AMD Radeon RX 7800M vs Intel Arc Pro B390 Comparison

Where Each One Wins

The recorded data presents an asymmetric comparison. The AMD Radeon RX 7800M has a full suite of ten benchmark scores across multiple test suites, while the Intel Arc Pro B390 has no recorded benchmark scores in the database. Consequently, the AMD part claims wins in every measured category by default, but the magnitude of those wins and the architectural context are what define the practical split.

The RX 7800M’s strongest recorded results come from the Geekbench compute tests, with 126,668 in Vulkan and 120,778 in OpenCL. These are its highest raw scores and indicate strong general compute throughput. The PassMark G3D score of 17,613 is the primary gaming-oriented metric, while the PassMark GPU Compute score of 9,342 covers non-graphics workloads. The 3DMark Steel Nomad DX12 result, 2,994, is the only modern DirectX 12 gaming benchmark recorded, and it is the lowest of the three major performance indicators relative to the others.

The Intel Arc Pro B390, with no benchmarks in the database, cannot claim a single measured win. Its percentile ranking of 50 places it at the median of all GPUs in the database, whereas the RX 7800M sits at the 73rd percentile. This gap alone suggests that the AMD part is positioned well above the median performer, while the Intel part is exactly average.

The use-case split, therefore, is not about which card wins which test. It is about which workloads each part is designed for. The RX 7800M carries a 180 W TDP with a dedicated 12 GB GDDR6 memory subsystem and a 192-bit bus, making it suitable for sustained, high-bandwidth tasks such as gaming and compute rendering. The Arc Pro B390 operates at 80 W with system-shared memory and an integrated form factor, indicating it targets low-power, portable, or embedded scenarios where the host system provides memory bandwidth.

Architecture Differences

The two parts diverge fundamentally in silicon design. The AMD Radeon RX 7800M uses the Navi 32 chip built on RDNA 3.0 architecture, fabricated on a 5 nm TSMC process. It packs 28,100 million transistors into a 346 mm² die, yielding a transistor density of 81.2 million transistors per square millimeter. The Intel Arc Pro B390 uses the Panther Lake chip with Xe3-LPG architecture on a 3 nm Intel process. Intel does not disclose transistor count or die size in the database, so no direct density comparison is possible.

The execution resource disparity is substantial. The RX 7800M contains 3,840 shading units, 240 texture mapping units, 96 ROPs, and 60 ray tracing cores. The Arc Pro B390 has 1,536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores. The AMD part has 2.5 times the shading units, 5 times the TMUs, 4 times the ROPs, and 5 times the ray tracing cores. These ratios directly explain the compute throughput gap.

Clock behavior differs sharply. The RX 7800M runs a 1,295 MHz base clock with a 2,335 MHz boost and a 2,145 MHz game clock. The Arc Pro B390 has a 300 MHz base clock and a 2,500 MHz boost. The Intel part boosts higher, but its base clock is far lower, and its much smaller execution array means the higher boost cannot compensate for the resource deficit.

Memory architecture separates the two completely. The RX 7800M has 12 GB of dedicated GDDR6 memory on a 192-bit bus, delivering 432.0 GB/s of bandwidth. The Arc Pro B390 uses system-shared memory with system-dependent bandwidth, meaning its performance scales with the host platform's memory configuration rather than a fixed specification. The AMD memory clock is 2,250 MHz (18 Gbps effective), while the Intel memory clock is listed simply as "System Shared."

Both parts support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Neither has a dedicated tensor core count listed. The FP32 throughput is 35.87 TFLOPS for the RX 7800M versus 7.680 TFLOPS for the Arc Pro B390, a 4.7-fold difference. The FP16 figures are 35.87 TFLOPS (1:1 ratio) for AMD and 15.36 TFLOPS (2:1 ratio) for Intel. The AMD part maintains full FP16 rate at 1:1, while Intel halves its rate at 2:1.

The bus interface also differs. The RX 7800M uses PCIe 4.0 x16, while the Arc Pro B390 is integrated with an IGP bus interface. Both are listed as IGP slot width with no dedicated power connectors. The RX 7800M carries a 180 W TDP, the Arc Pro B390 an 80 W TDP.

The Verdict

The data supports a clear verdict for raw performance: the AMD Radeon RX 7800M dominates every measurable benchmark category. Its average benchmark score of 27,883 places it at the 73rd percentile of all GPUs, while the Intel Arc Pro B390 has no recorded average score and sits at the 50th percentile. The RX 7800M's nearest rival, the AMD Radeon Pro Vega 20, scores 27,839, a 0.2% margin that is essentially a statistical tie. The RX 7800M also edges past the Radeon Pro W5500X (27,973, -0.3%), the NVIDIA GeForce GTX 980 Ti (28,020, -0.5%), and the AMD FirePro S7150 (28,117, -0.8%). These deltas are all within a single percentage point, indicating that the RX 7800M competes at the level of a high-end last-generation desktop GPU.

The Arc Pro B390 has no rivals listed, no benchmarks recorded, and an average score of zero. The database cannot validate any performance claim for it. Its 50th percentile ranking is the only positional data, and that is a median placement, not a sign of competitive strength.

The practical verdict: the RX 7800M is the choice for any workload where measured performance matters. Its 12 GB dedicated memory, 432.0 GB/s bandwidth, and 35.87 TFLOPS FP32 throughput support gaming, compute, and rendering tasks. The Arc Pro B390, with system-shared memory and a 7.680 TFLOPS FP32 ceiling, is a low-power integrated solution for portable devices where the host system supplies memory and where 80 W power draw is acceptable. The database shows no scenario where the Intel part outperforms the AMD part in a recorded test.

FAQ

Q: Which GPU has the higher FP32 compute throughput?

A: The AMD Radeon RX 7800M delivers 35.87 TFLOPS FP32, while the Intel Arc Pro B390 delivers 7.680 TFLOPS, a 4.7-fold advantage for AMD.

Q: How much memory bandwidth does the RX 7800M provide?

A: The RX 7800M has 12 GB of GDDR6 memory on a 192-bit bus with 432.0 GB/s bandwidth. The Arc Pro B390 uses system-shared memory with system-dependent bandwidth.

Q: What are the transistor and die size figures for each GPU?

A: The RX 7800M has 28,100 million transistors on a 346 mm² die with 81.2 million transistors per square millimeter. The Arc Pro B390 has unknown transistor count and die size in the database.

Q: Which GPU has a higher boost clock?

A: The Intel Arc Pro B390 boosts to 2,500 MHz, higher than the RX 7800M's 2,335 MHz boost. The RX 7800M has a 1,295 MHz base clock versus the Intel part's 300 MHz base.

Q: How do the ray tracing core counts compare?

A: The RX 7800M has 60 ray tracing cores, while the Arc Pro B390 has 12, a 5-to-1 ratio in favor of AMD.

Q: What percentile rankings do the two GPUs hold?

A: The RX 7800M is at the 73rd percentile of all GPUs in the database, while the Arc Pro B390 is at the 50th percentile.

Head-to-Head Benchmarks

The head-to-head benchmark list is empty in the database, meaning no direct paired test results were recorded. The comparison must be built from the RX 7800M's individual scores against the Arc Pro B390's absence of scores. That absence is itself the most significant data point.

The RX 7800M's 3DMark Steel Nomad DX12 score of 2,994 is its only modern gaming benchmark. There is no counterpart for the Arc Pro B390. In Geekbench OpenCL, the RX 7800M scores 120,778, and in Geekbench Vulkan it scores 126,668. These two results bracket the compute capability: Vulkan is slightly ahead of OpenCL by 4.9% (126,668 versus 120,778, a difference of 5,890 points). The PassMark G3D score of 17,613 is the gaming-oriented metric, while PassMark GPU Compute at 9,342 is roughly half of that, indicating that the card is stronger in graphics rendering than in pure compute under PassMark's methodology.

The PassMark DirectX scores show an interesting pattern. DirectX 11 scores 176, DirectX 9 scores 174, DirectX 10 scores 99, and DirectX 12 scores 89. The DirectX 11 and 9 results are nearly twice the DirectX 10 and 12 results. This suggests the RX 7800M's driver or hardware optimization favors legacy DirectX paths in PassMark's specific workload, or that the DirectX 12 test is more demanding. The PassMark G2D score of 892 is the only 2D metric, and it is far lower than the G3D score, which is expected for a dedicated GPU.

The average benchmark score of 27,883 is a weighted composite of these results. Its nearest rival, the AMD Radeon Pro Vega 20, is 0.2% behind at 27,839. The Radeon Pro W5500X is 0.3% ahead at 27,973, the GeForce GTX 980 Ti is 0.5% ahead at 28,020, and the FirePro S7150 is 0.8% ahead at 28,117. The RX 7800M sits in a tight cluster where all four rivals are within one percentage point of its average. This cluster represents a performance tier where the RX 7800M is competitive but not dominant, and the differences are within measurement noise.

The Arc Pro B390 has no such cluster. Its average benchmark score is zero, its nearest rivals list is empty, and its percentile is exactly 50. The data cannot place it in any performance tier. The only absolute statements available are its architectural specifications: 1,536 shading units, 48 TMUs, 24 ROPs, 12 ray tracing cores, 7.680 TFLOPS FP32, and 80 W TDP. Every one of those figures is lower than the corresponding RX 7800M figure, except boost clock (2,500 MHz versus 2,335 MHz) and process node (3 nm versus 5 nm). Those two advantages do not translate into any recorded benchmark win.

The largest measurable gap in the head-to-head comparison is the missing data itself. The RX 7800M has ten recorded benchmarks; the Arc Pro B390 has zero. The database's 73rd versus 50th percentile gap is the only direct positional comparison available, and it favors AMD by 23 percentile points. The RX 7800M's performance tier, defined by its four nearest rivals, is a known quantity. The Arc Pro B390's tier is undefined. Any buyer or system designer relying on the database must treat the Intel part as unproven in every measured workload, while the AMD part has a complete record of results.