AMD Radeon HD 8790M vs NVIDIA GeForce GTX 760M Comparison

FAQ

Q: Which GPU wins in OpenCL compute performance?

A: The NVIDIA GeForce GTX 760M scores 5604 in Geekbench OpenCL, which is 10.5% higher than the AMD Radeon HD 8790M's 5017. This is a clear win for NVIDIA in OpenCL workloads.

Q: How do the two compare in Vulkan graphics performance?

A: The AMD Radeon HD 8790M is significantly faster in Vulkan, scoring 6365 compared to the GTX 760M's 4868. That is a 30.8% advantage for AMD in this API.

Q: What are the average benchmark scores for each GPU?

A: The AMD Radeon HD 8790M has an average benchmark score of 5691, placing it in the 33rd percentile of all GPUs. The NVIDIA GeForce GTX 760M averages 5236, sitting in the 31st percentile.

Q: How does the AMD card compare to its nearest rivals?

A: The AMD Radeon HD 8790M is 1.6% ahead of the NVIDIA Quadro M500M (which scores 5604), and 0.4% behind the Intel Iris Pro Graphics P6300 (5712). It trails the GTX 670MX by 0.5% and the GTX 550 Ti by 0.7%.

Q: Where does the GTX 760M stand relative to its closest competitors?

A: The GTX 760M is 0.5% ahead of the NVIDIA Quadro 4000M (5211) and 1.5% ahead of the AMD Radeon R7 M260X (5161). It sits 0.9% behind the GeForce 940M (5284) and 1.3% behind the GTX 980M (5308).

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

A: The NVIDIA chip packs 2,540 million transistors, while the AMD chip has 950 million. That is a substantial difference in silicon complexity.

Architecture Differences

The AMD Radeon HD 8790M is built on GCN 1.0 architecture using the Mars chip, manufactured on a 28 nm process at TSMC. The NVIDIA GeForce GTX 760M uses the Kepler architecture with the GK106S chip, also on a 28 nm TSMC process. Despite the same node, the chips diverge significantly in physical design.

The AMD chip contains 950 million transistors on a 77 mm² die, yielding a transistor density of 12.3 million per mm². The NVIDIA chip is much larger, with 2,540 million transistors on a 221 mm² die, resulting in a lower density of 11.5 million per mm². The NVIDIA part has nearly three times the transistor count, but the AMD part is more densely packed per square millimeter.

The compute layout differs sharply. AMD fields 384 shading units, 24 texture mapping units, and 8 raster operation pipelines. NVIDIA doubles many of these counts: 768 shading units, 64 TMUs, and 16 ROPs. This gives the GTX 760M a theoretical pixel rate of 11.50 GPixel/s and a texture rate of 46.02 GTexel/s, while the HD 8790M manages 7.200 GPixel/s and 21.60 GTexel/s respectively.

The memory subsystems are nearly identical in capacity and type. Both use 2 GB of GDDR5 on a 128-bit bus. The AMD card has a bandwidth of 64.00 GB/s, while the NVIDIA card is marginally higher at 64.13 GB/s. Memory clocks are also close: 1000 MHz for AMD versus 1002 MHz for NVIDIA, both achieving 4 Gbps effective.

Clock behavior differs notably. The AMD GPU runs a base clock of 850 MHz with a boost up to 900 MHz. The NVIDIA GPU has a lower base of 628 MHz and boosts to 719 MHz. Even with lower clocks, the NVIDIA part produces higher raw throughput due to its wider execution resources. The FP32 compute rating tells the story: 1,104.4 GFLOPS for NVIDIA versus 691.2 GFLOPS for AMD.

API support shows a generation split. AMD lists DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. NVIDIA lists DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The Vulkan version is slightly newer on the NVIDIA side, but AMD has a higher DirectX feature level.

The NVIDIA GPU carries a 55 W TDP rating, while the AMD card has no TDP listed in the database. Both use MXM modules with no power connectors. The bus interface differs: AMD uses MXM-A (3.0), while NVIDIA uses PCIe 3.0 x16. Both are end-of-life products, with AMD releasing on 2013-03-31 and NVIDIA on 2013-05-29.

Head-to-Head Benchmarks

The database records two head-to-head benchmark comparisons. In Geekbench OpenCL, the NVIDIA GeForce GTX 760M wins with a score of 5604 against the AMD Radeon HD 8790M's 5017. The delta is 10.5% in NVIDIA's favor. This aligns with the NVIDIA card's larger compute footprint and double the shading units.

In Geekbench Vulkan, the tables turn completely. The AMD Radeon HD 8790M scores 6365, while the GTX 760M manages only 4868. The 30.8% margin is substantial and indicates that AMD's GCN architecture handles Vulkan workloads more efficiently in this comparison, despite having fewer raw resources on paper.

Each GPU wins exactly one benchmark in this head-to-head set. The split is clean: NVIDIA takes OpenCL, AMD takes Vulkan. This pattern suggests the choice between these two depends heavily on the API used by the target applications.

Looking at the wider percentile rankings, the AMD card sits at the 33rd percentile of all GPUs, slightly above the NVIDIA card's 31st percentile. The average benchmark score also favors AMD: 5691 versus 5236. That is an 8.7% gap in average performance, driven largely by the large Vulkan win.

The nearest rivals for each card reinforce the picture. The AMD card's closest competitor, the Intel Iris Pro Graphics P6300, scores 5712, just 0.4% above it. The NVIDIA card's closest rival, the Quadro 4000M, scores 5211, which is 0.5% below it. Neither card dominates its immediate competitive set; both sit in a tight cluster of mobile GPUs from the same era.

Specification Differences

The two GPUs differ across nearly every core specification. The AMD Radeon HD 8790M uses the Mars chip with GCN 1.0 architecture, while the NVIDIA GeForce GTX 760M uses GK106S with Kepler. AMD is in the Solar System generation (HD 8700M), NVIDIA is in the GeForce 700M generation.

Transistor count is a major differentiator: 950 million for AMD versus 2,540 million for NVIDIA. Die size also diverges, 77 mm² versus 221 mm². Transistor density is slightly higher on the AMD side at 12.3M per mm² versus 11.5M per mm².

Shader resources differ by a factor of two. AMD has 384 shading units, NVIDIA has 768. TMU counts are 24 versus 64. ROP counts are 8 versus 16. These are structural differences that directly impact fill rate and texture throughput.

Clock speeds run in opposite directions. AMD has a base clock of 850 MHz boosting to 900 MHz. NVIDIA has a base of 628 MHz boosting to 719 MHz. The AMD card runs about 25% faster in clock terms, but the NVIDIA card compensates with far more execution units.

Memory is effectively identical: 2 GB GDDR5, 128-bit bus, and bandwidth of 64.00 GB/s versus 64.13 GB/s. The memory clock difference is trivial, 1000 MHz versus 1002 MHz.

Pixel rate favors NVIDIA at 11.50 GPixel/s versus AMD's 7.200 GPixel/s. Texture rate also favors NVIDIA at 46.02 GTexel/s versus 21.60 GTexel/s. FP32 compute is 1,104.4 GFLOPS for NVIDIA versus 691.2 GFLOPS for AMD.

The NVIDIA card has a rated TDP of 55 W, while the AMD card has no TDP figure in the database. Both use MXM modules with no external power connectors. The bus interface differs: MXM-A (3.0) for AMD, PCIe 3.0 x16 for NVIDIA.

API feature sets show minor differences. AMD supports DirectX 12 (11_1) and Vulkan 1.2.170. NVIDIA supports DirectX 12 (11_0) and Vulkan 1.2.175. OpenGL 4.6 is common to both.

Release dates are close, with AMD launching 2013-03-31 and NVIDIA on 2013-05-29. Both are end-of-life. The predecessor and successor names differ: AMD's predecessor is London and successor is Gem System, while NVIDIA's predecessor is GeForce 600M and successor is GeForce 800M.

Where Each One Wins

The AMD Radeon HD 8790M wins in Vulkan-based workloads. Its Geekbench Vulkan score of 6365 is 30.8% above the GTX 760M, a commanding margin. For applications that use Vulkan, whether games or compute tasks, the AMD card is the stronger choice based on recorded data.

The AMD card also holds a higher average benchmark score overall, 5691 versus 5236, and a slightly better percentile position, 33rd versus 31st. For general-purpose mixed workloads, the database average favors AMD.

The NVIDIA GeForce GTX 760M wins in OpenCL compute. Its score of 5604 beats AMD's 5017 by 10.5%. OpenCL is used in many productivity and scientific applications, so this matters for users running those workloads.

The NVIDIA card also has substantially higher theoretical throughput metrics. Its FP32 compute of 1,104.4 GFLOPS is nearly 60% higher than AMD's 691.2 GFLOPS. Pixel rate and texture rate are also much higher, at 11.50 GPixel/s and 46.02 GTexel/s respectively. These numbers suggest NVIDIA has more raw headroom for workloads that scale with shader count and texture units.

The NVIDIA card's 768 shading units and 64 TMUs give it a structural advantage in geometry-heavy and texture-bound scenarios, even though its lower clocks reduce the realized advantage. The 16 ROPs versus 8 also help with fill-rate-limited scenes.

For users who care about power draw, the NVIDIA card has a published TDP of 55 W, while AMD has no such figure in the database. This makes the NVIDIA part easier to plan around for thermal budgets in compact laptops.

The Verdict

The data shows a split decision. The AMD Radeon HD 8790M is the better choice for Vulkan-based applications and has the higher average benchmark score. The NVIDIA GeForce GTX 760M is the better choice for OpenCL workloads and offers significantly higher theoretical compute throughput.

In the head-to-head record, each GPU wins one test. The AMD card wins by a large 30.8% margin in Vulkan, while the NVIDIA card wins by a more modest 10.5% in OpenCL. The AMD card's average score of 5691 versus NVIDIA's 5236 reflects that the Vulkan win outweighs the OpenCL loss in aggregate terms.

For a professional user running OpenCL compute tasks, the GTX 760M is the safer pick. Its 10.5% lead in that specific benchmark, combined with double the shading units and triple the transistors, suggests it will handle compute-heavy workloads more consistently. The 55 W TDP also provides a known power envelope.

For a gamer or user focused on Vulkan titles, the HD 8790M is clearly ahead. A 30.8% advantage in the Vulkan benchmark is decisive. The AMD card also holds a slight edge in overall percentile ranking, 33rd versus 31st, and a higher average score.

For a balanced user who does a mix of everything, the AMD card's higher average benchmark score makes it the default recommendation. The 5691 average versus 5236 is an 8.7% gap that reflects better all-around performance in the recorded tests. However, the NVIDIA card's superior raw specifications, including FP32 compute, pixel rate, and texture rate, should not be ignored for applications that leverage those capabilities.

The final call depends on the workload split. If OpenCL matters, take NVIDIA. If Vulkan or general average performance matters, take AMD. The database does not show a single dominant winner, only two GPUs with complementary strengths.