GPU Comparison

The NVIDIA GeForce GT 730M and NVIDIA GeForce 920M are two end-of-life mobile GPUs that, despite their different generational labels, land in the exact same performance percentile. Both sit at the 20th percentile among all GPUs, with average benchmark scores of 3316 and 3287, respectively, a gap of less than one percent. The data shows a near-total dead heat in aggregate, but the two chips diverge sharply when individual workloads are examined. The GeForce 920M wins one benchmark decisively; the GT 730M wins the other even more decisively. Which one is "better" depends entirely on the application.

Head-to-Head Benchmarks

The most striking result in the head-to-head data is the GeForce 920M's victory in Geekbench OpenCL. Here, the 920M scores 3725 against the GT 730M's 3107, a delta of -16.6 percent from the GT 730M's perspective. That translates to the 920M being roughly 20 percent faster in OpenCL compute workloads. This is not a marginal win; it is a commanding lead that places the 920M well outside the margin of error. For any task that leverages OpenCL, video encoding, physics simulation, or GPU-accelerated productivity apps, the 920M is the clear choice.

The tables turn completely in Geekbench Vulkan. The GT 730M scores 3524, while the 920M manages only 2849. That gives the GT 730M a 23.7 percent advantage, an even larger margin than the 920M's OpenCL lead. Vulkan is the modern graphics API used by a growing number of games and game engines, so this result is not academic. In Vulkan-based titles, the older GT 730M outperforms its newer rival by nearly a quarter. This is the single biggest performance gap between the two cards in either direction.

The average benchmark scores reflect this split. The GT 730M's average of 3316 is slightly higher than the 920M's 3287, a difference of just 0.9 percent. The nearest-rival data confirms how close these two are in aggregate: from the GT 730M's side, the 920M sits only 0.9 percent behind; from the 920M's side, the GT 730M is 0.9 percent ahead. Both cards also trail the Intel HD Graphics 530 by roughly half a percent to 1.4 percent, and the Intel HD Graphics P4600 by 2.2 to 3 percent. In plain terms, these two NVIDIA parts are interchangeable in overall throughput, but they are not interchangeable in specific workloads.

Architecture Differences

The GT 730M is built on the GK107 chip, using the original Kepler architecture. The 920M uses the GK208B chip with Kepler 2.0. Both are manufactured on the same 28 nm process at TSMC, but the silicon differs in size and complexity. The GK107 packs 1,270 million transistors into a 118 mm² die, yielding a transistor density of 10.8 million per square millimeter. The GK208B is smaller: 1,020 million transistors on an 87 mm² die, with a higher density of 11.7 million per square millimeter. The 920M's chip is more compact and denser, which is a sign of a later design iteration, but it also has less total transistor budget to work with.

The shading unit count is identical, both cards feature 384 shading units and 32 texture mapping units. The differences appear elsewhere. The GT 730M has 16 ROPs, while the 920M has only 8. That is a full half reduction in raster output units, which directly impacts pixel throughput in certain rendering paths. However, the 920M compensates with higher clocks. The GT 730M runs at a base and boost of 725 MHz, while the 920M runs at 954 MHz for both. This clock advantage is substantial, over 31 percent higher, and it shows up in the fill rates. The GT 730M produces 5.800 GPixel/s and 23.20 GTexel/s, whereas the 920M produces 7.632 GPixel/s and 30.53 GTexel/s.

Memory is another major divergence. Both cards use 2 GB of DDR3, but the GT 730M has a 128-bit memory bus while the 920M is limited to 64-bit. This doubles the GT 730M's memory bandwidth: 28.80 GB/s versus just 14.40 GB/s for the 920M. The memory clock is identical at 900 MHz (1800 Mbps effective), so the bandwidth gap is purely a function of bus width. This is a classic trade-off: the 920M has a narrower memory bus but a faster core, while the GT 730M has a wider bus but a slower core. The TDP is the same at 33 W for both, and neither requires power connectors. The GT 730M uses an MXM module slot, while the 920M is an IGP (integrated graphics processor) form factor. The bus interface also differs: PCIe 3.0 x16 for the GT 730M, PCIe 3.0 x8 for the 920M.

Where Each One Wins

The GeForce 920M wins in raw compute and fill-rate-bound scenarios. Its higher clock speed delivers superior pixel and texture throughput, and its OpenCL score of 3725 is the standout number in this comparison. If the workload is OpenCL-accelerated, think video transcoding, image processing, or any GPGPU task that isn't Vulkan-based, the 920M is the faster part. It also has the advantage of being a newer design (released in March 2015 versus January 2013 for the GT 730M), which may matter for software compatibility and driver maturity in niche applications. The 920M's higher transistor density also suggests a more refined implementation of the same process node.

The GT 730M wins in Vulkan rendering. Its score of 3524 in Geekbench Vulkan is 23.7 percent higher than the 920M's, and this is the more relevant metric for a GPU that will be used in gaming or game-like applications. The GT 730M's wider 128-bit memory bus gives it a substantial bandwidth advantage (28.80 GB/s versus 14.40 GB/s), which likely explains its Vulkan superiority, modern graphics APIs are often bandwidth-sensitive. The GT 730M also has double the ROP count (16 versus 8), which helps in pixel-heavy scenes. For anyone running Vulkan-based titles or engines, the GT 730M is the better choice despite being the older product.

The overall average benchmark score favors the GT 730M, but only by a hair: 3316 versus 3287. That is a 0.9 percent difference, which is effectively noise. Both cards occupy the same percentile bracket (20th) and both are end-of-life products. The data does not support declaring a universal winner. Instead, the split is clean: OpenCL belongs to the 920M, Vulkan belongs to the GT 730M. Users who know their primary workload can pick accordingly; users who need a general-purpose mobile GPU will find either card equally adequate.

FAQ

Q: Which GPU is faster overall, the GT 730M or the 920M?

A: They are nearly identical in aggregate. The GT 730M has an average benchmark score of 3316, while the 920M scores 3287, a difference of just 0.9 percent. Both rank at the 20th percentile among all GPUs.

Q: Why does the 920M win in OpenCL by such a large margin?

A: The 920M scores 3725 in Geekbench OpenCL versus 3107 for the GT 730M, a 16.6 percent advantage. This is likely due to its significantly higher clock speed: 954 MHz versus 725 MHz for the GT 730M.

Q: Why does the GT 730M win in Vulkan?

A: The GT 730M scores 3524 in Geekbench Vulkan versus 2849 for the 920M, a 23.7 percent lead. The GT 730M has a much wider 128-bit memory bus (versus 64-bit) and double the ROPs (16 versus 8), which helps in bandwidth-sensitive graphics workloads.

Q: Do both cards have the same amount of memory?

A: Yes, both feature 2 GB of DDR3 memory. However, the GT 730M has 28.80 GB/s of bandwidth, while the 920M has only 14.40 GB/s due to its narrower bus.

Q: Are these GPUs the same architecture?

A: Not exactly. The GT 730M uses the original Kepler architecture (GK107 chip), while the 920M uses Kepler 2.0 (GK208B chip). Both are on TSMC's 28 nm process, but the 920M has a smaller die (87 mm² versus 118 mm²) and fewer transistors (1,020 million versus 1,270 million).

Q: Which card is better for modern gaming?

A: For Vulkan-based games, the GT 730M is clearly superior, with a 23.7 percent higher score. The 920M's advantage is limited to OpenCL compute workloads, which are less common in gaming. Both cards are end-of-life and rank in the 20th percentile, so neither is suited for demanding titles.

Specification Differences

| Specification | NVIDIA GeForce GT 730M | NVIDIA GeForce 920M |

|---|---|---|

| Chip | GK107 | GK208B |

| Architecture | Kepler | Kepler 2.0 |

| Generation | GeForce 700M | GeForce 900M |

| Transistors | 1,270 million | 1,020 million |

| Die Size | 118 mm² | 87 mm² |

| Transistor Density | 10.8M / mm² | 11.7M / mm² |

| Base Clock | 725 MHz | 954 MHz |

| Boost Clock | 725 MHz | 954 MHz |

| Memory Bus Width | 128 bit | 64 bit |

| Memory Bandwidth | 28.80 GB/s | 14.40 GB/s |

| ROPs | 16 | 8 |

| Pixel Rate | 5.800 GPixel/s | 7.632 GPixel/s |

| Texture Rate | 23.20 GTexel/s | 30.53 GTexel/s |

| FP32 | 556.8 GFLOPS | 732.7 GFLOPS |

| Slot Width | MXM Module | IGP |

| Bus Interface | PCIe 3.0 x16 | PCIe 3.0 x8 |

| Release Date | 2013-01-19 | 2015-03-12 |

| Predecessor | GeForce 600M | GeForce 800M |

| Successor | GeForce 800M | GeForce 10 Mobile |

The shading units (384) and TMUs (32) are identical, as are the memory size (2 GB), memory type (DDR3), memory clock (900 MHz / 1800 Mbps effective), TDP (33 W), power connectors (None), display outputs (Portable Device Dependent), and API support (DirectX 12 (11_0), OpenGL 4.6, Vulkan 1.2.175). Neither card has a launch MSRP listed.