROPs on the HD 8870M. Pixel rate is 16.51 GPixel/s versus 12.40 GPixel/s, and texture rate is 66.05 GTexel/s versus 31.00 GTexel/s. The NVIDIA card’s texture rate more than doubles AMD’s, a reflection of its 64 TMUs.

Power and form factor tell another story. The GTX 650 Ti Boost is a dual-slot card rated at 134 W TDP, requiring a 1x 6-pin power connector and a 300 W suggested PSU. It measures 241 mm (9.5 inches) in length. The HD 8870M has no listed TDP, slot width, or power connector data — consistent with it being a mobile-oriented part. The HD 8870M also has no listed display outputs, while the GTX 650 Ti Boost offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

API support is close but not identical. The HD 8870M supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The GTX 650 Ti Boost supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. Both are end-of-life products, with release dates six days apart in late March 2013 — the GTX 650 Ti Boost on March 25, the HD 8870M on March 31. The HD 8870M’s predecessor is London and successor is Gem System; the GTX 650 Ti Boost’s predecessor is GeForce 500 and successor is GeForce 700.

# FAQ

Q: Which GPU has higher raw compute performance?

A: The NVIDIA GeForce GTX 650 Ti Boost. It delivers 1.585 TFLOPS FP32 versus the AMD Radeon HD 8870M’s 992.0 GFLOPS, and it wins the Geekbench OpenCL test 9,302 to 8,462.

Q: Does the AMD card win any benchmark comparison?

A: No. In the single head-to-head benchmark available (Geekbench OpenCL), the HD 8870M loses by 9%. There are zero benchmark wins for the AMD card in the data.

Q: How do memory bandwidth figures compare?

A: The GTX 650 Ti Boost offers 144.2 GB/s over a 192-bit bus, exactly double the HD 8870M’s 72.00 GB/s over a 128-bit bus. Both use 2 GB of GDDR5 memory.

Q: Are these cards from the same generation?

A: No. The HD 8870M is from the Solar System generation (HD 8800M series) using GCN 1.0 architecture, while the GTX 650 Ti Boost is from the GeForce 600 generation using Kepler. Both were released within six days of each other in late March 2013.

Q: Which GPU has better API compatibility?

A: The GTX 650 Ti Boost has a slightly newer Vulkan version (1.2.175 versus 1.2.170). The HD 8870M supports DirectX 12 (11_1) versus the GTX 650 Ti Boost’s DirectX 12 (11_0). Both support OpenGL 4.6.

Q: What is the transistor density difference?

A: The AMD Venus chip packs 1,500 million transistors into 123 mm² for a density of 12.2M per mm². The NVIDIA GK106 packs 2,540 million into 221 mm² for 11.5M per mm². AMD’s chip is denser despite being smaller overall.

# Specification Differences

| Specification | AMD Radeon HD 8870M | NVIDIA GeForce GTX 650 Ti Boost |

|---|---|---|

| Chip | Venus | GK106 |

| Architecture | GCN 1.0 | Kepler |

| Generation | Solar System (HD 8800M) | GeForce 600 |

| Transistors | 1,500 million | 2,540 million |

| Die Size | 123 mm² | 221 mm² |

| Transistor Density | 12.2M / mm² | 11.5M / mm² |

| Base Clock | 725 MHz | 980 MHz |

| Boost Clock | 775 MHz | 1032 MHz |

| Memory Clock | 1125 MHz (4.5 Gbps effective) | 1502 MHz (6 Gbps effective) |

| Memory Bus Width | 128 bit | 192 bit |

| Memory Bandwidth | 72.00 GB/s | 144.2 GB/s |

| Shading Units | 640 | 768 |

| TMUs | 40 | 64 |

| ROPs | 16 | 24 |

| Pixel Rate | 12.40 GPixel/s | 16.51 GPixel/s |

| Texture Rate | 31.00 GTexel/s | 66.05 GTexel/s |

| FP32 | 992.0 GFLOPS | 1.585 TFLOPS |

| TDP | Not listed | 134 W |

| Slot Width | Not listed | Dual-slot |

| Power Connectors | Not listed | 1x 6-pin |

| Suggested PSU | Not listed | 300 W |

| Length | Not listed | 241 mm (9.5 inches) |

| Display Outputs | Not listed | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 |

| DirectX | 12 (11_1) | 12 (11_0) |

| Vulkan | 1.2.170 | 1.2.175 |

| Predecessor | London | GeForce 500 |

| Successor | Gem System | GeForce 700 |

| Release Date | March 31, 2013 | March 25, 2013 |

| Launch MSRP | Not listed | 169 USD |

# The Verdict

The benchmark data is clear: the NVIDIA GeForce GTX 650 Ti Boost is the stronger GPU. It wins the only head-to-head test with a 9% margin, carries 128 more shading units, doubles memory bandwidth, and more than doubles texture throughput. Its 1.585 TFLOPS FP32 performance is 60% higher than the HD 8870M’s 992.0 GFLOPS. The NVIDIA card also holds a higher percentile versus all GPUs in its nearest rival group — it sits at the 42nd percentile with an average benchmark score of 8,067, while the AMD card sits at the 43rd percentile with an average of 8,462. Interestingly, the AMD part has a higher average score across its own rival group, but that does not overcome the direct comparison deficit.

The HD 8870M’s advantages are architectural efficiency and compactness. Its 28 nm chip is denser (12.2M transistors per mm² versus 11.5M), and its GCN 1.0 design achieves 43rd percentile standing despite lower absolute specs. For a mobile part with no listed TDP, this efficiency is notable. But efficiency without a benchmark win is a consolation prize. The GTX 650 Ti Boost’s 134 W TDP and dual-slot cooler requirement are the cost of its performance — a cost that desktop users with a 300 W PSU can easily absorb.

# Where Each One Wins

NVIDIA GeForce GTX 650 Ti Boost wins in every measurable performance category. It dominates in raw compute with 1.585 TFLOPS FP32, doubles memory bandwidth at 144.2 GB/s, and nearly doubles texture rate at 66.05 GTexel/s. Its 16.51 GPixel/s pixel rate outpaces the AMD card’s 12.40 GPixel/s. For desktop users building a system around OpenCL compute, the GTX 650 Ti Boost’s 9,302 OpenCL score is the stronger choice. It also benefits from a 192-bit memory bus, which historically helps in bandwidth-sensitive scenarios like high-resolution texture streaming and multi-monitor setups. Its display output suite (2x DVI, HDMI 1.4a, DisplayPort 1.2) gives it immediate connectivity options that the AMD card lacks entirely in the data.

AMD Radeon HD 8870M wins in efficiency-oriented contexts. Its smaller 123 mm² die and higher transistor density suggest better thermal behavior per unit area. With no TDP listed, it is positioned as a mobile or low-power part, making it suitable for compact systems where the GTX 650 Ti Boost’s 134 W draw and dual-slot cooler would be prohibitive. The HD 8870M also achieves a higher average benchmark score (8,462) than the GTX 650 Ti Boost’s average (8,067) when measured across their respective nearest rival groups — the AMD card’s average rivals the NVIDIA GeForce MX330 and AMD Radeon 880M, while the GTX 650 Ti Boost’s average clusters with the GRID K2 and GTX 650 Ti. In the Geekbench OpenCL test specifically, the HD 8870M’s 8,462 score is within 0.1% of the MX330 and 0.3% of the Radeon 880M, showing it remains competitive in its class despite losing to the GTX 650 Ti Boost.

The choice depends on the platform. Desktop builders with power headroom should pick the GTX 650 Ti Boost without hesitation — it wins the benchmark, has superior specs, and offers a full display output array. Mobile or space-constrained users should consider the HD 8870M, which delivers competitive OpenCL performance in a denser package. The data does not support any scenario where the AMD card outperforms the NVIDIA card in raw speed, but it does support the HD 8870M as a more efficient alternative for systems where the GTX 650 Ti Boost’s power and size requirements are deal-breakers.