AMD Radeon HD 8730M vs NVIDIA GeForce GTX 580M Comparison

Where Each One Wins

The benchmark record splits cleanly in favor of the NVIDIA GeForce GTX 580M. In the single recorded OpenCL workload, the GTX 580M posts a score of 6389, while the AMD Radeon HD 8730M trails at 5955. That is a 7.3% advantage for the NVIDIA part. The GTX 580M claims the only win in the head-to-head comparison, with a 1-0 record.

This places the GTX 580M in the 37th percentile of all GPUs in the database, while the HD 8730M sits in the 34th percentile. The gap is modest in absolute terms, but the direction is consistent: the older NVIDIA part outperforms the newer AMD chip in this particular compute workload. The GTX 580M's nearest rival is the NVIDIA GeForce GTX 460 SE, which scores identically at 6389, and it edges out the AMD Radeon Pro WX 4100 by 0.9%. The AMD Radeon R7 M350 is 1% behind. On the AMD side, the HD 8730M sits nearly level with the NVIDIA Quadro K620M, which scores 5957 (a 0% delta), and it beats the Intel UHD Graphics 730 by 0.4%. The NVIDIA Quadro K4000 is 0.5% ahead of the HD 8730M.

The practical takeaway: if the workload is OpenCL compute, the GTX 580M is the stronger choice. The HD 8730M is not without merit, but its recorded performance slots it just below the GTX 580M, and both parts are near the bottom of the overall distribution. Neither is a high-flyer; the GTX 580M is merely less low.

Architecture Differences

The two GPUs come from different architectural generations and process nodes. The GTX 580M uses the GF114 chip, built on Fermi 2.0 architecture, fabricated on a 40 nm process at TSMC. It packs 1,950 million transistors on a 332 mm² die, yielding a transistor density of 5.9 million per square millimeter. The HD 8730M uses the Mars chip, built on GCN 1.0 architecture, also at TSMC but on a 28 nm process. It has 950 million transistors on a 77 mm² die, giving a density of 12.3 million per square millimeter. The AMD part is far more transistor-dense, but the NVIDIA part has over twice the transistor count and a die more than four times larger.

The memory subsystems diverge sharply. The GTX 580M uses 2 GB of GDDR5 on a 256-bit bus, delivering 96.00 GB/s of bandwidth. The HD 8730M also has 2 GB, but it is DDR3 on a 128-bit bus, capping at 28.80 GB/s. That is a more than 3x bandwidth deficit for the AMD part, which directly impacts texture-heavy and memory-bound workloads.

The compute resources differ in configuration. Both have 384 shading units, but the GTX 580M has 64 texture mapping units and 32 ROPs, while the HD 8730M has 24 TMUs and 8 ROPs. The pixel rate for the GTX 580M is 9.920 GPixel/s versus 5.600 GPixel/s for the HD 8730M. Texture rate is 39.68 GTexel/s versus 16.80 GTexel/s. FP32 throughput is 952.3 GFLOPS versus 537.6 GFLOPS. The NVIDIA part leads in every throughput metric.

Clock behavior also differs. The GTX 580M lists a memory clock of 750 MHz with 3 Gbps effective, but no base or boost clock is recorded. The HD 8730M has a base clock of 650 MHz and a boost of 700 MHz, with memory at 900 MHz (1800 Mbps effective). The AMD part has explicit clock ranges; the NVIDIA part does not, likely due to its mobile MXM form factor and variable OEM implementation.

API support shows a split: both support DirectX 12 and OpenGL 4.6, but the HD 8730M adds Vulkan 1.2.170, while the GTX 580M has no Vulkan entry. The GTX 580M lists DirectX 12 (11_0) while the HD 8730M lists DirectX 12 (11_1). The bus interface is MXM-B (3.0) for the NVIDIA part and PCIe 3.0 x8 for the AMD part. The GTX 580M is rated at 100 W TDP with no power connectors, and its display outputs are portable-device dependent. The HD 8730M has no TDP listed, no slot width, and no display output data.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The NVIDIA GeForce GTX 580M scores 6389, which is 7.3% higher than the AMD Radeon HD 8730M's 5955.

Q: How does the memory bandwidth compare between the two?

A: The GTX 580M offers 96.00 GB/s with a 256-bit GDDR5 interface, while the HD 8730M provides 28.80 GB/s over a 128-bit DDR3 bus.

Q: Which GPU supports Vulkan?

A: Only the AMD Radeon HD 8730M lists Vulkan support, specifically version 1.2.170. The NVIDIA GTX 580M has no Vulkan entry in the database.

Q: Are the shading unit counts identical?

A: Yes, both GPUs have 384 shading units. However, the GTX 580M has 64 TMUs and 32 ROPs, while the HD 8730M has 24 TMUs and 8 ROPs.

Q: What are the process nodes and die sizes?

A: The GTX 580M is on a 40 nm process with a 332 mm² die, while the HD 8730M is on a 28 nm process with a 77 mm² die.

Q: Which GPU has a higher pixel fill rate?

A: The GTX 580M achieves 9.920 GPixel/s, nearly double the HD 8730M's 5.600 GPixel/s.

Specification Differences

| Specification | NVIDIA GeForce GTX 580M | AMD Radeon HD 8730M |

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

| Chip | GF114 | Mars |

| Architecture | Fermi 2.0 | GCN 1.0 |

| Process Node | 40 nm | 28 nm |

| Transistors | 1,950 million | 950 million |

| Die Size | 332 mm² | 77 mm² |

| Transistor Density | 5.9M / mm² | 12.3M / mm² |

| Memory Type | GDDR5 | DDR3 |

| Memory Bus Width | 256 bit | 128 bit |

| Memory Bandwidth | 96.00 GB/s | 28.80 GB/s |

| Base Clock | Not listed | 650 MHz |

| Boost Clock | Not listed | 700 MHz |

| Memory Clock | 750 MHz (3 Gbps effective) | 900 MHz (1800 Mbps effective) |

| TMUs | 64 | 24 |

| ROPs | 32 | 8 |

| Pixel Rate | 9.920 GPixel/s | 5.600 GPixel/s |

| Texture Rate | 39.68 GTexel/s | 16.80 GTexel/s |

| FP32 | 952.3 GFLOPS | 537.6 GFLOPS |

| TDP | 100 W | Not listed |

| Slot Width | MXM Module | Not listed |

| Power Connectors | None | Not listed |

| Bus Interface | MXM-B (3.0) | PCIe 3.0 x8 |

| Display Outputs | Portable Device Dependent | Not listed |

| Vulkan | Not listed | 1.2.170 |

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

| Release Date | 2011-06-27 | 2013-03-31 |

| Predecessor | GeForce 400M | London |

| Successor | GeForce 600M | Gem System |

Head-to-Head Benchmarks

The only recorded comparison is the Geekbench OpenCL test. The GTX 580M finishes with 6389, the HD 8730M with 5955. The 7.3% delta is the entire story in raw scores, but the underlying specification differences explain why.

Memory bandwidth is the largest chasm. The GTX 580M's 96.00 GB/s dwarfs the HD 8730M's 28.80 GB/s. A 3.33x advantage in bandwidth does not translate to a 3.33x score advantage, because OpenCL workloads are not purely bandwidth-bound. But it does push the NVIDIA part ahead in any kernel that streams data. The GTX 580M also has 64 TMUs versus 24, and 32 ROPs versus 8. That means the NVIDIA part can fill textures and write pixels at more than double the rate: 39.68 GTexel/s versus 16.80 GTexel/s, and 9.920 GPixel/s versus 5.600 GPixel/s. FP32 compute is 952.3 GFLOPS versus 537.6 GFLOPS, a 77% lead.

The HD 8730M counters with a newer process and more modern architecture. GCN 1.0 at 28 nm is denser (12.3M transistors per mm² versus 5.9M) and adds Vulkan support. But the smaller die and lower transistor count (950 million versus 1,950 million) limit its ceiling. The HD 8730M's boost clock reaches 700 MHz, but with only 24 TMUs and 8 ROPs, it cannot translate that clock into competitive throughput.

The percentile ranks confirm the gap: 37th for the GTX 580M versus 34th for the HD 8730M. The nearest rivals for each part reinforce the picture. The GTX 580M ties the GTX 460 SE exactly, sits 0.3% behind the RTX PRO 5000 72 GB Blackwell, and leads the Radeon Pro WX 4100 by 0.9% and the Radeon R7 M350 by 1%. The HD 8730M ties the Quadro K620M, sits 0.3% behind the HD 8750M, leads the UHD Graphics 730 by 0.4%, and trails the Quadro K4000 by 0.5%. Both GPUs are clustered in a narrow band of low-to-mid performance, but the GTX 580M sits at the top of that band.

The release dates matter contextually. The GTX 580M launched in June 2011, the HD 8730M in March 2013. Despite being nearly two years older, the NVIDIA part still outpaces the AMD part in this workload. That reflects the GTX 580M's wider memory bus and larger execution resources. The HD 8730M's architectural improvements, GCN 1.0 and 28 nm, do not compensate for the smaller memory interface and fewer ROPs.

In practical terms, the GTX 580M is the pick for OpenCL compute. Its 7.3% edge is not transformative, but it is consistent with every throughput metric in the database. The HD 8730M remains a functional mobile GPU, but its recorded performance places it below the older Fermi part. For buyers choosing between these two in a used or legacy system, the data says the NVIDIA card delivers more compute per benchmark point.