NVIDIA GeForce GT 640 vs NVIDIA GeForce GT 740 Comparison

The NVIDIA GeForce GT 740 and GT 640 are close relatives, sharing the same GK107 chip and Kepler architecture, yet they occupy different tiers in NVIDIA’s product stack. Both cards are now end-of-life, but the benchmark data reveals a consistent performance hierarchy, with the GT 740 leading across every measured workload. The average benchmark score for the GT 740 is 3431, placing it in the 21st percentile of all GPUs, while the GT 640 averages 3210, sitting in the 20th percentile. This slim percentile gap underscores how tightly matched these two cards are, but the individual test results show where the newer card’s advantages materialize.

Head-to-Head Benchmarks

The GT 740 wins all three head-to-head comparisons, but the margins are not uniform. The largest gap appears in the Geekbench Metal test, where the GT 740 scores 2363 against the GT 640’s 2086, a 13.3% advantage. This is the clearest differentiator between the two cards, suggesting that the GT 740’s memory subsystem provides a significant boost in compute-heavy graphics workloads that leverage the Metal API. The GT 640, by contrast, trails by a wide margin here, which aligns with its older memory configuration.

In the Geekbench Vulkan test, the GT 740 again leads, scoring 4083 versus 3809, a 7.2% delta. This result is notable because Vulkan is a low-overhead API that often exposes raw hardware capabilities; the GT 740’s higher effective memory clock appears to translate directly into better performance in this scenario. The smallest win for the GT 740 comes in Geekbench OpenCL, where it scores 3847 against 3736, a modest 3% improvement. OpenCL workloads tend to be more sensitive to shading unit throughput than memory bandwidth, and since both cards share the same 384 shading units, the gap narrows considerably.

Looking at the broader context, the GT 740’s average score of 3431 puts it just 0.1% behind the NVIDIA Quadro 2000M (3434) and 1.2% ahead of the Intel HD Graphics P4600 (3389). It also sits 2.7% below the GeForce 920MX (3528) and 3% above the Intel HD Graphics 530 (3332). The GT 640, meanwhile, is 0.9% ahead of the Intel Arc Pro B60 (3182) and 1.5% ahead of the Quadro P1000 (3163), but trails the GeForce 920M (3287) by 2.3% and the GT 730M (3316) by 3.2%. These rivalries show that both cards are firmly in entry-level territory, but the GT 740’s margins over its closest competitors are slightly more comfortable than the GT 640’s.

Architecture Differences

Both the GT 740 and GT 640 are built on the GK107 chip using TSMC’s 28 nm process node, with a die size of 118 mm² and 1,270 million transistors, yielding a transistor density of 10.8M per mm². They share the same Kepler architecture, which means the core compute configuration is identical: 384 shading units, 32 texture mapping units, and 16 raster operations pipelines. Neither card features ray tracing cores or tensor cores, and both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

The fundamental architectural divergence lies in the memory subsystem. The GT 740 ships with 1024 MB of GDDR5 memory on a 128-bit bus, running at an effective speed of 5 Gbps, producing a bandwidth of 80.19 GB/s. The GT 640, in contrast, uses 2 GB of DDR3 memory on the same 128-bit bus, but its effective speed is only 1782 Mbps, resulting in a bandwidth of just 28.51 GB/s. This is a 2.8x difference in raw bandwidth, which explains why the GT 740 consistently outperforms the GT 640 in memory-sensitive benchmarks.

Pixel and texture fill rates also differ slightly due to clock variations. The GT 740 achieves a pixel rate of 7.944 GPixel/s and a texture rate of 31.78 GTexel/s, while the GT 640 manages 7.216 GPixel/s and 28.86 GTexel/s. These are modest differences of roughly 10% and 10.1%, respectively, reflecting that the GT 740’s core clock is somewhat higher. Floating-point performance follows suit: the GT 740 delivers 762.6 GFLOPS of FP32 throughput, whereas the GT 640 provides 692.7 GFLOPS, a 10.1% gap that mirrors the texture rate delta.

FAQ

Q: Which card has more memory, and does that matter for benchmarks?

A: The GT 640 has 2 GB of DDR3 memory, while the GT 740 has 1 GB of GDDR5. Despite having half the capacity, the GT 740’s GDDR5 memory provides 80.19 GB/s of bandwidth versus the GT 640’s 28.51 GB/s, which is why the GT 740 wins all head-to-head tests despite lower capacity.

Q: Do both cards consume similar power?

A: Yes, the power draw is nearly identical. The GT 740 has a TDP of 64 W, while the GT 640 is rated at 65 W. Both suggest a 250 W power supply, and neither card requires a power connector beyond the GT 740’s single 6-pin.

Q: Are these cards capable of running modern API workloads?

A: Both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, so they can run applications using these APIs. However, their low percentile rankings (21st for the GT 740, 20th for the GT 640) indicate they are entry-level parts.

Q: What is the GT 740’s closest rival, and how does it compare?

A: The GT 740’s nearest rival is the NVIDIA Quadro 2000M, which scores 3434 on average, just 0.1% higher. The GT 740 also trails the GeForce 920MX by 2.7% but beats the Intel HD Graphics P4600 by 1.2% and the Intel HD Graphics 530 by 3%.

Q: What is the GT 640’s closest rival, and how does it compare?

A: The GT 640’s nearest rival is the Intel Arc Pro B60, which scores 3182, putting the GT 640 0.9% ahead. It also leads the Quadro P1000 by 1.5% but falls 2.3% behind the GeForce 920M and 3.2% behind the GT 730M.

Q: Which card has better display output options?

A: The GT 740 offers 2x DVI and 1x mini-HDMI 1.4a, while the GT 640 provides 1x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The GT 640 has more variety, but the GT 740 has dual DVI ports.

Specification Differences

The two cards share identical core counts, process node, die size, transistor count, architecture, and API support. The differences are confined to memory, clocks, power, and display outputs. The GT 740 uses 1024 MB of GDDR5 memory with a bus width of 128 bit and bandwidth of 80.19 GB/s, while the GT 640 uses 2 GB of DDR3 with the same bus width but only 28.51 GB/s bandwidth. The memory clock for the GT 740 is 1253 MHz (5 Gbps effective), whereas the GT 640 runs at 891 MHz (1782 Mbps effective). Pixel rate differs: 7.944 GPixel/s for the GT 740 versus 7.216 GPixel/s for the GT 640. Texture rate is 31.78 GTexel/s versus 28.86 GTexel/s. FP32 performance is 762.6 GFLOPS versus 692.7 GFLOPS. The GT 740 has a TDP of 64 W and requires a single 6-pin power connector, while the GT 640 has a TDP of 65 W and requires no power connector. Display outputs also differ: the GT 740 has 2x DVI and 1x mini-HDMI, while the GT 640 has 1x DVI, 1x HDMI, and 1x DisplayPort.

Where Each One Wins

The GT 740 is the clear winner in raw performance, taking all three benchmark tests. Its biggest victory is in Geekbench Metal, where the 13.3% lead over the GT 640 is substantial. This suggests the GT 740 is better suited for workloads that leverage Metal, such as macOS graphics acceleration or compute tasks optimized for that API. The GT 740’s GDDR5 memory also gives it a decisive edge in bandwidth-bound scenarios, which is why its Vulkan score is 7.2% higher. For users prioritizing maximum frame rates or compute throughput in entry-level gaming or productivity apps, the GT 740 is the stronger choice.

The GT 640, despite losing every benchmark, has its own niche. Its 2 GB of DDR3 memory provides double the capacity of the GT 740, which can be beneficial for applications that require larger framebuffers, such as high-resolution textures in older games or multi-monitor setups with lower memory pressure. The GT 640 also offers a more flexible display output configuration with HDMI and DisplayPort, which may be more convenient for certain monitor setups. Its slightly lower TDP (65 W versus 64 W) is negligible, but the absence of a power connector means it can be installed in systems with limited power supply headers, making it a more drop-in option for pre-built machines. The GT 640’s 3.2% deficit to the GT 730M in rival comparisons shows it is competitive within its class, even if it trails its newer sibling.

Ultimately, the data positions the GT 740 as the performance pick, while the GT 640 serves as a capacity-oriented alternative. The GT 740’s 13.3% Metal win and 7.2% Vulkan win are the most compelling reasons to choose it, whereas the GT 640’s larger memory pool and simpler power requirements are its only tangible advantages. For any benchmark-driven decision, the GT 740 is the logical choice; for scenarios where memory capacity outweighs bandwidth, the GT 640 remains relevant.