AMD Radeon R5 M420 vs NVIDIA GeForce 920M Comparison

The AMD Radeon R5 M420 and NVIDIA GeForce 920M are two entry-level mobile GPUs from the mid-2010s that now sit near the bottom of the historical performance pile, and the recorded data shows a narrow but consistent edge for the AMD card. Both are end-of-life products built on TSMC's 28 nm process, both connect over PCIe 3.0 x8, and both use DDR3 memory on a 64-bit bus, so the differences between them come down to core counts, clocks, and how each architecture converts those resources into measured throughput. In the one direct comparison available in the database, the Geekbench OpenCL test, the R5 M420 scores 3956 against the 920M's 3725, a 6.2 percent advantage for AMD. That single data point, combined with the cards' respective percentile placements, tells most of the story: the R5 M420 sits at the 23rd percentile of all GPUs, the 920M at the 20th, and the gap between them is real but small.

The Verdict

The data supports a straightforward recommendation. If the choice is strictly between these two GPUs, the Radeon R5 M420 is the pick based on measured compute performance: it wins the only head-to-head benchmark recorded, leading by 6.2 percent in Geekbench OpenCL, and it holds a higher percentile placement against the full GPU population. It also carries twice the memory capacity, 4 GB versus 2 GB, which matters for workloads that exceed a 2 GB footprint even at entry-level performance tiers. The GeForce 920M's counterargument is raw specification, not measurement: it has more shading units (384 versus 320), more texture units (32 versus 20), a higher clock (954 MHz versus a boost of 850 MHz), and substantially higher theoretical texture rate (30.53 GTexel/s versus 17.00 GTexel/s) and FP32 throughput (732.7 GFLOPS versus 544.0 GFLOPS). Yet the benchmark data does not convert that paper advantage into a win, at least in the OpenCL compute test available. Buyers choosing on measured results should take the R5 M420; the 920M's case rests on theoretical texture and shading numbers that the database cannot confirm in the tests recorded here. Neither card is competitive by modern standards, sitting in the low twenties of percentile rankings, so both are relevant only as legacy hardware.

FAQ

Q: Which GPU is faster in the recorded benchmarks?

A: The AMD Radeon R5 M420. In the Geekbench OpenCL test it scores 3956 versus 3725 for the GeForce 920M, a 6.2 percent lead. It is the only benchmark where both cards appear, and the R5 M420 wins it.

Q: How do the two compare against the wider GPU landscape?

A: The R5 M420 sits in the 23rd percentile of all GPUs in the database; the 920M sits in the 20th. Both are firmly entry-level, near the bottom of the historical distribution.

Q: Does the GeForce 920M have any benchmark result the R5 M420 lacks?

A: Yes. The 920M has a recorded Geekbench Vulkan score of 2849, a test for which no R5 M420 result exists in the database. This covers modern API graphics workloads, but it cannot be compared directly since the AMD card has no matching entry.

Q: Which card has more memory?

A: The R5 M420, with 4 GB of DDR3 versus 2 GB on the 920M. Both use a 64-bit bus, but the AMD card's memory runs faster in effective terms (2 Gbps versus 1800 Mbps), giving it 16.00 GB/s of bandwidth versus 14.40 GB/s.

Q: Which GPUs are these cards' closest rivals?

A: The R5 M420's nearest rivals are the NVIDIA GeForce 830M, GeForce GT 745M, Quadro K2000, and Radeon HD 6850 X2, all with average scores within about half a percent of its own. The 920M's rivals are the GeForce GT 730M, Intel HD Graphics 530, GeForce GT 640, and Intel HD Graphics P4600.

Q: Are both cards still in production?

A: No. Both are listed as end-of-life. The R5 M420 was released in May 2016 and the 920M in March 2015.

Architecture Differences

The two cards represent different design philosophies from the same era. The R5 M420 uses AMD's GCN 1.0 architecture on the Jet chip, part of the Gem System generation that carried the R5 M400 naming. The 920M uses NVIDIA's Kepler 2.0 architecture on the GK208B chip, part of the GeForce 900M series. Both were fabricated by TSMC on a 28 nm process, so neither holds a node advantage. The implementation differs considerably, though. The R5 M420's die is smaller, 56 mm² with 690 million transistors, yielding a density of 12.3 million transistors per square millimeter. The 920M's die is larger at 87 mm² with 1,020 million transistors, but slightly less dense at 11.7 million per square millimeter. In short, NVIDIA spent more silicon on this performance tier than AMD did.

Feature support is close, with a subtle edge to AMD in feature level. Both report DirectX 12 capability, but the R5 M420 lists feature level 11_1 versus 11_0 on the 920M. OpenGL support is identical at 4.6. Vulkan versions are nearly matched, 1.2.170 for the R5 M420 and 1.2.175 for the 920M. Neither card has RT cores or tensor cores, as expected for hardware of this class and vintage. Lineage also differs: the R5 M420 succeeded AMD's Solar System generation and was itself followed by Polaris Mobile, while the 920M followed the GeForce 800M series and was succeeded by GeForce 10 Mobile.

Specification Differences

The cards differ meaningfully across their core configurations despite sharing a platform envelope. Core counts favor NVIDIA: the 920M has 384 shading units and 32 texture units against the R5 M420's 320 and 20 respectively. Raster output units are tied at 8. Clocks also favor NVIDIA, with the 920M running a flat 954 MHz base and boost, while the R5 M420 runs 780 MHz base and 850 MHz boost. These combine into clear theoretical throughput advantages for the 920M: 732.7 GFLOPS FP32 versus 544.0, 30.53 GTexel/s texture rate versus 17.00, and 7.632 GPixel/s pixel rate versus 6.800. Memory is where AMD pulls ahead. The R5 M420 carries 4 GB of DDR3 at an effective 2 Gbps on its 64-bit bus, producing 16.00 GB/s of bandwidth. The 920M has 2 GB of DDR3 at an effective 1800 Mbps on the same bus width, producing 14.40 GB/s. Power data is only recorded for the 920M, with a 33 W TDP and no power connectors; the database has no TDP entry for the R5 M420, so thermal comparisons cannot be made numerically. Both are IGP-slot-class parts with portable-device-dependent display outputs and the same PCIe 3.0 x8 interface.

Head-to-Head Benchmarks

Only one test contains results for both cards, but it is decisive within its scope. In Geekbench OpenCL, a general GPU compute benchmark, the R5 M420 scores 3956 and the 920M scores 3725. That is a 6.2 percent advantage for AMD, which is modest in absolute terms but meaningful at this performance tier, where rivals cluster tightly. The database confirms how tight that clustering is: the R5 M420's nearest rivals, the GeForce 830M, GT 745M, Quadro K2000, and Radeon HD 6850 X2, all score within roughly half a percent of it, so a 6.2 percent win over the 920M is several times larger than the gaps separating it from its own peer group. The 920M's peer group sits lower: the GT 730M, HD Graphics 530, GT 640, and HD Graphics P4600 average between 3210 and 3389, all below the 920M's OpenCL figure and its 3287 overall average. The 920M does hold a result the R5 M420 lacks, a Geekbench Vulkan score of 2849, which documents its modern-API graphics capability, but with no counterpart entry for the AMD card it stands as context rather than comparison. The tally in the database is one win for the R5 M420 and none for the 920M.

Where Each One Wins

The R5 M420 wins where measured compute and memory capacity matter. Its OpenCL victory makes it the better choice for general GPU compute tasks reflected in that test, and its 4 GB frame buffer plus higher bandwidth suit workloads or games that need more than 2 GB of memory, a hard limit on the 920M that no clock speed can overcome. Its higher percentile placement, 23rd versus 20th, reinforces that it is the stronger performer in the database's own ranking. The 920M wins on paper in texture-heavy and shading-bound scenarios, at least theoretically: its 79 percent higher texture rate and 35 percent higher FP32 throughput suggest an advantage in fill-rate-limited situations, though the recorded benchmark data does not include a test that demonstrates this. It also offers a documented Vulkan result, evidence of tested modern-API functionality, and a defined 33 W thermal envelope that system builders can plan around, something the database cannot provide for the AMD card. On the evidence available, the R5 M420 is the measured winner and the sensible default; the 920M is the choice only where its theoretical texture and shading numbers, or its documented Vulkan and power data, specifically align with the workload.