AMD Radeon R5 Graphics vs NVIDIA GeForce GTX 650 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon R5 Graphics edges out the NVIDIA GeForce GTX 650 with an average benchmark score of 3883 versus 3823. The delta is small — the GTX 650 trails by roughly 1.5% in average score. Both sit in the bottom quartile of all GPUs, with the R5 at the 23rd percentile and the GTX 650 at the 22nd.

Q: How do the two compare in OpenCL performance?

A: The AMD Radeon R5 Graphics wins the OpenCL test decisively, scoring 5183 against the GTX 650’s 4545. That is a 14% advantage for the AMD part. This is the R5’s single biggest win in head-to-head testing.

Q: Which GPU wins in Vulkan performance?

A: The NVIDIA GeForce GTX 650 dominates in Vulkan, scoring 4524 versus the Radeon R5’s 2582. The delta is -42.9% from AMD’s perspective, meaning the GTX 650 is roughly 75% faster in this API. This is a massive reversal from the OpenCL result.

Q: What are the memory configurations of these two GPUs?

A: The AMD Radeon R5 Graphics uses system shared memory, with size, type, bus width, and bandwidth all listed as "System Shared" or "System Dependent." The NVIDIA GeForce GTX 650 has 1024 MB of dedicated GDDR5 memory on a 128-bit bus, delivering 80.00 GB/s of bandwidth at 1250 MHz (5 Gbps effective).

Q: Which card has more shading units and texture mapping units?

A: The NVIDIA GeForce GTX 650 has 384 shading units and 32 TMUs, while the AMD Radeon R5 Graphics has 256 shading units and 16 TMUs. The GTX 650 also has 16 ROPs versus the R5’s 4 ROPs.

Q: What is the power draw difference between the two?

A: The AMD Radeon R5 Graphics is rated at 15 W TDP and is an integrated GPU (IGP) with no power connectors. The NVIDIA GeForce GTX 650 is a 65 W card requiring a single 6-pin power connector and a suggested 250 W power supply.

Architecture Differences

The AMD Radeon R5 Graphics is built on GCN 2.0 architecture, specifically the "Spectre SL" chip, fabricated by GlobalFoundries on a 28 nm process. It belongs to the GCN 2.0 IGP generation (Kaveri) and integrates 2,410 million transistors on a 245 mm² die, giving a transistor density of 9.8M per mm². The GTX 650, in contrast, uses NVIDIA’s Kepler architecture with the GK106 chip, manufactured by TSMC on the same 28 nm node. It packs 2,540 million transistors into a smaller 221 mm² die, achieving a higher density of 11.5M per mm².

Memory architecture is a fundamental split. The R5 is an IGP with system-shared memory — no dedicated VRAM, no fixed bus width, and bandwidth that depends entirely on the host system’s RAM. The GTX 650 has 1024 MB of GDDR5 on a 128-bit bus with a fixed 80.00 GB/s bandwidth. This is a structural difference: the R5’s memory performance is variable and system-dependent, while the GTX 650’s is deterministic.

Compute resources differ sharply. The R5 has 256 shading units, 16 TMUs, and only 4 ROPs. The GTX 650 has 384 shading units, 32 TMUs, and 16 ROPs. These resource counts explain the GTX 650’s higher theoretical pixel rate (8.464 GPixel/s versus 3.032 GPixel/s) and texture rate (33.86 GTexel/s versus 12.13 GTexel/s). FP32 throughput also favors NVIDIA: 812.5 GFLOPS versus 388.1 GFLOPS.

API support is a notable divergence. The R5 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.2.170. The GTX 650 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The DirectX feature level differs — 12_0 for AMD versus 11_0 for NVIDIA — which affects modern API capabilities.

Physical and integration differences are stark. The R5 is an IGP with motherboard-dependent display outputs and no slot width. The GTX 650 is a single-slot card measuring 147 mm (5.8 inches) in length, with 1x DVI and 2x DisplayPort 1.2 outputs, and connects via PCIe 3.0 x16.

The Verdict

The data presents a split decision. In raw compute throughput and traditional rasterization metrics, the GTX 650 is the clear hardware winner — it has 50% more shading units, double the TMUs, four times the ROPs, and more than double the FP32 throughput. Its dedicated GDDR5 memory with 80.00 GB/s bandwidth is a categorical advantage over the R5’s system-shared memory.

However, benchmark results tell a more nuanced story. The R5 wins OpenCL decisively (5183 versus 4545, a 14% margin), while the GTX 650 crushes the R5 in Vulkan (4524 versus 2582, a 42.9% swing). The average benchmark scores are nearly identical — 3883 for the R5 versus 3823 for the GTX 650 — placing both in the same performance tier.

Who should pick which? If the workload is OpenCL-centric and the system has fast shared memory, the R5 is competitive despite being an IGP. If the workload uses Vulkan or requires consistent memory bandwidth, the GTX 650 is the safer choice. For gaming or GPU compute in a desktop with a dedicated power supply, the GTX 650’s hardware resources and memory subsystem make it the more capable all-rounder. For an ultra-low-power (15 W) integrated solution in a laptop or compact system, the R5 is the only option that fits.

Specification Differences

| Specification | AMD Radeon R5 Graphics | NVIDIA GeForce GTX 650 |

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

| Architecture | GCN 2.0 | Kepler |

| Chip | Spectre SL | GK106 |

| Process Node | 28 nm (GlobalFoundries) | 28 nm (TSMC) |

| Transistors | 2,410 million | 2,540 million |

| Die Size | 245 mm² | 221 mm² |

| Transistor Density | 9.8M / mm² | 11.5M / mm² |

| Shading Units | 256 | 384 |

| TMUs | 16 | 32 |

| ROPs | 4 | 16 |

| Memory Size | System Shared | 1024 MB |

| Memory Type | System Shared | GDDR5 |

| Memory Bus Width | System Shared | 128 bit |

| Memory Bandwidth | System Dependent | 80.00 GB/s |

| Memory Clock | System Shared | 1250 MHz (5 Gbps effective) |

| Pixel Rate | 3.032 GPixel/s | 8.464 GPixel/s |

| Texture Rate | 12.13 GTexel/s | 33.86 GTexel/s |

| FP32 | 388.1 GFLOPS | 812.5 GFLOPS |

| TDP | 15 W | 65 W |

| Slot Width | IGP | Single-slot |

| Power Connectors | None | 1x 6-pin |

| Suggested PSU | None | 250 W |

| Bus Interface | IGP | PCIe 3.0 x16 |

| Display Outputs | Motherboard Dependent | 1x DVI, 2x DisplayPort 1.2 |

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

| Vulkan | 1.2.170 | 1.2.175 |

| Length | N/A | 147 mm (5.8 inches) |

| Release Date | 2014-09-16 | 2013-11-26 |

| Predecessor | TeraScale 3 IGP | GeForce 500 |

| Successor | GCN 3.0 IGP | GeForce 700 |

Head-to-Head Benchmarks

The head-to-head data contains only two benchmark entries, and they split cleanly between the two GPUs.

OpenCL — AMD Radeon R5 Graphics wins. The R5 scores 5183, the GTX 650 scores 4545. The delta is 14% in AMD’s favor. This is a significant margin, especially considering the GTX 650’s superior hardware specifications — more shading units, more TMUs, more ROPs, and a dedicated memory bus. The R5’s win here suggests that its GCN 2.0 architecture is better optimized for OpenCL workloads, or that the system-shared memory configuration benefits from lower latency in certain compute patterns. Notably, the R5’s nearest rivals in average score include the NVIDIA Quadro 2000 (3898, -0.4% delta) and Quadro K2000D (3919, -0.9% delta), placing it in solid workstation-class company.

Vulkan — NVIDIA GeForce GTX 650 wins. The GTX 650 scores 4524, the R5 scores 2582. The delta is -42.9% from AMD’s perspective. This is a landslide victory for NVIDIA. The GTX 650’s Vulkan score is nearly double the R5’s, and the margin is roughly three times larger than the R5’s OpenCL advantage. This suggests Kepler’s Vulkan driver implementation is far more mature or efficient than GCN 2.0’s on this workload. The GTX 650’s nearest rivals include the GeForce MX110 (3834, -0.3% delta) and the Intel UHD Graphics 710 (3792, 0.8% delta), showing that it sits in the same performance band as modern entry-level parts in average terms.

The average benchmark score for each GPU reflects these mixed results. The R5 averages 3883; the GTX 650 averages 3823. The R5’s nearest rival is the Quadro 2000 at 3898 (-0.4% delta), while the GTX 650’s nearest rival is the MX110 at 3834 (-0.3% delta). Both GPUs sit in the bottom quarter of all GPUs — 23rd percentile for the R5, 22nd for the GTX 650 — meaning neither is a high-performance part by modern standards, but they are closely matched overall.

Where Each One Wins

AMD Radeon R5 Graphics wins in OpenCL compute workloads. The 14% OpenCL advantage is the R5’s only head-to-head victory, but it is a meaningful one for users running OpenCL-based applications — image processing, physics simulation, or general-purpose GPU compute that leverages this API. The R5 also wins on power efficiency from a system perspective: at 15 W TDP with no power connectors, it fits into designs where a 65 W card with a 6-pin connector is impossible. For ultra-portable or low-power systems, the R5 is the only viable choice between these two. Its GCN 2.0 architecture with 256 shading units and 388.1 GFLOPS is sufficient for light compute tasks, and the system-shared memory means no dedicated VRAM allocation is required. The R5 also holds a slight edge in average benchmark score (3883 versus 3823) and percentile ranking (23rd versus 22nd).

NVIDIA GeForce GTX 650 wins in Vulkan applications and raw rasterization throughput. The Vulkan score of 4524 versus 2582 is a 75% advantage — a massive gap that suggests the GTX 650 is the better choice for Vulkan-based games or compute. Its hardware resources are categorically superior: 384 shading units, 32 TMUs, 16 ROPs, 812.5 GFLOPS FP32, 8.464 GPixel/s pixel rate, and 33.86 GTexel/s texture rate. The dedicated 1024 MB GDDR5 on a 128-bit bus with 80.00 GB/s bandwidth eliminates the memory bottleneck that system-shared memory imposes on the R5. The GTX 650 also has a higher transistor density (11.5M per mm² versus 9.8M) and a slightly newer Vulkan version (1.2.175 versus 1.2.170). For desktop users with a power supply capable of 250 W, the GTX 650 is the stronger all-around performer for gaming and graphics-intensive tasks. Its 147 mm single-slot form factor with DVI and dual DisplayPort 1.2 outputs provides flexible display connectivity, whereas the R5’s outputs depend entirely on the motherboard.