Intel Arc B390 vs NVIDIA GeForce RTX 5070 Ti SUPER Comparison

Head-to-Head Benchmarks

The single recorded comparison between the Intel Arc B390 and the NVIDIA GeForce RTX 5070 Ti SUPER is the 3DMark Steel Nomad DX12 test. The NVIDIA card scores 6269.5 points, while the Intel integrated part scores 1482 points. This results in a delta of -76.4% for the Intel Arc B390, meaning the RTX 5070 Ti SUPER is roughly four times faster in this workload. The margin is decisive, and the data does not show any test where the Intel part closes the gap.

In percentile terms, the RTX 5070 Ti SUPER sits at the 36th percentile of all GPUs in the database, while the Arc B390 sits at the 9th percentile. That 27-point percentile gap underscores how far apart these two parts are in raw 3D performance. The nearest rivals for the RTX 5070 Ti SUPER include the NVIDIA GeForce RTX 4070 Ti SUPER AD102 with an average score of 6270 and a delta of 0%, effectively a dead heat. The Quadro K620 is 0.2% behind, the AMD FirePro W600 is 0.8% behind, and the AMD Radeon R7 M350 is 0.9% ahead. These are extremely tight margins, indicating that the RTX 5070 Ti SUPER lands in a well-populated performance cluster around the 6200 to 6300 point range.

The Arc B390's nearest rivals are all much older, low-end discrete parts. The NVIDIA GeForce GT 520MX scores 1463, just 1.3% behind. The GeForce 800M scores 1460, 1.5% behind. The GT 625 OEM scores 1446, 2.5% behind, and the GT 710 scores 1443, 2.7% behind. The Intel part leads all four of these by small single-digit margins. This places the Arc B390 at the very low end of the performance spectrum, alongside entry-level GPUs from over a decade ago, rather than anywhere near modern desktop-class hardware.

The gap between the two parts is not incremental; it is categorical. The RTX 5070 Ti SUPER delivers 43.94 TFLOPS of FP32 compute, while the Arc B390 delivers 7.680 TFLOPS. That is a 5.7x difference in raw shader throughput. The NVIDIA card also has 8960 shading units against 1536 for Intel, 280 texture mapping units against 48, and 96 ROPs against 24. Every major hardware resource is heavily skewed toward the discrete card.

Where Each One Wins

The RTX 5070 Ti SUPER wins the only benchmark test that exists in the database, so in measured 3D performance it is the clear victor. Its 6269.5 point score in 3DMark Steel Nomad places it at the 36th percentile, and its nearest rivals are all within 1% either way. The data suggests this card is squarely in the upper-midrange of the database's GPU rankings, trading blows with the RTX 4070 Ti SUPER AD102, which scores 6270. The 896.0 GB/s memory bandwidth, 16 GB of GDDR7 memory on a 256-bit bus, and 280 tensor cores give it capabilities that the Arc B390 cannot approach on any metric recorded.

The Arc B390 wins in the sense that it leads its own nearest rivals, all of which are older NVIDIA entry-level parts. It beats the GT 520MX by 1.3%, the GeForce 800M by 1.5%, the GT 625 OEM by 2.5%, and the GT 710 by 2.7%. That is a narrow but consistent lead across four comparable low-end GPUs. The Intel part also has a much lower power draw, rated at 80 W versus 350 W for the NVIDIA card, and its integrated nature means it requires no power connectors and occupies no expansion slot. For systems where discrete graphics is impossible or undesirable, the Arc B390 is the only option of the two that can function at all.

The use-case split is therefore straightforward. The RTX 5070 Ti SUPER is for high-performance 3D workloads, given its 5.7x FP32 advantage, higher pixel rate (235.4 GPixel/s versus 60.00 GPixel/s), and higher texture rate (686.6 GTexel/s versus 120.0 GTexel/s). The Arc B390 is for low-power integrated graphics duties, where its 80 W TDP and system-shared memory are design constraints rather than weaknesses. The data does not show any workload where the Intel part outperforms the NVIDIA part, but it does show the Arc B390 holding its own among the weakest discrete GPUs in the database.

Architecture Differences

The two chips come from different manufacturers, fabs, and design generations. The Intel Arc B390 uses the Panther Lake chip with the Xe3-LPG architecture, built on Intel's 3 nm process. The NVIDIA GeForce RTX 5070 Ti SUPER uses the GB203 chip with the Blackwell 2.0 architecture, built on TSMC's 5 nm process. The Intel part belongs to the Arc Graphics-M (Panther Lake) generation, while the NVIDIA part belongs to the GeForce 50 series.

Transistor counts and die sizes are only recorded for the NVIDIA chip. The GB203 packs 45,600 million transistors on a 378 mm² die, yielding a transistor density of 120.6M per mm². The Intel chip's transistor count and die size are listed as unknown, so no direct density comparison is possible. The NVIDIA card uses 16 GB of GDDR7 memory on a 256-bit bus with 896.0 GB/s of bandwidth. The Arc B390 has no dedicated memory at all; its memory size, type, and bus width are all listed as system shared, with bandwidth described as system dependent. This is a fundamental architectural split: one part has its own high-speed memory pool, the other borrows from the host system.

The compute resources differ sharply. The Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores. The RTX 5070 Ti SUPER has 8960 shading units, 280 TMUs, 96 ROPs, and 70 ray tracing cores. The NVIDIA part also has 280 tensor cores, while the Intel part lists none. FP32 throughput is 43.94 TFLOPS for NVIDIA and 7.680 TFLOPS for Intel. FP16 throughput is 43.94 TFLOPS for NVIDIA at a 1:1 ratio, and 15.36 TFLOPS for Intel at a 2:1 ratio. The NVIDIA card's FP16 rate equals its FP32 rate, indicating full-rate half-precision execution, while the Intel part halves its rate when moving from FP32 to FP16.

Clock behavior also differs. The Arc B390 has a base clock of 300 MHz and a boost clock of 2500 MHz, a wide dynamic range consistent with a power-sensitive integrated part. The RTX 5070 Ti SUPER has a base clock of 2295 MHz and a boost clock of 2452 MHz, a much narrower range. Pixel rate for the NVIDIA card is 235.4 GPixel/s versus 60.00 GPixel/s for Intel. Texture rate is 686.6 GTexel/s versus 120.0 GTexel/s. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, though the NVIDIA card's feature set is backed by substantially more hardware resources.

Physical and power characteristics are equally divergent. The Arc B390 is an IGP with no power connectors, no slot width, and display outputs that are portable device dependent. The RTX 5070 Ti SUPER is a dual-slot card measuring 304 mm in length, 137 mm in height, and 48 mm in width, requiring a single 16-pin power connector. Its TDP is 350 W, more than four times the Intel part's 80 W. The NVIDIA card uses a PCIe 5.0 x16 bus interface, while the Intel part is hardwired to the system as an integrated GPU. The NVIDIA card offers 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs, a fixed set of display connections that the Intel part cannot match, since its outputs depend on the host portable device.

The Verdict

The benchmark data gives a clear answer: the RTX 5070 Ti SUPER is the overwhelmingly faster GPU. Its 6269.5 point score in 3DMark Steel Nomad is 76.4% higher than the Arc B390's 1482 points, and its 36th percentile standing places it in a different performance class entirely. The Intel part's nearest rivals are the GT 520MX, GeForce 800M, GT 625 OEM, and GT 710, all of which it beats by margins of 1.3% to 2.7%. Those are not modern competitors; they are legacy entry-level parts. The RTX 5070 Ti SUPER's nearest rivals, by contrast, include the RTX 4070 Ti SUPER AD102 at a 0% delta, meaning the two are statistically identical in this test. The NVIDIA card also has the hardware to back up its score: 16 GB of GDDR7, 896.0 GB/s of bandwidth, 8960 shading units, and 70 RT cores. The Arc B390 has none of those discrete resources, relying instead on system shared memory and a 1536-shader configuration.

For a user selecting a GPU for measured 3D performance, the RTX 5070 Ti SUPER is the only choice supported by the data. For a user constrained to integrated graphics, the Arc B390 at least outperforms the GT 710 and similar low-end parts in the database. The production status of both is active, and the NVIDIA card's launch MSRP is 749 USD. The release dates are close, with the RTX 5070 Ti SUPER released on 2025-12-31 and the Arc B390 on 2026-01-26, but the performance gap between them is not close at all.

FAQ

Q: Which GPU has the higher 3DMark Steel Nomad score?

A: The NVIDIA GeForce RTX 5070 Ti SUPER scores 6269.5 points, while the Intel Arc B390 scores 1482 points, a delta of -76.4% for the Intel part.

Q: How does the Arc B390 compare to its nearest rivals?

A: The Arc B390 leads the NVIDIA GeForce GT 520MX by 1.3%, the GeForce 800M by 1.5%, the GT 625 OEM by 2.5%, and the GT 710 by 2.7%.

Q: What are the memory configurations of each GPU?

A: The RTX 5070 Ti SUPER has 16 GB of GDDR7 on a 256-bit bus with 896.0 GB/s bandwidth. The Arc B390 uses system shared memory with system dependent bandwidth.

Q: How do the architectures differ?

A: The Arc B390 uses Intel's Xe3-LPG architecture on a 3 nm process, while the RTX 5070 Ti SUPER uses NVIDIA's Blackwell 2.0 architecture on a 5 nm process.

Q: What is the TDP of each card?

A: The Arc B390 is rated at 80 W and requires no power connectors. The RTX 5070 Ti SUPER is rated at 350 W and requires one 16-pin power connector.

Q: Which GPU has more ray tracing cores?

A: The RTX 5070 Ti SUPER has 70 ray tracing cores, compared to 12 ray tracing cores on the Arc B390.