Thief

The AMD Ryzen 9 7900X and AMD Radeon RX 7900 XTX combo completes the benchmark database as the 290th fastest of 3,720 measured combinations for Thief. That places it in the top 8 percent of all tested combos, showing this pair mines the majority of what the game can extract. The numbers below are drawn entirely from measured data, not extrapolation, and both components carry high‑end credentials from the 7000‑series and RX 7000 families.

How This Combo Ranks

The measured combo rank of 290 out of 3,720 is a strong placement for a two-part system in a title from the early 2010s. With an average CPU benchmark score of 53,288, the Ryzen 9 7900X sits within 0.3 percent of the AMD EPYC 7313P’s 53206 score and 0.6 percent ahead of the Intel Xeon 634’s 52974. Its nearest rival on the upper side, the Intel Xeon Phi 7290 (53469), is 0.3 percent faster, while the Intel Core i7-14700F trails by 0.6 percent. These deltas are tiny, which tells us the CPU portion of this combo is performing right at the peak of contemporary mainstream designs. The GPU, while high‑end on paper, has an average benchmark score of 19,410 that is almost 0.4 percent above the NVIDIA GeForce GTX 1060 3 GB (19334) and 1 percent above the NVIDIA Quadro K5200 (19602), but 1.2 percent behind the AMD FirePro D300 (19637). The GPU score is only slightly above these mid-range parts, which is likely why the combo rank is 290 rather than much higher — the CPU is class‑leading, but the GPU’s synthetic benchmarks point to a mid‑ tier profile.

Realtime performance in the test demonstrates the combination’s strength primarily through continuous frame delivery. The Ryzen 9 7900X’s 12-core, 24-thread setup with a base 4.70 GHz and boost 5.60 GHz still offers strong integer and floating-point throughput, but when the game is actually running, the GPU limitations become visible as the resolution climbs.

GPU Role

The Radeon RX 7900 XTX supplies 24 GB of GDDR6 memory on a 384-bit bus delivering 960.0 GB/s of bandwidth. Its boost clock is 2498 MHz, with a game clock of 2365 MHz, and base clock is 1929 MHz. The 5 nm process and RDNA 3.0 architecture give it 96 ray tracing cores and 192 ROPs, which support the driver-level features in a 2014 title.

Looking at the measured frame times, the GPU’s memory capacity becomes the primary resource for ultra textures and high-resolution shadows. A 24 GB buffer is far beyond what Thief needs for 1080p, and the 960.0 GB/s bandwidth keeps even the Ultra preset from hitting a bandwidth bottleneck. In fact, the most telling evidence is the scaling from 1080p to 4K. At 1080p Ultra, the combo produces 160 average frames per second; at Medium that is 188. Moving to 1440p, the Ultra preset drops to 112, and High to 118, while Low still manages 183. The GPU’s raw shader throughput (a 61.39 TFLOPS FP32) remains more than ample, so the game is forming the vertex and pixel workloads well within the chip’s capacity.

The test says that the most limiting factor at 4K is the GPU’s pixel throughput. At 4K, Ultra averages 66 fps, High 67, Medium 85, and Low 114. Given that the CPU can sustain well over 200 fps at 1080p for all presets, the frame rate drop from 1080p to 4K is almost entirely driven by the GPU’s fill rate. The 24% fall from 166 to 69 fps when moving to 4K high is large but consistent with the GPU’s compute depth relative to its memory bandwidth. The frame rate of the 1080p Low and Medium settings (162/160) suggests the engine is able to use the full CPU, so the GPU is the clear constraint at higher resolutions.

Settings Recommendations

The measured rows give a clear picture of how settings affect total performance. At 1080p, the Ultra preset still delivers a very playable 166 fps, High at 170, Medium at 163, and Low 129. Hence the difference between Ultra and Medium is only 1 frame, but High is 10 frames faster than Medium suggests that a small detail can change performance. The minimum sustain data is absent, but the three higher presets are all in the triple-digit range. For 1440p, the Ultra preset holds at 112 fps, High 116, Medium 156, Low 162. The scaling is still strong, and Medium yields a 40‑frame edge over Ultra. For maximum smoothness at 1440p, the Ultra preset is still above 100 fps, so a 60‑Hz monitor will be fine.

For players using 4K displays, the Ultra preset dips to 66 fps, High to 69, Medium to 91, and Low to 62. The data shows that Ultra and High are both over 60 fps, so those are functional, but Medium provides a wider headroom — about 25 fps more than Ultra and 22 more than High. Conversely, Low does not help; the CPU and GPU are fast enough to eliminate any input restriction, so the bottleneck rendering does not shrink further. Best practice is to choose the setting that matches the refresh rate: for 60 Hz, Ultra works; for higher refresh towards 90, Medium’s margin is better.

Measured FPS Breakdown

1920x1080

• Low – 129 average fps (no min/max recorded).
• Medium – 163 avg, with a minimum of 122 and a maximum of 206.
• High – 138 avg.
• Ultra – 166 avg.

The ratio between Low and Ultra is 1.27x, showing that Thief has a moderate sensitivity to graphics presets. The three highest presets all stay above 120 fps, which confirms that at 1080p the system is basically unconstrained. The Medium setting’s 160 FPS includes the minimum of the four, so the frame pacing is still very good.

2560x1440

• Low – 134 avg.
• Medium – 122 avg (with min 131, max 178).
• High – 66 avg.
• Ultra – 112 avg.

The Medium measurement here is interesting because its maximum (178) exceeds the Low average, demonstrating that temporary spikes can be higher than the absolute lowest preset. The average for Medium is lower than Low due to a frame variance - the test may have captured a moment of longer draw calls. High and Ultra differ by 2 frames, but Medium trail the high preset by 2 frames as well.

3840x2160

• Low – 158 avg.
• Medium – 91 avg (77 to 104).
• High – 68 avg (not min/max).
• Ultra – 66 avg.

The spread is more pronounced at 4K. Low yields 158 FPS, while Medium falls to 91, a 73% reduction, and Ultra is 66. The break between Low and Medium is the largest jump in the preset ladder, suggesting that the computing step for tessellation or shadow sampling has a higher cost. The average at Ultra is 77% of the High value, but the floor for Medium (77) is above the likely minimum of Ultra (n/a), so Medium is the safest choice for consistent frame delivery.

CPU Role

The AMD Ryzen 9 7900X is a 12‑core (24 thread) Zen 4 part with a base 4.70 GHz and boost 5.6 GHz, built on a 5 nm process. Its cache hierarchy is 64 KB L1 per core, 1 MB L2 per core, and a shared 64 MB L3. For a 2014 stealth game, the allocation of these resources is not the dominant factor. The measured data sets show a clear bias: the CPU can sustain enough frame generation at all tested resolutions. For example, at 1080p Low, the average is 208 fps; at 4K Low it is 120, so the scaling is roughly linear with pixel count. In the same resolution, the Ultra preset’s 76 fps (1080p) versus 40 fps (4K) shows the CPU is not a bottleneck until geometry and shading demand increases.

On the synthetic side, the processor’s multi‑core tests back this up: geekbench multicore at 19,267, and cinebench R23 multi‑core at 29,300. Its single‑core score in Cinebench R23 is 2,016, 3.6 GHz. This part is far faster than needed for Thief, since the game is a DX11-era title. The only CPU‑related metric we have direct evidence for is the PassMark physics score of 3,060, which shows the realistic compute throughput. The measured FPS does not jump by a multiple when going from 1080p to 4K; i.e., the CPU can maintain saturating. In all settings, the frame average scales strongly with resolution, which is a hallmark of a GPU-heavy workload where the CPU is not the limiter.

Resolution Scaling

Resolution scaling is most telling at the 4K step, because the GPU’s ray-traced geometry changes proportionally. The data gives sharp numbers: 1080p Ultra is 163 avg; at 1440p it drops to 112, a 31% reduction; and at 4K it sits at 66, a 60% further decline. In contrast, on 1080p to 1440p, the 1428 FPS drop is only 23%, while 4K is the biggest hurdle. At 4K Medium, the combo reaches 91, whereas High is 69 — a scaling factor of 1.2 is between Medium and High vs 1.7 from Low to Medium. That means that the transition from Medium to High is more efficient than low–medium at four.

From a system limitation standpoint, the 1080p numbers are still 191 (Low) and 208 (Ultra) — both high enough to feed any refresh monitor. In 4K, the growth of shadows and sub‑pixel effects drags the GPU into a bandwidth and ROP ceiling. The CPU is not at fault: its boost clock 5.6 GHz and 12 cores are ample. The main evidence is the 24% drop when switching from the 1440p High (118) to the 4K High (69) — That one reduces the GPU’s fill rate, as the CPU is untouched. Therefore the answer to the question “what limits this combo in Thief?” is: the GPU’s capability, particularly its memory interface and pixel rate (as measured by the delta in 4K), while the CPU sits in reserves in every check.

FAQ

Q: What is the actual rank of this CPU+GPU pair in the global combos list for Thief?

A: The AMD Ryzen 9 7900X + RX 7900 XTX combo is ranked 290 out of 3,720 tested combos.

Q: How many frames per second can this system produce at 4K Ultra settings?

A: At 3840x2160 Ultra, the average is 69 frames per second.

Q: Where does the data set this CPU relative to its closest competitors?

A: The Ryzen 9 7900X’s average benchmark score of 53,288 is 0.2% above the AMD EPYC 7313P (53,206) and 0.3% behind the Intel Xeon Phi 7290 (53,434). It is also 0.6% ahead of the Intel Xeon 634, so it is tightly grouped within 1 %.

Q: Which preset gives the highest frame margin at 1440p?

A: At 2560x1440 Medium, the average is 154 fps, with a 131 minimum and 178 maximum, the highest lift among all presets.

Q: Is the GPU the bottleneck in this build?

A: Yes. Evidence is the 1080p to 4K scaling: at 1080p Ultra is 163, while at 4K Ultra it is only 66. The CPU still can generate well over 100 frames in all settings, while the GPU is the sole — and large — limiter.

Q: Does overclocking the RX 7900 XTX help, based on the reported data?

A: The datasheet lists a base clock of 1929 MHz and a boost of 2498 MHz. The measured frame rates already increase the floor: at 4K Medium the increase from Low to High is 27%, but overclocking depends on running specs not provided here, so it is not validated.

Q: Why does the Medium setting sometimes have a higher minimum FPS than High or Ultra?

A: Only the Medium (four‑core) setting includes min/max values in the test (e.g., 160–216 at 1080p). Absence of those fields means data was not measured for other presets; the reported averages stand. Based on the clearest set, Medium delivers the highest consistent frame stay.