NVIDIA GeForce RTX 5070 vs NVIDIA P104-100 Comparison

FAQ

Q: How does the NVIDIA GeForce RTX 5070 compare to the NVIDIA P104-100 in raw benchmark performance?

A: The RTX 5070 wins all three recorded head-to-head tests. Its largest margin is in Geekbench Vulkan, where it scores 178,923 versus 45,165, a 296.2% advantage. In 3DMark Steel Nomad DX12, it scores 5,077 versus 1,413, a 259.3% lead. The smallest margin is in Geekbench OpenCL: 172,660 versus 52,368, a 229.7% difference.

Q: What is the average benchmark score for each GPU according to the database?

A: The RTX 5070 has an average benchmark score of 40,377, placing it in the 82nd percentile of all GPUs. The P104-100 has an average score of 32,982, placing it in the 77th percentile. The RTX 5070's closest rival is the AMD Radeon Pro 580, with a 0.1% delta, while the P104-100's closest rival is the NVIDIA T600 Mobile, with a 0.4% delta.

Q: Which GPU offers more memory and what type?

A: The RTX 5070 has 12 GB of GDDR7 memory on a 192-bit bus, delivering 672.0 GB/s bandwidth. The P104-100 has 4 GB of GDDR5X memory on a 256-bit bus, delivering 320.3 GB/s bandwidth. The RTX 5070 has roughly double the bandwidth despite the narrower bus.

Q: Are there any display output differences between the two cards?

A: Yes. The RTX 5070 features 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs. The P104-100 has no display outputs at all, as it is listed under the "Mining GPUs" generation. This makes the RTX 5070 the only option for direct video output.

Q: What is the difference in production status?

A: The RTX 5070 is listed as "Active" production, with a release date of 2025-03-03. The P104-100 is listed as "End-of-life" production, with a release date of 2017-12-11. The RTX 5070 also has a recorded launch MSRP of 549 USD, while the P104-100 has no recorded launch MSRP.

Q: Which GPU has a higher transistor density?

A: The RTX 5070 has a transistor density of 118.3M per mm², while the P104-100 has 22.9M per mm². This reflects the process node difference: the RTX 5070 is built on a 5 nm process, while the P104-100 uses a 16 nm process.

Architecture Differences

The two GPUs belong to completely different architectural eras. The RTX 5070 uses the Blackwell 2.0 architecture, built on a 5 nm process at TSMC. The P104-100 uses the older Pascal architecture, also from TSMC but on a 16 nm process. This process shrink is a major factor in the performance gap.

The RTX 5070's chip, GB205, packs 31,100 million transistors onto a 263 mm² die. The P104-100's chip, GP104, contains 7,200 million transistors on a larger 314 mm² die. This means the RTX 5070 achieves far higher transistor density: 118.3M per mm² versus 22.9M per mm². The newer process allows more than four times the transistor count on a smaller physical footprint.

The RTX 5070 includes 48 RT cores and 192 tensor cores, features entirely absent from the P104-100, which has neither. This reflects the RTX 5070's support for hardware ray tracing and AI acceleration. The P104-100, designed for mining workloads, lacks these specialized units entirely. The shading units also differ dramatically: the RTX 5070 has 6,144 shading units, 192 TMUs, and 80 ROPs, while the P104-100 has 1,920 shading units, 120 TMUs, and 64 ROPs.

Clock behavior differs as well. The RTX 5070 runs at a base clock of 2325 MHz and a boost clock of 2512 MHz. The P104-100 has a base clock of 1607 MHz and a boost clock of 1733 MHz. Memory clocks also differ: the RTX 5070 uses 1750 MHz (28 Gbps effective) GDDR7, while the P104-100 uses 1251 MHz (10 Gbps effective) GDDR5X.

The API support shows the generational gap. The RTX 5070 supports DirectX 12 Ultimate (12_2), while the P104-100 supports DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4. The bus interfaces differ significantly: the RTX 5070 uses PCIe 5.0 x16, while the P104-100 uses PCIe 1.0 x4, which severely limits data transfer rates.

Power delivery also differs. The RTX 5070 has a TDP of 250 W, a 1x 16-pin power connector, and a suggested PSU of 600 W. The P104-100 has no recorded TDP, uses a 1x 8-pin connector, and has a suggested PSU of 200 W. Both are dual-slot cards, but the RTX 5070 is shorter at 245 mm versus the P104-100's 267 mm.

Head-to-Head Benchmarks

The recorded data shows a decisive sweep in favor of the RTX 5070 across all three shared tests. The largest relative gap appears in Geekbench Vulkan. The RTX 5070 scores 178,923 versus the P104-100's 45,165, a delta of 296.2%. This test stresses modern API features, which the Blackwell 2.0 architecture handles far better than Pascal. The RTX 5070's Vulkan score is roughly 3.96 times higher, a margin that reflects both the newer architecture and the higher core count.

The 3DMark Steel Nomad DX12 test shows a similar pattern. The RTX 5070 scores 5,077, while the P104-100 scores 1,413. This is a 259.3% delta, meaning the RTX 5070 delivers about 3.59 times the performance. Steel Nomad is a demanding DX12 workload, and the RTX 5070's 12_2 feature level support gives it an edge over the P104-100's 12_1 support. The RTX 5070 also has far more shading units (6,144 versus 1,920) and a much higher pixel rate (201.0 GPixel/s versus 110.9 GPixel/s).

Geekbench OpenCL shows the smallest relative margin, but it is still a substantial win. The RTX 5070 scores 172,660 versus 52,368, a 229.7% delta. This means the RTX 5070 is about 3.30 times faster in this compute-oriented test. The OpenCL result is notable because the P104-100 was originally designed for mining, which is a compute-heavy workload. Even in that domain, the newer card dominates. The RTX 5070's FP32 throughput is 30.87 TFLOPS versus the P104-100's 6.655 TFLOPS, a 4.64 times advantage.

The wins total 3 for the RTX 5070 and 0 for the P104-100 in the shared test suite. There are no recorded tests where the P104-100 comes out ahead. The average benchmark scores reinforce this: the RTX 5070 sits at 40,377, compared to 32,982 for the P104-100. The percentile rankings also show separation, with the RTX 5070 at 82 and the P104-100 at 77.

Specification Differences

The two cards differ across nearly every specification category. The RTX 5070 uses a Blackwell 2.0 architecture on a 5 nm process, while the P104-100 uses Pascal on 16 nm. The RTX 5070's chip is GB205 with 31,100 million transistors and a 263 mm² die. The P104-100's chip is GP104 with 7,200 million transistors and a 314 mm² die. Transistor density is 118.3M per mm² versus 22.9M per mm².

Memory configuration differs in size, type, and bandwidth. The RTX 5070 has 12 GB of GDDR7 on a 192-bit bus, delivering 672.0 GB/s. The P104-100 has 4 GB of GDDR5X on a 256-bit bus, delivering 320.3 GB/s. The bus width is wider on the older card, but the newer memory technology gives the RTX 5070 more than double the bandwidth.

Core counts are dramatically different. The RTX 5070 has 6,144 shading units, 192 TMUs, and 80 ROPs. The P104-100 has 1,920 shading units, 120 TMUs, and 64 ROPs. The RTX 5070 also adds 48 RT cores and 192 tensor cores, which the P104-100 lacks entirely. Pixel rate is 201.0 GPixel/s versus 110.9 GPixel/s, and texture rate is 482.3 GTexel/s versus 208.0 GTexel/s.

Compute throughput shows a wide gap. The RTX 5070 delivers 30.87 TFLOPS FP32 and 30.87 TFLOPS FP16 (1:1). The P104-100 delivers 6.655 TFLOPS FP32 and 104.0 GFLOPS FP16 (1:64). The FP16 ratios are particularly telling: the RTX 5070 has a 1:1 ratio, while the P104-100 has a 1:64 ratio, meaning FP16 performance is heavily cut on the older card.

Power and connectivity differ as well. The RTX 5070 has a TDP of 250 W, a 1x 16-pin connector, and a suggested PSU of 600 W. The P104-100 has no TDP recorded, a 1x 8-pin connector, and a suggested PSU of 200 W. The RTX 5070 uses PCIe 5.0 x16, while the P104-100 uses PCIe 1.0 x4. Display outputs exist only on the RTX 5070 (1x HDMI 2.1b, 3x DisplayPort 2.1b); the P104-100 has none.

Physical dimensions also differ. The RTX 5070 is 245 mm long, 115 mm tall, and 40 mm wide. The P104-100 is 267 mm long, with no recorded height or width. Production status is Active for the RTX 5070 and End-of-life for the P104-100. Release dates are 2025-03-03 and 2017-12-11, respectively.

Where Each One Wins

The RTX 5070 wins in every recorded benchmark category. In 3DMark Steel Nomad DX12, it is 259.3% faster. In Geekbench OpenCL, it is 229.7% faster. In Geekbench Vulkan, it is 296.2% faster. There are no recorded tests where the P104-100 outperforms the RTX 5070.

The use cases for the RTX 5070 are broad. It supports modern display outputs, so it can drive monitors directly. It has hardware ray tracing and tensor cores, making it suitable for gaming, 3D rendering, and AI workloads. Its 12 GB of GDDR7 memory and 672.0 GB/s bandwidth support high-resolution textures and large datasets. Its FP16 1:1 ratio makes it efficient for mixed-precision compute.

The P104-100 has no display outputs, so it cannot be used for any visual output tasks. Its 4 GB of GDDR5X memory and 320.3 GB/s bandwidth are sufficient for its original mining purpose but limiting elsewhere. It has no RT or tensor cores, so it lacks acceleration for ray tracing and AI inference. Its FP16 1:64 ratio means half-precision workloads run at a severe disadvantage.

For gaming, the RTX 5070 is the clear choice due to its DirectX 12 Ultimate support and higher core counts. For compute tasks, the RTX 5070's FP32 throughput of 30.87 TFLOPS dwarfs the P104-100's 6.655 TFLOPS. For any workload involving modern APIs, the RTX 5070's PCIe 5.0 x16 interface beats the P104-100's PCIe 1.0 x4. The P104-100's only advantage is its lower suggested PSU of 200 W versus 600 W, though it has no TDP recorded to confirm actual power draw.

The Verdict

The data supports the RTX 5070 as the superior product across every measurable dimension. It wins all three head-to-head tests with deltas between 229.7% and 296.2%. Its average benchmark score of 40,377 places it in the 82nd percentile, while the P104-100 sits at 32,982 in the 77th percentile. The RTX 5070's nearest rivals are all within 2% of its average score, indicating strong competition at its performance tier; the P104-100's nearest rivals are similarly close, but at a much lower absolute level.

The architectural differences explain the performance gap. The RTX 5070 uses a 5 nm process with 31,100 million transistors, while the P104-100 uses 16 nm with 7,200 million. The RTX 5070 has 48 RT cores and 192 tensor cores, which the P104-100 lacks. The RTX 5070 supports DirectX 12 Ultimate, while the P104-100 only reaches DirectX 12 (12_1). The newer card also has display outputs, making it usable for any visual task.

The P104-100 is an end-of-life product from 2017, designed for mining. It has no display outputs, no RT or tensor cores, and limited memory. Its PCIe 1.0 x4 interface is a bottleneck for modern workloads. While its 256-bit memory bus is wider, the older GDDR5X technology delivers less bandwidth than the RTX 5070's GDDR7.

For a user selecting between these two, the choice is straightforward. The RTX 5070, with its launch MSRP of 549 USD, offers modern features, active production status, and benchmark results that are roughly three to four times higher across the board. The P104-100 offers no recorded advantages in any shared test. The RTX 5070 is the only option for gaming, rendering, or any workload requiring video output. The P104-100, given its end-of-life status and lack of outputs, has no clear use case in a modern system. The data indicates the RTX 5070 is the definitive choice for anyone comparing these two GPUs.