AMD Radeon HD 8750M vs AMD Radeon R7 M340 Comparison

Head-to-Head Benchmarks

The only directly comparable benchmark in the database is Geekbench OpenCL, and the data shows a decisive advantage for the AMD Radeon HD 8750M. The HD 8750M scores 5970 points, while the Radeon R7 M340 scores 4932 points, producing a 21% delta in favor of the older part. This is a substantial margin for two mobile GPUs aimed at similar notebook segments, and it flips the expected script, as the R7 M340 is the newer design by a wide margin in release timing.

In the broader database rankings, the HD 8750M sits at the 34th percentile of all GPUs, while the R7 M340 sits at the 30th percentile. The HD 8750M's average benchmark score is 5970, the R7 M340's is 5063 when averaging its OpenCL and Vulkan results. The OpenCL gap alone, 21%, is larger than the percentile gap suggests, meaning the HD 8750M is not just slightly ahead but meaningfully faster in compute workloads.

Looking at the nearest rivals for each card reinforces the positioning. The HD 8750M is essentially tied with the NVIDIA Quadro K4000 (5982, delta -0.2%), the Quadro K620M (5957, delta 0.2%), and the AMD Radeon HD 8730M (5955, delta 0.3%). It also edges out the Quadro K4000M (5986, delta -0.3%). This cluster of scores within 0.3% shows the HD 8750M sits in a tight competitive band, roughly 0.2% slower than the Quadro K4000M and roughly 0.3% faster than the HD 8730M.

The R7 M340, by contrast, is exactly tied with the AMD Radeon R7 240 (5063, delta 0%), and only slightly ahead of the FirePro W4170M (5034, delta 0.6%), R5 M430 (5018, delta 0.9%), and Radeon R7 Graphics (4998, delta 1.3%). None of these deltas exceed 1.3%, indicating the R7 M340 is a mid-pack performer among its immediate peers. The HD 8750M's 21% lead over the R7 M340 is far larger than any delta within either card's rival group, making the head-to-head result unambiguous.

There is no second benchmark where the R7 M340 wins. The R7 M340 does have a Vulkan score of 5193 in the database, but the HD 8750M has no Vulkan result recorded, so no direct comparison is possible. For OpenCL, the only shared test, the HD 8750M wins outright with a 21% margin.

Architecture Differences

The two GPUs come from different generations of AMD's Graphics Core Next architecture. The HD 8750M uses GCN 1.0, while the R7 M340 uses GCN 3.0. This is a foundational difference, not a minor revision. GCN 1.0 debuted in the Solar System generation (HD 8700M series), while GCN 3.0 belongs to the Gem System generation (R7 M300 series). The R7 M340's predecessor is listed as Solar System, directly linking it to the HD 8750M's generation, while the HD 8750M's predecessor is London and its successor is Gem System, meaning the R7 M340 is the later product in the same lineage.

The chip designs differ substantially. The HD 8750M uses the Mars chip, a 28 nm part fabricated by TSMC with 950 million transistors on a 77 mm² die. The R7 M340 uses the Meso chip, also 28 nm and TSMC, but with 1,550 million transistors on a 125 mm² die. That is roughly 63% more transistors and a 62% larger die area for the newer part. Transistor density is nearly identical: 12.3 million per mm² for the HD 8750M versus 12.4 million per mm² for the R7 M340, which makes sense given the same process node and foundry.

Clock speeds favor the R7 M340. Its base clock is 943 MHz and boost is 1021 MHz, versus 775 MHz base and 825 MHz boost for the HD 8750M. That is a 168 MHz higher base and a 196 MHz higher boost. The memory clock also favors the newer part: 1000 MHz (2 Gbps effective) versus 900 MHz (1800 Mbps effective). However, the memory configuration is where the HD 8750M strikes back.

The HD 8750M has 1024 MB of DDR3 on a 128-bit bus, yielding 28.80 GB/s of bandwidth. The R7 M340 has 2 GB of DDR3 on a 64-bit bus, yielding only 16.00 GB/s. That is a 12.80 GB/s deficit for the R7 M340, or about 44% less bandwidth. Despite having double the memory capacity, the narrower bus cripples throughput. This is likely the single largest factor explaining the OpenCL result, as compute workloads often saturate memory bandwidth.

Compute unit counts are lower on the R7 M340. The HD 8750M has 384 shading units, 24 texture mapping units, and 8 ROPs. The R7 M340 has 320 shading units, 20 TMUs, and 8 ROPs. So the HD 8750M has 64 more shading units and 4 more TMUs, while ROP counts are equal. Pixel rate and texture rate reflect this: the HD 8750M achieves 6.600 GPixel/s and 19.80 GTexel/s, while the R7 M340 achieves 8.168 GPixel/s and 20.42 GTexel/s. The R7 M340 wins on pixel throughput by 1.568 GPixel/s and on texture throughput by 0.62 GTexel/s, thanks to its higher clocks, despite having fewer units.

FP32 compute is very close. The HD 8750M delivers 633.6 GFLOPS, the R7 M340 delivers 653.4 GFLOPS, a difference of 19.8 GFLOPS in favor of the newer part. The R7 M340 also lists FP16 at 653.4 GFLOPS (1:1), while the HD 8750M has no FP16 data. The near-identical FP32 figures make the 21% OpenCL gap even more striking, pointing to bandwidth and driver efficiency as the separating factors.

Where Each One Wins

The HD 8750M wins in raw OpenCL compute performance, with a 21% lead in the only shared benchmark. Its 128-bit memory bus and 28.80 GB/s bandwidth give it a clear advantage in memory-bound workloads. For tasks like OpenCL-based image processing, physics simulations, or any general-purpose GPU compute that relies on frequent memory access, the data indicates the HD 8750M is the stronger choice. Its lower shading unit count is offset by more units overall (384 versus 320) and a wider memory path. The 34th percentile ranking versus the R7 M340's 30th also places it higher in the global distribution.

The R7 M340 wins on several specification-based metrics that do not translate into a benchmark victory in this dataset. It has a higher pixel rate (8.168 GPixel/s versus 6.600 GPixel/s) and a higher texture rate (20.42 GTexel/s versus 19.80 GTexel/s). Its FP32 output is slightly higher at 653.4 GFLOPS versus 633.6 GFLOPS, and it supports FP16 at a 1:1 ratio, which the HD 8750M does not record. For scenarios where clock speed drives performance and memory bandwidth is less critical, the R7 M340 has a theoretical edge.

The R7 M340 also has double the memory capacity (2 GB versus 1 GB). In applications that require larger working sets, such as holding larger textures or datasets in VRAM, the R7 M340 can accommodate more data without spilling to system memory. However, its 16.00 GB/s bandwidth means moving that data is slower.

In API support, the R7 M340 is ahead. It lists DirectX 12 (12_0), while the HD 8750M lists DirectX 12 (11_1). Both support OpenGL 4.6 and Vulkan 1.2.170. The newer DirectX feature level on the R7 M340 could matter for titles that use DirectX 12 features not available under 11_1. The R7 M340 also has a recorded Vulkan score of 5193, while the HD 8750M has none, suggesting the newer part is better positioned for Vulkan workloads, though no direct comparison exists.

For gaming, the R7 M340's higher pixel rate and texture rate, combined with more VRAM, make it potentially better for fill-rate-bound scenarios at lower resolutions, assuming a game does not bottleneck on memory bandwidth. The HD 8750M, with its wider bus, would likely hold up better in higher-resolution scenarios or with anti-aliasing enabled, where bandwidth demand increases. The data does not include gaming benchmarks, so this remains an inference from the specification differences.

FAQ

Q: Which GPU is faster in OpenCL compute?

A: The AMD Radeon HD 8750M. In the Geekbench OpenCL test, it scores 5970 versus 4932 for the Radeon R7 M340, a 21% advantage.

Q: How much memory bandwidth does each card have?

A: The HD 8750M has 28.80 GB/s from a 128-bit bus with DDR3. The R7 M340 has 16.00 GB/s from a 64-bit bus with DDR3, despite having 2 GB of memory versus 1 GB.

Q: Does the R7 M340 have higher clock speeds?

A: Yes. The R7 M340 runs at 943 MHz base and 1021 MHz boost, while the HD 8750M runs at 775 MHz base and 825 MHz boost. Memory is also faster: 1000 MHz (2 Gbps effective) versus 900 MHz (1800 Mbps effective).

Q: Which card has more shading units?

A: The HD 8750M has 384 shading units and 24 TMUs, while the R7 M340 has 320 shading units and 20 TMUs. Both have 8 ROPs.

Q: What are the DirectX feature levels?

A: The HD 8750M supports DirectX 12 (11_1), while the R7 M340 supports DirectX 12 (12_0). Both support OpenGL 4.6 and Vulkan 1.2.170.

Q: Which card ranks higher in the database?

A: The HD 8750M is at the 34th percentile of all GPUs, while the R7 M340 is at the 30th percentile. The HD 8750M's average benchmark score is 5970; the R7 M340's is 5063.

Specification Differences

| Field | AMD Radeon HD 8750M | AMD Radeon R7 M340 |

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

| Architecture | GCN 1.0 | GCN 3.0 |

| Generation | Solar System (HD 8700M) | Gem System (R7 M300) |

| Chip | Mars | Meso |

| Process node | 28 nm | 28 nm |

| Foundry | TSMC | TSMC |

| Transistors | 950 million | 1,550 million |

| Die size | 77 mm² | 125 mm² |

| Transistor density | 12.3M / mm² | 12.4M / mm² |

| Base clock | 775 MHz | 943 MHz |

| Boost clock | 825 MHz | 1021 MHz |

| Memory clock | 900 MHz (1800 Mbps effective) | 1000 MHz (2 Gbps effective) |

| Memory size | 1024 MB | 2 GB |

| Memory type | DDR3 | DDR3 |

| Memory bus width | 128 bit | 64 bit |

| Memory bandwidth | 28.80 GB/s | 16.00 GB/s |

| Shading units | 384 | 320 |

| TMUs | 24 | 20 |

| ROPs | 8 | 8 |

| Pixel rate | 6.600 GPixel/s | 8.168 GPixel/s |

| Texture rate | 19.80 GTexel/s | 20.42 GTexel/s |

| FP32 | 633.6 GFLOPS | 653.4 GFLOPS |

| FP16 | Not recorded | 653.4 GFLOPS (1:1) |

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

| OpenGL | 4.6 | 4.6 |

| Vulkan | 1.2.170 | 1.2.170 |

| Release date | 2013-02-25 | 2015-05-04 |

| Production status | End-of-life | End-of-life |

| Predecessor | London | Solar System |

| Successor | Gem System | Polaris Mobile |

| Geekbench OpenCL score | 5970 | 4932 |

| Geekbench Vulkan score | Not recorded | 5193 |

| Avg benchmark score | 5970 | 5063 |

| Percentile vs all GPUs | 34 | 30 |

The specification table shows a clear generational split. The R7 M340 is a larger, newer chip with more transistors, higher clocks, and more memory capacity, but it is hamstrung by a 64-bit memory bus that halves bandwidth relative to the HD 8750M. The HD 8750M counters with more shading units, a wider bus, and a higher percentile ranking. The 21% OpenCL win for the HD 8750M, despite lower clocks and an older architecture, underscores how memory bandwidth and compute unit count can outweigh architectural generation and clock speed in real workloads.