NVIDIA GeForce GT 740M vs NVIDIA GeForce GT 745M Comparison

The NVIDIA GeForce GT 745M and NVIDIA GeForce GT 740M are both end-of-life Kepler mobile graphics solutions from the GeForce 700M generation, but benchmark data reveals they are not interchangeable. The GT 745M wins decisively in Vulkan performance with a 59.1% advantage, while the GT 740M counters with a 9.9% lead in OpenCL. The GT 745M posts an average benchmark score of 3953 against the GT 740M's 3717, a difference of roughly 6.3%, placing the former at the 23rd percentile and the latter at the 22nd percentile of all GPUs. These are close siblings, yet their architectural choices—particularly memory type and bus width—create distinct performance profiles that matter depending on the workload.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GT 745M averages 3953 across its benchmark suite, while the NVIDIA GeForce GT 740M averages 3717. That puts the GT 745M about 6.3% ahead overall, and it also sits one percentile higher (23rd vs 22nd) among all GPUs.

Q: How do the two compare in OpenCL performance?

A: The GT 740M wins in OpenCL, scoring 3974 versus the GT 745M's 3580. That is a 9.9% margin in favor of the GT 740M, making it the stronger choice for OpenCL-based compute workloads.

Q: What about Vulkan performance?

A: The GT 745M dominates Vulkan, scoring 5502 against the GT 740M's 3459. This is a massive 59.1% difference, the single largest gap between the two in any benchmark, indicating the GT 745M is far better suited for Vulkan API tasks.

Q: Do both GPUs use the same memory type?

A: No. The GT 745M uses 2 GB of GDDR5 memory on a 128-bit bus, delivering 64.00 GB/s of bandwidth. The GT 740M uses 2 GB of DDR3 memory on a 64-bit bus, yielding only 14.40 GB/s of bandwidth—a substantial reduction in memory throughput.

Q: Are there differences in the chip design despite similar core counts?

A: Yes. The GT 745M uses the GK107 chip (Kepler architecture) with 1,270 million transistors on a 118 mm² die. The GT 740M uses the GK208 chip (Kepler 2.0) with 1,020 million transistors on an 87 mm² die. Both have 384 shading units and 32 TMUs, but the GT 740M has only 8 ROPs versus 16 on the GT 745M.

Q: Which GPU has a higher pixel and texture rate?

A: The GT 740M has the higher rates: 8.264 GPixel/s pixel rate and 33.06 GTexel/s texture rate, compared to the GT 745M's 4.392 GPixel/s and 17.57 GTexel/s. The GT 740M also leads in FP32 performance at 793.3 GFLOPS versus 421.6 GFLOPS.

Where Each One Wins

The GT 745M is the clear winner in Vulkan-based scenarios. Its 5502 Vulkan score versus 3459 for the GT 740M represents a 59.1% advantage, which is not a marginal edge but a decisive one. For any application, game, or compute task that leverages the Vulkan API, the GT 745M is the superior part. This is further supported by its higher memory bandwidth (64.00 GB/s vs 14.40 GB/s), which likely contributes to its ability to feed data more efficiently to the shader cores in modern API workloads.

The GT 740M, however, takes the crown in OpenCL. Its 3974 OpenCL score beats the GT 745M's 3580 by 9.9%. The GT 740M also has significantly higher raw compute rates: 793.3 GFLOPS FP32, 33.06 GTexel/s texture fill, and 8.264 GPixel/s pixel throughput. These figures are roughly double those of the GT 745M (which sits at 421.6 GFLOPS, 17.57 GTexel/s, and 4.392 GPixel/s). Thus, for OpenCL compute tasks that are not bandwidth-limited, the GT 740M's higher shader and texture throughput gives it a measurable performance edge.

The split is clean: Vulkan favors the GT 745M strongly; OpenCL favors the GT 740M moderately. The GT 745M's overall average score is higher, but that average is buoyed by the massive Vulkan lead. In a head-to-head comparison, each GPU wins one benchmark, so the "best" choice depends entirely on the target API.

Architecture Differences

Both GPUs are built on TSMC's 28 nm process and belong to the Kepler family, but they diverge in implementation. The GT 745M uses the GK107 chip, which is the larger die at 118 mm² with 1,270 million transistors, resulting in a transistor density of 10.8M per mm². The GT 740M uses the GK208 chip, a smaller die at 87 mm² with 1,020 million transistors, yielding a higher density of 11.7M per mm². The GK208 is labeled as Kepler 2.0 architecture, while the GK107 is plain Kepler, suggesting the GT 740M's chip is a later revision.

Core counts are identical on paper: both have 384 shading units and 32 texture mapping units. The critical architectural split is in the ROP count and memory subsystem. The GT 745M has 16 ROPs, while the GT 740M has only 8. However, the GT 740M compensates with a much higher clock speed—980 MHz base and 1033 MHz boost—compared to the GT 745M, which has no listed base or boost clocks in the data. That clock advantage is why the GT 740M achieves higher pixel and texture rates despite fewer ROPs.

Memory architecture is the other major divergence. The GT 745M pairs its GK107 with a 128-bit GDDR5 interface running at 1000 MHz (4 Gbps effective), producing 64.00 GB/s of bandwidth. The GT 740M uses a 64-bit DDR3 interface at 900 MHz (1800 Mbps effective), producing just 14.40 GB/s. This is a 4.4x difference in memory bandwidth, which explains why the GT 745M excels in Vulkan—a modern API that often stresses memory throughput. The GT 740M's higher compute rates (793.3 GFLOPS vs 421.6 GFLOPS) suggest it is designed for raw shader throughput, but its narrow memory bus becomes a bottleneck in bandwidth-heavy scenarios.

Specification Differences

The two GPUs differ in several key specification fields. The chip is GK107 on the GT 745M versus GK208 on the GT 740M, with architecture listed as Kepler versus Kepler 2.0. Transistor count is 1,270 million versus 1,020 million, and die size is 118 mm² versus 87 mm². Transistor density is 10.8M per mm² versus 11.7M per mm².

Clock speeds differ: the GT 740M has a base clock of 980 MHz and a boost clock of 1033 MHz, while the GT 745M has no base or boost clock listed. Memory clocks are 1000 MHz (4 Gbps effective) for the GT 745M and 900 MHz (1800 Mbps effective) for the GT 740M. Memory type is GDDR5 versus DDR3, with bus widths of 128 bit versus 64 bit. Bandwidth is 64.00 GB/s versus 14.40 GB/s.

ROP count is 16 versus 8. Pixel rate is 4.392 GPixel/s versus 8.264 GPixel/s, and texture rate is 17.57 GTexel/s versus 33.06 GTexel/s. FP32 performance is 421.6 GFLOPS versus 793.3 GFLOPS. TDP is 45 W versus 33 W. Slot width is IGP versus MXM Module, and the GT 740M has no power connectors listed while the GT 745M's connectors are null. Bus interface is PCIe 3.0 x16 versus PCIe 3.0 x8. Both have 2 GB memory, 384 shading units, 32 TMUs, and identical API support (DirectX 12 (11_0), OpenGL 4.6, Vulkan 1.2.175).

Head-to-Head Benchmarks

The two GPUs meet in two benchmark tests, and each takes one. In Geekbench OpenCL, the GT 740M scores 3974 against the GT 745M's 3580, a 9.9% delta in favor of the GT 740M. This aligns with the GT 740M's higher FP32 throughput (793.3 GFLOPS) and texture rate (33.06 GTexel/s), which are roughly double the GT 745M's figures. The GT 740M's 980 MHz base clock and 1033 MHz boost clock give it a compute advantage that shows up in OpenCL's parallel workload patterns.

In Geekbench Vulkan, the result reverses dramatically. The GT 745M scores 5502, while the GT 740M manages only 3459. That is a 59.1% lead for the GT 745M, the largest performance gap in the entire comparison. The GT 745M's GDDR5 memory with 64.00 GB/s bandwidth is the likely differentiator here. Vulkan workloads often involve frequent memory access and data movement, and the GT 740M's 14.40 GB/s DDR3 interface is a severe constraint. Despite having fewer ROPs and lower compute rates, the GT 745M's memory subsystem allows it to dominate in this API.

The average benchmark scores reflect this split: the GT 745M averages 3953, while the GT 740M averages 3717. The GT 745M's nearest rivals include the AMD Radeon R5 M420 (3956, -0.1%) and NVIDIA GeForce 830M (3957, -0.1%), all within a tight band. The GT 740M's nearest rivals include the NVIDIA Quadro 3000M (3718, 0%) and GeForce 825M (3694, 0.6%), showing it sits in a similar performance tier. The deltaPct values are small, confirming both GPUs are closely matched against their peers.

The Verdict

The data points to a clear but conditional verdict: pick the GT 745M if Vulkan performance or memory bandwidth is a priority; pick the GT 740M if OpenCL compute throughput matters more. The GT 745M's 59.1% Vulkan lead is the headline number—it is not a small edge but a dominant one, and it drives the GT 745M's higher average score (3953 vs 3717) and better percentile ranking (23rd vs 22nd). For modern gaming or applications that use Vulkan, the GT 745M is the superior choice, and its 64.00 GB/s GDDR5 bandwidth is a critical asset.

The GT 740M, however, offers a compelling counter-case for OpenCL users. Its 9.9% OpenCL win is backed by superior raw specs: 793.3 GFLOPS FP32, 33.06 GTexel/s texture rate, and 8.264 GPixel/s pixel rate. It also consumes less power (33 W vs 45 W TDP), making it more efficient for laptops where thermals and battery life are concerns. The GT 740M's higher clocks (980 MHz base, 1033 MHz boost) and Kepler 2.0 architecture suggest a more refined compute design, even if its narrow 64-bit DDR3 memory bus (14.40 GB/s) limits it in bandwidth-bound scenarios.

For a general-purpose mobile GPU, the GT 745M's higher average score and Vulkan dominance make it the safer recommendation. But for users specifically targeting OpenCL workloads—or those prioritizing lower power draw—the GT 740M is the better fit. The two GPUs are evenly matched in head-to-head wins (1-1), but the magnitude of the GT 745M's Vulkan victory outweighs the GT 740M's modest OpenCL edge. Ultimately, the GT 745M is the overall leader, with the GT 740M serving as the efficient OpenCL specialist.