Intel Arc B370 vs NVIDIA RTX PRO 6000 Blackwell Server Comparison

Intel Arc B370 and NVIDIA RTX PRO 6000 Blackwell Server occupy opposite ends of the GPU spectrum, one a low-power integrated part for portable devices, the other a dual-slot server accelerator. The recorded data, based on the 3DMark Steel Nomad DX12 test, shows a decisive performance gap between them.

Head-to-Head Benchmarks

The only direct comparison available in the database is the 3DMark Steel Nomad DX12 benchmark. In this test, the Intel Arc B370 scores 1184 points, while the NVIDIA RTX PRO 6000 Blackwell Server scores 5996 points. This gives the NVIDIA part a delta of -80.3% from the perspective of the Intel chip, meaning the Intel Arc B370 delivers only about one-fifth of the NVIDIA card’s score. The NVIDIA RTX PRO 6000 Blackwell Server is the clear winner in this head-to-head matchup.

The Intel Arc B370’s raw score of 1184 places it at the 5th percentile of all GPUs in the database. Its nearest rivals are all older or lower-tier parts: the ATI Mobility Radeon HD 5570 scores 1186, a delta of -0.2%; the ATI Radeon HD 5770 scores 1190, a delta of -0.5%; the AMD Radeon HD 7650M scores 1192, a delta of -0.7%; and the AMD FirePro M2000 scores 1168, a delta of 1.4%. These deltas are all within 1.5% of the Arc B370, indicating that the Intel part performs in the same class as these legacy mobile and entry-level workstation cards. The Arc B370 is effectively competitive with a decade-old Radeon HD 5770, which is a notable result for a modern 3nm integrated GPU.

The NVIDIA RTX PRO 6000 Blackwell Server, by contrast, scores 5996 and sits at the 34th percentile of all GPUs. Its nearest rivals include the NVIDIA GeForce GTX 770M at 6000, a delta of -0.1%; the AMD Radeon RX 6400 at 6001, a delta of -0.1%; the AMD FirePro W4100 at 5987, a delta of 0.2%; and the NVIDIA Quadro K4000M at 5986, a delta of 0.2%. The fact that this server-class Blackwell accelerator scores within 0.2% of a Radeon RX 6400, a low-profile desktop card, and the GeForce GTX 770M, a mobile part from years ago, suggests that the Steel Nomad workload may not be well suited to the NVIDIA card’s strengths, or that the benchmark’s scaling favors other architectures. Still, the absolute score is 5 times higher than the Intel Arc B370’s result.

The Verdict

The data does not support any scenario where the Intel Arc B370 outperforms the NVIDIA RTX PRO 6000 Blackwell Server. The NVIDIA card wins the sole recorded benchmark by a wide margin, and its 126.0 TFLOPS FP32 throughput versus the Intel part’s 6.144 TFLOPS confirms the scale of the gap. The RTX PRO 6000 Blackwell Server also has 24,064 shading units, 752 texture mapping units, 192 ROPs, 188 ray tracing cores, and 752 tensor cores, while the Arc B370 has 1,280 shading units, 40 TMUs, 20 ROPs, and 10 ray tracing cores. Every compute resource on the NVIDIA card is an order of magnitude larger.

The Intel Arc B370 is an integrated graphics processor for portable devices, with system-shared memory and no dedicated VRAM. It draws 25 W and uses no power connectors. The NVIDIA part is a dual-slot server card with a 600 W TDP, a 16-pin power connector, and a suggested PSU of 1000 W. These are not competing products in any practical sense; the database simply records both for comparative reference.

For workloads that stress raw graphics throughput, the NVIDIA RTX PRO 6000 Blackwell Server is the only viable option of the two. The Intel part can only be considered for the lowest-end portable or embedded use cases where its 25 W envelope and integrated nature are the primary constraints.

Where Each One Wins

The Intel Arc B370 wins exclusively on power efficiency and integration. Its 25 W TDP is the only recorded power figure for either card, and it requires no external power connector. It is an IGP, meaning it is built into the processor and shares system memory, which makes it suitable for compact, low-power portable devices. Its 300 MHz base clock and 2400 MHz boost clock operate within that 25 W envelope, and its display outputs are portable-device dependent. The Arc B370 also uses the 3 nm process node from Intel foundry, which is smaller than the NVIDIA card’s 5 nm TSMC node, but the database does not record transistor counts or die size for the Intel part, so a density comparison is not possible.

The NVIDIA RTX PRO 6000 Blackwell Server wins every performance-related category. Its 96 GB of GDDR7 memory on a 512-bit bus delivers 1.79 TB/s of bandwidth, compared to the Arc B370’s system-shared memory with system-dependent bandwidth. The NVIDIA card’s 1,968.0 GTexel/s texture rate and 502.5 GPixel/s pixel rate dwarf the Intel part’s 96.00 GTexel/s and 48.00 GPixel/s. The FP16 throughput also favors NVIDIA: 126.0 TFLOPS at 1:1 ratio versus 12.29 TFLOPS at a 2:1 ratio for Intel. The NVIDIA card uses a PCIe 5.0 x16 bus interface and provides four DisplayPort 2.1b outputs, while the Intel part uses an IGP bus interface and depends on the host device for display output.

The NVIDIA card is built on the GB202 chip with the Blackwell 2.0 architecture, uses 92,200 million transistors on a 750 mm² die, and has a transistor density of 122.9 million per square millimeter. Its predecessor is recorded as Server Hopper and its successor as Server Rubin. The Intel chip is Panther Lake with Xe3-LPG architecture, belongs to the Arc Graphics-M generation, and has no recorded predecessor or successor. The Intel part’s release date is January 2026, while the NVIDIA card’s release date is March 2025.

FAQ

Q: Which GPU scores higher in 3DMark Steel Nomad DX12?

A: The NVIDIA RTX PRO 6000 Blackwell Server scores 5996, versus 1184 for the Intel Arc B370, a delta of -80.3% for the Intel part.

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

A: The Arc B370 scores 1184, which is within 1.5% of the ATI Mobility Radeon HD 5570 (1186, -0.2%), the ATI Radeon HD 5770 (1190, -0.5%), the AMD Radeon HD 7650M (1192, -0.7%), and the AMD FirePro M2000 (1168, +1.4%). It sits at the 5th percentile of all GPUs.

Q: What are the memory specifications of the NVIDIA RTX PRO 6000 Blackwell Server?

A: It has 96 GB of GDDR7 memory on a 512-bit bus, with 1.79 TB/s bandwidth and a 1750 MHz memory clock at 28 Gbps effective.

Q: Does the Intel Arc B370 have dedicated VRAM?

A: No, it uses system-shared memory with system-dependent bandwidth, and its memory type and bus width are both listed as system shared.

Q: What is the TDP of each card?

A: The Intel Arc B370 has a 25 W TDP and uses no power connectors. The NVIDIA RTX PRO 6000 Blackwell Server has a 600 W TDP, uses one 16-pin power connector, and has a suggested PSU of 1000 W.

Q: Which card has more shading units?

A: The NVIDIA RTX PRO 6000 Blackwell Server has 24,064 shading units, while the Intel Arc B370 has 1,280.

Q: What architectures do the two GPUs use?

A: The Intel Arc B370 uses Xe3-LPG architecture on the Panther Lake chip, built on a 3 nm node at Intel. The NVIDIA RTX PRO 6000 Blackwell Server uses Blackwell 2.0 architecture on the GB202 chip, built on a 5 nm node at TSMC.

Architecture Differences

The two GPUs are built on fundamentally different architectures and process nodes. The Intel Arc B370 uses the Xe3-LPG architecture, which is the low-power graphics variant of Intel’s Xe3 family, implemented on the Panther Lake chip. It is fabricated on a 3 nm process at Intel’s foundry. The NVIDIA RTX PRO 6000 Blackwell Server uses the Blackwell 2.0 architecture on the GB202 chip, fabricated on a 5 nm process at TSMC. The Intel chip’s transistor count and die size are not recorded in the database, while the NVIDIA chip uses 92,200 million transistors on a 750 mm² die, giving a transistor density of 122.9 million per square millimeter.

The compute resources differ by roughly an order of magnitude. The Intel Arc B370 has 1,280 shading units, 40 TMUs, 20 ROPs, and 10 ray tracing cores. The NVIDIA RTX PRO 6000 Blackwell Server has 24,064 shading units, 752 TMUs, 192 ROPs, 188 ray tracing cores, and 752 tensor cores. The NVIDIA card also has dedicated tensor cores, while the Intel part has no recorded tensor core count. Clock speeds show a similar pattern: the Intel part runs at a 300 MHz base and 2400 MHz boost, while the NVIDIA card runs at 1590 MHz base and 2617 MHz boost.

Memory architecture is another major divergence. The Intel Arc B370 uses system-shared memory with no dedicated VRAM, making its bandwidth system-dependent. The NVIDIA card has 96 GB of GDDR7 on a 512-bit interface with 1.79 TB/s bandwidth. The NVIDIA card’s pixel rate is 502.5 GPixel/s and its texture rate is 1,968.0 GTexel/s, versus 48.00 GPixel/s and 96.00 GTexel/s for the Intel part. FP32 compute is 126.0 TFLOPS for NVIDIA and 6.144 TFLOPS for Intel. FP16 compute is 126.0 TFLOPS at 1:1 ratio for NVIDIA and 12.29 TFLOPS at 2:1 ratio for Intel.

The physical and power characteristics are equally distinct. The Intel Arc B370 is an IGP with no slot width, no power connectors, and a 25 W TDP. The NVIDIA RTX PRO 6000 Blackwell Server is a dual-slot card measuring 267 mm in length, 111 mm in height, and 40 mm in width, with a 600 W TDP, one 16-pin power connector, and a suggested 1000 W PSU. The NVIDIA card uses a PCIe 5.0 x16 bus interface and provides four DisplayPort 2.1b outputs. The Intel part uses an IGP bus interface and its display outputs are portable-device dependent. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA card’s predecessor is Server Hopper and its successor is Server Rubin; the Intel part has no recorded predecessor or successor. The NVIDIA card was released in March 2025, while the Intel part’s release date is January 2026.