GPU Comparison

Head-to-Head Benchmarks

The recorded data shows a single head-to-head benchmark result, and it is an unambiguous victory for the NVIDIA GeForce GTX 670MX. In the Geekbench OpenCL test, the GTX 670MX scores 6125, while the GeForce 930A scores 5317. This represents a 15.2% performance advantage for the older Kepler-based mobile part.

This margin is substantial in real-world terms. The GTX 670MX delivers a score that sits 7.6% above the 930A's entire average benchmark score. The gap is large enough that the 930A would need a significant architectural efficiency improvement to close it, and the data shows it does not have one.

The GeForce 930A does have one counterpoint: it supports the Vulkan API at version 1.4, whereas the GTX 670MX supports Vulkan 1.2.175. That software feature difference, however, does not translate into a compute benchmark win. The OpenCL result is the only direct comparison available, and it favors the GTX 670MX by a clear margin.

The average benchmark scores reinforce the same conclusion. The GTX 670MX averages 5721 across its recorded tests, while the 930A averages 5317. That is a 404-point gap, or roughly 7.6% in favor of the GTX 670MX. The percentile rankings agree: the GTX 670MX sits at the 33rd percentile of all GPUs, while the 930A sits at the 31st. Neither card is a high-end performer, but the GTX 670MX is measurably stronger.

Architecture Differences

The two GPUs come from different NVIDIA design generations and use different chips. The GTX 670MX is built on the GK104 chip, part of the Kepler architecture, and belongs to the GeForce 600M generation. The 930A uses the GM108 chip, part of the Maxwell architecture, and belongs to the GeForce 900A generation.

Both are manufactured by TSMC on the same 28 nm process node. The transistor counts differ sharply: the GTX 670MX packs 3,540 million transistors on a 294 mm² die, yielding a transistor density of 12.0 million per square millimeter. The 930A has 1,020 million transistors on a 77 mm² die, with a density of 13.2 million per square millimeter. The Maxwell chip achieves a higher density, but it is much smaller in absolute terms.

The compute resources show a major gap. The GTX 670MX has 960 shading units, 80 texture mapping units, and 24 ROPs. The 930A has 384 shading units, 24 TMUs, and 8 ROPs. The GTX 670MX has 2.5 times the shader count, 3.33 times the texture units, and 3 times the ROPs. These are not small differences; they define the performance ceiling for each part.

Clock speeds favor the 930A. Its base clock is 928 MHz and boost clock is 941 MHz, compared to 601 MHz for both base and boost on the GTX 670MX. The Maxwell chip runs 54% faster at base clock. But the GTX 670MX's massive resource advantage more than compensates.

Memory configurations diverge completely. The GTX 670MX has 3 GB of GDDR5 on a 192-bit bus, delivering 67.20 GB/s of bandwidth. The 930A has 2 GB of DDR3 on a 64-bit bus, delivering only 16.02 GB/s. That is a 4.19 times bandwidth advantage for the GTX 670MX. The memory type difference matters: GDDR5 is far faster than DDR3 for graphics workloads, and the wider bus amplifies that gap.

The bus interface also differs. The GTX 670MX uses PCIe 3.0 x16, while the 930A uses PCIe 3.0 x8. The x16 interface provides twice the bandwidth to the host system, which can benefit data transfer in some workloads.

Power consumption shows the 930A's efficiency advantage. The GTX 670MX is rated at 75 W, while the 930A is rated at 33 W. The newer Maxwell part draws less than half the power of the older Kepler part. This is the 930A's strongest architectural selling point.

The GTX 670MX uses the GK104 chip, which is a larger, more power-hungry design. It has no power connectors, relying on the motherboard slot, same as the 930A. The 930A has a slot width of IGP (integrated graphics processor), suggesting it is designed for compact or low-power systems.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 670MX averages 5721 across its recorded benchmarks, while the NVIDIA GeForce 930A averages 5317. The GTX 670MX leads by 404 points, or about 7.6%.

Q: What is the performance difference in the only direct head-to-head test?

A: In the Geekbench OpenCL test, the GTX 670MX scores 6125 against 5317 for the 930A. The GTX 670MX wins by 15.2%.

Q: Does the GeForce 930A have any advantage in API support?

A: Yes. The 930A supports Vulkan 1.4, while the GTX 670MX supports Vulkan 1.2.175. Both support DirectX 12 (11_0) and OpenGL 4.6.

Q: How do the memory bandwidth figures compare?

A: The GTX 670MX has a memory bandwidth of 67.20 GB/s using 3 GB of GDDR5 on a 192-bit bus. The 930A has 16.02 GB/s using 2 GB of DDR3 on a 64-bit bus. The GTX 670MX has 4.19 times the bandwidth.

Q: Which GPU has more shading units?

A: The GTX 670MX has 960 shading units, compared to 384 on the 930A. That is a 2.5 times advantage for the GTX 670MX.

Q: What is the power consumption difference?

A: The GTX 670MX is rated at 75 W, while the 930A is rated at 33 W. The 930A consumes less than half the power of the GTX 670MX.

The Verdict

The data points to a clear winner for raw compute performance: the NVIDIA GeForce GTX 670MX. Its 15.2% lead in the OpenCL benchmark, combined with a 404-point advantage in average score, makes it the stronger GPU for any task that relies on shader throughput or memory bandwidth.

The GTX 670MX has 2.5 times the shading units, 3.33 times the texture units, and 3 times the ROPs compared to the 930A. It also has 4.19 times the memory bandwidth. These are structural advantages that clock speed cannot overcome. The 930A's higher clocks (928 MHz base vs 601 MHz) are not enough to close the resource gap.

The 930A's case rests on efficiency and modern features. It draws 33 W versus 75 W, making it suitable for systems with tight power budgets. It supports Vulkan 1.4, a newer API version. Its smaller die (77 mm² vs 294 mm²) and lower transistor count (1,020 million vs 3,540 million) mean it is a simpler, cheaper part to produce.

For users who need maximum performance from a mobile GPU, the GTX 670MX is the correct choice. The benchmark data shows it wins the only direct comparison and holds a higher percentile ranking (33rd vs 31st). The 930A is only preferable when power draw is the primary constraint and performance is secondary.

The GTX 670MX sits among rivals like the Intel Iris Pro Graphics P6300 (0.2% ahead), the NVIDIA GeForce GTX 550 Ti (0.2% behind), and the AMD Radeon HD 8790M (0.5% ahead). The 930A sits near the NVIDIA GeForce 840M (0.1% behind), the NVIDIA GeForce GTX 980M (0.2% ahead), and the NVIDIA GeForce 940M (0.6% ahead). Both cards are firmly in the mid-to-low end of the performance spectrum, but the GTX 670MX is the stronger of the two.

Specification Differences

The two GPUs differ in nearly every major specification. The GTX 670MX uses the GK104 chip with the Kepler architecture, while the 930A uses the GM108 chip with Maxwell. The GTX 670MX belongs to the GeForce 600M generation, and the 930A is part of the GeForce 900A generation.

The GTX 670MX has 3,540 million transistors on a 294 mm² die. The 930A has 1,020 million transistors on a 77 mm² die. Transistor density is 12.0M per mm² for the GTX 670MX and 13.2M per mm² for the 930A.

Clock speeds: the GTX 670MX runs at 601 MHz base and 601 MHz boost, with memory at 700 MHz (2.8 Gbps effective). The 930A runs at 928 MHz base and 941 MHz boost, with memory at 1001 MHz (2 Gbps effective).

Memory: the GTX 670MX has 3 GB of GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth. The 930A has 2 GB of DDR3 on a 64-bit bus with 16.02 GB/s bandwidth.

Compute resources: the GTX 670MX has 960 shading units, 80 TMUs, and 24 ROPs. The 930A has 384 shading units, 24 TMUs, and 8 ROPs.

Pixel and texture rates: the GTX 670MX achieves 12.02 GPixel/s and 48.08 GTexel/s. The 930A achieves 7.528 GPixel/s and 22.58 GTexel/s. FP32 performance is 1,153.9 GFLOPS for the GTX 670MX and 722.7 GFLOPS for the 930A.

Power: the GTX 670MX is rated at 75 W, and the 930A is rated at 33 W. The GTX 670MX uses PCIe 3.0 x16, while the 930A uses PCIe 3.0 x8. The 930A has a slot width of IGP. Both have no power connectors.

API support: both support DirectX 12 (11_0) and OpenGL 4.6. The GTX 670MX supports Vulkan 1.2.175, while the 930A supports Vulkan 1.4.

Release timing: the GTX 670MX was released on 2012-09-30, and the 930A on 2015-03-12. Both are end-of-life products.

Where Each One Wins

The GTX 670MX wins in every performance metric recorded. It has higher raw compute in OpenCL (6125 vs 5317), a higher average benchmark score (5721 vs 5317), and a higher percentile ranking (33rd vs 31st). It wins the head-to-head by 15.2%.

The GTX 670MX is the better choice for any workload that stresses the GPU's compute resources. Its 960 shading units and 80 TMUs handle parallel workloads more effectively. Its 67.20 GB/s memory bandwidth is essential for texture-heavy scenarios, and its 3 GB GDDR5 frame buffer provides more capacity for high-resolution assets. The 192-bit bus is a major advantage over the 64-bit bus on the 930A.

The GTX 670MX also wins on interface width. PCIe 3.0 x16 provides twice the host bandwidth of the 930A's PCIe 3.0 x8. This can matter for workloads that transfer data between CPU and GPU frequently.

The 930A wins on power efficiency. Its 33 W TDP is less than half the GTX 670MX's 75 W. This makes it suitable for thin-and-light laptops or systems with limited cooling and battery capacity. The 930A also has a newer Vulkan implementation (1.4 vs 1.2.175), which could matter for applications that specifically leverage newer Vulkan features.

The 930A also has a higher clock speed (941 MHz boost vs 601 MHz boost), but this does not translate into a performance win in the recorded data. The architectural resource deficit is too large.

The GTX 670MX is the pick for performance. The 930A is the pick for efficiency. There is no benchmark result where the 930A outperforms the GTX 670MX, so any decision favoring the 930A must be based on power constraints or software feature requirements, not on measured speed.