AMD Radeon HD 8790M vs NVIDIA GeForce GTX 765M Comparison

# Head-to-Head Benchmarks

The benchmark data presents a clear hierarchy between these two mobile graphics processors. In the Geekbench OpenCL test, the NVIDIA GeForce GTX 765M delivers a score of 7,176, while the AMD Radeon HD 8790M manages 5,017. That represents a 30.1% deficit for the AMD part — a substantial margin that places it firmly behind in compute workloads. The gap narrows considerably in the Geekbench Vulkan test, where the GTX 765M scores 6,714 against the HD 8790M's 6,365, a difference of just 5.2%. This suggests that while the NVIDIA part holds the overall advantage, the AMD architecture is comparatively more competitive in API-specific workloads that leverage Vulkan's lower-level hardware access.

Looking at the broader context, the HD 8790M's average benchmark score of 5,691 places it just ahead of its closest rival, the Intel Iris Pro Graphics P6300, which averages 5,712 — a delta of only -0.4%. The GTX 765M, by contrast, averages 5,501, sitting slightly below the NVIDIA GeForce MX130 (5,508, -0.1%) and marginally above the AMD Radeon R7 M440 (5,483, +0.3%). These percentile positions tell an interesting story: despite the GTX 765M's decisive OpenCL victory, its average score across all benchmarks is actually lower than the HD 8790M's average. This happens because the GTX 765M's Vulkan result of 6,714 and OpenCL result of 7,176 are averaged against its Metal score of 2,612, which drags the overall figure down significantly. The HD 8790M has no Metal benchmark in the data, so its average of 5,691 reflects only its two stronger results.

The head-to-head table confirms that the GTX 765M wins both recorded benchmark comparisons, giving it a 2-0 record against the HD 8790M. However, the magnitude of those wins is uneven — the OpenCL gap of 30.1% is a dominant performance, while the Vulkan margin of 5.2% is closer to a narrow victory. For users prioritizing compute-heavy tasks that rely on OpenCL, the GTX 765M is the clear choice. For Vulkan-centric workloads, the two parts are much more evenly matched.

# Architecture Differences

The architectural divide between these two GPUs is substantial. The AMD Radeon HD 8790M is built on the Mars chip using the GCN 1.0 architecture, manufactured on a 28 nm process at TSMC. It packs 950 million transistors onto a 77 mm² die, yielding a transistor density of 12.3 million per mm². The NVIDIA GeForce GTX 765M, meanwhile, uses the GK106 chip based on the Kepler architecture, also fabricated on TSMC's 28 nm node. The GK106 is a much larger die at 221 mm², housing 2,540 million transistors — nearly 2.7 times the transistor count of the Mars chip — though its transistor density of 11.5M / mm² is slightly lower than AMD's.

The compute resources differ dramatically. The HD 8790M contains 384 shading units, 24 texture mapping units, and 8 raster output units. The GTX 765M doubles or more than doubles each of these: 768 shading units, 64 TMUs, and 16 ROPs. This hardware advantage translates directly into throughput figures. The GTX 765M achieves a pixel rate of 13.81 GPixel/s and a texture rate of 55.23 GTexel/s, compared to the HD 8790M's 7.20 GPixel/s and 21.60 GTexel/s. In raw floating-point performance, the GTX 765M delivers 1,325.6 GFLOPS of FP32 compute, while the HD 8790M offers 691.2 GFLOPS — the NVIDIA part is roughly 92% faster on paper.

Clock speeds tell a different story. The HD 8790M runs at a base clock of 850 MHz and boosts to 900 MHz, while the GTX 765M starts lower at 797 MHz and boosts to 863 MHz. The AMD part's higher clocks partially compensate for its smaller shader count, but the sheer scale of the NVIDIA GPU's resource allocation outweighs the clock advantage. Memory configurations are nearly identical: both use 2 GB of GDDR5 on a 128-bit bus, with the HD 8790M offering 64.00 GB/s of bandwidth and the GTX 765M at 64.13 GB/s — a negligible difference. Both run their memory at effectively 4 Gbps effective.

The API support reveals a notable divergence. The HD 8790M supports DirectX 12 (11_1), while the GTX 765M supports DirectX 12 (11_0) — the AMD part has a slight edge in the feature level. Both support OpenGL 4.6. For Vulkan, the GTX 765M lists version 1.2.175, while the HD 8790M lists 1.2.170 — a minor revision difference. The GTX 765M also has a Metal benchmark score of 2,612, which the HD 8790M lacks entirely, indicating Apple's API is available on the NVIDIA part but not the AMD one. The GTX 765M carries a TDP of 75 W, while the HD 8790M's TDP is not listed in the data.

# Where Each One Wins

The GTX 765M wins decisively in OpenCL compute. Its score of 7,176 versus 5,017 represents a 30.1% advantage, making it the preferred option for applications that offload parallel processing to the GPU through OpenCL. This includes tasks like video encoding, physics simulations, and general-purpose compute workloads. The NVIDIA part also edges out the AMD card in Vulkan, scoring 6,714 against 6,365, though the 5.2% margin is far less commanding.

The HD 8790M's strengths are more subtle but still present in the data. Its average benchmark score of 5,691 exceeds the GTX 765M's average of 5,501, despite losing both head-to-head comparisons. This is entirely due to the GTX 765M's low Metal score of 2,612, which pulls its average down. If a workload relies on Metal — primarily Apple macOS environments — the GTX 765M would likely struggle relative to its other results, though the HD 8790M has no Metal score to compare directly. The HD 8790M's higher base clock of 850 MHz versus 797 MHz and boost clock of 900 MHz versus 863 MHz could also translate to better performance in lightly-threaded or latency-sensitive tasks where clock speed matters more than raw throughput.

For gaming, the GTX 765M's doubled TMU count (64 versus 24) and doubled ROP count (16 versus 8) suggest it can handle texture-heavy scenes and higher resolutions with less bottlenecking. The pixel rate of 13.81 GPixel/s versus 7.20 GPixel/s is a direct indicator of fill-rate capability, which matters for high-resolution rendering. The HD 8790M, with its lower TDP and smaller die, may be better suited to thinner laptops with limited cooling, though the data does not include thermal or power figures for the AMD part.

# FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon HD 8790M averages 5,691 across its benchmarks, while the NVIDIA GeForce GTX 765M averages 5,501. The AMD part is about 3.5% higher in this aggregate measure.

Q: How much faster is the GTX 765M in OpenCL?

A: The GTX 765M scores 7,176 in Geekbench OpenCL, versus 5,017 for the HD 8790M, making it 30.1% faster in that specific test.

Q: Do both GPUs support the same memory configuration?

A: Yes, both have 2 GB of GDDR5 memory on a 128-bit bus. The GTX 765M offers 64.13 GB/s bandwidth, while the HD 8790M provides 64.00 GB/s — a difference of just 0.13 GB/s.

Q: What is the transistor count difference between the two chips?

A: The GTX 765M's GK106 chip contains 2,540 million transistors, while the HD 8790M's Mars chip has 950 million. The NVIDIA part has roughly 2.7 times more transistors.

Q: Does the HD 8790M have a higher clock speed?

A: Yes, the HD 8790M runs at 850 MHz base and 900 MHz boost, compared to the GTX 765M's 797 MHz base and 863 MHz boost. The AMD part is about 6.7% higher at base clock and 4.3% higher at boost.

Q: Which GPU supports a higher DirectX feature level?

A: The HD 8790M supports DirectX 12 (11_1), while the GTX 765M supports DirectX 12 (11_0). The AMD part has a slightly higher feature level, though both are DirectX 12 compatible.

# The Verdict

Based strictly on the benchmark data, the NVIDIA GeForce GTX 765M is the stronger performer in direct competition. It wins both head-to-head tests, with a commanding 30.1% lead in OpenCL and a more modest 5.2% lead in Vulkan. Its hardware specifications — 768 shading units, 64 TMUs, 16 ROPs, and 1,325.6 GFLOPS of FP32 compute — establish it as the more capable compute and graphics processor on paper. The GTX 765M also holds a 75 W TDP rating, giving it a defined power envelope for system designers to plan around.

The AMD Radeon HD 8790M, however, should not be dismissed. Its average benchmark score of 5,691 is higher than the GTX 765M's 5,501, a reversal driven entirely by the NVIDIA part's weak Metal result of 2,612. For users who never touch Metal workloads, the GTX 765M's average would be significantly higher, but the data as presented shows the AMD card delivering more consistent results across its available benchmarks. The HD 8790M also offers higher clock speeds (850/900 MHz versus 797/863 MHz) and a slightly better DirectX feature level (11_1 versus 11_0).

The choice between these two depends on workload priorities. For OpenCL-heavy applications, the GTX 765M is the obvious pick — its 30.1% lead is too large to ignore. For Vulkan-based tasks, the GTX 765M still wins, but the margin is narrow enough that other factors like driver maturity or system integration could tip the balance. For users who value consistent average performance across mixed workloads, the HD 8790M's higher aggregate score makes it a defensible alternative, particularly if the GTX 765M's Metal deficiency is a concern. Both GPUs are end-of-life products, so this comparison is purely academic for those evaluating used or legacy hardware.

# Specification Differences

| Field | AMD Radeon HD 8790M | NVIDIA GeForce GTX 765M |

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

| Chip | Mars | GK106 |

| Architecture | GCN 1.0 | Kepler |

| Generation | Solar System (HD 8700M) | GeForce 700M |

| Transistors | 950 million | 2,540 million |

| Die Size | 77 mm² | 221 mm² |

| Transistor Density | 12.3M / mm² | 11.5M / mm² |

| Base Clock | 850 MHz | 797 MHz |

| Boost Clock | 900 MHz | 863 MHz |

| Memory Bandwidth | 64.00 GB/s | 64.13 GB/s |

| Shading Units | 384 | 768 |

| TMUs | 24 | 64 |

| ROPs | 8 | 16 |

| Pixel Rate | 7.200 GPixel/s | 13.81 GPixel/s |

| Texture Rate | 21.60 GTexel/s | 55.23 GTexel/s |

| FP32 Performance | 691.2 GFLOPS | 1,325.6 GFLOPS |

| TDP | Not listed | 75 W |

| Bus Interface | MXM-A (3.0) | MXM-B (3.0) |

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

| Vulkan Version | 1.2.170 | 1.2.175 |

| Release Date | 2013-03-31 | 2013-05-29 |

| Predecessor | London | GeForce 600M |

| Successor | Gem System | GeForce 800M |

| Avg Benchmark Score | 5,691 | 5,501 |

| Percentile vs All GPUs | 33 | 32 |