AMD FirePro W4170M vs NVIDIA GeForce GTX 670MX Comparison

Where Each One Wins

The benchmark data splits cleanly between these two mobile GPUs. The NVIDIA GeForce GTX 670MX wins the only head-to-head test recorded in the database, taking the Geekbench OpenCL result by a substantial margin. The AMD FirePro W4170M does not win any direct comparison, and in fact the database shows no benchmark category where it outperforms the GTX 670MX.

Looking at the broader context around each card, the GTX 670MX sits in the 33rd percentile of all GPUs tracked, with an average benchmark score of 5721. Its nearest rivals include the Intel Iris Pro Graphics P6300 at 5712 (just 0.2% behind), the NVIDIA GeForce GTX 550 Ti at 5731 (0.2% ahead), the AMD Radeon HD 8790M at 5691 (0.5% behind), and the NVIDIA Quadro M500M at 5604 (2.1% behind). This places the GTX 670MX in a tight cluster where small percentage differences separate competitors.

The FirePro W4170M lands in the 30th percentile with an average score of 5034. Its nearest rivals are the AMD Radeon R5 M430 at 5018 (0.3% behind), the AMD Radeon R7 M340 at 5063 (0.6% ahead), the AMD Radeon R7 240 at 5063 (0.6% ahead), and the AMD Radeon R7 Graphics at 4998 (0.7% behind). The FirePro sits in a similarly dense competitive band, but that band is noticeably lower than the one occupied by the GTX 670MX.

The practical interpretation is straightforward. If the workload is general GPU compute, OpenCL acceleration, or any task that the database's benchmark suite captures, the GTX 670MX is the stronger part. The FirePro W4170M is not without merit in its own weight class, but it does not challenge the GTX 670MX in any recorded metric. For a user choosing between these two, the data consistently points one direction.

Architecture Differences

The two chips come from different architectural lineages and different design philosophies. The GTX 670MX uses NVIDIA's Kepler architecture on the GK104 chip, fabricated on a 28 nm process at TSMC. The die is large at 294 mm² and packs 3,540 million transistors, which works out to a transistor density of 12.0 million per square millimeter. The FirePro W4170M uses AMD's GCN 1.0 architecture on the Opal chip, also fabricated on 28 nm at TSMC. The die is much smaller at 77 mm² with 950 million transistors, giving a density of 12.3 million per square millimeter.

These die size and transistor count differences reflect a fundamental design split. The GK104 is a wide, high-throughput design built for raw shading and texturing throughput. The Opal chip is a compact implementation with far fewer resources, aimed at fitting into constrained mobile chassis while still delivering GCN compute features.

The execution resources show the scale of the gap. The GTX 670MX has 960 shading units, 80 texture mapping units, and 24 raster output units. The FirePro W4170M has 384 shading units, 24 TMUs, and 8 ROPs. That is 2.5 times the shading units, over 3.3 times the TMUs, and 3 times the ROPs on the NVIDIA side. Pixel rate and texture rate follow accordingly: the GTX 670MX records 12.02 GPixel/s and 48.08 GTexel/s, while the FirePro W4170M manages 7.200 GPixel/s and 21.60 GTexel/s. FP32 throughput is 1,153.9 GFLOPS versus 691.2 GFLOPS.

Memory configurations also diverge. The GTX 670MX uses 3 GB of GDDR5 on a 192-bit bus, producing 67.20 GB/s of bandwidth. The FirePro W4170M uses 2 GB of GDDR5 on a 128-bit bus, producing 64.00 GB/s. The bandwidth figures are close, but the NVIDIA card has 50% more capacity and a wider bus. Clock behavior differs as well: the GTX 670MX runs at a flat 601 MHz for both base and boost, while the FirePro W4170M has a base of 825 MHz and a boost of 900 MHz. The AMD part clocks higher, but its smaller resource pool cannot overcome the NVIDIA card's raw width.

Interface and power delivery also differ. The GTX 670MX uses PCIe 3.0 x16 and draws 75 W with no power connectors required. The FirePro W4170M uses PCIe 3.0 x8 and is packaged as an MXM module, also with no power connectors. Both are marked end-of-life in the database, and both use Portable Device Dependent display outputs, meaning their video connections depend entirely on the laptop or workstation they are installed in.

Head-to-Head Benchmarks

The database contains one direct comparison between these two GPUs: the Geekbench OpenCL test. The GTX 670MX scores 6125, while the FirePro W4170M scores 5034. That is a delta of 21.7% in favor of the NVIDIA part. In practical terms, this means the GTX 670MX delivers roughly a fifth more OpenCL compute performance than the FirePro W4170M in the same workload.

Putting that score in context helps. The GTX 670MX's 6125 OpenCL result is above its own average benchmark score of 5721, meaning it performs slightly better in OpenCL than in its overall blended average. The FirePro W4170M's 5034 OpenCL score exactly matches its average benchmark score, so OpenCL is representative of its general performance profile.

Against their respective nearest rivals, the standings differ. The GTX 670MX's average of 5721 puts it just 0.2% behind the GeForce GTX 550 Ti and 0.2% ahead of the Iris Pro P6300, showing that it trades blows with desktop-class parts from its era. The FirePro W4170M's average of 5034 puts it 0.6% behind both the Radeon R7 M340 and Radeon R7 240, and 0.3% ahead of the Radeon R5 M430. The FirePro is competitive within its own peer group, but that group sits roughly 700 points lower in average score than the GTX 670MX's group.

The single head-to-head result tells the whole story. In the only recorded test where both cards ran the same benchmark, the GTX 670MX won by 21.7%. That margin is large enough to be decisive for compute-oriented workloads, and there is no recorded counterexample where the FirePro W4170M takes the lead.

FAQ

Q: Which GPU has higher raw compute throughput?

A: The GTX 670MX. It records 1,153.9 GFLOPS of FP32 performance versus 691.2 GFLOPS for the FirePro W4170M, a difference driven by its 960 shading units against 384.

Q: Do these cards support similar graphics APIs?

A: Both support DirectX 12 and OpenGL 4.6. The GTX 670MX supports Vulkan 1.2.175, while the FirePro W4170M supports Vulkan 1.2.170. The NVIDIA card's DirectX support is listed as 12 (11_0), while the AMD card lists 12 (11_1).

Q: How do their memory bandwidth numbers compare?

A: They are close. The GTX 670MX has 67.20 GB/s over a 192-bit bus with 3 GB of GDDR5. The FirePro W4170M has 64.00 GB/s over a 128-bit bus with 2 GB of GDDR5. The NVIDIA card has slightly more bandwidth and 50% more capacity.

Q: What is the transistor density difference between the two chips?

A: Minimal. The GTX 670MX's GK104 chip has 12.0 million transistors per square millimeter, while the FirePro W4170M's Opal chip has 12.3 million per square millimeter. Both are built on TSMC's 28 nm process.

Q: Which GPU sits higher in the overall database rankings?

A: The GTX 670MX, at the 33rd percentile of all GPUs, with an average benchmark score of 5721. The FirePro W4170M sits at the 30th percentile with an average score of 5034.

Q: Was the OpenCL head-to-head close?

A: No. The GTX 670MX scored 6125 against the FirePro W4170M's 5034, a 21.7% margin. That is a decisive win in the only direct comparison recorded.

Specification Differences

The two GPUs differ across nearly every major specification category. The GTX 670MX uses NVIDIA's Kepler architecture on the GK104 chip, while the FirePro W4170M uses AMD's GCN 1.0 architecture on the Opal chip. Transistor counts are 3,540 million versus 950 million, and die sizes are 294 mm² versus 77 mm². Transistor density is nearly identical at 12.0M per mm² versus 12.3M per mm².

Clock speeds favor the AMD part: the GTX 670MX runs at 601 MHz base and boost, while the FirePro W4170M runs at 825 MHz base and 900 MHz boost. Memory clocks also favor AMD, with the FirePro using 1000 MHz (4 Gbps effective) versus 700 MHz (2.8 Gbps effective) for the GTX 670MX. However, the NVIDIA card has more memory (3 GB versus 2 GB), a wider bus (192-bit versus 128-bit), and slightly higher bandwidth (67.20 GB/s versus 64.00 GB/s).

Compute resources heavily favor NVIDIA: 960 shading units versus 384, 80 TMUs versus 24, and 24 ROPs versus 8. Pixel rate is 12.02 GPixel/s versus 7.200 GPixel/s, texture rate is 48.08 GTexel/s versus 21.60 GTexel/s, and FP32 throughput is 1,153.9 GFLOPS versus 691.2 GFLOPS. The TDP for the GTX 670MX is listed as 75 W, while the FirePro W4170M has no TDP listed in the database.

Bus interfaces differ: PCIe 3.0 x16 for the GTX 670MX versus PCIe 3.0 x8 for the FirePro W4170M. The FirePro is packaged as an MXM module, while the GTX 670MX has no slot width listed. Both use no power connectors and portable-device-dependent display outputs. Release dates also differ, with the GTX 670MX launching in 2012 and the FirePro W4170M in 2015.

The Verdict

The data points to a clear choice for most users. The GTX 670MX wins the only direct benchmark comparison by 21.7%, has more than double the shading units, more than triple the texture units and ROPs, higher pixel and texture rates, higher FP32 throughput, more memory, and a wider memory bus. Its average benchmark score of 5721 places it at the 33rd percentile, while the FirePro W4170M's 5034 places it at the 30th percentile. In every measured category that matters for compute and graphics throughput, the NVIDIA part is ahead.

The FirePro W4170M does have some points in its favor. It clocks significantly higher, with a 900 MHz boost versus 601 MHz for the GTX 670MX. It also has a slightly higher transistor density, and its GCN architecture brings its own feature set. For a user who specifically needs AMD's mobile workstation feature set or who is constrained to an MXM module form factor, the FirePro W4170M could still be the right fit. But those are compatibility and ecosystem considerations, not performance ones.

For anyone choosing purely on recorded benchmark performance, the GTX 670MX is the stronger GPU. It wins the head-to-head, it holds a higher percentile ranking, and its nearest rivals are consistently faster than the FirePro W4170M's nearest rivals. The 21.7% OpenCL margin is the single most informative number in this comparison, and it favors NVIDIA without qualification.