GPU Comparison

The NVIDIA GeForce GT 740 and the NVIDIA GeForce 920M are both end-of-life Kepler-based parts, but the benchmark data shows they are not equals. The GT 740 wins both head-to-head tests, with its advantage ranging from a narrow 3.3% in OpenCL to a commanding 43.3% in Vulkan. While the 920M is a more recent mobile design with a lower power draw, the desktop GT 740 delivers superior raw compute performance across the board, placing it in the 21st percentile of all GPUs compared to the 920M’s 20th percentile.

Head-to-Head Benchmarks

The most decisive result in the head-to-head comparison is the Geekbench Vulkan test. Here, the GT 740 scores 4083, while the 920M manages only 2849. That is a delta of 43.3%, meaning the GT 740 is nearly 45% faster in this API. This is a massive gap, not a marginal one. The Vulkan result indicates that the GT 740’s memory subsystem and rasterization throughput scale far better with modern, low-level graphics APIs. The 920M’s score of 2849 places it well behind, and the data suggests that any workload leveraging Vulkan will be severely bottlenecked on the mobile chip.

The OpenCL test tells a much closer story, but the winner is still the GT 740. The GT 740 posts 3847 points against the 920M’s 3725, a delta of just 3.3%. While this is a win, the margin is small enough that real-world differences in OpenCL compute tasks would be difficult to perceive. Both GPUs are within 122 points of each other, which is less than the variance seen between some of their nearest rivals. For instance, the GT 740’s closest rival, the NVIDIA Quadro 2000M, scores 3434, just 0.1% behind. The 920M’s closest rival, the GeForce GT 730M, scores 3316, 0.9% behind. The 3.3% delta between the two contenders is more significant than these rival gaps, but it is not a generational leap.

Looking at the overall average benchmark scores, the GT 740’s 3431 average is 4.4% higher than the 920M’s 3287. This aggregate figure confirms the head-to-head results: the GT 740 is consistently faster, but the margin is modest in compute-heavy tasks. The GT 740’s percentile ranking of 21 versus the 920M’s 20 reflects this narrow but real performance separation. In practical terms, the GT 740 sits just above the GeForce 920MX (which scores 3528) and well above the Intel HD Graphics 530 (3332), while the 920M is sandwiched between the GT 730M (3316) and the GT 640 (3210).

The texture and pixel rate figures reinforce the GT 740’s advantage. The GT 740 delivers 31.78 GTexel/s and 7.944 GPixel/s, while the 920M offers 30.53 GTexel/s and 7.632 GPixel/s. These are differences of about 4% and 4.1%, respectively, which align with the OpenCL margin. However, the Vulkan gap is far larger than any hardware specification would suggest, pointing to architectural or driver-level advantages for the desktop part.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GT 740 has an average benchmark score of 3431, while the NVIDIA GeForce 920M scores 3287. This gives the GT 740 a 4.4% lead in aggregate performance.

Q: How much faster is the GT 740 in the Vulkan test?

A: The GT 740 scores 4083 in Geekbench Vulkan, compared to the 920M’s 2849. This represents a 43.3% performance advantage for the GT 740.

Q: Is the GeForce 920M competitive in OpenCL workloads?

A: The 920M scores 3725 in Geekbench OpenCL, which is only 3.3% behind the GT 740’s 3847. While it loses, the margin is small enough that the 920M remains viable for basic OpenCL compute tasks.

Q: What are the memory specifications of each GPU?

A: The GT 740 uses 1024 MB of GDDR5 memory on a 128-bit bus, delivering 80.19 GB/s of bandwidth. The 920M uses 2 GB of DDR3 on a 64-bit bus, providing 14.40 GB/s of bandwidth.

Q: Which GPU has a higher transistor density?

A: The GeForce 920M has a transistor density of 11.7M / mm², which is higher than the GT 740’s 10.8M / mm². However, the GT 740 has a larger die size of 118 mm² versus 87 mm².

Q: How do these GPUs compare to their nearest rivals?

A: The GT 740’s closest rival is the NVIDIA Quadro 2000M, which is 0.1% behind. The 920M’s closest rival is the NVIDIA GeForce GT 730M, which is 0.9% behind. Both GPUs also sit near the Intel HD Graphics 530, with the GT 740 being 3% ahead and the 920M being 1.4% behind.

Where Each One Wins

The NVIDIA GeForce GT 740 wins in every benchmark category where both are tested. Its strongest showing is in Vulkan, where the 43.3% delta makes it the clear choice for any application or game using that API. The GT 740 also wins in OpenCL, albeit by a slim 3.3% margin. For users running compute tasks that rely on OpenCL, the GT 740 is the safer bet, but the 920M is not far behind.

The GeForce 920M, despite losing both head-to-head tests, has its own advantages that are not captured by raw performance scores. It has a TDP of only 33 W, compared to the GT 740’s 64 W. This makes the 920M suitable for thin and light laptops where power draw and heat are critical constraints. The 920M is also an integrated graphics processor (IGP), meaning it requires no power connectors and has a bus interface of PCIe 3.0 x8, whereas the GT 740 is a single-slot card requiring a 1x 6-pin power connector and a 250 W suggested PSU.

The 920M also offers double the memory capacity at 2 GB versus 1 GB, though this comes with a severe bandwidth penalty. The GT 740’s GDDR5 memory provides 80.19 GB/s, which is 5.6 times higher than the 920M’s 14.40 GB/s. For texture-heavy workloads or high-resolution rendering, the GT 740’s bandwidth is a decisive factor. The 920M’s larger frame buffer is only useful if the workload fits within its limited bandwidth, which is a rare scenario.

In terms of production status, both are end-of-life. The GT 740 was released on 2014-05-28, while the 920M came later on 2015-03-12. The GT 740 belongs to the GeForce 700 generation, and the 920M belongs to the GeForce 900M generation, but the 920M’s newer generation does not translate into better performance.

Specification Differences

The two GPUs differ in almost every major specification except for shading units, TMUs, and API support. Both have 384 shading units and 32 TMUs, and both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The differences begin with the chip: the GT 740 uses GK107, while the 920M uses GK208B.

The GT 740 has a process node of 28 nm, a die size of 118 mm², and 1,270 million transistors. The 920M also uses a 28 nm process but has a smaller die at 87 mm² and fewer transistors at 1,020 million. The transistor density is higher on the 920M at 11.7M / mm² versus 10.8M / mm², but the GT 740’s larger die allows for more total resources.

Memory is a major divergence. The GT 740 has 1024 MB of GDDR5 on a 128-bit bus, yielding 80.19 GB/s. The 920M has 2 GB of DDR3 on a 64-bit bus, yielding 14.40 GB/s. The GT 740’s memory clock is 1253 MHz (5 Gbps effective), while the 920M’s is 900 MHz (1800 Mbps effective). The 920M has a base and boost clock of 954 MHz, whereas the GT 740’s clocks are not listed.

The ROP count differs: the GT 740 has 16 ROPs, while the 920M has only 8. This directly impacts pixel rate, with the GT 740 achieving 7.944 GPixel/s versus 7.632 GPixel/s. The GT 740 also has a higher texture rate at 31.78 GTexel/s versus 30.53 GTexel/s, and a higher FP32 throughput at 762.6 GFLOPS versus 732.7 GFLOPS.

Power and physical specifications diverge sharply. The GT 740 has a TDP of 64 W, is single-slot, requires a 1x 6-pin power connector, and has a suggested PSU of 250 W. The 920M has a TDP of 33 W, is an IGP, requires no power connectors, and has no suggested PSU. The GT 740 uses a PCIe 3.0 x16 interface, while the 920M uses PCIe 3.0 x8. Display outputs also differ: the GT 740 has 2x DVI and 1x mini-HDMI 1.4a, while the 920M’s outputs are portable device dependent.

Architecture Differences

Both GPUs are built on NVIDIA’s Kepler architecture, but they are not identical. The GT 740 uses the original Kepler design, while the 920M is based on Kepler 2.0. The GT 740’s chip is GK107, and the 920M’s is GK208B. The GK208B is a smaller, more power-efficient chip, which explains the 920M’s lower TDP and lack of external power connectors.

The GT 740 has a larger die (118 mm²) and more transistors (1,270 million) than the 920M (87 mm² and 1,020 million). The 920M compensates with a higher transistor density, but the GT 740’s additional silicon translates into more ROPs (16 versus 8) and a wider memory bus (128-bit versus 64-bit). The GT 740’s memory type is GDDR5, which is fundamentally faster than the 920M’s DDR3.

The 920M’s Kepler 2.0 architecture does not introduce any feature that offsets its hardware deficits. Both GPUs support the same DirectX, OpenGL, and Vulkan versions. The GT 740’s higher pixel rate, texture rate, and FP32 throughput are all direct consequences of its larger chip and faster memory. The 920M’s only architectural advantage is its lower power consumption, which is a product of its smaller die and reduced ROP count.

The GT 740’s predecessor is the GeForce 600 series, and its successor is the GeForce 900 series. The 920M’s predecessor is the GeForce 800M series, and its successor is the GeForce 10 Mobile series. This places the 920M in a later mobile lineage, but the data shows that generational position does not guarantee performance superiority.

The Verdict

The NVIDIA GeForce GT 740 is the clear performance winner. It beats the GeForce 920M in both head-to-head benchmarks, with a 3.3% OpenCL advantage and a 43.3% Vulkan advantage. Its average benchmark score of 3431 is higher than the 920M’s 3287, and it ranks in the 21st percentile of all GPUs versus the 920M’s 20th. If the priority is raw compute or gaming performance, the GT 740 is the only choice.

The GeForce 920M is the better option for mobile, low-power systems. Its 33 W TDP, IGP form factor, and lack of power connectors make it suitable for laptops where the GT 740’s 64 W TDP and single-slot card design are impractical. The 920M also has 2 GB of memory, which is double the GT 740’s 1 GB, though this is offset by a much narrower 64-bit bus and DDR3 memory.

The data does not support the 920M for performance-critical tasks. Its Vulkan score of 2849 is abysmal compared to the GT 740’s 4083, and its nearest rivals include the Intel HD Graphics 530, which is only 1.4% behind. The GT 740, by contrast, sits 3% ahead of the same Intel integrated GPU. For desktop users with access to a PCIe x16 slot and a 250 W PSU, the GT 740 is the superior investment. For laptop buyers who need basic acceleration without power draw, the 920M is the only viable option, but its performance is firmly in integrated-graphics territory.