NVIDIA GeForce GT 555M vs NVIDIA GeForce GTX 670M Comparison

Where Each One Wins

The benchmark data presents an unusually tight contest. In the single available OpenCL test, the NVIDIA GeForce GTX 670M scores 6513, while the NVIDIA GeForce GT 555M scores 6493. The GTX 670M wins the head-to-head by a margin of 0.3%. That is the entirety of the victory split: one win for the GTX 670M, zero wins for the GT 555M. The practical implication is that for general-purpose compute workloads measured by Geekbench OpenCL, the two parts are effectively interchangeable in performance, with the GTX 670M holding a razor-thin edge.

The more interesting story emerges from the rival context. The GTX 670M sits at the 38th percentile among all GPUs, while the GT 555M sits at the 37th percentile. Both are mid-pack mobile parts. Against its nearest rivals, the GTX 670M is 0.3% ahead of the GT 555M, 0.5% ahead of the Quadro M5000M, 0.6% ahead of the AMD Radeon Vega 10 Mobile, and 0.6% behind the Intel UHD Graphics P750. The GT 555M, meanwhile, is 0.2% ahead of the Quadro M5000M, 0.3% ahead of the AMD Radeon Vega 10 Mobile, 0.3% behind the GTX 670M, and 0.9% behind the Intel UHD Graphics P750. These deltas are so small that they fall within normal run-to-run variance. Neither card wins any category decisively; they are clustered with integrated graphics and mid-range mobile parts in a narrow performance band.

Where the GTX 670M wins is on paper architecture: it carries more shading units, texture mapping units, and ROPs. Where the GT 555M wins is on power efficiency, with a 35 W TDP versus 75 W for the GTX 670M. In a laptop context, that power difference may translate into thermal headroom, battery life, or chassis design flexibility. The data does not show a compute performance advantage for either part that would justify choosing one over the other on speed alone.

Architecture Differences

The two GPUs share the Fermi lineage but diverge in implementation. The GTX 670M uses the GF114 chip, built on Fermi 2.0 architecture, while the GT 555M uses the GF106 chip with the original Fermi architecture. Both are manufactured by TSMC on a 40 nm process node, which means the lithography is identical. The transistor counts tell a story of scale: the GF114 packs 1,950 million transistors on a 332 mm² die, yielding a transistor density of 5.9 million per square millimeter. The GF106 contains 1,170 million transistors on a 238 mm² die, with a density of 4.9 million per square millimeter. The GTX 670M is thus a substantially larger and more complex chip, with 66.7% more transistors and a 39.5% larger die area.

The compute resources reflect that gap. The GTX 670M has 336 shading units, 56 texture mapping units, and 24 ROPs. The GT 555M has 144 shading units, 24 texture mapping units, and 16 ROPs. That is a 2.33x difference in shading units, a 2.33x difference in TMUs, and a 1.5x difference in ROPs. The raw throughput numbers follow accordingly. Pixel rate on the GTX 670M is 8.372 GPixel/s versus 3.540 GPixel/s on the GT 555M, a 2.36x advantage. Texture rate is 33.49 GTexel/s versus 14.16 GTexel/s, a 2.37x advantage. FP32 compute is 803.7 GFLOPS versus 339.8 GFLOPS, a 2.37x advantage. On every architectural throughput metric, the GTX 670M is roughly 2.3 to 2.4 times the GT 555M.

Memory architecture diverges as well. The GTX 670M uses 1536 MB of GDDR5 on a 192-bit bus, delivering 72.00 GB/s of bandwidth. The GT 555M uses 1024 MB of DDR3 on a 128-bit bus, delivering 28.80 GB/s. That is a 2.5x bandwidth advantage for the GTX 670M. The memory clock is listed as 750 MHz (3 Gbps effective) for the GTX 670M and 900 MHz (1800 Mbps effective) for the GT 555M, but the wider bus and faster memory type make the GTX 670M far superior in memory throughput. The GT 555M also uses a different bus interface: PCIe 2.0 x16, whereas the GTX 670M uses MXM-B (3.0). The GTX 670M is an MXM module, while the GT 555M has no slot width listed.

Both parts support DirectX 12 (11_0) and OpenGL 4.6. Neither lists Vulkan support. Both are end-of-life products. The GTX 670M was released on 2012-03-21, and the GT 555M on 2011-07-01, so the GTX 670M is roughly eight months newer. The GTX 670M belongs to the GeForce 600M generation, while the GT 555M belongs to the GeForce 500M generation. The GTX 670M's predecessor is GeForce 500M, and its successor is GeForce 700M. The GT 555M's predecessor is GeForce 400M, and its successor is GeForce 600M.

Head-to-Head Benchmarks

The only head-to-head benchmark available is Geekbench OpenCL. The GTX 670M scores 6513, and the GT 555M scores 6493. The winner is the GTX 670M by 0.3%. That 20-point absolute difference is minuscule. To put it in perspective, the GTX 670M's lead over the GT 555M is the same as its lead over the Quadro M5000M (0.5% versus 0.3%) and smaller than its lead over the AMD Radeon Vega 10 Mobile (0.6%). The GT 555M is actually closer to the Quadro M5000M (0.2% ahead) and the AMD Radeon Vega 10 Mobile (0.3% ahead) than it is to the GTX 670M. The Intel UHD Graphics P750 beats both NVIDIA parts, sitting 0.6% above the GTX 670M and 0.9% above the GT 555M.

Given the architectural disparity — a 2.37x difference in FP32 compute and a 2.5x difference in memory bandwidth — the near-identical OpenCL scores are surprising. The data suggests that the Geekbench OpenCL workload does not scale with raw shading unit count or memory bandwidth. It may be memory-latency bound, or it may not fully utilize the additional resources. Alternatively, the GT 555M's higher memory clock (900 MHz versus 750 MHz) could help close the gap in certain access patterns, though the GTX 670M's GDDR5 and wider bus should still dominate. This is a case where the specification sheet promises a lopsided contest, but the benchmark result shows parity. The data implies that for typical OpenCL compute tasks in this era, the extra silicon on the GTX 670M did not translate into proportional performance gains.

The percentile rankings reinforce the parity. The GTX 670M is at the 38th percentile, the GT 555M at the 37th. Both are below the midpoint of all GPUs, which is consistent with mobile parts from 2011-2012. The deltaPct values among the nearest rivals are all under 1%, indicating that this entire cluster of GPUs — the GTX 670M, GT 555M, Quadro M5000M, Radeon Vega 10 Mobile, and Intel UHD Graphics P750 — are statistically indistinguishable in OpenCL performance. The GTX 670M's win is real in the data but practically negligible.

Specification Differences

The following fields differ between the two parts:

• Chip: GF114 (GTX 670M) versus GF106 (GT 555M)
• Architecture: Fermi 2.0 versus Fermi
• Generation: GeForce 600M versus GeForce 500M
• Transistors: 1,950 million versus 1,170 million
• Die Size: 332 mm² versus 238 mm²
• Transistor Density: 5.9M / mm² versus 4.9M / mm²
• Memory Clock: 750 MHz (3 Gbps effective) versus 900 MHz (1800 Mbps effective)
• Memory Size: 1536 MB versus 1024 MB
• Memory Type: GDDR5 versus DDR3
• Memory Bus Width: 192 bit versus 128 bit
• Memory Bandwidth: 72.00 GB/s versus 28.80 GB/s
• Shading Units: 336 versus 144
• TMUs: 56 versus 24
• ROPs: 24 versus 16
• Pixel Rate: 8.372 GPixel/s versus 3.540 GPixel/s
• Texture Rate: 33.49 GTexel/s versus 14.16 GTexel/s
• FP32: 803.7 GFLOPS versus 339.8 GFLOPS
• TDP: 75 W versus 35 W
• Slot Width: MXM Module versus null
• Power Connectors: None versus null
• Bus Interface: MXM-B (3.0) versus PCIe 2.0 x16
• Release Date: 2012-03-21 versus 2011-07-01
• Predecessor: GeForce 500M versus GeForce 400M
• Successor: GeForce 700M versus GeForce 600M

Fields that are identical or null on both: series, codename, process node (40 nm), foundry (TSMC), base/boost/game clocks, RT cores, tensor cores, FP16, suggested PSU, dimensions, display outputs, DirectX (12 (11_0)), OpenGL (4.6), Vulkan (null), production status (End-of-life), and launch MSRP (null).

FAQ

Q: Which GPU is faster in the benchmark data?

A: The NVIDIA GeForce GTX 670M wins the Geekbench OpenCL test with a score of 6513 versus 6493 for the NVIDIA GeForce GT 555M. The delta is 0.3%.

Q: How does the GTX 670M compare to the GT 555M in raw compute specifications?

A: The GTX 670M has 336 shading units, 56 TMUs, and 24 ROPs, with FP32 performance of 803.7 GFLOPS. The GT 555M has 144 shading units, 24 TMUs, and 16 ROPs, with FP32 performance of 339.8 GFLOPS. The GTX 670M is roughly 2.37x faster on paper.

Q: What is the memory bandwidth difference between the two cards?

A: The GTX 670M uses 1536 MB of GDDR5 on a 192-bit bus, delivering 72.00 GB/s. The GT 555M uses 1024 MB of DDR3 on a 128-bit bus, delivering 28.80 GB/s. The GTX 670M has 2.5x the bandwidth.

Q: Which GPU draws less power?

A: The GT 555M has a TDP of 35 W, while the GTX 670M has a TDP of 75 W. The GT 555M is the more power-efficient part.

Q: How do these GPUs rank against all other GPUs?

A: The GTX 670M is at the 38th percentile, and the GT 555M is at the 37th percentile. Both are in the lower-middle range of the overall GPU distribution.

Q: Are these GPUs still in production?

A: No. Both are listed as end-of-life products. The GTX 670M was released on 2012-03-21, and the GT 555M was released on 2011-07-01.

The Verdict

The data presents a paradox. On every architectural metric, the GTX 670M is overwhelmingly superior: 2.37x the shading units, 2.37x the texture rate, 2.36x the pixel rate, 2.5x the memory bandwidth, and 66.7% more transistors. Yet the only benchmark result shows a 0.3% difference. This suggests that for the specific workload captured by Geekbench OpenCL, the two GPUs perform essentially identically. Anyone choosing between them on the basis of compute benchmarks would be selecting on a coin flip, with the GTX 670M holding a negligible edge.

For a user who values raw specification headroom — for future workloads, for higher resolutions, or for tasks that are not represented by this single benchmark — the GTX 670M is the obvious pick. It offers more of everything that matters on paper: memory bandwidth, shading power, and texture throughput. The 75 W TDP is a cost, but for a laptop designed to handle it, the GTX 670M provides a more capable foundation.

For a user who prioritizes power efficiency, the GT 555M is the rational choice. At 35 W, it draws less than half the power of the GTX 670M while delivering statistically identical OpenCL performance. In a thin-and-light chassis, that power difference could be decisive. The GT 555M also uses a standard PCIe 2.0 x16 interface, whereas the GTX 670M requires an MXM module, which limits upgrade paths and platform compatibility.

The verdict is situational. If performance per watt matters, the GT 555M wins. If maximum theoretical capability matters, the GTX 670M wins. The benchmark data alone cannot separate them; the specification sheet can. The GTX 670M is the more powerful part that fails to demonstrate its advantage in the available test. The GT 555M is the efficient part that punches above its weight. Neither is a bad choice, but they serve different priorities.