GPU Comparison
AMD Radeon HD 8790M
RTX A400
PERFORMANCE BENCHMARKS
Analysis: AMD Radeon HD 8790M vs NVIDIA RTX A400
The NVIDIA RTX A400 and AMD Radeon HD 8790M occupy opposite ends of the hardware timeline and performance spectrum. The RTX A400, a modern workstation entry from 2024, delivers overwhelming compute advantages, while the HD 8790M, a 2013-era mobile part, trails by massive margins in every measurable test. The data shows a categorical generational gap rather than a competitive matchup.
Where Each One Wins
The RTX A400 wins every benchmark category where both products have recorded scores. In OpenCL compute workloads, it scores 22,844 versus the HD 8790M's 5,017, a 355.3% advantage. In Vulkan graphics tests, the A400 posts 22,237 against 6,365, representing a 249.4% lead. These are not marginal victories; they are complete dominance across heterogeneous compute and modern graphics APIs.
The HD 8790M has no wins in the head-to-head comparison. It is a legacy mobile GPU designed for portability in an era before ray tracing, tensor cores, or modern API support. Its only benchmarks available are OpenCL and Vulkan, both of which it loses decisively. For users seeking a card to handle modern workstation applications, the RTX A400's feature set and raw throughput make it the only viable choice. For those specifically needing legacy compatibility with older GCN-era drivers or MXM module form factors, the HD 8790M retains a niche, but it cannot compete on performance.
FAQ
Q: Which GPU is faster in OpenCL compute tasks?
A: The NVIDIA RTX A400 scores 22,844 in Geekbench OpenCL, which is 355.3% higher than the AMD Radeon HD 8790M's score of 5,017.
Q: How do the two cards compare in Vulkan graphics performance?
A: The RTX A400 achieves 22,237 in Geekbench Vulkan, a 249.4% improvement over the HD 8790M's 6,365.
Q: What is the memory configuration difference?
A: The RTX A400 features 4 GB of GDDR6 memory on a 64-bit bus with 96.00 GB/s bandwidth. The HD 8790M has 2 GB of GDDR5 on a 128-bit bus with 64.00 GB/s bandwidth.
Q: Does the RTX A400 support modern APIs that the HD 8790M lacks?
A: Yes. The RTX A400 supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, while the HD 8790M is limited to DirectX 12 (11_1) and Vulkan 1.2.170.
Q: What are the production statuses of these cards?
A: The RTX A400 is listed as Active production, released on 2024-04-15. The HD 8790M is End-of-life, released on 2013-03-31.
Q: Which GPU has a higher average benchmark score?
A: The RTX A400 averages 6,078 across all benchmark tests, compared to the HD 8790M's 5,691, placing the A400 at the 35th percentile of all GPUs versus the HD 8790M's 33rd percentile.
Head-to-Head Benchmarks
The head-to-head results are stark. In Geekbench OpenCL, the RTX A400 scores 22,844 against the HD 8790M's 5,017. That is a delta of 355.3%, meaning the A400 delivers over four and a half times the compute throughput. This gap reflects the architectural chasm between a 2024 Ampere workstation chip and a 2013 GCN mobile part. The A400's 2.706 TFLOPS of FP32 performance and 768 shading units dwarf the HD 8790M's 691.2 GFLOPS and 384 shading units. Even the A400's texture rate of 42.29 GTexel/s is roughly double the HD 8790M's 21.60 GTexel/s.
In Geekbench Vulkan, the margin narrows slightly but remains overwhelming. The A400 posts 22,237 versus 6,365 for the HD 8790M, a 249.4% difference. Vulkan's lower-level overhead benefits the newer architecture more, as the A400's dedicated tensor cores and RT cores can accelerate certain workloads, though the HD 8790M lacks both. The A400's pixel rate of 28.19 GPixel/s is nearly four times the HD 8790M's 7.200 GPixel/s, further demonstrating fill-rate dominance. Across these two tests, the A400 wins both, with the average benchmark score reflecting a 387-point gap (6,078 vs 5,691).
Specification Differences
The most glaring difference is process technology: the RTX A400 uses Samsung's 8 nm node, while the HD 8790M uses TSMC's 28 nm. This translates to transistor density of 43.5M per mm² for the A400 versus 12.3M per mm² for the HD 8790M. The A400 packs 8,700 million transistors on a 200 mm² die, whereas the HD 8790M fits 950 million on 77 mm².
Memory differs fundamentally. The A400 has 4 GB of GDDR6 at 1500 MHz with 12 Gbps effective speed, yielding 96.00 GB/s bandwidth over a 64-bit bus. The HD 8790M has 2 GB of GDDR5 at 1000 MHz with 4 Gbps effective speed, yielding 64.00 GB/s over a 128-bit bus. The A400's narrower bus is compensated by faster memory technology and double the capacity.
Clock speeds favor the A400: base 1417 MHz and boost 1762 MHz versus the HD 8790M's base 850 MHz and boost 900 MHz. The A400's boost clock is nearly double the HD 8790M's base clock. Shading units count 768 for the A400 versus 384 for the HD 8790M. Texture mapping units are equal at 24, but ROPs differ: 16 for the A400 versus 8 for the HD 8790M.
The A400 also introduces features absent on the HD 8790M: 6 RT cores and 24 tensor cores. Power and form factor differ too: the A400 is a single-slot, 50 W card requiring no power connectors and a 250 W suggested PSU, while the HD 8790M is an MXM module with no listed TDP. The bus interface is PCIe 4.0 x8 for the A400 versus MXM-A (3.0) for the HD 8790M. Display outputs are 4x mini-DisplayPort 1.4a on the A400, while the HD 8790M's outputs are "Portable Device Dependent."
Architecture Differences
The RTX A400 is built on NVIDIA's Ampere architecture, the same design language as the workstation Ax000 series. It uses the GA107 chip, a 2024 release that succeeds Quadro Turing and precedes Workstation Ada. The HD 8790M uses AMD's GCN 1.0 architecture on the Mars chip, from the Solar System generation (HD 8700M). GCN 1.0 dates to 2013, succeeding London and preceding Gem System.
Ampere brings hardware-accelerated ray tracing via 6 RT cores and AI acceleration via 24 tensor cores. GCN 1.0 has neither. The A400 supports DirectX 12 Ultimate (12_2), enabling features like variable rate shading and mesh shaders. The HD 8790M is limited to DirectX 12 (11_1), lacking those modern features. Vulkan support also differs: the A400 runs Vulkan 1.4, while the HD 8790M runs Vulkan 1.2.170.
The A400's FP16 performance equals its FP32 at 2.706 TFLOPS (1:1 ratio), a hallmark of Ampere's unified compute design. The HD 8790M has no listed FP16 capability, reflecting its older GCN design that treated half-precision as secondary. The A400's foundry is Samsung at 8 nm, while the HD 8790M uses TSMC at 28 nm. This process advantage enables the A400's higher clocks and transistor density, but also explains its larger die (200 mm² vs 77 mm²) despite the HD 8790M having fewer features. The A400's 50 W TDP is remarkable given its performance, though the HD 8790M's TDP is unlisted, making direct power efficiency comparisons impossible. The architectural gap is not incremental; it is a full generation leap in both compute capability and feature support.