NVIDIA P106-100 vs NVIDIA TITAN Xp Comparison

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA P106-100 has a higher average benchmark score of 23,249 compared to the NVIDIA TITAN Xp's 19,177. This places the P106-100 at the 68th percentile of all GPUs, while the TITAN Xp sits at the 64th percentile.

Q: Why does the TITAN Xp win every head-to-head benchmark if its average score is lower?

A: The average score includes different test sets. The TITAN Xp has additional PassMark results (DirectX 9, 10, 11, 12, G2D, G3D, and compute) that drag its average down. In the three shared benchmarks, the TITAN Xp wins all of them: 3DMark Steel Nomad DX12 (2,372 vs 899), Geekbench OpenCL (72,585 vs 35,951), and Geekbench Vulkan (87,180 vs 32,897).

Q: What are the nearest rivals for each card?

A: For the P106-100, the closest rivals are the AMD Radeon Pro Vega 16 (0% delta), AMD Radeon RX 6600M (-0.1%), AMD Radeon R9 M290X (-0.1%), and AMD Radeon AI PRO R9700 (-0.3%). For the TITAN Xp, the nearest rivals are the NVIDIA GeForce GTX 780 (0.1% delta), NVIDIA Tesla K20m (0.5%), NVIDIA GeForce RTX 4050 Mobile (0.7%), and AMD Radeon RX 6600 (0.7%).

Q: What is the memory configuration difference?

A: The P106-100 uses 6 GB of GDDR5 on a 192-bit bus, yielding 192.2 GB/s bandwidth. The TITAN Xp uses 12 GB of GDDR5X on a 384-bit bus, yielding 547.6 GB/s bandwidth, which is roughly 2.8 times higher.

Q: Do both cards support the same APIs?

A: Yes, both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. Neither has ray tracing or tensor cores.

Q: What is the TITAN Xp's launch MSRP?

A: The TITAN Xp had a launch MSRP of 1,199 USD. The P106-100 has no listed launch MSRP.

Architecture Differences

Both cards are built on NVIDIA's Pascal architecture and use TSMC's 16 nm process, but they are fundamentally different chips. The P106-100 uses the GP106 die with 4,400 million transistors on a 200 mm² die, giving a transistor density of 22.0M per mm². The TITAN Xp uses the GP102 die with 11,800 million transistors on a 471 mm² die, giving a density of 25.1M per mm². That is a 2.7 times transistor count difference and a 2.35 times die area difference.

The P106-100 belongs to the "Mining GPUs" generation, which explains its most striking architectural feature: no display outputs at all. It is a compute-only card. In contrast, the TITAN Xp is part of the GeForce 10 generation and offers 1x HDMI 2.0 and 3x DisplayPort 1.4a outputs.

The compute core counts diverge sharply. The P106-100 has 1,280 shading units, 80 texture mapping units, and 48 ROPs. The TITAN Xp has 3,840 shading units, 240 TMUs, and 96 ROPs. That is exactly 3x the shading units, 3x the TMUs, and 2x the ROPs. Neither card has dedicated ray tracing or tensor cores, which aligns with their Pascal generation.

Clock speeds tell an interesting story. The P106-100 runs a base clock of 1506 MHz and boost of 1709 MHz, which is higher than the TITAN Xp's base of 1405 MHz and boost of 1582 MHz. The smaller GP106 chip runs hotter clocks but with far fewer cores, so total throughput still favors the TITAN Xp massively.

The memory subsystems are completely different in type and bandwidth. The P106-100 uses GDDR5 at 2002 MHz (8 Gbps effective) while the TITAN Xp uses GDDR5X at 1426 MHz (11.4 Gbps effective). With a 384-bit bus versus 192-bit, the TITAN Xp achieves 547.6 GB/s, nearly three times the P106-100's 192.2 GB/s.

The bus interface also differs: PCIe 1.0 x16 on the P106-100 versus PCIe 3.0 x16 on the TITAN Xp. The TITAN Xp also has a predecessor (GeForce 900) and successor (GeForce 20), while the P106-100 has neither listed.

The Verdict

The data presents a clear split. The TITAN Xp wins every shared benchmark by large margins. In 3DMark Steel Nomad DX12 it scores 2,372 versus 899, a 62.1% advantage. In Geekbench OpenCL it scores 72,585 versus 35,951, a 50.5% advantage. In Geekbench Vulkan it scores 87,180 versus 32,897, a 62.3% advantage. If the workload is any of these three tests, the TITAN Xp is the obvious choice.

However, the P106-100 has the higher average benchmark score (65,249 vs 19,277) and higher percentile ranking (68th vs 64th). The P106-100's nearest rivals include the AMD Radeon RX 6600M and AMD Radeon R9 M290X, while the TITAN Xp's nearest rivals include the older NVIDIA GeForce GTX 780 and the RTX 4050 Mobile. The P106-100 averages roughly 21% higher than the TITAN Xp across all recorded results.

The TITAN Xp costs a launch MSRP of 1,199 USD. The P106-100 has no price listed. For users who need maximum raw compute in the shared tests, the TITAN Xp is unmatched. For users who look at the overall database average, the P106-100 is ranked higher.

The P106-100 is a mining-oriented card with no display outputs, which limits its use to compute or mining workloads. The TITAN Xp is a full-featured GPU with display outputs, PCIe 3.0, and a much larger memory pool. The data suggests the TITAN Xp is the faster part in direct comparisons, but the P106-100 has a better overall standing in the database, likely due to different benchmark weighting.

Specification Differences

The two cards differ in nearly every hardware specification. Process node is identical (16 nm TSMC), but the die sizes differ: 200 mm² for the P106-100 versus 471 mm² for the TITAN Xp. Transistor count is 4,400 million versus 11,800 million. Transistor density is 22.0M/mm² versus 25.1M/mm².

Clocks differ in both base and boost. P106-100 runs at 1506 MHz base and 1709 MHz boost. TITAN Xp runs lower at 1405 MHz base and 1582 MHz boost. Memory clocks are 2002 MHz (8 Gbps effective) versus 1426 MHz (11.4 Gbps effective).

Memory configuration is completely different: 6 GB GDDR5 on 192-bit versus 12 GB GDDR5X on 384-bit. Bandwidth is 192.2 GB/s versus 547.6 GB/s. The compute units scale as 1,280 vs 3,840 shading units, 80 vs 240 TMUs, and 48 vs 96 ROPs. Pixel rate is 82.03 GPixel/s versus 151.9 GPixel/s. Texture rate is 136.7 GTexel/s versus 379.7 GTexel/s. FP32 is 4.375 TFLOPS versus 12.15 TFLOPS. FP16 is 68.36 GFLOPS (1:64) versus 189.8 GFLOPS (1:64).

TDP is 120 W versus 250 W. The P106-100 uses a single 6-pin connector, while the TITAN Xp uses a 6-pin plus an 8-pin. Suggested PSU is 300 W versus 600 W. The bus interface is PCIe 5.0 x16 versus PCIe 3.0 x16. Display outputs: none versus 1x HDMI 2.0 and 3x DisplayPort 1.4a. The TITAN Xp has dimensions of 267 mm x 112 mm x 40 mm, while the P106-100 is 250 mm long with no height or width listed. Release dates differ by about two months (2017-06-18 vs 2017-04-05). The TITAN Xp has a launch MSRP of 1,199 USD; the P106-100 has none.

Head-to-Head Benchmarks

The three shared benchmarks all go to the TITAN Xp. The 3DMark Steel Nomad DX12 test shows the most extreme delta: 2,372 for the TITAN Xp versus 899 for the P106-100, a difference of 62.1%. That is more than 2.6 times the score. This test likely stresses rasterization and memory bandwidth, both of which heavily favor the TITAN Xp's larger core and memory subsystem.

Geekbench Vulkan shows a 62.3% delta, the largest of all: 87,180 for the TITAN Xp versus 32,897 for the P106-100. Vulkan performance scales with shader count and bandwidth, and the TITAN Xp has 3x shading units and 2.8x bandwidth. This result is consistent with the compute core ratio.

Geekbench OpenCL shows a 50.5% delta, with 72,585 versus 35,951. Interestingly, this is the smallest relative gap of the three, but still a massive absolute difference. The TITAN Xp's FP32 output of 12.15 TFLOPS versus 4.375 TFLOPS (2.8x) does not fully translate to the 2x OpenCL score, possibly due to driver or workload characteristics.

The P106-100 wins zero head-to-head benchmarks. All three go to the TITAN Xp. The delta percentages are all negative for the P106-100: -62.1%, -50.5%, and -62.3%. There is no benchmark in which the P106-100 comes close to matching the TITAN Xp.

When placed among their nearest rivals, the TITAN Xp's average score of 19,177 sits within 0.7% of the GeForce GTX 780 and RTX 4050 Mobile. The P106-100's average of 65,249 is within 0.3% of the AMD Radeon AI PRO R9700. The two cards occupy completely different performance neighborhoods in the database.

Where Each One Wins

The TITAN Xp wins every shared benchmark. The single biggest win is in Geekbench Vulkan, with a 62.3% advantage. The 3DMark Steel Nomad DX12 follows at 62.1%, and Geekbench OpenCL at 50.5%. If your workload is any of these three, the TITAN Xp is the clear winner. The TITAN Xp also has a full display output set (1x HDMI, 3x DP), so it can drive monitors. It has 12 GB of memory, double the P106-100, and a 384-bit bus with 547.6 GB/s bandwidth. It also supports PCIe 3.0, which is more standard than the P106-100's PCIe 5.0 x16 bus (an oddity for a mining card).

The P106-100 wins only in the aggregate average benchmark score. With an average of 65,249 versus 19,177, it sits at the 68th percentile versus 64th. Its nearest rivals are all AMD Radeon parts (Pro Vega 16, RX 6600M, R9 M290X, AI PRO R9700), and it has a 0.3% delta from the top rival. The P106-100 also has a much lower TDP (120 W vs 250 W) and only a single 6-pin power connector, with a 300 W suggested PSU. This makes it far easier to power. It runs at higher boost clocks (1709 MHz vs 1582 MHz), which helps in compute tasks that are clock-sensitive.

For users who prioritize the database's overall ranking, the P106-100 is the higher-ranked GPU. For users who prioritize any specific benchmark in the head-to-head set, the TITAN Xp is the only choice. The P106-100 also has no display outputs, so it cannot drive a monitor at all; the TITAN Xp can. The TITAN Xp is the more versatile and faster card in every direct test, while the P106-100 is a lower-power compute card that happens to average better across the full database.