Intel Arc Pro B390 vs NVIDIA RTX PRO 6000 Blackwell Server Comparison

Head-to-Head Benchmarks

The recorded data contains only one benchmark entry, and it belongs exclusively to the NVIDIA RTX PRO 6000 Blackwell Server. In the 3DMark Steel Nomad DX12 test, the NVIDIA card posts a score of 5996. The Intel Arc Pro B390 has no benchmark scores recorded in the database, so no direct numerical comparison between the two is possible from the available measurements. The Intel part carries an average benchmark score of 0, while the NVIDIA part averages 5996.

The NVIDIA RTX PRO 6000 Blackwell Server's nearest rivals in the database are not the Intel Arc Pro B390. Instead, the NVIDIA card sits within 0.2% of several much older or lower-tier parts. The GeForce GTX 770M scores 6000, which places it 0.1% above the RTX PRO 6000 Blackwell Server. The Radeon RX 6400 also scores 6001, again 0.1% higher. On the flip side, the FirePro W4100 scores 5987, which is 0.2% lower, and the Quadro K4000M scores 5986, also 0.2% lower. These delta values indicate that the RTX PRO 6000 Blackwell Server's single recorded benchmark result places it in a cluster of widely divergent hardware, from mobile GPUs to low-end desktop parts, which is unusual for a server-class accelerator.

The Intel Arc Pro B390 has no nearest rivals listed and no benchmark entries, meaning the database holds no performance evidence for it. The wins tally reflects this asymmetry: the NVIDIA card wins zero direct head-to-head matchups because no head-to-head benchmarks exist, and the Intel card also records zero wins. The only measurable performance signal in this comparison is the NVIDIA card's Steel Nomad score of 5996, and even that score is nearly identical to the four rivals listed around it, with deltas of just 0.1% or 0.2% in either direction. The data does not support any claim of dominance for either product.

The Verdict

From the recorded data, the NVIDIA RTX PRO 6000 Blackwell Server is the only product with a verified benchmark result. Its 5996 score in 3DMark Steel Nomad DX12 is the sole quantitative performance data point in this comparison. The Intel Arc Pro B390 has no scores, no rivals, and no measurable output, so the database cannot substantiate any performance claim for it. A buyer seeking a server GPU with a confirmed benchmark footprint would have to select the NVIDIA part based on the available evidence.

The percentile ranking further complicates the picture. The NVIDIA RTX PRO 6000 Blackwell Server sits at the 34th percentile among all GPUs in the database, meaning 66% of recorded parts score higher. Its nearest rivals, as noted, are the GTX 770M, RX 6400, FirePro W4100, and Quadro K4000M, all of which score within 0.2% of it. This suggests that the single Steel Nomad result does not reflect the NVIDIA card's actual compute capability, given its massive shading unit count and memory bandwidth, but the data cannot be ignored. The Intel Arc Pro B390, at the 50th percentile with an average score of zero, is statistically undefined in this context.

For a server deployment, the data points to the NVIDIA RTX PRO 6000 Blackwell Server as the only option with a recorded performance metric. The Intel part, being an integrated graphics processor with no benchmark entries, cannot be evaluated on the same terms. The specification sheet for Intel shows a 3 nm process, 80 W TDP, and system-shared memory, while NVIDIA offers a 600 W dual-slot card with 96 GB of GDDR7 and a 512-bit bus. The choice, strictly by the numbers, favors NVIDIA for any workload where the Steel Nomad score or the raw memory and compute specs matter.

Architecture Differences

The two GPUs diverge at the architectural level. Intel's Arc Pro B390 uses the Panther Lake chip with Xe3-LPG architecture, built on a 3 nm process at Intel's own foundry. It belongs to the Arc Graphics-WM (Panther Lake) generation. NVIDIA's RTX PRO 6000 Blackwell Server uses the GB202 chip with Blackwell 2.0 architecture, produced on a 5 nm process at TSMC. The node difference is significant: Intel's 3 nm process is newer and denser on paper, while NVIDIA's 5 nm process is older but paired with a much larger physical implementation.

The transistor counts cannot be compared directly. Intel lists its transistor count as unknown, while NVIDIA specifies 92,200 million transistors on a 750 mm² die, yielding a transistor density of 122.9 million per square millimeter. The Intel die size is also unknown, so no density calculation is possible. The NVIDIA die is physically massive, which aligns with its 24,064 shading units, 752 texture mapping units, and 192 raster operation units. Intel counters with 1,536 shading units, 48 TMUs, and 24 ROPs, a fraction of NVIDIA's resource pool.

Ray tracing and tensor cores further separate the architectures. Intel includes 12 ray tracing cores but lists no tensor cores. NVIDIA includes 188 RT cores and 752 tensor cores, the latter being essential for AI and deep learning workloads. The API support is identical across both parts: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The compute rates tell the story: Intel delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 (2:1 ratio), while NVIDIA delivers 126.0 TFLOPS FP32 and 126.0 TFLOPS FP16 (1:1 ratio). NVIDIA's FP16 throughput matches its FP32, whereas Intel halves its FP16 rate, indicating different design priorities.

Specification Differences

The specification sheets show stark contrasts. Intel's Arc Pro B390 is an integrated graphics processor (IGP) with no slot width, no power connectors, and a bus interface of IGP. NVIDIA's RTX PRO 6000 Blackwell Server is a dual-slot card measuring 267 mm in length, 111 mm in height, and 40 mm in width, requiring a single 16-pin power connector and a suggested 1000 W power supply. The TDP difference is 80 W for Intel versus 600 W for NVIDIA, an 8x gap in power draw.

Memory configurations are entirely different. Intel uses system-shared memory with a system-dependent bandwidth, meaning the GPU has no dedicated VRAM. NVIDIA has 96 GB of GDDR7 on a 512-bit bus with 1.79 TB/s of bandwidth, clocked at 1750 MHz (28 Gbps effective). The clock speeds differ as well: Intel runs at a 300 MHz base and 2500 MHz boost, while NVIDIA runs at 1590 MHz base and 2617 MHz boost. NVIDIA's boost clock is only 117 MHz higher than Intel's, but its base clock is 1290 MHz higher, reflecting the much larger and more power-hungry design.

Display outputs also differ. Intel's outputs are portable device dependent, reflecting its IGP nature. NVIDIA provides 4x DisplayPort 2.1b. The pixel and texture rates scale accordingly: Intel achieves 60.00 GPixel/s and 120.0 GTexel/s, while NVIDIA achieves 502.5 GPixel/s and 1,968.0 GTexel/s. The release dates show NVIDIA launched on 2025-03-17, while Intel's part arrived on 2026-01-26, nearly ten months later. NVIDIA's predecessor is Server Hopper and its successor is Server Rubin, while Intel's predecessor is HD Graphics-WM with no successor listed. Both are marked as active in production status.

FAQ

Q: Which GPU has a higher benchmark score in the database?

A: The NVIDIA RTX PRO 6000 Blackwell Server has a recorded 3DMark Steel Nomad DX12 score of 5996. The Intel Arc Pro B390 has no benchmark scores recorded, with an average score of 0.

Q: How does the NVIDIA RTX PRO 6000 Blackwell Server compare to its nearest rivals?

A: Its nearest rivals are the GeForce GTX 770M (6000, 0.1% higher), Radeon RX 6400 (6001, 0.1% higher), FirePro W4100 (5987, 0.2% lower), and Quadro K4000M (5986, 0.2% lower). All four are within 0.2% of the NVIDIA card's 5996 score.

Q: What is the memory configuration of each GPU?

A: The Intel Arc Pro B390 uses system-shared memory with a system-dependent bandwidth and no dedicated VRAM. The NVIDIA RTX PRO 6000 Blackwell Server has 96 GB of GDDR7 on a 512-bit bus with 1.79 TB/s bandwidth.

Q: Which GPU has more shading units?

A: The NVIDIA RTX PRO 6000 Blackwell Server has 24,064 shading units, compared to 1,536 for the Intel Arc Pro B390. NVIDIA also has 752 TMUs and 192 ROPs, versus 48 TMUs and 24 ROPs for Intel.

Q: What are the power requirements for each GPU?

A: The Intel Arc Pro B390 has an 80 W TDP, no power connectors, and is an integrated graphics processor. The NVIDIA RTX PRO 6000 Blackwell Server has a 600 W TDP, requires a single 16-pin power connector, and a suggested 1000 W power supply.

Q: What process nodes do the two GPUs use?

A: The Intel Arc Pro B390 uses a 3 nm process at Intel's foundry. The NVIDIA RTX PRO 6000 Blackwell Server uses a 5 nm process at TSMC.