AMD Radeon HD 8790M vs AMD Radeon R7 M370 Comparison

Where Each One Wins

The benchmark data is unambiguous: the AMD Radeon R7 M370 wins both recorded head-to-head tests, making it the stronger mobile GPU in this pairing. The R7 M370 takes the Geekbench OpenCL test with a score of 7063 versus 5017 for the HD 8790M, a decisive 40.8% advantage. In the Vulkan test, the margin narrows dramatically: 6465 versus 6365, a 1.6% lead. So while the M370 is the winner in both categories, the nature of those wins differs sharply. The OpenCL result indicates a substantial compute-throughput advantage, while the Vulkan result suggests the two GPUs are nearly interchangeable in API-level rendering tasks.

The HD 8790M has no benchmark win to claim. Its average benchmark score sits at 5691, which places it in the 33rd percentile of all GPUs in the database. The M370, by contrast, averages 6764 and occupies the 38th percentile. In practical terms, the M370 is the pick for compute-heavy workloads such as OpenCL accelerating applications, while the Vulkan gap is small enough that the HD 8790M remains competitive in Vulkan-based titles, just not ahead.

Architecture Differences

Both GPUs share the same fundamental architecture: GCN 1.0, built on TSMC's 28 nm process. Both use 950 million transistors on a 77 mm² die, giving an identical transistor density of 12.3M per mm². The chip names differ, with the M370 using "Litho" and the HD 8790M using "Mars," but the underlying hardware is structurally the same. Both have 384 shading units, 24 texture mapping units, and 8 ROPs.

The core clock speeds differ. The M370 runs at a base of 875 MHz and boosts to 960 MHz. The HD 8790M runs at 850 MHz base and 900 MHz boost. That 25 MHz base and 60 MHz boost advantage for the M370 translates directly into higher pixel and texture rates: 7.680 GPixel/s and 23.04 GTexel/s versus 7.200 GPixel/s and 21.60 GTexel/s. FP32 compute is 737.3 GFLOPS for the M370 versus 691.2 GFLOPS for the HD 8790M.

Memory configuration differs in speed but not in capacity or bus width. Both use 2 GB of GDDR5 on a 128-bit bus, but the M370 runs its memory at 900 MHz (3.6 Gbps effective) yielding 57.60 GB/s, while the HD 8790M runs at 1000 MHz (4 Gbps effective) yielding 64.00 GB/s. The HD 8790M actually has a 11.1% bandwidth advantage, which helps explain why the Vulkan gap is so small; the M370's core-clock advantage is offset by the HD 8790M's memory clock advantage in memory-bound scenarios.

The bus interfaces differ: the M370 uses PCIe 3.0 x8, while the HD 8790M uses MXM-A (3.0), a module form factor. The HD 8790M is listed as an MXM Module with no power connectors, and its display outputs are portable-device dependent. The M370 has no listed slot width or power connector details. Both support DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170, so API-level feature parity is complete.

Head-to-Head Benchmarks

The largest single win belongs to the M370 in Geekbench OpenCL. Its score of 7063 versus 5017 is a 40.8% advantage. This is not a marginal edge; it is a categorical difference in compute throughput. The M370's higher core clocks, combined with the same 384 shader count, yield 737.3 GFLOPS versus 691.2 GFLOPS. The OpenCL test appears to scale closely with theoretical FP32, and the M370's boost clock advantage of 60 MHz over the HD 8790M's boost clock is the primary driver.

The Vulkan test tells a different story. The M370 scores 6465, the HD 8790M scores 6365, a mere 1.6% difference. This is within the range of run-to-run variance. The HD 8790M's higher memory bandwidth (64.00 GB/s versus 57.60 GB/s) appears to offset much of the M370's compute edge in Vulkan workloads, which often involve texture fetching and frame-buffer access. The result is that, in Vulkan, the two GPUs are effectively peers, with the M370 holding only a nominal lead.

For context on the M3700's standing, its nearest rivals in the database include the NVIDIA GeForce GTX 675M (avg 6946, 2.6% lower), the AMD FirePro M5100 (avg 6830, 1% lower), and the NVIDIA GeForce GT 1010 (avg 6698, 1% higher). The M370 edges past the GTX 675M and the FirePro M5100, but trails the GT 1010 by a single percent. For the HD 8790M, its nearest rivals are the NVIDIA GeForce GTX 550 Ti (avg 5731, 0.7% higher), the NVIDIA GeForce GTX 670MX (avg 5721, 0.5% higher), and the Intel Iris Pro Graphics P6300 (avg 5712, 0.4% higher). The HD 8790M sits just below these three, and 1.6% ahead of the NVIDIA Quadro M500M.

The database also records win counts: the M3700 wins 2 head-to-head tests, the HD 8790M wins 0. The verdict is consistent across metrics.

FAQ

Q: Which GPU is faster in OpenCL compute workloads?

A: The AMD Radeon R7 M370 is significantly faster, scoring 7063 versus 5017 in Geekbench OpenCL, a 40.8% advantage.

Q: Is the Vulkan performance difference meaningful?

A: No. The M3700 scores 6465 and the HD 8790M scores 6365, a 1.6% difference that is well within measurement noise.

Q: Do the two GPUs have the same hardware configuration?

A: Yes, both are GCN 1.0 on 28 nm with 950 million transistors, 384 shading units, 24 TMUs, and 8 ROPs. The differences are in clock speeds and memory clock.

Q: Why does the HD 8790M have higher memory bandwidth?

A: Because its memory runs at 1000 MHz (4.0 Gbps effective) versus 900 MHz (3.6 Gbps effective) for the M3700, yielding 64.00 GB/s versus 57.60 GB/s.

Q: How do these GPUs compare to their nearest rivals?

A: The M3700 is about 1% faster than the GT 1010 and 2.6% faster than the GTX 675M, while the HD 8790M is about 0.5% to 0.7% slower than the GTX 670MX and GTX 550 Ti.

Q: Which GPU has a higher overall average benchmark score?

A: The M3700 averages 6764, while the HD 8790M averages 5691. The M3700 also sits in the 38th percentile of all GPUs versus the 33rd for the HD 8790M.

The Verdict

The data points to a single conclusion: choose the AMD Radeon R7 M3700 for any workload where compute performance matters. It wins both tests, with a decisive 40.8% lead in OpenCL and a narrow but positive 1.6% lead in Vulkan. Its higher core clocks (875/960 MHz versus 850/900 MHz) give it a 6.7% FP32 advantage (737.3 GFLOPS versus 691.2 GFLOPS), and its pixel and texture rates are also higher (7.680 GPixel/s and 23.04 GTexel/s versus 7.200 and 21.60).

The only area where the HD 8790M holds a technical edge is memory bandwidth: 64.00 GB/s versus 57.60 GB/s. This makes the Vulkan test nearly a tie, but it does not translate into a benchmark win. For memory-bound scenarios, the HD 8790M may close the gap, but the recorded tests show the M3700 still prevails.

The HD 8790M is not a bad GPU. It sits at the 33rd percentile, close to the GTX 550 Ti and GTX 670MX. But the M3700 is a generation ahead in release timing (2015 versus 2013) and delivers a substantial compute uplift. If the choice is between these two for a system, the M3700 is the superior pick. The HD 8790M remains acceptable for Vulkan-era titles where the 1.6% gap is invisible, but the OpenCL workload is where the M3700 separates itself by a wide margin.

Specification Differences

| Specification | AMD Radeon R7 M370 | AMD Radeon HD 8790M |

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

| Chip | Litho | Mars |

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

| Base Clock | 875 MHz | 850 MHz |

| Boost Clock | 960 MHz | 900 MHz |

| Memory Clock | 900 MHz (3.6 Gbps effective) | 1000 MHz (4 Gbps effective) |

| Memory Bandwidth | 57.60 GB/s | 64.00 GB/s |

| Pixel Rate | 7.680 GPixel/s | 7.200 GPixel/s |

| Texture Rate | 23.04 GTexel/s | 21.60 GTexel/s |

| FP32 | 737.3 GFLOPS | 691.2 GFLOPS |

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

| Slot Width | Not listed | MXM Module |

| Power Connectors | Not listed | None |

| Display Outputs | Not listed | Portable Device Dependent |

| Release Date | 2015-05-04 | 2013-03-31 |

| Predecessor | Solar System | London |

| Successor | Polaris Mobile | Gem System |