AMD Radeon HD 8790M vs Intel UHD Graphics 730 Comparison

# Intel UHD Graphics 730 vs AMD Radeon HD 8790M

The Intel UHD Graphics 730 and AMD Radeon HD 8790M occupy the same 33rd percentile in the benchmark database, yet they achieve that standing through completely different design philosophies and performance profiles. The Intel part, an integrated GPU from the Rocket Lake generation, averages 5929 across all benchmark tests, while the AMD discrete mobile GPU, based on GCN 1.0, averages 5691. The head-to-head results show a split decision: Intel wins Geekbench OpenCL by 19.4%, while AMD wins Geekbench Vulkan by 7.8%. This creates a nuanced comparison where the "better" GPU depends entirely on the workload and API in question.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The Intel UHD Graphics 730 posts an average benchmark score of 5929, which is approximately 4.2% higher than the AMD Radeon HD 8790M's 5691 average.

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

A: Intel wins decisively, scoring 5988 against AMD's 5017, representing a 19.4% advantage for the Intel UHD Graphics 730 in this compute API test.

Q: What happens in Geekbench Vulkan?

A: The tables turn. AMD scores 6365 versus Intel's 5870, giving the Radeon HD 8790M a 7.8% lead in this graphics API benchmark.

Q: Which GPU has more shading units?

A: The AMD Radeon HD 8790M features 384 shading units, exactly double the 192 shading units found in the Intel UHD Graphics 730.

Q: What are the memory configurations of these two GPUs?

A: The AMD Radeon HD 8790M has 2 GB of dedicated GDDR5 memory on a 128-bit bus with 64.00 GB/s bandwidth, while the Intel UHD Graphics 730 uses system-shared memory with bandwidth described as "System Dependent."

Q: Are both GPUs still in production?

A: No, both have reached end-of-life status. The Intel UHD Graphics 730 was released in 2021-03-29, while the AMD Radeon HD 8790M launched earlier on 2013-03-31.

Architecture Differences

The architectural gap between these two GPUs is substantial, reflecting their different release timelines and design goals. The Intel UHD Graphics 730 is built on Intel's Generation 12.1 architecture, implemented on a 14 nm+++ process node at Intel's own foundry. It derives from the Rocket Lake chip, positioning it as a modern integrated solution. The AMD Radeon HD 8790M, in contrast, uses the older GCN 1.0 architecture on a 28 nm process from TSMC, based on the Mars chip. This process disparity – 14 nm+++ versus 28 nm – explains some of the efficiency and feature differences.

The AMD GPU is built around the Mars chip with 950 million transistors on a 77 mm² die, yielding a transistor density of 12.3M per mm². The Intel UHD Graphics 730's transistor count and die size are not specified in the data, but its integration into the Rocket Lake processor means it shares silicon with CPU cores rather than existing as a standalone chip. This fundamental difference in packaging – IGP versus MXM Module – dictates how each GPU interacts with the rest of the system.

Shader resources differ dramatically. The AMD Radeon HD 8790M fields 384 shading units, 24 texture mapping units, and 8 raster operation units. The Intel UHD Graphics 730 counters with 192 shading units, 12 TMUs, and 8 ROPs. AMD has exactly twice the shading and texture hardware, while ROP counts match at 8 each. Clock speeds tell a different story: Intel runs at a base of 300 MHz with a boost of 1300 MHz, while AMD operates at an 850 MHz base and 900 MHz boost. The Intel part's higher boost clock partially compensates for its reduced shader count.

Memory architecture is another major divergence. AMD employs 2 GB of GDDR5 on a 128-bit bus with 64.00 GB/s of dedicated bandwidth and a 1000 MHz memory clock (4 Gbps effective). Intel's solution is entirely system-shared, with no dedicated VRAM, a system-dependent memory bus, and bandwidth that scales with the host platform's memory subsystem. This gives AMD a massive bandwidth advantage in memory-bound workloads, though Intel's shared memory can benefit from modern high-speed system RAM.

API support shows subtle differences. Intel supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. AMD supports DirectX 12 (11_1) – a slightly lower feature level – but matches OpenGL 4.6 and supports Vulkan 1.2.170. The generation gap in Vulkan support (1.4 versus 1.2.170) reflects Intel's newer architecture. Neither GPU includes ray tracing or tensor cores.

Head-to-Head Benchmarks

The two benchmark results in the head-to-head comparison tell a story of API-dependent performance. In Geekbench OpenCL, the Intel UHD Graphics 730 scores 5988 against AMD's 5017. This 19.4% margin is Intel's most decisive victory, and it aligns with Intel's position in the nearest rivals list, where it sits just 0.4% behind the AMD Radeon HD 8730M (5955) and 0.5% behind the NVIDIA Quadro K620M (5957). The Intel GPU's compute performance in OpenCL appears to punch above its modest hardware specifications, likely benefiting from the modern architecture and higher boost clock.

Geekbench Vulkan flips the outcome. The AMD Radeon HD 8790M scores 6365 versus Intel's 5870, a 7.8% advantage for AMD. This result places AMD's Vulkan performance notably higher than its OpenCL showing, suggesting the GCN architecture handles the Vulkan API particularly well. AMD's nearest rivals in this metric include the Intel Iris Pro Graphics P6300 at 5712 (0.4% behind) and the NVIDIA GeForce GTX 670MX at 5721 (0.5% behind), with the Quadro M500M at 5604 (1.6% behind). Notably, AMD's Vulkan score of 6365 exceeds all of its listed rivals' averages, while its OpenCL score of 5017 falls below them.

The average benchmark scores reflect these dynamics. Intel's 5929 average is anchored by two relatively close scores (5988 and 5870), showing consistency across APIs. AMD's 5691 average masks significant API variance: its Vulkan score of 6365 is 12.4% above its own average, while its OpenCL score of 5017 is 11.8% below it. This inconsistency suggests that AMD's GPU performance is heavily workload-dependent, whereas Intel delivers more uniform results.

Specification Differences

The specification sheets reveal fundamental differences across nearly every major category. The process nodes diverge sharply: Intel uses 14 nm+++ while AMD uses 28 nm, reflecting their respective fabrication generations. Foundries differ as well – Intel fabricates its own chip, while AMD relies on TSMC. The AMD GPU's transistor count is documented at 950 million on a 77 mm² die, whereas Intel's transistor count and die size are not specified in the data.

Clock speeds show Intel's strategy of lower base with higher boost: 300 MHz base up to 1300 MHz boost, versus AMD's 850 MHz base to 900 MHz boost. The memory subsystems are completely different: Intel uses system-shared memory with type, bus width, and bandwidth all listed as "System Shared" or "System Dependent," while AMD has 2 GB of GDDR5 with a 128-bit bus and a fixed 64.00 GB/s bandwidth. Memory clock differs accordingly, with AMD running at 1000 MHz (4 Gbps effective) and Intel having no dedicated memory clock.

Compute resources favor AMD in raw counts: 384 shading units versus 192, and 24 TMUs versus 12, with ROPs tied at 8. However, the fill rates tell a more nuanced story. Intel's pixel rate is 10.40 GPixel/s, which is higher than AMD's 7.200 GPixel/s, despite the equal ROP counts – a consequence of Intel's higher boost clock. AMD's texture rate of 21.60 GTexel/s surpasses Intel's 15.60 GTexel/s, leveraging its doubled TMU count. Floating-point performance similarly splits: AMD's FP32 of 691.2 GFLOPS exceeds Intel's 499.2 GFLOPS, but Intel offers FP16 at 998.4 GFLOPS (2:1 ratio) while AMD lists no FP16 capability.

Power and physical specifications differ by design philosophy. Intel's TDP is 15 W as an IGP, while AMD's TDP is not specified. Intel uses a Ring Bus interface and is an IGP slot type, while AMD uses an MXM-A (3.0) bus interface and an MXM Module form factor with no power connectors. Display outputs follow the same pattern: Intel is "Motherboard Dependent" while AMD is "Portable Device Dependent." The release dates are separated by eight years – Intel on 2021-03-29 and AMD on 2013-03-31 – and both are marked end-of-life.

Where Each One Wins

The Intel UHD Graphics 730 wins in OpenCL compute workloads, delivering a 19.4% advantage over the AMD Radeon HD 8790M. This makes it the stronger choice for applications that leverage OpenCL for general-purpose GPU computing, such as certain video encoding, image processing, or scientific workloads. Intel's consistency across both benchmark tests – scores of 5988 and 5870 – indicates more predictable performance regardless of API selection. The higher pixel rate of 10.40 GPixel/s also suggests an edge in fill-rate-bound scenarios, despite having only 8 ROPs. Additionally, Intel's FP16 support at 998.4 GFLOPS opens up half-precision compute paths that AMD cannot access. The 15 W TDP makes it suitable for power-constrained integrated designs, and its modern Vulkan 1.4 support ensures compatibility with the latest graphics API features.

The AMD Radeon HD 8790M wins in Vulkan-based workloads, outperforming Intel by 7.8% with a score of 6365. The 384 shading units and 24 TMUs provide substantial raw compute throughput, reflected in its FP32 performance of 691.2 GFLOPS and texture rate of 21.60 GTexel/s. The dedicated 2 GB GDDR5 memory with 64.00 GB/s bandwidth is a decisive advantage in memory-intensive applications, eliminating the contention for system memory bandwidth that integrated GPUs face. Games and applications that scale well with shader count and memory bandwidth will favor AMD. The 850 MHz base clock (versus Intel's 300 MHz) means AMD maintains respectable performance even without boost behavior, which can be beneficial in sustained workloads. For mobile platforms with MXM slots, AMD's discrete solution offers dedicated video memory and a self-contained module design.

The split is clean: Intel for OpenCL compute, API currency, and fill-rate-bound tasks; AMD for Vulkan graphics, shader-heavy workloads, and memory-bandwidth-hungry applications. Both GPUs sit at the 33rd percentile of all GPUs, indicating they target similar overall performance tiers despite their architectural differences. Users should select based on the specific API and workload profile they expect to encounter most frequently.