AMD Radeon R7 M260X vs NVIDIA GeForce GTX 670MX Comparison

# NVIDIA GeForce GTX 670MX vs AMD Radeon R7 M260X

The mobile GPU comparison in this database entry pairs two discrete-class parts from the same era of notebook computing. The NVIDIA GeForce GTX 670MX is a high-end Kepler part built for larger, performance-oriented laptops, while the AMD Radeon R7 M260X is a mainstream GCN chip designed for thinner, more power-conscious machines. The recorded data shows a clear performance hierarchy, with the NVIDIA part taking a decisive lead across every measured workload.

FAQ

Q: What is the average benchmark score of the NVIDIA GeForce GTX 670MX?

A: The average benchmark score for the GTX 670MX is 5,721. This places the GPU in the 33rd percentile of all tested GPUs in the database, meaning it outperforms roughly one third of all recorded parts.

Q: How does the NVIDIA GeForce GTX 670MX compare to the AMD Radeon R7 M260X in raw compute?

A: In the OpenCL benchmark, the GTX 670MX scores 6,125, which is 7.6% higher than the R7 M260X's score of 5,690. This indicates a substantial lead in general compute workloads, likely due to the much larger core configuration (960 shading units versus 384).

Q: What are the key architectural differences that explain the performance gap?

A: The NVIDIA part uses the GK104 chip on a 28 nm process with 3,540 million transistors, while the AMD part uses the Opal chip on the same 28 nm node with 950 million transistors. The NVIDIA die size is 294 mm² compared to AMD's 77 mm², giving the NVIDIA a significantly higher transistor density of 12.0M per mm² versus AMD's 12.3M per mm².

Q: Which GPU has the higher memory bandwidth?

A: The NVIDIA GTX 670MX has a clear advantage here. It has a 192-bit memory bus and GDDR5 memory running at 700 MHz, providing 67.20 GB/s of bandwidth. The AMD Radeon R7 M260X uses a narrower 128-bit bus with GDDR5 memory at 1000 MHz, achieving 64.00 GB/s.

Q: What is the thermal design power (TDP) of each card?

A: The database lists a 75W TDP for the NVIDIA GeForce GTX 670MX. No TDP is recorded for the AMD Radeon R7 M260X, so a direct power draw comparison is not possible from the available data.

Architecture Differences

The two GPUs represent fundamentally different design philosophies from their respective architects. The NVIDIA GeForce GTX 670MX is built on the Kepler architecture, using the GK104 chip manufactured by TSMC on a 28nm process. The die is large at 294 mm², containing 3,540 million transistors, which results in a density of 12.0 million transistors per square millimeter. The GPU is equipped with 960 shading units, 80 texture mapping units, and 24 ROPs.

The AMD Radeon R7 M260X, in contrast, is based on the older GCN 1.0 architecture, using the Opal chip, also at TSMC on a 28nm node. The physical die is a compact 77 mm², housing 950 million transistors, for a density of 12.3M per mm². The compute array is smaller: 384 shading units, 24 TMUs, and 8 ROPs.

Clock behavior differs notably. The NVIDIA part runs at a base and boost frequency of 601 MHz, with no separate game clock listed. Its memory runs at 700 MHz. The AMD part has a higher base clock of 620 MHz, a higher boost clock of 715 MHz, and a memory clock of 1000 MHz. Despite the higher clocks on the AMD side, the substantially larger execution resource pool on the NVIDIA chip wins out.

API support shows modern feature sets from both vendors. The NVIDIA GPU supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The AMD part supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170.

The Verdict

The benchmark data is unequivocal. The NVIDIA GeForce GTX 670MX wins both recorded head-to-head tests by a wide margin. In the OpenCL test it scores 6,125 against 5,690, a 7.6% advantage. In the Vulkan test, the margin is larger: NVIDIA scores 5,316 versus AMD's 4,631, a 14.8% advantage.

The root cause lies in the hardware resource allocation. The NVIDIA part has 2.5 times as many shading units (960 versus 384), 3.3 times as many TMUs (80 versus 24), and 3 times as many ROPs (24 versus 8). The AMD part attempts to compensate with higher clock speeds (620 MHz base, 715 MHz boost versus 601 MHz/601 MHz) and faster memory clocks, but it cannot overcome the sheer execution width deficit.

The NVIDIA GTX 670MX also has a 3 GB GDDR5 frame buffer with a 192-bit interface, meaning significantly more memory capacity for large frame buffers and textures. The AMD Radeon R7 M260X offers 1024 MB of GDDR5 on a 128-bit bus.

For any workload in the database that is shader-bound or heavily parallel, the NVIDIA GPU is the clear choice from the numbers. The AMD part is for lighter, lower-power notebooks where modest gaming is secondary. The AMD R7 M260X would be the pick if power draw is the primary constraint, as its TDP is not listed while the NVIDIA is listed at 75 W, though this must be weighed against the lesser performance.

Specification Differences

Between the two record entries, the most meaningful differences are in the execution core and memory subsystem. The NVIDIA GeForce GTX 670MX uses the GK104 chip with 960 shaders, 80 TMUs, and 24 ROPs. It carries 3 GB of GDDR5 across a 192-bit bus with a 700 MHz memory clock, yielding 67.20 GB/s of bandwidth.

The AMD Radeon R7 M260X uses the Opal chip with 384 shaders, 24 TMUs, and 8 ROPs. Its memory subsystem is 1024 MB of GDDR5 across a 128-bit bus, but the memory clock runs at 1000 MHz, resulting in 64.00 GB/s of bandwidth.

The NVIDIA GPU has a die size of 294 mm² and a transistor count of 3,540 million. The AMD die is 77 mm² with 950 million transistors. The NVIDIA part has a density of 12.0M transistors per mm², the AMD 12.3M per mm². The AMD chip is the denser of the two, but the NVIDIA part has far more functional units and memory bandwidth.

The NVIDIA part features a 75W TDP, while the AMD part has no TDP in the database. The NVIDIA is an end-of-life product, as is the AMD part, but they belong to different generations: NVIDIA is the successor to the GeForce 500M and the predecessor to the GeForce 700M. The AMD part follows the Solar System as the successor to the Polaris Mobile.

Head-to-Head Benchmarks

The database includes two head-to-head benchmark runs for these specific parts, meaning they were evaluated on the same test session and are directly comparable.

Geekbench OpenCL: The NVIDIA GeForce GTX 670MX scores 6.125, the AMD Radeon R7 M260X scores 5.690. The NVIDIA wins by 7.6%. This is a significant margin, and it is explained by the NVIDIA chip having 2.5 times the shading units and 3 times the ROP count. The AMD chip has a higher core clock (620 MHz base, 715 MHz boost) vs the NVIDIA's 601 MHz fixed, but the NVIDIA has 3 GB of VRAM against the AMD's 1 GB. The result is a combination of pure compute, memory capacity, and memory bandwidth, all of which favor NVIDIA.

Geekbench Vulkan: The gap widens in the Vulkan workload. The NVIDIA GeForce GTX 670MX scores 5,316, the AMD Radeon R7 M260X scores 4,631. That is a 14.8% advantage. This test tends to scale with the available shading units and geometry throughput, and the NVIDIA has 960 shaders vs 384 on the AMD, plus 80 TMUs vs 24. The AMD's higher boost clock of 715 MHz does not compensate for the 2.5x shader core advantage.

In the two recorded head-to-head runs, the NVIDIA GPU wins both. The average result for the NVIDIA is 5,721, and the AMD 5,161. The delta between the two is 10.8% in favor of NVIDIA, which aligns with the consistent wins in each individual test. The database also records the NVIDIA GPU at the 33rd percentile of all GPUs, while the AMD sits at the 30th percentile, meaning both are below the median of the database.