Intel Iris Pro Graphics P6300 vs NVIDIA GeForce GTX 670MX Comparison

The NVIDIA GeForce GTX 670MX and Intel Iris Pro Graphics P6300 occupy the same average performance tier despite their radically different designs. The benchmark data shows the GTX 670MX holds the only head-to-head win, but by a margin that places both firmly in the same competitive class. In the sole direct comparison available, the Geekbench OpenCL test, the GTX 670MX scores 6,125 against the Iris Pro’s 5,712, a 7.2% advantage. The average benchmark scores tell a tighter story: the GTX 670MX averages 5,721, while the Iris Pro averages 5,712, a difference of only 0.2%. Both GPUs sit at the 33rd percentile of all GPUs, indicating that neither is a performance leader in the broader market.

Head-to-Head Benchmarks

The available head-to-head data consists of a single Geekbench OpenCL run, which the NVIDIA GeForce GTX 670MX wins decisively. The GTX 670MX posts 6,125 points, while the Intel Iris Pro Graphics P6300 manages 5,712 points, yielding a 7.2% delta in favor of the discrete NVIDIA part. This is the only direct comparison in the dataset, and it indicates that in raw compute workloads measured by OpenCL, the Kepler-based mobile GPU holds a clear edge.

However, the context from the nearest rivals list tempers that victory. The GTX 670MX’s average score of 5,721 places it just 0.2% above the Iris Pro’s 5,712 average. The delta percentages against other rivals reinforce how close these parts are: the GTX 670MX is 0.2% ahead of the Iris Pro, 0.2% behind the GTX 550 Ti (which scores 5,731), and 0.5% ahead of the AMD Radeon HD 8790M (5,691). For the Iris Pro, the picture is nearly identical in reverse: it sits 0.1% behind the GTX 670MX, 0.3% behind the GTX 550 Ti, and 0.4% ahead of the Radeon HD 8790M. The single head-to-head win is real, but the aggregate data shows these GPUs are effectively peers in overall benchmark performance.

The 7.2% OpenCL delta is the largest margin between the two in any metric, yet it does not translate into a dominant overall lead. The GTX 670MX’s higher raw compute throughput is evident in its 1,153.9 GFLOPS FP32 rating versus the Iris Pro’s 614.4 GFLOPS, but the average scores suggest that real-world workloads do not always scale linearly with theoretical peak. The GTX 670MX also wins on pixel rate (12.02 GPixel/s vs. 4.80 GPixel/s) and texture rate (48.08 GTexel/s vs. 38.40 GTexel/s), yet the final average scores remain within 0.2% of each other.

Architecture Differences

The two GPUs come from entirely different design philosophies. The NVIDIA GeForce GTX 670MX uses the GK104 chip on the Kepler architecture, built on a 28 nm process at TSMC. The die measures 294 mm² and contains 3,540 million transistors, yielding a transistor density of 12.0 million per mm². The Intel Iris Pro Graphics P6300, in contrast, uses the Broadwell GT3e chip on the Generation 8.0 architecture, fabricated on Intel’s 14 nm process. Intel does not disclose transistor count or die size for this part, but the process node advantage is clear: 14 nm versus 28 nm.

Clock speeds differ substantially. The GTX 670MX runs at a base and boost clock of 601 MHz, which is relatively modest for a discrete GPU. The Iris Pro starts at 300 MHz base but boosts to 800 MHz, a higher peak clock that partially compensates for its smaller execution resources. Memory configurations are fundamentally different: the GTX 670MX has 3 GB of dedicated GDDR5 on a 192-bit bus, delivering 67.20 GB/s of bandwidth, while the Iris Pro uses system shared memory with system-dependent bandwidth. The NVIDIA part’s memory clock is 700 MHz (2.8 Gbps effective), whereas the Intel solution relies on the host system’s memory.

Execution resources also diverge sharply. The GTX 670MX packs 960 shading units, 80 texture mapping units, and 24 ROPs. The Iris Pro has 384 shading units, 48 TMUs, and only 6 ROPs. The NVIDIA part’s larger ROP count drives its 12.02 GPixel/s fill rate, while the Intel part manages 4.80 GPixel/s. Texture rate is closer: 48.08 GTexel/s versus 38.40 GTexel/s, a 25% gap. The FP32 compute figures reflect the shader count difference: 1,153.9 GFLOPS for NVIDIA versus 614.4 GFLOPS for Intel, a 1.88x advantage.

Power draw is another major architectural differentiator. The GTX 670MX is rated at 75 W TDP and requires no power connectors, while the Iris Pro is an IGP with a 15 W TDP. The Intel part’s bus interface is a Ring Bus, typical for integrated graphics, whereas the NVIDIA part uses PCIe 3.0 x16. Display outputs are portable-device-dependent for the GTX 670MX and motherboard-dependent for the Iris Pro, reflecting their intended deployment scenarios.

Where Each One Wins

The GTX 670MX wins the only direct benchmark in the dataset, the Geekbench OpenCL test, by 7.2%. Its architectural strengths suggest it is better suited for workloads that exploit high pixel throughput (12.02 GPixel/s) and large dedicated memory pools (3 GB GDDR5). The FP32 compute rating of 1,153.9 GFLOPS is nearly double the Intel part’s 614.4 GFLOPS, indicating an advantage in compute-heavy tasks. The GTX 670MX also has more texture units (80 vs. 48) and more ROPs (24 vs. 6), which should help in traditional rasterization workloads.

The Iris Pro Graphics P6300 does not win any head-to-head benchmark, but it is not far behind. Its average score of 5,712 is within 0.2% of the GTX 670MX’s 5,721. The Intel part’s higher boost clock (800 MHz versus 601 MHz) and modern 14 nm process allow it to approach the NVIDIA part’s performance with far fewer resources. The Iris Pro’s 15 W TDP makes it suitable for systems where power draw is a primary constraint, and its system-shared memory model eliminates the need for dedicated VRAM. In use cases where the host system provides fast memory, the Iris Pro can deliver competitive OpenCL results despite its 384 shading units.

The nearest rivals data show that both GPUs sit in a tight cluster. The GTX 670MX is 0.2% behind the GTX 550 Ti, while the Iris Pro is 0.3% behind the same card. Both are ahead of the AMD Radeon HD 8790M, with the GTX 670MX leading by 0.5% and the Iris Pro by 0.4%. The NVIDIA Quadro M500M is farther behind at 5,604, with the GTX 670MX ahead by 2.1% and the Iris Pro by 1.9%. Neither GPU can claim a meaningful overall victory in this peer group.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 670MX has an average benchmark score of 5,721, while the Intel Iris Pro Graphics P6300 averages 5,712, a difference of 0.2%.

Q: What is the performance difference in the head-to-head OpenCL test?

A: The GTX 670MX scores 6,125 in Geekbench OpenCL, while the Iris Pro scores 5,712, giving the NVIDIA part a 7.2% win.

Q: How do the two GPUs compare on power consumption?

A: The GTX 670MX has a 75 W TDP, while the Iris Pro is rated at 15 W TDP, a five-fold difference.

Q: Which GPU has more shading units?

A: The GTX 670MX has 960 shading units, compared to 384 on the Iris Pro.

Q: What type of memory does each GPU use?

A: The GTX 670MX uses 3 GB of dedicated GDDR5 on a 192-bit bus, while the Iris Pro uses system shared memory with system-dependent bandwidth.

Q: How does the FP32 compute performance compare?

A: The GTX 670MX delivers 1,153.9 GFLOPS, while the Iris Pro delivers 614.4 GFLOPS, a 1.88x advantage for NVIDIA.

Specification Differences

The two GPUs differ across nearly every specification category. The process node is 28 nm for the GTX 670MX and 14 nm for the Iris Pro. The NVIDIA part uses the GK104 chip with 3,540 million transistors on a 294 mm² die, while Intel does not disclose those figures. The GTX 670MX runs at 601 MHz base and boost, while the Iris Pro runs at 300 MHz base and 800 MHz boost. Memory is 3 GB GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth for NVIDIA, versus system shared memory for Intel. The NVIDIA part has 960 shading units, 80 TMUs, and 24 ROPs, while the Iris Pro has 384 shading units, 48 TMUs, and 6 ROPs. Pixel rate is 12.02 GPixel/s versus 4.80 GPixel/s, texture rate is 48.08 GTexel/s versus 38.40 GTexel/s, and FP32 is 1,153.9 GFLOPS versus 614.4 GFLOPS. TDP is 75 W versus 15 W. The GTX 670MX uses PCIe 3.0 x16, while the Iris Pro uses a Ring Bus. API support differs: the GTX 670MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, while the Iris Pro supports DirectX 12 (11_1), OpenGL 4.4, and Vulkan 1.0. Release dates are September 2012 for NVIDIA and September 2014 for Intel.

The Verdict

The data supports a nuanced conclusion. The NVIDIA GeForce GTX 670MX wins the only head-to-head benchmark and offers superior theoretical specifications in nearly every category: double the FP32 throughput, four times the ROPs, and dedicated GDDR5 memory. Its 7.2% OpenCL victory is the largest gap in the dataset, and its average score is higher than the Iris Pro’s by a hair. For users who prioritize raw compute or pixel fill rate, the GTX 670MX is the clear choice.

The Intel Iris Pro Graphics P6300, however, achieves 99.8% of the GTX 670MX’s average performance while consuming only 20% of the power (15 W versus 75 W). Its 14 nm process and higher boost clock allow it to compete despite having fewer than half the shading units. For systems where power efficiency is paramount, the Iris Pro is the better option. The data shows no workload where the Iris Pro outright wins, but its average score of 5,712 places it in the same performance class as the GTX 670MX.

Ultimately, the choice depends on system constraints. The GTX 670MX is a discrete, end-of-life mobile GPU with a 75 W TDP and dedicated memory, suited for heavier workloads. The Iris Pro is an integrated part with a 15 W TDP and system-shared memory, designed for power-lean platforms. Benchmarks indicate they trade blows within a 0.2% average margin, with the NVIDIA part holding the only direct win. Neither GPU offers a compelling overall advantage; the GTX 670MX edges ahead on paper and in the single test, while the Iris Pro offers the same class of performance at a fraction of the power draw.