NVIDIA GeForce GTX 580M vs NVIDIA GeForce GTX 670MX Comparison

The Verdict

The data presents a surprisingly close contest between two mobile graphics solutions from different NVIDIA generations. The NVIDIA GeForce GTX 580M, built on the older Fermi 2.0 architecture, edges out the Kepler-based GTX 670MX in the only shared benchmark, the Geekbench OpenCL test, scoring 6389 against 6125, a 4.3% advantage. However, the GTX 670MX holds a significant edge in overall capability when considering its additional Vulkan benchmark result and its superior architectural fundamentals.

For users prioritizing raw compute performance in OpenCL workloads, the GTX 580M is the clear choice from the recorded data, as it wins the only direct head-to-head comparison. Yet the GTX 670MX counters with a much higher transistor count, a newer manufacturing process, and a lower thermal design power, which suggests it may offer better efficiency in sustained mobile workloads. The GTX 580M sits at the 37th percentile among all GPUs in the database, while the GTX 670MX sits at the 33rd percentile, meaning both are positioned in the lower-middle range of the overall performance spectrum, but the 580M's percentile ranking is slightly higher.

The practical verdict depends on workload priorities. The GTX 580M wins where OpenCL compute performance matters most, and its higher pixel fill rate of 9.920 GPixel/s versus 12.02 GPixel/s for the 670MX actually shows the 670MX ahead in that metric, so the 580M's win is not universal. The GTX 670MX, despite losing the OpenCL test, offers nearly double the shading units, 960 versus 384, and a more modern feature set including Vulkan support, making it the more future-proof option for applications leveraging newer graphics APIs. Users focused purely on the measured OpenCL score should pick the GTX 580M, while those valuing architectural modernization, lower power draw, and API compatibility should gravitate toward the GTX 670MX.

Architecture Differences

The two GPUs represent distinct generations of NVIDIA mobile graphics. The GTX 580M is built on the Fermi 2.0 architecture using the GF114 chip, manufactured on a 40 nm process at TSMC. The GTX 670MX transitions to the Kepler architecture with the GK104 chip, manufactured on a more advanced 28 nm process, also at TSMC. This process shrink allows the GTX 670MX to pack 3,540 million transistors onto a 294 mm² die, whereas the GTX 580M contains only 1,950 million transistors on a larger 332 mm² die. The transistor density tells the story clearly: the 670MX achieves 12.0 million transistors per square millimeter, more than double the 5.9 million per square millimeter of the 580M.

The compute architecture diverges significantly. The GTX 670MX features 960 shading units, 80 texture mapping units, and 24 ROPs, while the GTX 580M features only 384 shading units, 64 TMUs, and 32 ROPs. This means the 670MX has 2.5 times the shading units and 1.25 times the texture units, but fewer ROPs. The raw compute throughput reflects this: the 670MX delivers 1,153.9 GFLOPS of FP32 performance versus 952.3 GFLOPS for the 580M, a 21.2% advantage for the Kepler part.

Clock speeds differ as well. The GTX 670MX has a base and boost clock of 601 MHz, while the GTX 580M's clock data is not recorded in the database, but its memory runs at 750 MHz (3 Gbps effective) compared to 700 MHz (2.8 Gbps effective) for the 670MX. Memory configurations also diverge: the 580M uses 2 GB of GDDR5 on a 256-bit bus, yielding 96.00 GB/s of bandwidth, while the 670MX uses 3 GB of GDDR5 on a narrower 192-bit bus, yielding 67.20 GB/s. The 580M's wider memory bus gives it 42.9% more memory bandwidth despite having less capacity.

Thermal and interface differences are pronounced. The GTX 580M has a TDP of 100 W and uses an MXM-B (3.0) bus interface, while the GTX 670MX draws only 75 W and uses PCIe 3.0 x16. Both use portable-device-dependent display outputs and require no external power connectors. API support shows the generational gap: both support DirectX 12 (11_0) and OpenGL 4.6, but the GTX 670MX additionally supports Vulkan 1.2.175, while the GTX 580M has no recorded Vulkan support.

Where Each One Wins

The benchmark data is limited but revealing. In the Geekbench OpenCL test, the GTX 580M wins with a score of 6389 against 6125, a 4.3% delta. This is the only direct head-to-head measurement available, and it favors the older Fermi architecture. The GTX 580M also has a higher average benchmark score of 6389, which matches its single OpenCL result, while the GTX 670MX has an average of 5721 across its two recorded benchmarks (OpenCL at 6125 and Vulkan at 5316).

The GTX 670MX wins in areas not directly benchmarked but clearly visible in the specification data. Its FP32 compute throughput of 1,153.9 GFLOPS exceeds the 580M's 952.3 GFLOPS, indicating greater raw compute capacity despite the lower OpenCL score. Its texture rate of 48.08 GTexel/s outpaces the 580M's 39.68 GTexel/s, and its pixel rate of 12.02 GPixel/s exceeds the 580M's 9.920 GPixel/s. These metrics suggest the 670MX should perform better in texture-heavy and pixel-heavy workloads, even though the OpenCL test tells a different story.

The GTX 670MX also wins on efficiency. With a TDP of 75 W versus 100 W for the GTX 580M, the Kepler part delivers higher theoretical compute while drawing 25% less power. This makes it the preferable option for thinner, cooler-running mobile platforms. The 670MX also offers Vulkan support, opening the door to modern graphics workloads that the 580M cannot access.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The NVIDIA GeForce GTX 580M scores 6389 in Geekbench OpenCL, while the GTX 670MX scores 6125, giving the 580M a 4.3% advantage.

Q: Does the GTX 670MX support Vulkan?

A: Yes, the GTX 670MX supports Vulkan 1.2.175, while the GTX 580M has no recorded Vulkan support in the database.

Q: How do the memory bandwidth figures compare?

A: The GTX 580M has 96.00 GB/s of bandwidth from its 256-bit bus and 2 GB of GDDR5, while the GTX 670MX has 67.20 GB/s from a 192-bit bus and 3 GB of GDDR5, making the 580M's bandwidth 42.9% higher.

Q: Which GPU has more shading units?

A: The GTX 670MX has 960 shading units, compared to 384 for the GTX 580M, a 2.5-fold difference in favor of the Kepler part.

Q: What are the thermal design power ratings?

A: The GTX 580M has a TDP of 100 W, while the GTX 670MX has a TDP of 75 W, making the 670MX the more power-efficient option.

Q: Which GPU has the higher pixel fill rate?

A: The GTX 670MX has a pixel rate of 12.02 GPixel/s, exceeding the GTX 580M's 9.920 GPixel/s, despite the 580M winning the OpenCL benchmark.

Head-to-Head Benchmarks

The only direct benchmark comparison in the database is the Geekbench OpenCL test, where the GTX 580M records a score of 6389 against the GTX 670MX's 6125. The 4.3% delta in favor of the 580M is notable given the architectural differences. The GTX 670MX has 960 shading units and 1,153.9 GFLOPS of FP32 compute, yet it still trails the 384-shader GTX 580M in this particular workload. This suggests that OpenCL performance in this test does not scale linearly with shader count or theoretical FP32 throughput, possibly due to driver maturity, memory bandwidth constraints, or the specific characteristics of the workload.

The GTX 580M's memory bandwidth of 96.00 GB/s, compared to 67.20 GB/s for the 670MX, may play a role in its OpenCL victory. With a 256-bit bus versus 192-bit, the 580M can move data more quickly, which can be decisive in compute tasks. The 580M's memory clock of 750 MHz (3 Gbps effective) exceeds the 670MX's 700 MHz (2.8 Gbps effective), further reinforcing its bandwidth advantage.

However, the GTX 670MX counters with its Vulkan benchmark score of 5316, a test the 580M cannot run due to lack of Vulkan support. This additional data point demonstrates that the 670MX can handle modern graphics APIs, even if its OpenCL score is lower. The 670MX also shows theoretical superiority in texture rate, 48.08 GTexel/s versus 39.68 GTexel/s, and pixel rate, 12.02 GPixel/s versus 9.920 GPixel/s, suggesting that in gaming or rendering workloads that emphasize texture and pixel throughput, the Kepler part may overcome its OpenCL deficit.

The average benchmark scores tell a broader story. The GTX 580M's average is 6389, equal to its single OpenCL result, placing it at the 37th percentile among all GPUs. The GTX 670MX's average of 5721, pulled down by its lower Vulkan score, places it at the 33rd percentile. The 670MX's nearest rivals include the Intel Iris Pro Graphics P6300 at a 0.2% delta, the NVIDIA GeForce GTX 550 Ti at -0.2%, and the AMD Radeon HD 8790M at 0.5%, showing it sits in a tightly contested performance band. The 580M's nearest rivals include the GTX 460 SE at 0% delta, the RTX PRO 5000 72 GB Blackwell at -0.3%, and the AMD Radeon Pro WX 4100 at 0.9%, indicating a similar clustering around its score.

Specification Differences

The two GPUs differ across nearly every major specification category. The GTX 580M uses the GF114 chip with Fermi 2.0 architecture, while the GTX 670MX uses the GK104 chip with Kepler architecture. The manufacturing process advances from 40 nm to 28 nm, and the transistor count rises from 1,950 million to 3,540 million. The die size actually shrinks from 332 mm² to 294 mm², while transistor density more than doubles from 5.9 million per square millimeter to 12.0 million per square millimeter.

Memory configurations diverge substantially. The GTX 580M offers 2 GB of GDDR5 on a 256-bit bus with 96.00 GB/s bandwidth, while the GTX 670MX offers 3 GB of GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth. Memory clocks differ, with the 580M running at 750 MHz (3 Gbps effective) versus 700 MHz (2.8 Gbps effective) for the 670MX. The 580M's memory runs faster and wider, but the 670MX has more capacity.

Compute resources favor the 670MX heavily in some areas and the 580M in others. The 670MX has 960 shading units, 80 TMUs, and 24 ROPs, while the 580M has 384 shading units, 64 TMUs, and 32 ROPs. FP32 performance reaches 1,153.9 GFLOPS for the 670MX versus 952.3 GFLOPS for the 580M. Texture rate is 48.08 GTexel/s for the 670MX and 39.68 GTexel/s for the 580M. Pixel rate is 12.02 GPixel/s for the 670MX and 9.920 GPixel/s for the 580M.

Power and interface specifications also differ. The 580M has a TDP of 100 W, while the 670MX draws 75 W. The 580M uses an MXM-B (3.0) bus interface, while the 670MX uses PCIe 3.0 x16. Both require no external power connectors and both have portable-device-dependent display outputs. API support is identical for DirectX 12 (11_0) and OpenGL 4.6, but the 670MX adds Vulkan 1.2.172 support while the 580M has none recorded.

Release timing and lineage differ as well. The GTX 580M was released on June 27, 2011, as part of the GeForce 500M generation, succeeding the GeForce 400M and preceding the GeForce 600M. The GTX 670MX was released on September 30, 2012, as part of the GeForce 600M generation, succeeding the GeForce 500M and preceding the GeForce 700M. Both are end-of-life products with no recorded launch MSRP in the database.