AMD Radeon RX Vega 11 vs NVIDIA P106-090 Comparison

The AMD Radeon RX Vega 11 and the NVIDIA P106-090 are two very different GPUs separated by a wide performance gulf in the benchmarks provided. The data shows NVIDIA’s mining-oriented card is decisively faster in both shared tests, while AMD’s integrated part counters with a lower power draw and more flexible feature set. This analysis breaks down the head-to-head results, architectural differences, and use-case implications based strictly on the facts.

Head-to-Head Benchmarks

The comparison is lopsided. The NVIDIA P106-090 wins both benchmark tests, taking a clean 2-0 sweep. In Geekbench OpenCL, the P106-090 scores 21,304 against the Vega 11’s 12,525, a difference of 41.2% in NVIDIA’s favor. This is a massive gap, placing the P106-090 well ahead of the integrated AMD part in raw compute workloads. The Geekbench Vulkan test tells a similar story, though the margin narrows slightly. Here, the P106-090 posts 18,596 versus the Vega 11’s 13,543, giving NVIDIA a 27.2% lead.

These deltas are significant. A 41.2% advantage in OpenCL means the P106-090 is not just marginally better but in a different performance class for general-purpose compute. The Vulkan gap, while smaller, still represents a substantial win for NVIDIA in graphics API workloads. Interestingly, the Vega 11’s average benchmark score across all tests is 14,385, which is actually higher than the P106-090’s average of 13,470. This is because the P106-090’s average is dragged down by a weak result in the 3DMark Steel Nomad DX12 test (scoring only 509), while the Vega 11 has no corresponding DirectX 12 benchmark in its data set. So while the head-to-head tests favor NVIDIA, the broader average score paradoxically favors AMD, though that average includes tests not shared between the two.

Looking at the percentile rankings, both cards sit near the middle of the GPU pack. The Vega 11 lands at the 56th percentile, while the P106-090 sits at the 54th. This suggests that despite the P106-090’s dominance in these two specific tests, its overall standing is slightly lower due to that poor Steel Nomad result. In the nearest rival comparisons, the Vega 11 trades blows with the NVIDIA GeForce GTX TITAN (delta of 0.1%) and the GeForce GTX 965M (delta of -0.1%), showing it is right in that performance tier. The P106-090, meanwhile, is nearly identical to the GeForce GTX 570 (delta of -0.3%) and the AMD Radeon Pro 555 (delta of 0.5%).

Architecture Differences

The two GPUs come from entirely different eras and design philosophies. The AMD Radeon RX Vega 11 is an integrated graphics processor (IGP) built on the GCN 5.0 architecture, using the Picasso chip at a 12 nm node from GlobalFoundries. It packs 4,940 million transistors on a 210 mm² die, yielding a density of 23.5M / mm². The NVIDIA P106-090, by contrast, is a dedicated mining card based on the older Pascal architecture, using the GP106 chip at a 16 nm node from TSMC. It has 4,400 million transistors on a 200 mm² die, with a slightly lower density of 22.0M / mm². Despite being on a larger process node, the P106-090 still manages a comparable transistor count due to its dedicated design.

The memory subsystem is where the biggest practical differences emerge. The Vega 11 uses System Shared memory, meaning it has no dedicated VRAM and instead borrows from the system’s RAM. Its bandwidth is listed as System Dependent, which makes performance highly variable based on the host system. The P106-090, in contrast, has a fixed 3 GB of GDDR5 memory on a 192-bit bus, delivering a solid 192.2 GB/s of bandwidth. This dedicated memory is a major reason why the NVIDIA card wins in compute tasks—it does not have to contend with system memory latency or contention.

Compute unit counts are similar but not identical. The Vega 11 has 704 shading units, 44 texture mapping units (TMUs), and only 8 render output units (ROPs). The P106-090 has slightly more shading units at 768, plus 48 TMUs and a substantially higher 48 ROPs. This ROP count difference is stark—the P106-090 has six times the pixel-pushing capability, which explains its much higher pixel rate of 73.49 GPixel/s versus the Vega 11’s 11.20 GPixel/s. The texture rate favors NVIDIA as well, at 73.49 GTexel/s versus 61.60 GTexel/s.

Clock speeds and power draw tell the rest of the story. The Vega 11 runs at a base of 300 MHz with a boost of 1400 MHz, while the P106-090 starts at 1354 MHz and boosts to 1531 MHz. The P106-090’s higher clocks, combined with its superior memory and ROP counts, drive its FP32 performance to 2.352 TFLOPS versus the Vega 11’s 1.971 TFLOPS. However, the Vega 11 has a massive advantage in power efficiency: its TDP is just 15 W, compared to the P106-090’s 75 W. The NVIDIA card also requires a 1x 6-pin power connector and a 250 W suggested PSU, while the Vega 11 draws everything from the motherboard as an IGP.

Where Each One Wins

Given the benchmark data, the NVIDIA P106-090 is the clear winner for raw computational throughput. Its 41.2% OpenCL lead and 27.2% Vulkan lead make it the obvious choice for any workload that relies on GPU compute, such as OpenCL-based rendering, physics simulation, or Vulkan-accelerated applications. The dedicated 192.2 GB/s of bandwidth and 48 ROPs also make it better suited for traditional rasterization, provided the software does not require a display output—which is a critical caveat.

The AMD Radeon RX Vega 11 wins in scenarios where the P106-090 cannot even participate. The P106-090 has No outputs, meaning it cannot drive a monitor. It is purely a compute device. The Vega 11, as an IGP, has Motherboard Dependent display outputs, meaning it can run a desktop and handle everyday graphics tasks. Its 15 W TDP is also a fraction of the P106-090’s 75 W, making it far more suitable for low-power or mobile systems where a dedicated card is impractical.

For API coverage, the P106-090 supports Vulkan 1.4, which is newer than the Vega 11’s Vulkan 1.3, and both support DirectX 12 (12_1) and OpenGL 4.6. The Vega 11 also offers FP16 performance of 3.942 TFLOPS (2:1), which is significantly higher than the P106-090’s paltry 36.74 GFLOPS (1:64). For workloads that use half-precision math, the AMD part is dramatically better, despite losing in the full-precision FP32 race.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon RX Vega 11, with an average score of 14,385 versus the NVIDIA P106-090’s 13,470.

Q: Why does the P106-090 lose the average score comparison despite winning both head-to-head tests?

A: The P106-090’s average is lowered by its score of 509 in the 3DMark Steel Nomad DX12 test, which the Vega 11 does not have in its benchmark set.

Q: Can the NVIDIA P106-090 be used to connect a monitor?

A: No, the P106-090 has No outputs, making it unsuitable for display purposes.

Q: What is the memory configuration difference?

A: The Vega 11 uses System Shared memory with bandwidth that is System Dependent, while the P106-090 has 3 GB of GDDR5 on a 192-bit bus with 192.2 GB/s bandwidth.

Q: Which card is more power-efficient?

A: The AMD Radeon RX Vega 11, with a 15 W TDP, is far more efficient than the NVIDIA P106-090’s 75 W TDP.

Q: How do the cards compare in FP16 performance?

A: The Vega 11 offers 3.942 TFLOPS FP16 (2:1 ratio), while the P106-090 delivers only 36.74 GFLOPS (1:64 ratio), making the AMD card vastly superior for half-precision compute.

The Verdict

The data presents a clear, if narrow, picture. The NVIDIA P106-090 is the superior performer in every shared benchmark, with a commanding 41.2% lead in OpenCL and a 27.2% lead in Vulkan. Its dedicated 3 GB GDDR5 memory and 48 ROPs make it a far more capable compute engine. If the task is headless GPU compute—mining, rendering, or number-crunching—the P106-090 is the obvious choice, provided the system can supply 75 W and a 6-pin connector.

The AMD Radeon RX Vega 11, however, is the only viable option for a system that needs a display output. Its Motherboard Dependent outputs and 15 W TDP make it ideal for lightweight, integrated builds. It also beats the P106-090 in FP16 throughput by a factor of over 100x (3.942 TFLOPS vs 36.74 GFLOPS), making it the better pick for half-precision workloads. Users should also note that while the P106-090 wins the head-to-head, its overall percentile ranking (54th) is actually lower than the Vega 11’s (56th), due to the weak Steel Nomad result. Ultimately, the choice hinges on whether the workload requires a display: choose the P106-090 for raw compute, or the Vega 11 for a balanced, low-power IGP with output capabilities.