NVIDIA GeForce 940M vs NVIDIA RTX A400 Comparison

The NVIDIA RTX A400 and the NVIDIA GeForce 940M come from different eras and different product worlds, yet the database places them closer together than one might expect on overall average score: 6078 for the A400 versus 5284 for the 940M. That narrow average gap hides dramatic differences in the individual measurements, and questioning why the averages converge while the direct comparisons diverge is the key to understanding these two GPUs. One is an active workstation card built on Samsung's 8 nm process; the other is an end-of-life mobile Maxwell part on TSMC's 28 nm node. The data tells a story of nearly a decade of architectural change compressed into two small, power-frugal packages.

Where Each One Wins

The recorded head-to-head results are lopsided. The RTX A400 wins both shared benchmark entries, taking geekbench_opencl with 22844 against 6018 and geekbench_vulkan with 22237 against 4549. Those are not marginal wins. The opencl gap sits at 279.6 percent in the A400's favor, and the Vulkan gap widens to 388.8 percent. In compute-oriented API tests, the A400 is simply operating on a different level than the 940M, delivering roughly four times the throughput in the Vulkan case.

So where does the 940M win? Nowhere in the direct measurements, but context matters. Its percentile against all GPUs in the database is 31, which is not far below the A400's 35. The 940M's nearest rivals include the GeForce GTX 980M, GeForce 930A, GeForce 840M, and GeForce GTX 760M, all clustered within a point of its average score of 5284. This suggests the averaging method blends tests where the two cards land close together, likely because the A400's Passmark suite entries (passmark_g3d at 5983, passmark_gpu_compute at 2557) pull its average down in a way the 940M's sparse two-test record avoids. The database simply has more test coverage for the A400, and that asymmetry shapes the averages.

The practical read: for any workload measured here, general compute through OpenCL or graphics through Vulkan, the A400 is the stronger choice. The 940M's case rests entirely on its historical context as a mobile part whose rivals were other mobile GeForce chips of its generation.

Architecture Differences

The architectural gulf is stark. The RTX A400 uses the GA107 chip on the Ampere architecture, part of the Workstation Ampere (Ax000) generation, manufactured by Samsung on an 8 nm process. It packs 8,700 million transistors onto a 200 mm² die, yielding a density of 43.5M transistors per square millimeter. The GeForce 940M uses the GM107 chip on Maxwell, from the GeForce 900M generation, built by TSMC on a 28 nm process. Its transistor count is 1,870 million on a 148 mm² die, a density of 12.6M per square millimeter.

That density comparison is where the data gets interesting. The A400 achieves roughly three and a half times the transistor density of the 940M, which is exactly what one would expect moving from 28 nm to 8 nm manufacturing. Yet the die is only moderately larger in absolute terms, 200 mm² versus 148 mm², and the transistor budget has grown by a factor of about four and a half.

Feature-wise, the A400 carries 6 RT cores and 24 tensor cores, hardware the 940M lacks entirely, since Maxwell mobile chips predate both ray tracing and tensor acceleration. The A400 supports DirectX 12 Ultimate (feature level 12_2), while the 940M stops at DirectX 12 (feature level 11_0). Both report OpenGL 4.6 and Vulkan 1.4 support in the database.

Memory is another dividing line. The A400 pairs 4 GB of GDDR6 on a 64 bit bus, clocked at 1500 MHz with 12 Gbps effective signaling, for 96.00 GB/s of bandwidth. The 940M uses 2 GB of DDR3 on the same 64 bit bus width, at 900 MHz and 1800 Mbps effective, delivering just 14.40 GB/s. Same bus width, radically different memory technology: the A400 provides several multiples of the bandwidth, which helps explain why its compute scores scale so far ahead despite modest raw shader counts.

Power draw inverts expectations, though. The workstation A400, far faster, has a TDP of 50 W, while the older mobile 940M lists 75 W. Newer process technology lets the newer card do much more with less.

The Verdict

Based strictly on the recorded data, the RTX A400 is the better performer in every measured head-to-head test, and by wide margins. Anyone choosing between these two for compute or graphics work should pick the A400. It also has practical advantages that show up in the spec sheet: it is a single-slot desktop card measuring 163 mm long and 69 mm tall, needs no power connectors, supports PCIe 4.0 x8, and offers four mini-DisplayPort 1.4a outputs. Its suggested PSU rating is a modest 250 W.

The 940M has no recorded wins and is an end-of-life MXM module with portable-device-dependent display outputs. Its relevance is historical: it sits within a point of the GTX 980M, 930A, 840M, and GTX 760M in average score, marking it as a typical performer among its own generation's mobile parts. For anyone still weighing these two against each other, the data offers no scenario where the 940M measures ahead.

FAQ

Q: How much faster is the RTX A400 than the GeForce 940M in OpenCL compute?

A: The A400 scored 22844 in geekbench_opencl against the 940M's 6018, a 279.6 percent advantage for the A400.

Q: What about Vulkan performance?

A: The gap is even larger there. The A400 recorded 22237 versus 4549 for the 940M, a 388.8 percent difference in the A400's favor.

Q: Why are the two cards' average scores so close if the head-to-head gaps are so large?

A: The averages are 6078 for the A400 and 5284 for the 940M, and the database holds more test results for the A400, including Passmark entries such as passmark_g3d at 5983 and passmark_gpu_compute at 2557. The differing test mixes make the averages converge even though the shared direct comparisons favor the A400 heavily.

Q: Which card uses less power?

A: The A400, with a 50 W TDP, despite being far faster. The 940M lists a 75 W TDP.

Q: Do both cards support ray tracing hardware?

A: No. The A400 includes 6 RT cores and 24 tensor cores. The 940M has neither, as Maxwell mobile silicon predates these units.

Q: Are both cards still in production?

A: No. The A400's production status is Active, while the 940M is listed as End-of-life.

Head-to-Head Benchmarks

Only two tests appear in both cards' records, and the A400 sweeps them.

First, geekbench_opencl. The RTX A400 posts 22844; the GeForce 940M posts 6018. That is a 279.6 percent delta, meaning the A400 delivers close to four times the OpenCL throughput. Given the underlying hardware, this tracks: 2.706 TFLOPS of FP32 on the A400 against 1,124.4 GFLOPS on the 940M, plus 24 tensor cores and far greater memory bandwidth. The A400's 96.00 GB/s versus 14.40 GB/s means the older card is likely starved for data long before its shaders saturate.

Second, geekbench_vulkan. Here the A400 scores 22237 to the 940M's 4549, a 388.8 percent lead. Interestingly, the 940M's Vulkan result is notably lower than its own OpenCL result, dropping from 6018 to 4549, while the A400 stays nearly flat between the two APIs (22844 versus 22237). The data invites a question: does the 940M's feature level 11_0 DirectX support and lack of modern hardware features cost it disproportionately in the Vulkan path? The recorded numbers are consistent with that reading, though the database does not isolate the cause.

The win tally stands at 2 for the A400 and 0 for the 940M. There is no measured workload in the shared set where the older card competes.

Specification Differences

The cards diverge on nearly every specification. The A400 is built by Samsung on 8 nm; the 940M by TSMC on 28 nm. Transistor counts are 8,700 million versus 1,870 million, and die sizes are 200 mm² versus 148 mm², giving densities of 43.5M versus 12.6M per square millimeter.

Clocks favor the A400: 1417 MHz base and 1762 MHz boost, against 1020 MHz base and 1098 MHz boost. Memory differs in type (GDDR6 versus DDR3), capacity (4 GB versus 2 GB), and bandwidth (96.00 GB/s versus 14.40 GB/s), though both use a 64 bit bus.

Shader resources split in interesting ways. The A400 has 768 shading units and 24 TMUs; the 940M has 512 shading units but 32 TMUs. Both carry 16 ROPs. The 940M's higher texture unit count does not translate into a texture-rate win, however: the A400's clock advantage yields 42.29 GTexel/s against 35.14 GTexel/s, and its pixel rate of 28.19 GPixel/s beats the 940M's 17.57 GPixel/s.

The A400 lists FP16 throughput of 2.706 TFLOPS at 1:1 with FP32; the 940M has no FP16 figure recorded, and no RT or tensor cores. TDP is 50 W versus 75 W. Form factor separates them entirely: a single-slot desktop card versus an MXM module on an MXM-B (3.0) interface versus PCIe 4.0 x8. Release dates are 2024-04-15 for the A400 and 2015-03-12 for the 940M, and their market positioning differs accordingly, with the A400 succeeding Quadro Turing in the workstation line and the 940M followed by GeForce 10 Mobile in the consumer notebook line.