Intel Arc B370 vs Intel Arc B390 Comparison

Intel Arc B370 and Intel Arc B390 belong to the same Panther Lake chip family and share the Xe3-LPG architecture. Both are integrated graphics solutions within the Arc Graphics-M generation. The recorded data includes a single standardized DirectX 12 benchmark, 3DMark Steel Nomad, which provides the basis for comparison. The B390 finishes ahead in that test, and the margin is substantial.

Head-to-Head Benchmarks

The only benchmark result available for both parts is 3dmark_3dmark_steel_nomad_dx12. In that test, the Intel Arc B370 scores 1184 points, while the Intel Arc B390 scores 1482 points. That is a difference of 298 points. The delta percentage, calculated from the perspective of the B370, shows the B390 leading by 20.1 percent. In other words, the B390 outperforms the B370 by roughly one fifth in this workload.

The gap places the two products in different performance tiers relative to the broader database. The B370 sits at the 5th percentile of all GPUs, meaning 95 percent of recorded parts score higher. The B390 reaches the 9th percentile, which is still low overall but represents a meaningful step up. The B390’s score of 1482 also places it ahead of several NVIDIA parts in its nearest rival group. It beats the GeForce GT 520MX by 1.3 percent, the GeForce 800M by 1.5 percent, the GeForce GT 625 OEM by 2.5 percent, and the GeForce GT 710 by 2.7 percent. Those deltas are small, but they are consistent wins.

The B370’s nearest rivals are all AMD or ATI parts. It trails the Mobility Radeon HD 5570 by 0.2 percent, the Radeon HD 5770 by 0.5 percent, and the Radeon HD 7650M by 0.7 percent. It leads the FirePro M2000 by 1.4 percent. This places the B370 in a very tight cluster where a few points separate the cards. The B390, by contrast, sits in a slightly higher cluster where its nearest competitors are all NVIDIA parts, and it leads all of them.

The win count confirms the direction of the data. In the head-to-head comparison, the B390 records one win, while the B370 records zero. There is no benchmark in the database where the B370 comes out ahead. That does not mean the B370 is without merit; it simply means the recorded measurements favor the B390 in the single available test. The 20.1 percent delta is a large margin for two parts that share the same architecture and chip generation. The B390’s higher boost clock and larger execution resource counts explain the difference in raw throughput.

The average benchmark score mirrors the single result. The B370 has an average of 1184, and the B390 has an average of 1482. Because only one test is recorded, the average is identical to the individual score. This makes the comparison clean and unambiguous. The B390 is faster in the measured workload, and the margin is not marginal.

FAQ

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

A: The Intel Arc B390 scores 1482 points, which is 298 points higher than the Intel Arc B370’s 1184 points. The B390 leads by 20.1 percent in this test.

Q: How does the Intel Arc B370 compare to its nearest rivals?

A: The B370 is very close to several AMD parts. It trails the ATI Mobility Radeon HD 5570 by 0.2 percent, the ATI Radeon HD 5770 by 0.5 percent, and the AMD Radeon HD 7650M by 0.7 percent. It leads the AMD FirePro M2000 by 1.4 percent.

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

A: The B390 leads all four of its nearest recorded rivals. It is 1.3 percent ahead of the NVIDIA GeForce GT 520MX, 1.5 percent ahead of the NVIDIA GeForce 800M, 2.5 percent ahead of the NVIDIA GeForce GT 625 OEM, and 2.7 percent ahead of the NVIDIA GeForce GT 710.

Q: What percentile do these GPUs occupy in the database?

A: The Intel Arc B370 sits at the 5th percentile of all GPUs. The Intel Arc B390 sits at the 9th percentile. Both are low in the overall distribution, but the B390 is positioned higher.

Q: Do both GPUs support the same DirectX version?

A: Yes. Both the Intel Arc B370 and the Intel Arc B390 report DirectX 12 Ultimate (12_2) support. They also both support OpenGL 4.6 and Vulkan 1.4.

Q: What is the memory configuration for these GPUs?

A: Both use system-shared memory. The size, type, bus width, and bandwidth are all listed as system shared or system dependent. Neither has dedicated video memory.

The Verdict

The data points to a clear hierarchy. The Intel Arc B390 is the faster part in the only recorded benchmark, and the margin is large enough that the two should not be treated as interchangeable. The B390’s score of 1482 compares favorably to its nearest rivals, all of which it beats. The B370’s score of 1184 puts it in a much lower percentile and among rivals where it is roughly even, trailing three parts by less than one percent and leading one by just over one percent.

For a user choosing between these two integrated graphics solutions, the B390 delivers measurably higher performance in the Steel Nomad workload. The 20.1 percent delta is consistent with the B390 having more shading units, more texture mapping units, more raster operation units, more ray tracing cores, and a higher boost clock. The B370 is not a poor performer in absolute terms; it is simply lower in the recorded distribution. The B390’s 9th percentile, while still low, is nearly double the B370’s 5th percentile.

Neither part is a high-end solution. Both occupy the low end of the database, and both rely on system-shared memory. The B390’s lead over its nearest rivals is consistent but small, ranging from 1.3 to 2.7 percent. The B370’s position among its nearest rivals is even tighter, with deltas under one percent in either direction. The practical takeaway from the measurements is that the B390 is the stronger option within this pair, and the B370 is the weaker one.

Specification Differences

The two GPUs share several specifications but differ in execution resources and power limits. Both use a 3 nm process node from Intel, both are built on the Xe3-LPG architecture, and both belong to the Arc Graphics-M generation with the Panther Lake chip. Neither has a dedicated memory pool; both use system shared memory with system dependent bandwidth.

The shading unit count differs. The B370 has 1280 shading units, while the B390 has 1536. That is an increase of 256 units. Texture mapping units also differ: the B370 has 40, and the B390 has 48. Raster operation units scale similarly, with the B370 at 20 and the B390 at 24. Ray tracing cores increase from 10 on the B370 to 12 on the B390.

Clock speeds are not identical. Both have a 300 MHz base clock, but the boost clock differs. The B370 boosts to 2400 MHz, while the B390 boosts to 2500 MHz. That is a 100 MHz advantage for the B390. The pixel rate reflects the combined effect of clocks and ROPs: the B370 reaches 48.00 GPixel/s, and the B390 reaches 60.00 GPixel/s. Texture rate also scales: the B370 delivers 96.00 GTexel/s, and the B390 delivers 120.0 GTexel/s.

The FP32 throughput shows a clear difference. The B370 is rated at 6.144 TFLOPS, while the B390 is rated at 7.680 TFLOPS. FP16 performance doubles those figures using a 2:1 ratio: 12.29 TFLOPS for the B370 and 15.36 TFLOPS for the B390. These raw throughput numbers align with the benchmark delta of 20.1 percent. The B390’s FP32 figure is exactly 25 percent higher than the B370’s, which is consistent with the 1536 versus 1280 shading unit count.

Power consumption is a major differentiator. The B370 has a TDP of 25 W, while the B390 has a TDP of 80 W. That is a threefold increase. Both are IGP parts with no power connectors and no suggested PSU, and both use an IGP bus interface. Display outputs are portable device dependent for both.

Architecture Differences

Both GPUs are built on the same Xe3-LPG architecture and use the same Panther Lake chip. The process node is identical at 3 nm, and the foundry is Intel for both. Transistor count and die size are listed as unknown for both, so no comparison can be made there. The generation is also identical: Arc Graphics-M (Panther Lake).

The architectural difference lies in the scale of execution resources. The B390 has more shading units, more texture mapping units, more ROPs, and more ray tracing cores. Those additional units are the reason the B390’s pixel rate, texture rate, and FP32 throughput are all higher. The B390 also has a higher boost clock, which compounds the resource advantage. The architecture itself does not change between the two; the implementation is simply larger and clocked slightly higher in the B390.

Both parts support the same API set. DirectX 12 Ultimate (12_2) is available on both, as are OpenGL 4.6 and Vulkan 1.4. There are no tensor cores listed for either part, and no dedicated memory interface. The memory subsystem is entirely system shared, which means performance will depend on the host platform’s memory bandwidth. That is a shared constraint, not a differentiator.

The TDP difference is the most notable architectural consequence. The B390 needs 80 W to operate at its larger configuration and higher clocks, while the B370 runs at 25 W. That difference matters for thermal design and power budgets. Both are integrated parts, so the platform must supply the power. The B390’s higher power draw enables its higher throughput, and the B370’s lower power draw explains its more modest results.

Where Each One Wins

The recorded data shows no benchmark where the Intel Arc B370 wins. In the single available test, the B390 wins outright. That makes the use-case split straightforward. The B390 is the choice for workloads that rely on the measured DirectX 12 performance. Its higher scores in 3DMark Steel Nomad translate to better raw rendering capability. The B390 also leads its nearest rivals, all NVIDIA parts, by a consistent margin. That suggests it holds up well against lower-end discrete GPUs from the previous generation.

The B370’s position is different. It does not win against the B390, and it does not clearly win against its nearest rivals either. It trails three AMD parts by fractions of a percent and leads one by a slightly larger fraction. The B370’s strength is not performance; it is efficiency. At 25 W, it delivers a score of 1184, while the B390 needs 80 W to deliver 1482. The B370 offers more than 80 percent of the B390’s performance at less than one third of the power draw. For systems with strict thermal or power limits, the B370 has a clear advantage in efficiency, even if it loses in absolute throughput.

The B390’s 9th percentile placement, while low, is still higher than the B370’s 5th percentile. In the database distribution, the B390 is closer to mid-range parts than the B370. Its nearest rivals are all NVIDIA parts, and it beats each of them. That gives the B390 a use case as the better integrated GPU for demanding tasks. The B370’s nearest rivals are AMD parts from an older generation, and it is essentially tied with them. That places the B370 in a cluster where the choice between parts comes down to platform compatibility rather than performance.

The FP32 and FP16 numbers reinforce the split. The B390’s 7.680 TFLOPS FP32 throughput is 25 percent higher than the B370’s 6.144 TFLOPS. The B390’s texture rate of 120.0 GTexel/s is 25 percent higher than the B370’s 96.00 GTexel/s. The pixel rate difference is larger in percentage terms: 60.00 GPixel/s versus 48.00 GPixel/s is a 25 percent increase as well. The B390 simply has more of everything that matters for rendering throughput.

Power efficiency is the B370’s only measurable advantage. At 25 W, it uses 55 W less than the B390. That is a significant difference for integrated graphics, where the surrounding platform may have limited cooling. The B370 also has a lower boost clock, which reduces peak power draw. For a user who prioritizes battery life or quiet operation over raw frame rates, the B370 is the logical pick. For a user who wants the highest recorded score in the database, the B390 is the only choice.

The delta of 20.1 percent between the two parts is large enough to be meaningful. The B390 is not marginally faster; it is one fifth faster in the measured test. The B370 is not without purpose, but its purpose is efficiency, not performance. The data supports this split cleanly.