AMD Radeon R9 M395X vs AMD Radeon RX 6700S Comparison

AMD Radeon R9 M395X vs AMD Radeon RX 6700S: a comparison between a 2015 mobile graphics solution built on GCN 3.0 and a 2022 RDNA 2.0 part. The data shows a generational leap in performance, but the older card still has specific strengths in certain workloads. This analysis relies entirely on recorded benchmark scores and hardware specifications.

Head-to-Head Benchmarks

The database records two direct head-to-head benchmark comparisons between these GPUs: Geekbench Metal and Geekbench OpenCL. The AMD Radeon RX 6700S wins both tests decisively. In the Geekbench Metal test, the RX 6700S scores 72,864 points against the R9 M395X’s 33,953 points. This represents a delta of -53.4%, meaning the R9 M395X trails by 53.4%. In practical terms, the RX 6700S delivers more than double the Metal performance, a massive advantage for any macOS or Metal-based application.

The Geekbench OpenCL result is even more lopsided. The RX 6700S posts 70,965 points, while the R9 M395X manages only 17,829 points. The delta here is -74.9%, indicating the older card is nearly 75% slower in this compute-oriented test. This is a stark illustration of how far GPU architecture has advanced in seven years.

The R9 M395X records no wins in the head-to-head section. Its average benchmark score across all tests is 25,891 points, placing it in the 71st percentile of all GPUs. The RX 6700S, by contrast, has an average score of 22,154 points, which places it in the 67th percentile. This is a curious situation: the newer card wins the direct comparisons by huge margins, yet its overall average score is lower. The reason lies in the benchmark suite composition. The RX 6700S has results for eleven separate tests, including several Passmark sub-tests where scores are low (for instance, Passmark DirectX 9 at 195, Passmark DirectX 10 at 87, and Passmark DirectX 12 at 58). These low scores drag down its average, while the R9 M395X only has two benchmark entries, both of which are moderately high.

For context on the RX 6700S, its nearest rival in the database is the AMD Radeon RX 5700, which scores 22,170 points, a delta of only -0.1%. The NVIDIA GeForce RTX 5060 Mobile scores 22,435 points, putting it 1.3% ahead of the RX 6700S. The NVIDIA GeForce GTX 1060 6 GB sits 1.4% behind at 21,856 points. This places the RX 6700S in a performance tier comparable to desktop-class GPUs from a few generations prior.

The R9 M395X’s nearest rivals are a different story. Its average of 25,891 points puts it 0.1% ahead of the AMD FirePro W7100 (25,856 points) and 0.2% ahead of the AMD FirePro D700 (25,842 points). Interestingly, it is 0.5% behind the AMD Radeon RX 6750 XT (26,011 points) and 0.6% ahead of the NVIDIA GeForce RTX 3080 Ti Mobile (25,740 points). These tight margins show that the R9 M395X, despite its age, holds its own in the database’s average scoring metric, though this is heavily influenced by the limited number of tests it participated in.

Architecture Differences

The architectural gap between these two GPUs is fundamental. The R9 M395X uses the Amethyst chip based on GCN 3.0 architecture, manufactured on a 28 nm process at TSMC. The RX 6700S uses the Navi 23 chip based on RDNA 2.0, also from TSMC but on a 7 nm node. The process shrink is significant: the R9 M395X has 5,000 million transistors on a 366 mm² die, while the RX 6700S packs 11,060 million transistors into a much smaller 237 mm² die. Transistor density tells the story clearly: the R9 M395X achieves 13.7 million transistors per mm², whereas the RX 6700S reaches 46.7 million per mm², roughly 3.4 times denser.

Core configuration differs substantially. The R9 M395X has 2,048 shading units, 128 texture mapping units (TMUs), and 32 raster operations pipelines (ROPs). The RX 6700S has 1,792 shading units, 112 TMUs, and 64 ROPs. Despite having fewer shading units and TMUs, the RX 6700S outputs far more work. Its pixel rate is 128.0 GPixel/s versus 23.14 GPixel/s for the R9 M395X, and its texture rate is 224.0 GTexel/s versus 92.54 GTexel/s. This is a direct consequence of the much higher clock speeds on the newer card.

The R9 M395X has no base or boost clock listed, but its memory runs at 1250 MHz (5 Gbps effective). The RX 6700S has a base clock of 1700 MHz, a game clock of 1890 MHz, and a boost clock of 2000 MHz, with memory at 1750 MHz (14 Gbps effective). The memory subsystems also differ: both have 8 GB, but the R9 M395X uses GDDR5 on a 256-bit bus delivering 160.0 GB/s of bandwidth. The RX 6700S uses GDDR6 on a 128-bit bus, yet delivers 224.0 GB/s due to the much higher effective memory speed.

Feature support is another differentiator. The RX 6700S includes 28 ray tracing cores, a feature completely absent from the R9 M395X. The API support reflects this: the RX 6700S supports DirectX 12 Ultimate (12_2), while the R9 M395X only reaches DirectX 12 (12_0). Vulkan support is also newer on the RX 6700S (version 1.4 versus 1.2.170). Both cards support OpenGL 4.6. The RX 6700S has a 2:1 FP16 to FP32 ratio, offering 14.34 TFLOPS of half-precision compute, while the R9 M395X has a 1:1 ratio at 2.961 TFLOPS. In FP32, the RX 6700S delivers 7.168 TFLOPS versus 2.961 TFLOPS for the older card.

Power and physical specifications also differ. The R9 M395X has a TDP of 75 W and uses an MXM Module slot width. The RX 6700S has a TDP of 80 W and is classified as an IGP (integrated graphics processor) with a bus interface of PCIe 4.0 x8. The R9 M395X uses PCIe 3.0 x16. Neither card requires external power connectors, and both have display outputs described as portable device dependent.

Where Each One Wins

The RX 6700S wins in every direct benchmark comparison and in almost every practical scenario. Its lead in Geekbench Metal and OpenCL is so large that it is the clear choice for compute-heavy tasks, modern game engines that leverage DirectX 12 Ultimate features, and any workload that can use ray tracing. The 28 ray tracing cores give it a capability the R9 M395X simply does not have. Its higher pixel rate (128.0 GPixel/s versus 23.14 GPixel/s) means it can drive higher resolutions and refresh rates more effectively, and its 224.0 GB/s memory bandwidth, despite the narrower bus, supports these demands.

The R9 M395X’s only advantages are contextual. Its 256-bit memory bus is wider, which can be beneficial in certain bandwidth-sensitive legacy applications, but the actual bandwidth is lower than the RX 6700S. Its average benchmark score is higher (25,891 versus 22,154) and its percentile ranking is better (71st versus 67th), but this is an artifact of the limited test set, not a reflection of real-world superiority. In the two tests where both cards were measured, the R9 M395X is dramatically slower. The R9 M395X also has a lower TDP (75 W versus 80 W), a marginal difference. For users on legacy systems with MXM slots, the R9 M395X is the only one of the two that fits physically. The RX 6700S, as an IGP, is designed to be soldered onto a motherboard, so it is not a drop-in replacement for the older card.

FAQ

Q: Which GPU is faster in Geekbench Metal?

A: The AMD Radeon RX 6700S scores 72,864 points versus 33,953 for the R9 M395X, a delta of -53.4% in favor of the RX 6700S.

Q: How do the two cards compare in Geekbench OpenCL?

A: The RX 6700S scores 70,965 points, while the R9 M395X scores 17,829 points. The R9 M395X is 74.9% slower in this test.

Q: Does the R9 M395X support ray tracing?

A: No. The R9 M395X has no ray tracing cores. The RX 6700S includes 28 ray tracing cores.

Q: What is the memory configuration difference?

A: Both have 8 GB, but the R9 M395X uses GDDR5 on a 256-bit bus with 160.0 GB/s bandwidth, while the RX 6700S uses GDDR6 on a 128-bit bus with 224.0 GB/s bandwidth.

Q: Which card has a higher average benchmark score?

A: The R9 M395X has an average score of 25,891 and sits in the 71st percentile, while the RX 6700S has an average of 22,154 and sits in the 67th percentile. This is due to the RX 6700S having many more low-scoring benchmark entries.

Q: What are the process nodes for each GPU?

A: The R9 M395X is built on a 28 nm process, and the RX 6700S is built on a 7 nm process, both at TSMC.

Specification Differences

The following table lists only the fields where the two GPUs differ.

| Specification | AMD Radeon R9 M395X | AMD Radeon RX 6700S |

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

| Architecture | GCN 3.0 | RDNA 2.0 |

| Generation | Gem System (R9 M300) | Navi Mobile (RX 6000M) |

| Process Node | 28 nm | 7 nm |

| Transistors | 5,000 million | 11,060 million |

| Die Size | 366 mm² | 237 mm² |

| Transistor Density | 13.7M / mm² | 46.7M / mm² |

| Base Clock | Not listed | 1700 MHz |

| Game Clock | Not listed | 1890 MHz |

| Boost Clock | Not listed | 2000 MHz |

| Memory Clock | 1250 MHz (5 Gbps effective) | 1750 MHz (14 Gbps effective) |

| Memory Type | GDDR5 | GDDR6 |

| Bus Width | 256 bit | 128 bit |

| Memory Bandwidth | 160.0 GB/s | 224.0 GB/s |

| Shading Units | 2048 | 1792 |

| TMUs | 128 | 112 |

| ROPs | 32 | 64 |

| Ray Tracing Cores | None | 28 |

| Pixel Rate | 23.14 GPixel/s | 128.0 GPixel/s |

| Texture Rate | 92.54 GTexel/s | 224.0 GTexel/s |

| FP32 Performance | 2.961 TFLOPS | 7.168 TFLOPS |

| FP16 Performance | 2.961 TFLOPS (1:1) | 14.34 TFLOPS (2:1) |

| TDP | 75 W | 80 W |

| Slot Width | MXM Module | IGP |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

| DirectX Support | 12 (12_0) | 12 Ultimate (12_2) |

| Vulkan Support | 1.2.170 | 1.4 |

| Release Date | 2015-05-04 | 2022-01-03 |

| Predecessor | Solar System | Polaris Mobile |

| Successor | Polaris Mobile | None listed |

| Geekbench Metal Score | 33,953 | 72,864 |

| Geekbench OpenCL Score | 17,829 | 70,965 |

The Verdict

The data is unambiguous. The AMD Radeon RX 6700S is the superior GPU in every measurable performance category. It more than doubles the R9 M395X in Geekbench Metal and is nearly four times faster in Geekbench OpenCL. It brings ray tracing support, a modern 7 nm process, higher memory bandwidth, and significantly higher pixel and texture rates. Any user choosing between these two for a new system should select the RX 6700S without hesitation.

The R9 M395X retains relevance only in specific legacy scenarios. Its MXM Module form factor means it is the only one of the two that can be installed in older laptops designed for that interface. Its lower TDP of 75 W, while only 5 W less than the RX 6700S, might be a consideration in tightly constrained thermal envelopes. Its higher average benchmark score and percentile ranking, while misleading, do indicate that it is not a weak card by historical standards; it sits near the AMD Radeon RX 6750 XT and the NVIDIA GeForce RTX 3080 Ti Mobile in the database’s aggregate scoring.

For a builder with a modern laptop that supports an IGP-style discrete GPU, the RX 6700S is the only rational choice. It is faster, more feature-rich, and more efficient per unit of performance. For someone maintaining a legacy MXM-based machine, the R9 M395X is the only option of the two that will physically fit. The performance gap is enormous, but the choice is dictated by hardware compatibility rather than capability.