Intel Iris Xe MAX Graphics vs NVIDIA GeForce GTX 670 Comparison

Head-to-Head Benchmarks

The recorded data contains a single direct comparison between the Intel Iris Xe MAX Graphics and the NVIDIA GeForce GTX 670. That test is Geekbench OpenCL, where the NVIDIA GPU posts a score of 15341 against 14315 for the Intel part. This gives the GTX 670 a 6.7% advantage, a meaningful but not overwhelming margin in raw compute throughput.

The OpenCL result aligns with the broader benchmark averages in the database. The GTX 670 carries an average benchmark score of 12773, while the Iris Xe MAX averages 14315. Interestingly, the Intel part has a higher average score despite losing the head-to-head OpenCL run, which reflects that the GTX 670's average is dragged down by additional tests (Metal and Vulkan) where its scores are lower. The GTX 670's Metal result is 7424, and its Vulkan result is 15553, the latter being its strongest single measurement.

When placed against their respective nearest rivals, both GPUs sit near the middle of the pack. The Iris Xe MAX lands at the 56th percentile of all GPUs, while the GTX 670 sits at the 52nd percentile. The Intel part's closest competitor, the AMD Radeon Vega 11, scores 14352, just 0.3% higher, making the two statistically indistinguishable. The NVIDIA GeForce GTX 1070 Ti trails the Intel part by only 0.3% at 14277, and the GTX TITAN leads by 0.4% at 14373. The AMD Radeon RX Vega 11 is 0.5% ahead at 14385. For the GTX 670, the nearest rivals cluster within a narrow band: the GTX 590 scores 12830 (0.4% ahead), the AMD Radeon Pro 455 scores 12831 (0.4% ahead), the AMD Radeon RX 7600M XT scores 12710 (0.5% behind), and the AMD FirePro W5100 scores 12847 (0.6% ahead). Both cards effectively trade blows with their immediate competition.

The single head-to-head shows a clear winner for the GTX 670, but the context matters. The Intel Iris Xe MAX is an integrated-class part with a 25 W TDP, while the GTX 670 is a dual-slot discrete card with a 170 W TDP. The performance gap of 6.7% is far smaller than the power envelope difference would suggest. The GTX 670 also achieves its OpenCL score with an older Kepler architecture, while the Intel part uses Generation 12.1 Xe Graphics. In raw terms, the GTX 670 wins the only directly comparable test, but the Intel part is competitive enough to remain relevant in the same performance tier.

Where Each One Wins

The NVIDIA GeForce GTX 670 wins the only direct benchmark comparison, the Geekbench OpenCL test, with a 6.7% margin. This is its clear victory condition. The GTX 670 also demonstrates versatility by supporting three separate benchmark APIs: OpenCL, Metal, and Vulkan. Its Vulkan score of 15553 is the highest single measurement for either card, surpassing the Iris Xe MAX's OpenCL score of 14315 by 8.6%. That suggests the GTX 670 excels in cross-platform compute workloads, particularly those leveraging Vulkan for lower-level hardware access.

The Intel Iris Xe MAX Graphics, despite losing the head-to-head, secures wins in other respects. Its average benchmark score of 14315 exceeds the GTX 670's average of 12773 by 12.1%. This is because the Intel part has only one recorded benchmark (OpenCL), while the GTX 670's average includes the much lower Metal score of 7424, which pulls its average down. The Iris Xe MAX also sits at a higher percentile (56th versus 52nd), indicating that its single result places it slightly better relative to the entire GPU landscape.

In use-case terms, the GTX 670 is the choice for applications that can use Vulkan or Metal, since it has recorded results in both. The Iris Xe MAX has no recorded Vulkan or Metal scores, so its OpenCL performance is the only measurable strength. For compute workloads relying solely on OpenCL, the GTX 670 still leads by 6.7%, so the Intel part does not win any specific benchmark. However, the Iris Xe MAX's higher average and percentile suggest that for a user looking at overall database standing, the Intel part is not far behind despite the direct loss.

The GTX 670's wins are concentrated in raw compute throughput and API coverage. The Iris Xe MAX's advantages are structural: it has a higher average score, a better percentile, and a much lower power draw, though the latter is not a benchmark result. For pure performance, the GTX 670 is the winner in the measured tests. For efficiency and database standing, the Intel part holds its own.

FAQ

Q: Which GPU has a higher Geekbench OpenCL score?

A: The NVIDIA GeForce GTX 670 scores 15341, while the Intel Iris Xe MAX Graphics scores 14315. The GTX 670 leads by 6.7%.

Q: How does the Intel Iris Xe MAX compare to its nearest rival, the AMD Radeon Vega 11?

A: The AMD Radeon Vega 11 scores 14352, which is 0.3% higher than the Intel part's 14315. The difference is negligible.

Q: What is the GTX 670's strongest benchmark result?

A: The GTX 670's best recorded score is in Geekbench Vulkan at 15553, which is higher than its OpenCL score of 15341 and its Metal score of 7424.

Q: Which GPU has a better percentile ranking among all GPUs?

A: The Intel Iris Xe MAX sits at the 56th percentile, while the NVIDIA GeForce GTX 670 sits at the 52nd percentile. The Intel part ranks slightly higher.

Q: Why does the Intel Iris Xe MAX have a higher average benchmark score despite losing the head-to-head?

A: The Intel part has only one recorded benchmark (OpenCL at 14315), while the GTX 670 has three recorded benchmarks, including a low Metal score of 7424. That Metal result lowers the GTX 670's average to 12773, while the Intel part's average remains 14315.

Q: How close is the GTX 670 to its nearest rival, the AMD Radeon RX 7600M XT?

A: The AMD Radeon RX 7600M XT scores 12710, which is 0.5% lower than the GTX 670's average of 12773. The GTX 670 holds a slight edge.

Specification Differences

The two GPUs differ across nearly every major specification field. The Intel Iris Xe MAX uses a 10 nm process node fabricated by Intel, while the NVIDIA GeForce GTX 670 uses a 28 nm node from TSMC. The GTX 670 has a die size of 294 mm² and contains 3,540 million transistors, with a transistor density of 12.0M per mm². The Intel part has a die size of 95 mm², and no transistor count or density is recorded.

Memory configurations diverge sharply. The Intel part has 4 GB of LPDDR4X on a 128 bit bus, delivering 68.26 GB/s of bandwidth. The GTX 670 has 2 GB of GDDR5 on a 256 bit bus, delivering 192.3 GB/s, nearly three times the bandwidth. Clock speeds also differ: the Iris Xe MAX runs at a base of 300 MHz and boosts to 1650 MHz, while the GTX 670 runs at 915 MHz base and 980 MHz boost. The memory clock for the Intel part is 2133 MHz (4.3 Gbps effective), while the GTX 670's memory runs at 1502 MHz (6 Gbps effective).

The compute units show a different distribution. The Intel part has 768 shading units, 48 TMUs, and 24 ROPs. The GTX 670 has 1344 shading units, 112 TMUs, and 32 ROPs. This leads to different rates: the Intel part achieves 39.60 GPixel/s pixel rate and 79.20 GTexel/s texture rate, while the GTX 670 achieves 27.44 GPixel/s and 109.8 GTexel/s. The FP32 throughput is close, with the Intel part at 2.534 TFLOPS and the GTX 670 at 2.634 TFLOPS. The Intel part also lists FP16 at 5.069 TFLOPS (2:1), while no FP16 figure is recorded for the GTX 670.

Power and physical specs differ substantially. The Intel part has a TDP of 25 W, a suggested PSU of 200 W, and is an IGP with no display outputs. The GTX 670 has a TDP of 170 W, a suggested PSU of 450 W, is a dual-slot card with 2x 6-pin power connectors, and measures 241 mm by 111 mm by 38 mm. The bus interface also differs: the Intel part uses PCIe 4.0 x8, while the GTX 670 uses PCIe 3.0 x16. The GTX 670 has a launch MSRP of 399 USD. The Intel part has no recorded launch MSRP. The GTX 670 provides 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2 outputs, while the Intel part has no outputs.

Architecture Differences

The architectural split reflects a generational gap. The Intel Iris Xe MAX is built on the DG1 chip using Generation 12.1 architecture, part of the Xe Graphics generation. The NVIDIA GeForce GTX 670 is built on the GK104 chip using Kepler architecture, part of the GeForce 600 generation. The Intel part uses a 10 nm process, while the GTX 670 uses 28 nm, a significant node difference that explains the Intel part's much smaller die (95 mm² versus 294 mm²) and lower TDP (25 W versus 170 W).

Both cards lack dedicated ray tracing cores and tensor cores, so neither offers hardware-accelerated ray tracing or AI tensor operations. The Intel part has 768 shading units, 48 TMUs, and 24 ROPs, while the GTX 670 has 1344 shading units, 112 TMUs, and 32 ROPs, giving the NVIDIA card a substantial advantage in raw shader count and texture units. The Intel part compensates with a much higher boost clock (1650 MHz versus 980 MHz), which narrows the FP32 gap to just 4% (2.534 TFLOPS versus 2.634 TFLOPS).

Memory architecture differs fundamentally. The Intel part uses LPDDR4X with 4 GB capacity on a 128 bit bus, while the GTX 670 uses GDDR5 with 2 GB on a 256 bit bus. The GTX 670's bandwidth advantage is large at 192.3 GB/s versus 68.26 GB/s, but the Intel part offers double the capacity. The GTX 670's memory clock is 1502 MHz (6 Gbps effective), while the Intel part runs at 2133 MHz (4.3 Gbps effective), showing a trade-off between clock speed and bus width.

API support also differs. The Intel part supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The GTX 670 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The Intel part has a higher DirectX feature level and a newer Vulkan version, while the GTX 670 lags slightly in API versioning. The GTX 670 also records Metal and Vulkan benchmark scores, while the Intel part only records OpenCL, suggesting different API testing coverage in the database.

The production status for both is end-of-life. The Intel part was released on 2020-10-30, with its predecessor listed as Graphics and successor as Alchemist. The GTX 670 was released on 2012-05-09, with its predecessor as GeForce 500 and successor as GeForce 700. The Intel part is an integrated graphics processor with no outputs, while the GTX 670 is a discrete dual-slot card with multiple display outputs.

The Verdict

The data points to a narrow but consistent win for the NVIDIA GeForce GTX 670 in direct performance. The GTX 670 leads the only head-to-head benchmark, Geekbench OpenCL, by 6.7%, and it also posts a higher Vulkan score of 15553, which is the strongest single result between the two. For users prioritizing raw compute throughput in OpenCL or Vulkan workloads, the GTX 670 is the clear pick from the recorded measurements.

However, the Intel Iris Xe MAX Graphics has its own case. Its average benchmark score of 14315 is 12.1% higher than the GTX 670's average of 12773, and its 56th percentile ranking beats the GTX 670's 52nd. This is partly due to the GTX 670's low Metal score of 7424, which drags its average down. The Intel part also offers double the memory capacity (4 GB versus 2 GB) and a much lower TDP (25 W versus 170 W), though these are not benchmark results.

For a user building a system with a small power budget or limited space, the Intel part is the sensible choice given its IGP form factor and 25 W TDP. For a user seeking the highest compute scores in the database, the GTX 670 wins the direct comparison and offers broader API coverage with Metal and Vulkan results. The GTX 670 also carries a launch MSRP of 399 USD, which is the only price information available.

The GTX 670 is the better performer in the measured tests, with a 6.7% OpenCL lead and a higher peak Vulkan score. The Intel part is the better overall database entry, with a higher average and percentile, and it remains competitive despite losing the head-to-head. Neither card dominates decisively; the choice depends on whether the user values the GTX 670's raw speed and API coverage or the Intel part's efficiency and memory capacity. Based strictly on the recorded data, the GTX 670 is the performance winner, while the Iris Xe MAX is the more balanced all-rounder.