GPU Comparison

The NVIDIA P106-090 and the AMD Radeon 740M represent two fundamentally different approaches to graphics processing, separated by nearly seven years of technological evolution. The data reveals a clear performance hierarchy, but the story is more nuanced than simple wins and losses. Despite its age and mining-specific origins, the P106-090 consistently outperforms the modern integrated graphics solution in raw compute benchmarks, yet the 740M counters with architectural sophistication and efficiency that the older card cannot match.

Head-to-Head Benchmarks

The benchmark results are unambiguous in their verdict: the NVIDIA P106-090 dominates the AMD Radeon 740M in every recorded test. In the Geekbench OpenCL compute benchmark, the P106-090 scores 21,304 points against the 740M’s 10,172 points, translating to a staggering 109.4% performance advantage. This means the NVIDIA card delivers more than double the raw compute throughput of the AMD integrated processor in OpenCL workloads, a gap that speaks to the fundamental differences in their design philosophies.

The Vulkan benchmark tells a similar but less extreme story. Here, the P106-090 achieves 18,596 points while the 740M manages 15,568 points, giving NVIDIA a 19.5% lead. While narrower than the OpenCL gap, this still represents a decisive victory for the discrete card. The smaller delta in Vulkan suggests that the RDNA 3.0 architecture handles modern graphics APIs more efficiently relative to its raw compute capabilities, narrowing the performance gap when the workload leverages newer driver optimizations.

The average benchmark scores reinforce this hierarchy. The P106-090 posts an average score of 13,470, placing it at the 54th percentile of all GPUs, while the 740M averages 12,870, sitting just one percentile lower at 53rd. This places both cards in remarkably close proximity in the overall performance distribution, despite the head-to-head deltas showing significant differences in individual tests. The P106-090’s nearest rival is the NVIDIA GeForce GTX 570 with an average score of 13,515, a mere 0.3% difference, while the 740M’s closest competitor is the AMD Radeon RX 580 at 12,928, just 0.4% away.

Architecture Differences

The architectural chasm between these two GPUs is vast. The NVIDIA P106-090 uses the GP106 chip built on the Pascal architecture, manufactured on a 16 nm process at TSMC. This chip contains 4,400 million transistors packed into a 200 mm² die, yielding a transistor density of 22.0 million per square millimeter. In contrast, the AMD Radeon 740M employs the Phoenix2 chip with RDNA 3.0 architecture, built on a current-generation 4 nm process, also from TSMC. This smaller 137 mm² die houses an extraordinary 20,900 million transistors, achieving a density of 152.6 million per square millimeter, nearly seven times denser than the Pascal chip.

The implications of this density difference are profound. The P106-090 features 768 shading units, 48 texture mapping units, and 48 render output units, while the 740M has only 256 shading units, 16 TMUs, and 8 ROPs. Despite having three times fewer shading units, the AMD part achieves higher FP32 throughput at 2.867 TFLOPS compared to the NVIDIA’s 2.352 TFLOPS, thanks to its much higher boost clock of 2800 MHz versus 1531 MHz. The RDNA 3.0 architecture also delivers FP16 performance at a 1:1 ratio with FP32, reaching 2.867 TFLOPS, while the Pascal architecture’s FP16 throughput is severely restricted at 36.74 GFLOPS, a 1:64 ratio.

The 740M also introduces hardware ray tracing capabilities with 4 RT cores, a feature entirely absent from the Pascal-based P106-090. This architectural advantage, combined with DirectX 12 Ultimate support, positions the integrated GPU as significantly more future-proof despite its lower raw benchmark scores.

FAQ

Q: Why does the NVIDIA P106-090 score higher in OpenCL despite having fewer shading units than its competitor?

A: The P106-090’s OpenCL score of 21,304 versus the 740M’s 10,172 reflects its discrete design with dedicated 3 GB GDDR5 memory and 192.2 GB/s bandwidth. The 740M relies on system shared memory with bandwidth described as "system dependent," which likely bottlenecks compute workloads. Additionally, the P106-090’s 768 shading units, while individually less capable, provide more parallel execution lanes than the 740M’s 256 shading units.

Q: How does the transistor density difference affect performance?

A: The 740M’s 152.6 million transistors per square millimeter on a 4 nm process represents a massive manufacturing advantage over the P106-090’s 22.0 million per square millimeter on 16 nm. This density enables the smaller chip to pack 20,900 million transistors versus 4,400 million, allowing for more advanced features like ray tracing cores and higher clock speeds despite fewer shading units.

Q: Which card has better API support?

A: The AMD Radeon 740M supports DirectX 12 Ultimate (12_2), indicating full support for the latest DirectX features including ray tracing. The NVIDIA P106-090 supports DirectX 12 (12_1), which lacks some of the newer features. Both cards support OpenGL 4.6 and Vulkan 1.4.

Q: What is the significance of the 109.4% OpenCL delta?

A: This delta represents the P106-090 achieving more than double the 740M’s OpenCL performance. Such a large gap suggests that the NVIDIA card’s dedicated memory architecture provides a substantial advantage in compute-intensive workloads that benefit from high bandwidth and low latency, which integrated graphics with shared memory cannot match.

Q: Does the 740M’s higher FP32 throughput translate to better gaming performance?

A: Not necessarily. While the 740M’s 2.867 TFLOPS exceeds the P106-090’s 2.352 TFLOPS, the head-to-head Vulkan benchmark shows the P106-090 winning with an 19.5% higher score. Gaming performance depends on many factors beyond raw compute, including memory bandwidth, driver optimization, and architecture efficiency.

Q: How do these cards compare to their closest rivals?

A: The P106-090 sits within 0.5% of the GeForce GTX 570, Radeon Pro 555, Radeon HD 7770M, and Radeon RX 9070 XT in average score. The 740M similarly sits within 0.4% of the FirePro W5100, Radeon Pro 455, GeForce GTX 590, and Radeon RX 580, indicating both cards are competitively positioned in their respective performance tiers.

Specification Differences

The two GPUs differ in nearly every measurable specification. The P106-090 uses 3 GB of dedicated GDDR5 memory on a 192-bit bus with 192.2 GB/s bandwidth, while the 740M uses system shared memory with bus width and bandwidth described as "system dependent." The NVIDIA card operates at base and boost clocks of 1354 MHz and 1531 MHz respectively, while the AMD part runs at 800 MHz base and 2800 MHz boost.

Process technology separates them significantly: 16 nm for Pascal versus 4 nm for RDNA 3.0, resulting in transistor counts of 4,400 million versus 20,900 million on die sizes of 200 mm² versus 137 mm². The P106-090 has 768 shading units, 48 TMUs, and 48 ROPs, while the 740M has 256 shading units, 16 TMUs, and 8 ROPs, plus 4 RT cores. Pixel and texture rates favor the NVIDIA card at 73.49 GPixel/s and 73.49 GTexel/s versus 22.40 GPixel/s and 44.80 GTexel/s for the 740M.

Power characteristics differ dramatically: the P106-090 has a 75 W TDP with a 1x 6-pin power connector and requires a 250 W suggested PSU, while the 740M consumes just 45 W as an IGP with no power connectors. The P106-090 uses a PCIe 1.0 x1 interface (a curious limitation for a discrete card) and has no display outputs, while the 740M uses PCIe 4.0 x8 with motherboard-dependent display outputs. The P106-090 is dual-slot and end-of-life, while the 740M is an active IGP.

The Verdict

The data presents a clear performance winner: the NVIDIA P106-090 takes both head-to-head benchmarks and holds a higher average score. Its 109.4% OpenCL advantage and 19.5% Vulkan lead are decisive, and its 54th percentile ranking versus the 740M’s 53rd confirms its superior position. For compute-heavy workloads that leverage OpenCL, the P106-090 is unequivocally the stronger choice.

However, the 740M’s architectural advantages cannot be dismissed. Its 4 nm process, ray tracing support, DirectX 12 Ultimate compatibility, and dramatically lower power consumption of 45 W versus 75 W make it a more modern and efficient solution. The absence of display outputs on the P106-090 is a critical limitation, this mining-focused card cannot drive a monitor, making it entirely unsuitable for standard desktop use. The 740M, as an integrated GPU, requires no separate card and works with motherboard-dependent outputs.

The production status also matters: the P106-090 is end-of-life with no successor, while the 740M is active with both predecessor and successor identified. For users requiring raw compute throughput in a headless configuration, the P106-090’s benchmark superiority is compelling. For anyone needing a functional graphics solution with modern features and efficiency, the 740M is the only viable option despite its lower scores.

Where Each One Wins

The NVIDIA P106-090 wins in raw compute performance, dominating the Geekbench OpenCL benchmark by 109.4% and the Vulkan benchmark by 19.5%. Its dedicated 3 GB GDDR5 memory with 192.2 GB/s bandwidth provides consistent, high-throughput memory access that integrated solutions cannot replicate. Its higher pixel rate of 73.49 GPixel/s and texture rate of 73.49 GTexel/s suggest it excels in fill-rate-bound workloads.

The AMD Radeon 740M wins on architectural sophistication, featuring a 4 nm process with 152.6 million transistors per square millimeter, enabling ray tracing cores and DirectX 12 Ultimate support. Its 45 W TDP makes it dramatically more power-efficient, and its compact IGP form factor requires no additional power connections or slot space. The 740M’s higher FP32 throughput of 2.867 TFLOPS and 1:1 FP16 capability indicate potential advantages in workloads that leverage these features. Its PCIe 4.0 x8 interface provides significantly more bandwidth to the host system than the P106-090’s PCIe 1.0 x1 connection. The 740M also wins on currency: it is an active product with a successor, while the P106-090 is end-of-life. For integrated graphics users seeking modern gaming features and efficiency, the 740M represents the forward-looking choice, even as the benchmarks show the older discrete card winning the performance crown.