AMD Steam Machine GPU vs Intel Arc B390 Comparison

FAQ

Q: What is the core architectural difference between the AMD Steam Machine GPU and the Intel Arc B390?

A: The AMD Steam Machine GPU uses the RDNA 3.0 architecture on a 6 nm TSMC process with the Navi 33 chip, while the Intel Arc B390 uses the Xe3-LPG architecture on a 3 nm Intel process with the Panther Lake chip.

Q: How do the two compare in terms of raw compute performance?

A: The AMD Steam Machine GPU delivers 17.56 TFLOPS of FP32 compute, while the Intel Arc B390 delivers 7.680 TFLOPS of FP32. This gives the AMD part a 2.29x advantage in single-precision throughput.

Q: What is the memory configuration difference?

A: The AMD Steam Machine GPU has 8 GB of dedicated GDDR6 memory on a 128-bit bus with 288.0 GB/s bandwidth. The Intel Arc B390 uses System Shared memory, with bus width and bandwidth described as System Dependent.

Q: How do their power requirements differ?

A: The AMD Steam Machine GPU has a TDP of 110 W and requires no power connectors. The Intel Arc B390 has a TDP of 80 W and also uses no power connectors, as it is an integrated graphics processor (IGP).

Q: What benchmark data exists for the Intel Arc B390?

A: The Intel Arc B390 has one recorded benchmark: a 3DMark Steel Nomad DX12 score of 1482. This places it at the 9th percentile of all GPUs, with its nearest rival being the NVIDIA GeForce GT 520MX at 1463 points, a 1.3% difference.

Q: Why is there no benchmark data for the AMD Steam Machine GPU?

A: The recorded database shows no benchmark entries for the AMD Steam Machine GPU, with an average benchmark score of 0 and a 50th percentile ranking. The head-to-head benchmark comparison also contains no entries, and neither GPU has a recorded win in the comparison.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The AMD Steam Machine GPU is a discrete console-class part built on the RDNA 3.0 architecture, using the Navi 33 chip under the codename "Hotpink Bonefish". It belongs to the Console GPU (Valve) generation, indicating a purpose-built part for a specific hardware platform. The Intel Arc B390, by contrast, uses the Xe3-LPG architecture with the Panther Lake chip and is classified as Arc Graphics-M, an integrated graphics solution for mobile platforms.

Manufacturing processes separate them clearly. The AMD part uses a 6 nm process from TSMC with 13,300 million transistors on a 204 mm² die, yielding a transistor density of 65.2 million per mm². The Intel part uses a 3 nm process from Intel, though its transistor count and die size are recorded as unknown in the database. The process node difference suggests Intel is using a more advanced manufacturing technology, while AMD's larger, mature 6 nm design carries more transistors overall.

Compute resources differ substantially. The AMD Steam Machine GPU has 1792 shading units, 112 texture mapping units, 64 ROPs, and 28 ray tracing cores. The Intel Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores. AMD leads in every category, with 2.33x more TMUs and 2.67x more ROPs. The shading unit gap is narrower at 1.17x, suggesting Intel's architecture attempts to compensate with higher clock speeds.

Clock behavior reveals another difference. The AMD part has a base clock of 1720 MHz, a game clock of 2250 MHz, and a boost clock of 2450 MHz. The Intel part has a very low base clock of 300 MHz but a boost clock of 2500 MHz, slightly higher than AMD's boost. This large base-to-boost spread on the Intel part is characteristic of integrated graphics that scale dynamically based on thermal and power headroom.

FP16 throughput also differs in ratio. The AMD Steam Machine GPU achieves 17.56 TFLOPS of FP16 at a 1:1 ratio with FP32, meaning it does not accelerate half-precision separately. The Intel Arc B390 delivers 15.36 TFLOPS of FP16 at a 2:1 ratio, meaning it can process two FP16 operations per FP32 operation. Intel's advantage in FP16 relative to its own FP32 suggests better support for workloads that benefit from half-precision math.

The Intel part's integrated nature shows in its bus interface, recorded as IGP, and its slot width of IGP. The AMD part has physical dimensions of 156 mm in length, 152 mm in height, and 162 mm in width, confirming it is a discrete card. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API feature parity is complete despite the architectural differences.

The Verdict

The recorded data points to a clear performance hierarchy, but the evidence is one-sided. The AMD Steam Machine GPU has no benchmark scores in the database, while the Intel Arc B390 has exactly one: a 3DMark Steel Nomad DX12 score of 1482, placing it at the 9th percentile of all GPUs. That percentile ranking places it near the bottom of the performance distribution, with its nearest rivals being the NVIDIA GeForce GT 520MX (1463, 1.3% behind), NVIDIA GeForce 800M (1460, 1.5% behind), NVIDIA GeForce GT 625 OEM (1446, 2.5% behind), and NVIDIA GeForce GT 710 (1443, 2.7% behind).

The AMD Steam Machine GPU's 50th percentile ranking, despite having no recorded benchmarks, suggests the database assigns it a mid-range position based on its specifications. The hardware differences support this interpretation. The AMD part has 2.29x more FP32 throughput, 2.33x more TMUs, 2.67x more ROPs, 2.33x more ray tracing cores, and dedicated GDDR6 memory with 288.0 GB/s of bandwidth. The Intel part relies on system shared memory with bandwidth described as System Dependent, which typically imposes a performance penalty compared to dedicated memory.

Who should pick which depends entirely on the intended role. The Intel Arc B390 is an integrated part for portable devices, with an 80 W TDP and no power connectors, designed for systems where physical space and power draw are constrained. Its 3DMark Steel Nomad score of 1482 indicates it can handle basic 3D tasks but sits at the 9th percentile, meaning it outperforms only a small fraction of the GPU population. The AMD Steam Machine GPU is a discrete part with a 110 W TDP, also requiring no power connectors, and appears designed to deliver substantially higher throughput in a fixed console platform.

The data does not support a head-to-head benchmark comparison because the database contains no entries for such a matchup. The wins counter shows 0 for both parts. What the data does show is a wide specification gap favoring AMD in every compute category, while Intel leads only in boost clock (2500 MHz vs 2450 MHz), FP16 efficiency ratio (2:1 vs 1:1), and process node advancement (3 nm vs 6 nm). For users who need a dedicated GPU with high throughput and dedicated memory, the AMD Steam Machine GPU's specifications position it as the stronger part. For users who need an integrated solution in a portable device, the Intel Arc B390 is the only viable option of the two, as it is the only one classified as an IGP.

Specification Differences

Process Node: AMD uses 6 nm from TSMC. Intel uses 3 nm from Intel.

Transistors: AMD has 13,300 million. Intel's count is unknown.

Die Size: AMD measures 204 mm². Intel's size is unknown.

Base Clock: AMD runs at 1720 MHz. Intel runs at 300 MHz.

Boost Clock: AMD boosts to 2450 MHz. Intel boosts to 2500 MHz.

Memory Size: AMD has 8 GB GDDR6. Intel uses System Shared.

Memory Type: AMD uses GDDR6. Intel uses System Shared.

Memory Bus Width: AMD has 128 bit. Intel uses System Shared.

Memory Bandwidth: AMD has 288.0 GB/s. Intel uses System Dependent.

Shading Units: AMD has 1792. Intel has 1536.

TMUs: AMD has 112. Intel has 48.

ROPs: AMD has 64. Intel has 24.

Ray Tracing Cores: AMD has 28. Intel has 12.

Pixel Rate: AMD delivers 156.8 GPixel/s. Intel delivers 60.00 GPixel/s.

Texture Rate: AMD delivers 274.4 GTexel/s. Intel delivers 120.0 GTexel/s.

FP32: AMD delivers 17.56 TFLOPS. Intel delivers 7.680 TFLOPS.

FP16: AMD delivers 17.56 TFLOPS (1:1). Intel delivers 15.36 TFLOPS (2:1).

TDP: AMD is rated at 110 W. Intel is rated at 80 W.

Slot Width: AMD is not specified. Intel is IGP.

Bus Interface: AMD is not specified. Intel is IGP.

Display Outputs: AMD has 1x HDMI 2.1a and 1x DisplayPort 2.1. Intel is Portable Device Dependent.

Dimensions: AMD measures 156 mm by 152 mm by 162 mm. Intel has no recorded dimensions.

Release Date: AMD releases on 2026-06-28. Intel releases on 2026-01-26.

Head-to-Head Benchmarks

The database records no head-to-head benchmark entries between these two GPUs, and the wins counter shows 0 for both. The only direct performance evidence available is the Intel Arc B390's single 3DMark Steel Nomad DX12 score of 1482. That score places the Intel part at the 9th percentile of all GPUs, a position that puts it in the company of very old or very low-end parts. Its nearest rival, the NVIDIA GeForce GT 520MX, scores 1463, just 1.3% behind. The NVIDIA GeForce 800M follows at 1460 (1.5% behind), the NVIDIA GeForce GT 625 OEM at 1446 (2.5% behind), and the NVIDIA GeForce GT 710 at 1443 (2.7% behind). The tight clustering of these scores, all within 2.7% of each other, indicates the Intel Arc B390 operates in a performance band where even small differences matter.

The AMD Steam Machine GPU has no recorded benchmark scores at all. Its average benchmark score is 0, and its percentile ranking is 50. That 50th percentile is a database classification rather than a measured result, but it suggests the part is expected to perform in the middle of the GPU distribution. The specification data supports this expectation. The AMD part has 2.29x the FP32 throughput of the Intel part, 2.61x the pixel rate, 2.29x the texture rate, and dedicated memory bandwidth of 288.0 GB/s versus Intel's system-dependent figure. These are not marginal differences; they suggest the AMD part would outperform the Intel part by a wide margin in most measurable workloads.

The absence of head-to-head data means this comparison relies on the single available benchmark for Intel and the specification deltas for AMD. The Intel part's 1482 score at the 9th percentile is the only concrete performance number in the entire comparison. The AMD part's specifications, particularly its 17.56 TFLOPS FP32 and 288.0 GB/s memory bandwidth, place it in a different performance class entirely. The data implies a lopsided matchup, but the lack of a direct benchmark prevents a quantified statement of the margin.

Where Each One Wins

The AMD Steam Machine GPU wins in every recorded compute category. Its FP32 throughput of 17.56 TFLOPS is 2.29x the Intel part's 7.680 TFLOPS, which matters for general-purpose compute and traditional rasterization workloads. Its pixel rate of 156.8 GPixel/s is 2.61x Intel's 60.00 GPixel/s, directly affecting fill-rate-bound scenarios such as high-resolution rendering and heavy post-processing effects. Its texture rate of 274.4 GTexel/s is 2.29x Intel's 120.0 GTexel/s, which benefits texture-heavy scenes and anisotropic filtering. Its 28 ray tracing cores versus Intel's 12 suggest a 2.33x advantage in ray-traced workloads, assuming similar per-core efficiency. Its dedicated 8 GB GDDR6 memory with 288.0 GB/s bandwidth avoids the performance variability of Intel's system-shared memory, whose bandwidth is recorded as System Dependent.

The Intel Arc B390 wins in specific efficiency and integration categories. Its 3 nm process from Intel is more advanced than AMD's 6 nm from TSMC, which contributes to its lower 80 W TDP versus AMD's 110 W. Its boost clock of 2500 MHz edges out AMD's 2450 MHz, though its base clock of 300 MHz is dramatically lower. Its FP16 throughput of 15.36 TFLOPS at a 2:1 ratio means it processes half-precision math more efficiently relative to its FP32 capability, a feature that could benefit certain compute workloads that utilize FP16. Its classification as an IGP with no power connectors and no physical dimensions makes it suitable for portable devices where a discrete card cannot fit. Its earlier release date of 2026-01-26 versus AMD's 2026-06-28 means it reached the market five months earlier.

The use-case split follows the hardware roles. The AMD Steam Machine GPU, with its discrete form factor, dedicated memory, and higher throughput, is positioned for gaming or compute tasks that demand sustained performance. The Intel Arc B390, as an integrated part with system shared memory and an 80 W TDP, is positioned for portable devices where power efficiency and physical integration take priority over raw performance. Its single benchmark score of 1482 at the 9th percentile confirms it targets modest 3D workloads rather than high-end gaming. The data shows two parts designed for different segments, with AMD holding the performance advantage and Intel holding the efficiency and integration advantage.