AMD Radeon 680M vs AMD Radeon R9 370X Comparison

The AMD Radeon R9 370X and AMD Radeon 680M are two GPUs separated by nearly a decade of architectural evolution, yet their benchmark data reveals a surprisingly close overall contest. The R9 370X, a 2015 discrete card, edges out the 2023 integrated graphics solution in average score (15862 vs 15270), but the 680M dominates in modern API workloads. The data shows a split decision: the R9 370X wins the OpenCL compute race by 10.3%, while the 680M crushes the Vulkan test by 58.9%. The R9 370X holds a 1.2 percentile advantage over all GPUs (58th vs 57th), but the 680M's architectural efficiency and feature set make it the more forward-looking choice. The verdict depends entirely on workload: legacy compute favors the discrete card, while modern gaming and API-heavy tasks belong to the integrated RDNA 2 part.

The Verdict

Pick the AMD Radeon R9 370X if your priority is raw compute throughput in OpenCL applications. Its score of 25893 in Geekbench OpenCL is 10.3% higher than the 680M's 23468, and its average benchmark score of 15862 places it within 0.1% of the AMD Radeon RX 7700 (15852) and just 1.7% behind the NVIDIA GeForce RTX 3060 Ti (16129). This is a card that sits in the 58th percentile of all GPUs, outperforming the 680M's 57th percentile ranking. It offers 2 GB of dedicated GDDR5 memory on a 256-bit bus with 179.2 GB/s of bandwidth, which is critical for memory-intensive compute tasks.

Choose the AMD Radeon 680M if you need modern API support, efficiency, and Vulkan performance. Its Geekbench Vulkan score of 21965 is more than double the R9 370X's 9018, a 58.9% advantage. It also delivers 3.379 TFLOPS of FP32 compute compared to 2.637 TFLOPS for the older card, and it has 12 dedicated ray tracing cores. The 680M's nearest rivals include the NVIDIA GeForce RTX 2060 (15290, only 0.1% higher) and the RTX 3050 OEM (15199, 0.5% lower), which shows its integrated design competes with dedicated GPUs from previous generations. Its 50 W TDP versus 180 W for the R9 370X makes it dramatically more power-efficient, and it supports PCIe 4.0 x8 versus PCIe 3.0 x16.

The data does not support a single "winner." The R9 370X wins one head-to-head benchmark; the 680M wins the other. The R9 370X has a higher average score but a lower modern API score. The 680M has better peak throughput in FP32 and FP16, but its memory bandwidth is system-dependent, which is a wildcard. For legacy compatibility and raw OpenCL number-crunching, the R9 370X is the choice. For anything involving Vulkan, ray tracing, or power-sensitive deployments, the 680M is the clear pick.

Architecture Differences

The architectural gap between these two GPUs is generational. The R9 370X uses the Trinidad chip based on GCN 1.0 architecture, manufactured on a 28 nm process at TSMC with 2,800 million transistors on a 212 mm² die. The 680M uses the Rembrandt+ chip based on RDNA 2.0 architecture, built on a 6 nm process at the same foundry with 13,100 million transistors on a 208 mm² die. The transistor density tells the story: the 680M packs 63.0 million transistors per mm² versus 13.2 million for the R9 370X — a 4.8x density advantage.

The 680M features 768 shading units, 48 TMUs, and 32 ROPs, clocked at 2000 MHz base and 2200 MHz boost. The R9 370X has 1280 shading units, 80 TMUs, and 32 ROPs, but at much lower clocks: 980 MHz base and 1030 MHz boost. Despite fewer shading units, the 680M achieves higher throughput due to its clock speed advantage — its 70.40 GPixel/s pixel rate and 105.6 GTexel/s texture rate both exceed the R9 370X's 32.96 GPixel/s and 82.40 GTexel/s.

The 680M also brings features the R9 370X lacks: 12 ray tracing cores, FP16 support at 6.758 TFLOPS (2:1 ratio), and DirectX 12 Ultimate (12_2) support. The R9 370X is limited to DirectX 12 (11_1), though both support OpenGL 4.6. Vulkan support differs significantly: the 680M supports Vulkan 1.4, while the R9 370X is capped at 1.2.170. Memory architecture is fundamentally different — the R9 370X uses dedicated GDDR5, while the 680M relies on system-shared memory with bandwidth described as "System Dependent."

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon R9 370X scores 15862 on average, while the AMD Radeon 680M scores 15270. The R9 370X is 3.9% higher and ranks in the 58th percentile of all GPUs compared to the 680M's 57th percentile.

Q: How do they compare in Vulkan performance?

A: The 680M is dramatically faster, scoring 21965 in Geekbench Vulkan versus 9018 for the R9 370X. This represents a 58.9% advantage for the 680M, which supports Vulkan 1.4 compared to the R9 370X's 1.2.170.

Q: What about OpenCL compute workloads?

A: The R9 370X wins here, scoring 25893 in Geekbench OpenCL versus 23468 for the 680M — a 10.3% lead. This aligns with the R9 370X's higher average score and its position near the AMD Radeon RX 7700 (15852, 0.1% delta).

Q: Does the 680M support ray tracing?

A: Yes, the 680M includes 12 ray tracing cores, a feature entirely absent from the R9 370X. It also supports DirectX 12 Ultimate (12_2), while the R9 370X is limited to DirectX 12 (11_1).

Q: Which card has better memory bandwidth?

A: The R9 370X has a clear advantage with 179.2 GB/s of dedicated GDDR5 bandwidth on a 256-bit bus. The 680M's memory is system-shared with bandwidth described as "System Dependent," meaning it varies based on the host system's RAM configuration.

Q: What are the power consumption differences?

A: The 680M is rated at 50 W TDP and requires no power connectors, while the R9 370X has a 180 W TDP and needs two 6-pin connectors. The R9 370X also suggests a 450 W power supply, whereas the 680M has no such requirement.

Specification Differences

• Process Node: 28 nm (R9 370X) vs 6 nm (680M)
• Transistors: 2,800 million vs 13,100 million
• Die Size: 212 mm² vs 208 mm²
• Transistor Density: 13.2M / mm² vs 63.0M / mm²
• Base Clock: 980 MHz vs 2000 MHz
• Boost Clock: 1030 MHz vs 2200 MHz
• Memory Size: 2 GB GDDR5 vs System Shared
• Memory Bus Width: 256 bit vs System Shared
• Memory Bandwidth: 179.2 GB/s vs System Dependent
• Shading Units: 1280 vs 768
• TMUs: 80 vs 48
• Ray Tracing Cores: None vs 12
• Pixel Rate: 32.96 GPixel/s vs 70.40 GPixel/s
• Texture Rate: 82.40 GTexel/s vs 105.6 GTexel/s
• FP32: 2.637 TFLOPS vs 3.379 TFLOPS
• FP16: Not supported vs 6.758 TFLOPS (2:1)
• TDP: 180 W vs 50 W
• Slot Width: Dual-slot vs IGP
• Power Connectors: 2x 6-pin vs None
• Suggested PSU: 450 W vs None
• Bus Interface: PCIe 3.0 x16 vs PCIe 4.0 x8
• DirectX Support: 12 (11_1) vs 12 Ultimate (12_2)
• Vulkan Support: 1.2.170 vs 1.4
• Production Status: End-of-life vs Active
• Release Date: 2015-08-26 vs 2023-01-02
• Launch MSRP: 199 USD vs Not available
• Display Outputs: 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 vs Portable Device Dependent

Head-to-Head Benchmarks

The head-to-head data shows exactly one win for each GPU, making this a balanced matchup. In Geekbench OpenCL, the R9 370X scores 25893 against the 680M's 23468, a 10.3% victory for the older discrete card. This result aligns with the R9 370X's overall average score of 15862, which places it just 0.1% above the AMD Radeon RX 7700 (15852) and 1.2% above the AMD Radeon Pro W5500 (15679). The 680M's OpenCL result of 23468 is its weakest benchmark showing.

In Geekbench Vulkan, the tables turn completely. The 680M scores 21965, while the R9 370X manages only 9018. This 58.9% difference is the largest margin in either direction and reflects the 680M's modern architecture and Vulkan 1.4 support versus the R9 370X's older GCN design and Vulkan 1.2.170. The 680M's Vulkan score is its strongest, contributing to its average of 15270 and its position near the NVIDIA GeForce GTX 580 (15283, 0.1% lower) and RTX 2060 (15290, 0.1% lower).

The R9 370X also holds a third benchmark not part of the head-to-head: its Geekbench Metal score of 12676, which the 680M does not have a comparable result for. The 680M has an exclusive 3DMark Steel Nomad DX12 score of 378, which the R9 370X lacks. These exclusive tests further separate the two, but the shared tests show a 1-1 tie.

Where Each One Wins

AMD Radeon R9 370X wins in:

• OpenCL compute workloads: A 10.3% lead over the 680M (25893 vs 23468) makes it the better choice for GPGPU tasks that rely on OpenCL.
• Dedicated memory bandwidth: 179.2 GB/s on a 256-bit bus versus system-shared memory gives it predictable performance in memory-bound applications.
• Average benchmark score: 15862 vs 15270, placing it higher in the percentile rankings (58th vs 57th).
• Legacy compatibility: DirectX 12 (11_1) and older display outputs (DVI, HDMI 1.4a, DisplayPort 1.2) suit older systems and software.

AMD Radeon 680M wins in:

• Vulkan performance: A massive 58.9% advantage (21965 vs 9018) makes it the clear choice for Vulkan-based games and applications.
• Ray tracing: 12 dedicated RT cores versus none on the R9 370X enables hardware-accelerated ray tracing.
• Modern API support: DirectX 12 Ultimate (12_2) and Vulkan 1.4 versus the R9 370X's more limited support.
• Peak compute throughput: 3.379 TFLOPS FP32 and 6.758 TFLOPS FP16 versus 2.637 TFLOPS FP32 and no FP16 support.
• Efficiency: 50 W TDP versus 180 W, with no power connectors or PSU requirement, making it ideal for laptops and low-power builds.
• Pixel and texture rates: 70.40 GPixel/s and 105.6 GTexel/s versus 32.96 GPixel/s and 82.40 GTexel/s, indicating faster rasterization in modern workloads.
• PCIe 4.0 support: x8 interface at PCIe 4.0 speeds versus PCIe 3.0 x16, offering higher bandwidth potential per lane.

The data-driven conclusion: the R9 370X is a compute-oriented relic with a strong OpenCL showing, while the 680M is a modern, efficient, and feature-rich IGP that excels in Vulkan and ray-traced scenarios. Choose based on your API needs and power constraints.