AMD FirePro W4170M vs NVIDIA GeForce 930M Comparison

FAQ

Q: Which GPU has the higher Geekbench OpenCL score, the AMD FirePro W4170M or the NVIDIA GeForce 930M?

A: The NVIDIA GeForce 930M edges out the AMD FirePro W4170M with a score of 5046 versus 5034, a difference of 0.2% in favor of the NVIDIA part.

Q: How do the two compare in terms of memory bandwidth?

A: The AMD FirePro W4170M has a substantial advantage, offering 64.00 GB/s of bandwidth over a 128-bit bus, while the NVIDIA GeForce 930M delivers 12.80 GB/s over a 64-bit bus. This is a roughly 5x difference in raw bandwidth.

Q: What is the transistor count and die size for each chip?

A: The AMD chip, codenamed Opal, packs 950 million transistors on a 77 mm² die. The NVIDIA chip, GM108S, contains 1,020 million transistors on the same 77 mm² die size, giving it a slightly higher transistor density of 13.2M per mm² versus 12.3M per mm².

Q: Which GPU supports a newer version of DirectX?

A: The AMD FirePro W4170M supports DirectX 12 (11_1), while the NVIDIA GeForce 930M supports DirectX 12 (11_0). The AMD part has a slightly higher feature level.

Q: How do the pixel and texture rates compare?

A: The AMD FirePro W4170M achieves a pixel rate of 7.200 GPixel/s and a texture rate of 21.60 GTexel/s. The NVIDIA GeForce 930M trails with 4.392 GPixel/s and 13.18 GTexel/s, respectively.

Q: What is the average benchmark score for each GPU across all recorded tests?

A: The AMD FirePro W4170M has an average benchmark score of 5034, while the NVIDIA GeForce 930M has a lower average of 4388, due to its additional Vulkan test score of 3729 pulling the average down.

Architecture Differences

The architectural divide between these two mobile GPUs is clear. The AMD FirePro W4170M is built on GCN 1.0, a design that emphasizes compute throughput and balanced shader workloads. The NVIDIA GeForce 930M uses the Maxwell architecture, which focuses on efficiency and higher clocks per watt, though in this case the clocks are notably low.

Both chips are manufactured on the same 28 nm process at TSMC, and both use a 77 mm² die. However, NVIDIA fits more transistors into that space: 1,020 million versus AMD's 950 million, resulting in a density of 13.2M per mm² compared to AMD's 12.3M per mm². This density difference hints at a more compact or differently organized shader layout in the Maxwell design.

The memory subsystems could not be more different. The AMD part pairs its 2 GB frame buffer with GDDR5 memory on a 128-bit bus, yielding 64.00 GB/s of bandwidth. The NVIDIA part also has 2 GB, but uses DDR3 on a 64-bit bus, which caps bandwidth at just 12.80 GB/s. This is the single largest architectural gap between the two, and it drives most of the performance differences in bandwidth-sensitive workloads.

Clock behavior also diverges. AMD runs a base clock of 825 MHz with a boost up to 900 MHz, while NVIDIA runs a flat 549 MHz for both base and boost. The NVIDIA memory clock is also lower at 800 MHz (1600 Mbps effective) versus AMD's 1000 MHz (4 Gbps effective). The AMD GPU compensates for its older architecture with much higher clocks and a wider memory interface.

Feature support differs in the API stack. AMD lists Vulkan 1.2.170, while NVIDIA lists Vulkan 1.4, a newer revision. DirectX support favors AMD slightly with version 12 (11_1) versus NVIDIA's 12 (11_0). Both support OpenGL 4.6. The NVIDIA part also has a rated TDP of 33 W, while the AMD part has no TDP listed in the database. Slot widths differ too: AMD uses an MXM Module form factor, while NVIDIA is listed as IGP.

Head-to-Head Benchmarks

The only direct head-to-head benchmark recorded in the database is Geekbench OpenCL. In that test, the NVIDIA GeForce 930M scores 5046 against the AMD FirePro W4170M's 5034. The delta is a mere 0.2% in NVIDIA's favor. This is not a meaningful performance gap; it falls well within run-to-run variance for most OpenCL workloads.

The AMD part does not have a Vulkan score recorded, so the only comparison point is OpenCL. NVIDIA's Vulkan score of 3729 cannot be compared directly, as no AMD equivalent exists in the data. Still, the OpenCL result shows two GPUs that are effectively tied in raw compute throughput, despite their very different memory architectures and clock speeds.

What is striking is that the AMD GPU achieves parity with NVIDIA in OpenCL while operating with a much higher memory bandwidth (64.00 GB/s versus 12.80 GB/s). This suggests that the OpenCL workload is not bandwidth-limited, or that the NVIDIA architecture extracts more useful work per byte of memory traffic. The Maxwell design appears to be more efficient per clock and per memory transaction, compensating for its low clocks and narrow bus.

The delta percentages against nearest rivals are also telling. The AMD FirePro W4170M sits 0.3% above the AMD Radeon R5 M430 and 0.7% above the AMD Radeon R7 Graphics, but 0.6% below both the AMD Radeon R7 M340 and the AMD Radeon R7 240. The NVIDIA GeForce 930M is 0.5% below the NVIDIA GeForce GT 645M, 0.7% above the Intel Iris Pro Graphics 5200, 1.2% above the NVIDIA GeForce RTX 4070 GDDR6 (a surprising entry in the rival list), and 2.2% above the AMD FirePro W2100. These margins are all small, placing both GPUs in a dense cluster of similar-performance mobile parts.

Specification Differences

The following fields differ between the two GPUs in the database. The AMD FirePro W4170M has a chip named Opal, while the NVIDIA GeForce 930M uses GM108S. The architecture differs (GCN 1.0 versus Maxwell), as does the generation (FirePro Mobile Wx100M versus GeForce 900M).

Transistor counts differ: 950 million for AMD, 1,020 million for NVIDIA. Transistor density follows, at 12.3M per mm² versus 13.2M per mm². Base clocks are 825 MHz versus 549 MHz, and boost clocks are 900 MHz versus 549 MHz. Memory clocks are 1000 MHz (4 Gbps effective) versus 800 MHz (1600 Mbps effective). Memory type is GDDR5 versus DDR3, and bus width is 128 bit versus 64 bit. Memory bandwidth is 64.00 GB/s versus 12.80 GB/s.

Pixel rates are 7.200 GPixel/s versus 4.392 GPixel/s. Texture rates are 21.60 GTexel/s versus 13.18 GTexel/s. FP32 compute is 691.2 GFLOPS versus 421.6 GFLOPS. The NVIDIA part has a TDP of 33 W, while the AMD part has none listed. Slot width is MXM Module versus IGP. DirectX support is 12 (11_1) versus 12 (11_0). Vulkan support is 1.2.170 versus 1.4.

Release dates differ by about a month: the AMD part launched on 2015-04-22, while the NVIDIA part launched on 2015-03-12. The predecessor and successor names also differ: AMD lists FirePro Mobility as predecessor and Radeon Pro Mobile as successor, while NVIDIA lists GeForce 800M and GeForce 10 Mobile. The AMD part has a single benchmark score (5034) and a percentile of 30, while the NVIDIA part has two scores (5046 and 3729), an average of 4388, and a percentile of 26.

Where Each One Wins

The NVIDIA GeForce 930M wins the only head-to-head benchmark, the Geekbench OpenCL test, by 0.2%. It also has a higher peak score in that test. The NVIDIA part offers Vulkan 1.4 support, which is a newer API revision than AMD's 1.2.170. Its lower TDP of 33 W may make it easier to integrate into thin-and-light designs, though the AMD part's TDP is simply not listed.

The AMD FirePro W4170M wins decisively in raw specification categories. Its FP32 compute of 691.2 GFLOPS is 64% higher than the NVIDIA part's 421.6 GFLOPS. Pixel rate is 64% higher, texture rate is 64% higher, and memory bandwidth is exactly 5 times higher. The AMD part also has a wider memory bus (128 bit versus 64 bit) and faster memory type (GDDR5 versus DDR3). The AMD GPU supports DirectX 12 (11_1), a slightly higher feature level than NVIDIA's 12 (11_0).

In terms of percentile ranking, the AMD part sits at the 30th percentile of all GPUs, while the NVIDIA part sits at the 26th percentile. This suggests that AMD's single OpenCL score places it slightly higher in the overall distribution, despite losing the head-to-head matchup.

The nearest rival data reinforces the AMD part's position. The AMD FirePro W4170M is within 0.7% of the AMD Radeon R7 Graphics and within 0.6% of the AMD Radeon R7 240, both of which are integrated or low-end desktop parts. The NVIDIA GeForce 930M is 2.2% faster than the AMD FirePro W2100 and 1.2% faster than the NVIDIA GeForce RTX 4070 GDDR6 in this specific benchmark cluster, which is an unusual grouping that reflects the narrow score range.

The Verdict

The benchmark data shows two GPUs that are effectively tied in compute performance, despite radically different approaches to memory and clock design. The NVIDIA GeForce 930M wins the OpenCL head-to-head by 0.2%, which is negligible. The AMD FirePro W4170M counters with vastly superior memory bandwidth, higher pixel and texture rates, and 64% more FP32 compute. These specifications suggest the AMD part should excel in workloads that are bandwidth-bound or that leverage high fill rates, such as texture-heavy rendering or compute tasks that stream large datasets.

The NVIDIA GeForce 930M, by contrast, achieves near-parity with a fraction of the memory bandwidth and much lower clocks. This points to a more efficient architecture that extracts more performance per byte and per clock. For users whose workloads are latency-sensitive or where the API support matters, the NVIDIA part's Vulkan 1.4 support and lower TDP of 33 W could be deciding factors.

The AMD part is the better choice for anyone running OpenCL compute that can utilize its bandwidth advantage, or for applications that benefit from DirectX 12 (11_1) features. The NVIDIA part is the better choice for users who need Vulkan 1.4 support, who prioritize lower power draw, or who are building into an IGP-form-factor system. The data does not support a clear overall winner; it supports a split decision based on workload characteristics.

Both GPUs are end-of-life products, and both sit near the 26th to 30th percentile of all GPUs, indicating they are low-end parts by modern standards. The recorded benchmark scores show they belong to the same performance tier, with the AMD part offering more headroom in memory-bound tasks and the NVIDIA part offering better architectural efficiency and newer API support. Choose accordingly based on the specific application requirements, not on the near-tied OpenCL scores.