SYSTEM ANALYZER

Rate My PC: Intel Core Ultra 9 285K + NVIDIA GeForce RTX 5070 SUPER

Get a comprehensive performance analysis of your gaming rig with detailed benchmarks, bottleneck detection, and upgrade recommendations

Build Configuration

Select your components

PROCESSOR (CPU)

▼

GRAPHICS CARD (GPU)

▼

88 Score

Top 12% Ranking

88 / 100

HIGH-END

Power Build

Top 12% of systems. Excellent for 1440p Ultra or 4K High gaming.

1440p Ultra4K High

System Balance Analysis

CPU vs GPU performance ratio

Well Balanced

CPU

97%

GPU

79%

PROCESSOR

Intel Core Ultra 9 285K

83,807 Benchmark Score

Top 3% Market Ranking

View Full Specs →

GRAPHICS CARD

NVIDIA GeForce RTX 5070 SUPER

2,690 Benchmark Score

Top 21% Market Ranking

View Full Specs →

Market Position

How your build compares to others

Budget

0-30

Mid-Range

30-60

High-End

60-85

Enthusiast

85-100

Your Build

Game Performance Benchmarks

Real-world 4K FPS in popular titles

View All Games →

Performance Insights

Tips to maximize your system

Strong Performance

Excellent for 1440p gaming. Most games will run at high/ultra settings smoothly.

Compatible Games See what you can play Compare CPUs Find upgrades Compare GPUs Find upgrades

Performance Tiers Explained

90-100

Ultimate

4K Ultra gaming, VR ready, ray tracing enabled, professional workloads

4K 60+ FPS VR Ready

70-89

High-End

1440p Ultra or 4K High settings, excellent for modern AAA titles

1440p Ultra 4K High

50-69

Mid-Range

1080p Ultra or 1440p Medium, great value for most gamers

1080p Ultra 1440p Med

30-49

Entry Level

1080p Medium settings, suitable for eSports and older titles

1080p Med eSports

0-29

Legacy

Basic gaming, older titles, consider upgrading for modern games

720p-1080p Low Older Games

# GPU Analysis — VRAM, bandwidth, clocks, RT/tensor hardware, what the benchmark scores mean for rendering

The NVIDIA GeForce RTX 5070 SUPER is built on the Blackwell 2.0 architecture with a GB205 chip, fabricated on TSMC's 5 nm process with 31,100 million transistors on a 263 mm² die. The GPU operates with a base clock of 2325 MHz and a boost clock of 2512 MHz, while memory runs at 1750 MHz (28 Gbps effective). It carries 18 GB of GDDR7 memory across a 192-bit bus, yielding 672.0 GB/s of bandwidth — a substantial pool for high-resolution textures and large datasets.

The compute configuration is substantial: 6400 shading units, 200 texture mapping units, and 80 raster operations pipelines. The pixel rate is 201.0 GPixel/s and the texture rate is 502.4 GTexel/s, indicating strong fill-rate capabilities for both geometric and texture-heavy scenes. For ray tracing, there are 50 dedicated RT cores, and for AI-accelerated workloads, 200 tensor cores are present. FP32 throughput is rated at 32.15 TFLOPS, with FP16 also at 32.15 TFLOPS (1:1 ratio), making the card equally adept at standard graphics and mixed-precision compute tasks.

In the 3DMark Steel Nomad DX12 benchmark, the RTX 5070 SUPER scores 2690. This places it at the 18th percentile among all GPUs, which means the vast majority of GPUs in the database outperform it. The nearest rivals are telling: NVIDIA Quadro K1100M (2664, 1% slower), GeForce GT 1030 (2662, 1.1% slower), Intel Arc Pro B50 (2660, 1.1% slower), and GeForce GT 440 (2645, 1.7% slower). These are all low-end or legacy parts, and the deltaPct values are small — the 5070 SUPER is only fractionally ahead of them. This is a striking result for a card with 18 GB of GDDR7 and 50 RT cores; the benchmark score suggests that the Steel Nomad workload does not translate the card's theoretical capabilities into proportional performance, or that the card is severely constrained in this particular test.

For rendering workloads, the 50 RT cores and 200 tensor cores are the relevant hardware. The ray tracing performance is not directly measured in the provided benchmark, but the presence of dedicated RT hardware means that real-time ray-traced scenes will have hardware acceleration. The tensor cores enable DLSS-style upscaling and AI denoising, which can significantly reduce render times in supported applications. The 672.0 GB/s bandwidth and 18 GB capacity are generous for 4K texture sets and complex scenes, but the 18th percentile score suggests that the card may not deliver the expected throughput in compute-heavy rendering tasks. The 32.15 TFLOPS FP32 figure is moderate for modern GPUs, and the 1:1 FP16 ratio indicates that half-precision workloads will not receive a throughput boost, which could be a limitation for AI training or certain scientific compute tasks that rely on FP16 acceleration.

# Benchmark Performance — exact CPU and GPU scores, percentile positions, and what the combined picture is

The CPU benchmarks for the Intel Core Ultra 9 285K are extensive and consistently strong. In Cinebench R15, the multi-core score is 6494 and single-core is 359. Cinebench R20 shows 24003 multi-core and 3388 single-core. Cinebench R23 delivers 42522 multi-core and 2377 single-core. Geekbench results are 26702 multi-core and 2870 single-core. PassMark tests further illustrate the CPU's capabilities: multi-thread 67260, single-thread 5087, integer math 172379, floating-point math 224324, extended instructions 62277, data compression 790052, data encryption 57745, physics 3938, find prime numbers 541, and random string sorting 94927.

The CPU's average benchmark score is 83807, placing it at the 96th percentile among all CPUs — a top-tier result. The nearest rivals show a tight cluster: Intel Core Ultra 9 290K Plus (84003, -0.2% delta, meaning the 285K is slightly behind), AMD EPYC 4584PX (83090, 0.9% ahead), AMD EPYC 9135 (82980, 1% ahead), and AMD EPYC 7F72 (85072, -1.5% behind). The deltas are remarkably small, all within 1.5%, which indicates that the 285K sits at the very top of desktop CPU performance, trading places with high-end server and workstation parts.

The GPU's benchmark score of 2690 in 3DMark Steel Nomad places it at the 18th percentile, as noted. The combined percentile for this CPU+GPU pairing is 57th, reflecting the massive imbalance between the two components. The CPU is in the top 4% of all CPUs, while the GPU is in the bottom 18% — the pairing is heavily skewed toward processor performance.

The combined picture is one of a workstation-class CPU paired with a GPU that benchmarks like an entry-level part. For CPU-bound workloads such as compilation, video encoding, or physics simulation, this system would excel. For GPU-bound tasks like gaming or 3D rendering, the RTX 5070 SUPER's low percentile suggests it would be the limiting factor, potentially bottlenecking the CPU's capabilities. The data shows a system built for compute-heavy, CPU-centric tasks rather than graphics-intensive applications.

# CPU Analysis — cores, clocks, architecture, what the benchmark scores mean for real workloads

The Intel Core Ultra 9 285K is a 24-core, 24-thread processor from the Core Ultra Series 2, based on the Arrow Lake architecture (Arrow Lake-S). It uses the Intel Socket 1851 and is fabricated on a 3 nm process by TSMC, with 17,800 million transistors on a 243 mm² die. The base clock is 3.70 GHz, boosting to 5.70 GHz. The TDP is 125 W, and the multiplier is unlocked for overclocking. The cache hierarchy includes 192 KB L1 per core, 3 MB L2 per core, and 36 MB shared L3. Memory support is DDR5 over a dual-channel bus with 102.4 GB/s bandwidth, and ECC memory is supported. The CPU provides PCIe Gen 5 with 20 lanes (CPU only) and includes integrated Arc Xe-LPG Graphics with 64 execution units.

The 96th percentile ranking and average score of 83807 put it in elite company. The nearest rivals are all EPYC server parts or a higher-tier Ultra 9, and the 285K trades blows with them within a 1.5% window. For multi-core workloads, the Cinebench R23 score of 42522 is indicative of very strong parallel performance, suitable for video rendering, scientific simulation, and heavy compilation tasks. The PassMark multi-thread score of 67260 corroborates this. The single-core performance is also excellent: Cinebench R23 single-core of 2377 and Geekbench single-core of 2870 show that lightly-threaded tasks like office productivity and legacy software will run with high responsiveness.

The 24 threads are all physical (no hyperthreading), which means each core is dedicated to a single thread. This can be advantageous for workloads that benefit from predictable, consistent core performance, but it may reduce peak throughput compared to a similarly sized CPU with simultaneous multithreading. The 17,800 million transistor count and 36 MB shared L3 cache are substantial, aiding in data-heavy workloads like database operations, which is reflected in the PassMark data compression score of 790052 and encryption score of 57745.

The integrated graphics are present but minimal — 64 execution units of Arc Xe-LPG — which is suitable for basic display output but not for gaming or compute. The ECC memory support is a notable feature for workstation reliability, and the dual-channel DDR5 with 102.4 GB/s bandwidth is standard for a desktop platform, though it is lower than what server parts with 8-channel memory would offer. The 20 PCIe Gen 5 lanes from the CPU are sufficient for a single high-end GPU and a couple of NVMe drives.

# Who Should Build It — target users and industries tied strictly to the measured performance

The 96th percentile CPU and 18th percentile GPU create a specific profile. This system is for users whose primary workload is CPU-bound and who need occasional graphics output, but not high-end GPU acceleration. Content creators working with video encoding or batch image processing would benefit from the Cinebench R23 multi-core score of 42522 and PassMark floating-point math of 224324, which indicate rapid processing of large datasets. Software developers compiling large codebases would see fast build times from the multi-thread score of 67260, and the single-thread score of 5087 in PassMark ensures snappy UI responsiveness during development.

Students in computer science or engineering fields would find value in the CPU's 96th percentile performance for running simulations, compiling projects, and handling virtual machines, while the GPU's 18th percentile is adequate for basic graphical output in IDEs and documentation tools. Small business workstations handling accounting, spreadsheets, and database operations would benefit from the data compression score of 790052 and data encryption score of 57745, which speed up data manipulation and secure transactions.

Gamers at 1080p or 1440p with low graphics settings might find the system usable, but the GPU's 18th percentile means frame rates will be low in modern titles. The 50 RT cores and 200 tensor cores are present, but the low benchmark score suggests they are not performing to the level of the CPU. This is not a gaming-first build; it is a compute-first build with a GPU that is merely present for display output and light acceleration. For 3D rendering or GPU-accelerated machine learning, the system would be severely underpowered on the GPU side, and users in those fields should look elsewhere.

# Balance and Bottleneck — which component limits which workload, using percentiles and FPS scaling as evidence

The imbalance between the 96th percentile CPU and 18th percentile GPU is stark. In CPU-bound workloads — such as code compilation, data compression, or physics calculations — the Core Ultra 9 285K will operate at its full potential. The PassMark integer math score of 172379 and multi-thread score of 67260 are strong, and with no GPU involvement, the system performs at the top 4% level.

In GPU-bound workloads such as gaming or 3D rendering, the RTX 5070 SUPER's 18th percentile becomes the limiting factor. The 3DMark Steel Nomad score of 2690 places it in the company of the GeForce GT 1030 and GT 440, which are entry-level parts. Even if the CPU is capable of feeding the GPU with data at high rates, the GPU cannot render frames quickly enough to leverage the CPU's power. The FPS scaling will be poor: at lower resolutions like 1080p, the CPU can prepare frames fast, but the GPU will still take the same amount of time to rasterize them, so frame rates will be limited by the GPU's fill rate of 201.0 GPixel/s and texture rate of 502.4 GTexel/s.

The bottleneck is clear: for any graphics-intensive task, the GPU is the constraint. The 18 GB of VRAM and 672.0 GB/s bandwidth are generous, but the raw compute of 32.15 TFLOPS FP32 is not sufficient to drive high frame rates in modern games. For CPU-bound tasks, the system is well-balanced — the CPU is the star, and the GPU is idle. For a balanced system, the GPU would need to be several tiers higher to match the CPU's percentile, but the data shows no such pairing.

# Upgrade Path and Platform — socket, memory support, PCIe, PSU headroom from suggestedPsu/tdp, what a sensible next upgrade looks like

The platform is based on the Intel Socket 1851, which is the current socket for Arrow Lake processors. The CPU supports DDR5 memory over a dual-channel bus, with ECC support as a bonus for workstation reliability. The memory bandwidth is 102.4 GB/s, which is standard for dual-channel DDR5. The CPU provides PCIe Gen 5 with 20 lanes, allowing for a high-speed GPU or NVMe storage. The TDP is 125 W for the CPU, and the GPU has a TDP of 275 W, with a single 16-pin power connector. The suggested PSU is not provided in the data, so PSU headroom cannot be quantified, but the combined TDP of 400 W is a consideration for PSU sizing.

A sensible next upgrade would be to replace the GPU with a higher-performing model that better matches the CPU's 96th percentile. The current GPU's 18th percentile is the weakest link, and upgrading to a GPU with a higher 3DMark score would balance the system. The PCIe Gen 5 x16 bus is future-proof and will not bottleneck any modern GPU. The 18 GB of GDDR7 memory on the current GPU is ample, but the low compute performance suggests that a GPU with higher FP32 throughput and a better benchmark score would be the priority.

Alternatively, adding more memory or faster storage would not address the bottleneck — the CPU and memory subsystem are already top-tier. The ECC memory support means that users can opt for error-correcting DDR5 if their workloads require it. The dual-channel memory bus is a limitation for memory-bandwidth-hungry tasks, but for most workloads, the 102.4 GB/s is sufficient. The upgrade path is clear: keep the CPU, replace the GPU.

# Usage Scenarios — grounded in the scores

High-refresh gaming: The RTX 5070 SUPER scores 2690 in 3DMark Steel Nomad, at the 18th percentile. This is far below what is needed for high-refresh gaming (144 Hz or higher). Frame rates will be low even at 1080p, and the system is not suitable for competitive gaming. The CPU's single-thread score of 5087 in PassMark and 2870 in Geekbench ensures that the CPU will not be the bottleneck at low resolutions, but the GPU will cap FPS well below high-refresh targets.

Streaming: Streaming involves both CPU encoding (via x264) and GPU rendering. The CPU's multi-thread score of 67260 in PassMark and Cinebench R23 score of 42522 are excellent for software encoding, which can produce high-quality streams. The GPU's 18th percentile will limit the game's frame rate, but the streaming encoder can keep up. The 18 GB of VRAM is ample for game capture, and the 50 RT cores could handle any streaming overlays.

Video editing: Video editing is a mix of CPU and GPU tasks. The CPU's Cinebench R23 multi-core score of 42522 and PassMark floating-point math of 224324 will handle timeline scrubbing and effect processing well. The GPU's 18th percentile will slow down GPU-accelerated effects and encoding — the 32.15 TFLOPS FP32 is moderate, but the low benchmark score suggests it will not accelerate renders significantly. For 4K editing, the system is CPU-limited, which is a positive.

3D rendering: In 3D rendering, the GPU is critical for viewport performance and final renders. The RTX 5070 SUPER's 3DMark score of 2690 at the 18th percentile is poor, meaning viewport interaction will be sluggish and GPU-based renderers (like Blender Cycles or Octane) will be slow. The 50 RT cores are present, but their performance is not reflected in the benchmark score. CPU-based rendering would be better, leveraging the 42522 Cinebench R23 score, but this is rare in modern workflows.

Software development: This is where the system shines. The CPU's 96th percentile and multi-thread score of 67260 will compile code quickly, and the single-thread score of 5087 ensures fast IDE responses. The 36 MB L3 cache and 24 cores handle parallel builds efficiently. The GPU is irrelevant for most development tasks, and the system is well-suited for developers working on large codebases or running multiple virtual machines.

Student and office work: For general productivity, the system is overkill but effective. The CPU's single-thread performance makes document editing, web browsing, and spreadsheet work instantaneous. The GPU is more than enough for 2D desktop rendering and office applications. The 18 GB of VRAM is unused, but the system will handle multitasking with ease, thanks to the 24 cores and 36 MB L3 cache.

# FAQ

Q: What is the CPU's percentile ranking among all CPUs?

A: The Intel Core Ultra 9 285K is at the 96th percentile, with an average benchmark score of 83807.

Q: How does the GPU compare to its nearest rivals in the 3DMark Steel Nomad test?

A: The RTX 5070 SUPER scores 2690, which is 1% higher than the NVIDIA Quadro K1100M (2664), 1.1% higher than the GeForce GT 1030 (2662), and 1.1% higher than the Intel Arc Pro B50 (2660).

Q: What is the memory configuration of the RTX 5070 SUPER?

A: It has 18 GB of GDDR7 memory on a 192-bit bus, with a bandwidth of 672.0 GB/s and an effective memory clock of 28 Gbps.

Q: Does the CPU support ECC memory?

A: Yes, the Intel Core Ultra 9 285K supports ECC memory, which is notable for workstation reliability.

Q: What are the Cinebench R23 scores for the CPU?

A: The multi-core score is 42522 and the single-core score is 2377.

Q: What is the combined percentile for this CPU+GPU pairing?

A: The combined percentile is 57th, reflecting the significant gap between the CPU's 96th percentile and the GPU's 18th percentile.

Q: What is the GPU's FP32 performance?

A: The RTX 5070 SUPER has an FP32 throughput of 32.15 TFLOPS, with FP16 also at 32.15 TFLOPS (1:1 ratio).

# Gaming Performance

No measured FPS data exists for this exact CPU+GPU combination — the FACT PACK contains no measuredFps rows. All FPS figures below are estimates derived from the benchmark scores.

The RTX 5070 SUPER's 3DMark Steel Nomad score of 2690 places it at the 18th percentile among all GPUs. This is a very low position, and the nearest rivals (GT 1030, GT 440) are entry-level parts. For gaming, this translates to low frame rates across all modern titles. At 1080p with ultra settings, the GPU would struggle to maintain 30 FPS in most AAA games, and esports titles might reach 60 FPS only at lower settings. The CPU's single-thread score of 5087 in PassMark and 2870 in Geekbench ensures that the CPU is not the bottleneck at 1080p — the GPU is the sole limiter.

At 1440p, the situation worsens. The GPU's 201.0 GPixel/s pixel rate and 502.4 GTexel/s texture rate are insufficient for high-detail rendering at this resolution. Frame rates would drop below 30 FPS in most titles, and the 18 GB of VRAM is largely unused because the GPU cannot push enough pixels to fill it. At 4K, the GPU is effectively unusable for gaming — the 18th percentile score means it is not designed for high-resolution rendering.

The 50 RT cores and 200 tensor cores are present, but their performance is not validated by the benchmark score. Ray tracing would likely cause severe frame rate drops, and DLSS upscaling (using tensor cores) might help, but the base rendering performance is too low to start from. The 672.0 GB/s bandwidth is generous, but it is wasted on a GPU that cannot process data fast enough. Gaming with this system is not recommended; the CPU's power is irrelevant when the GPU cannot keep up.

# Build Overview

This is a desktop build combining the Intel Core Ultra 9 285K CPU and the NVIDIA GeForce RTX 5070 SUPER GPU. The CPU is a 24-core, 24-thread Arrow Lake processor on the Intel Socket 1851 platform, with a 96th percentile ranking and an average benchmark score of 83807. The GPU is a Blackwell 2.0 part with 18 GB of GDDR7 memory, 6400 shading units, and 50 RT cores, but it ranks at the 18th percentile with a 3DMark Steel Nomad score of 2690.

The combined percentile is 57th, which accurately reflects the system's character: it is a top-tier CPU paired with a low-tier GPU. The CPU is in the top 4% of all processors, capable of handling the most demanding compute workloads, while the GPU is in the bottom 18%, suitable only for basic display output and light acceleration. The system is a workstation-class machine for CPU-centric tasks, not a gaming or graphics workstation. The 125 W CPU TDP and 275 W GPU TDP indicate a moderate power draw, and the PCIe Gen 5 x16 bus is future-proof. The overall tier is high for compute, low for graphics, and the system should be viewed as a CPU-first build.

Rate My PC: Intel Core Ultra 9 285K + NVIDIA GeForce RTX 5070 SUPER

Power Build

System Balance Analysis

Intel Core Ultra 9 285K

NVIDIA GeForce RTX 5070 SUPER

Market Position

Game Performance Benchmarks

Cyberpunk 2077

Red Dead Redemption 2

Microsoft Flight Simulator

Control

Call of Duty: Warzone

Fortnite

Performance Insights

Strong Performance

Performance Tiers Explained

Ultimate

High-End

Mid-Range

Entry Level

Legacy

4K Gaming Performance

Cyberpunk 2077

Red Dead Redemption 2

Microsoft Flight Simulator

Control

Call of Duty: Warzone

Fortnite

1440p Gaming Performance

Cyberpunk 2077

Red Dead Redemption 2

Microsoft Flight Simulator

Control

Call of Duty: Warzone

Fortnite

1080p Gaming Performance

Cyberpunk 2077

Red Dead Redemption 2

Microsoft Flight Simulator

Control

Call of Duty: Warzone

Fortnite

Related Processors

Related Graphics Cards

Compare Your CPU

Compare Your GPU

Related PC Builds

Same CPU with Different GPUs

Same GPU with Different CPUs

Other Popular Builds