Thief

# Thief on Intel Core i5-9400 + NVIDIA GeForce GTX 1070

Thief, the stealth classic released in 2014, runs on the tested combination of Intel Core i5-9400 and NVIDIA GeForce GTX 1070 across four settings (Low, Medium, High, Ultra) at three resolutions (1920x1080, 2560x1440, 3840x2160). The measured frame rates show a wide spread—from a peak of 119 FPS at 1080p Low to a floor of 28 FPS at 4K Ultra. The combo ranks 3116th out of 3720 tested combos for this game, placing it in the lower portion of the database. Both the CPU and GPU sit at the 47th percentile of all tested parts, suggesting a balanced mid-range pairing, yet the game itself appears to heavily favor the GPU—the FPS scaling directly tracks resolution changes more than settings changes within each resolution.

Resolution Scaling

When resolution steps from 1080p to 1440p and then to 4K, average FPS declines consistently across every settings preset. At Low, the averages are 119 (1080p), 88 (1440p), and 51 (4K). Medium shows 97, 66, and 39; High reaches 74, 51, and 30; Ultra dips to 70, 48, and 28. The gap between 1080p and 4K is roughly double or more: Low goes from 119 down to 51, Medium from 97 to 39, High from 74 to 30, and Ultra from 70 to 28. In each case, going from 1080p to 1440p removes about 25-40 FPS, and the move from 1440p to 4K removes another 20-25 FPS.

This strong resolution dependence points to the GPU as the primary limiting factor. The CPU’s load stays roughly the same when the resolution changes—the scene’s geometry and logic remain identical—while rendering work rises with pixel count. The GTX 1070 has a high fill-rate ceiling (107.7 GPixel/s) and memory bandwidth of 256.3 GB/s, so the raw capability of the GPU defines the framerate. Consider that at 4K even the lowest settings returns only 51 FPS, while 1080p Low hits 119. The difference of over 80 FPS directly from pixel count reduction confirms that the i5-9400 is able to feed the game well enough that shrinking the rendering workload hands back essentially all performance headroom.

At 1440p, the combination is still solidly playable: High settings average 51 FPS, Ultra 48, Medium 66, and Low 88. But only at 1080p does every setting stay above 70 FPS with Ultra at 70, High at 74, Medium at 97, and Low at 119. So the data suggests the combo is best at 1080p, reasonably smooth at 1440p for most settings, and only viable at 4K with Low or Medium (51 and 39 FPS respectively). The 4K row with Ultra yields 28 FPS—which not most games would consider acceptable.

How This Combo Ranks

In the overall database, the i5-9400 + GTX 1070 combination is ranked 1,116th out of 3,720 tested combos for Thief. That means it sits in the highest 30% of all tested pairs. But note the context: Thief is an older title (2014) with modest requirements, and the database includes many modern GPUs that will dominate. To put the rank in perspective, the GTX 1070’s own percentile is 47% of all GPUs—meaning roughly in the bottom half of the GPUs tested in the database. Its measured point produces a rank that is below the median candidate set.

The CPU also holds the 47th percentile across all CPUs. Thus each component individually is near the middle of its group. The combined rank of 1,177/3,720 means the pairing performs slightly worse than the sum of its parts might suggest. This is likely because Thief benefits more from GPU raw power than from CPU threads. However, the rank itself is still within the play field of most software: 4 out of 5 combos tested are better, but the performance at 1080p (High 74, Ultra 70) is perfectly acceptable for a mechanical stealth game.

Core i5-9400’s average benchmark score across all treats is 2254, and its nearest rivals are like the Intel Core i7-1068NG7 (2259, delta −0.2%), Intel Core i7-10710U (2248, +0.3%), and AMD Ryzen 5 PRO 1500 (2246, +0.4%). These are all within a rounding percent, meaning CPU performance is effectively identical to those parts. The generous CPU headroom shown by the resolution scaling—stable FPS at 1080p across settings—confirms the CPU isn’t a bottleneck for this game.

GPU Role

The NVIDIA GeForce GTX 1070 is the centerpiece here: it’s from the Pascal generation, built on 16nm TSMC, with 7,200 million transistors on a 314 mm² die. Its boost clock is 1683 MHz (base 1506), memory runs at 2002 MHz (8 Gbps effective) on a 256-bit GDDR5 bus that provides 256.3 GB/s bandwidth. There are 1920 shading units, 120 TMUs, 64 ROPs, and peak pixel throughput of 107.7 GPixel/s. The GPU’s own average benchmark score is 9780, placing it at the 47th percentile of all GPUs.

In the measured FPS rows, the memory bandwidth and clock combination directly shapes how each resolution scales. At 1080p, the GPU produces framerates between 70 (Ultra) and 119 (Low). At 4K, the same cards deliver 28–51 FPS. The GTX 1070’s 8GB VRAM is plenty for Thief, which released in 2014 and likely never needs more than a 2GB of buffer; the limiter is raw pixel fill and shader power. Also note that the GPU’s PassMark G3D score is 13498, while its direct competitors in the database (FirePro W5000, Quadro M2000M, Tesla M10, Tesla C2070) all have average scores between 9716 and 9832, within 0.7% of the GTX 1070’s 9780. So in absolute terms, the GTX 1070 is a mid-range card, which is consistent with its 47th percentile.

In game, the data shows that Medium and High settings cause about a 20–25% framerate hit at each resolution compared with Low, but the Ultra settings is even heavier. For a 60+ frame target, only 1080p and 1440p Medium/High reach that bar; 4K never reaches it except Low.

FAQ

Q: What is the fastest measured setting at 1080p?

A: At 1920x1080, the Low preset delivers 119 FPS average, followed by Medium at 97, High at 74, and Ultra at 70.

Q: Can the combination handle 4K UHD?

A: According to the data, at 3840x2160: Low yields 51 FPS, Medium 39 FPS (with 33 min and 45 max), High 30 FPS, and Ultra 28 FPS—only Low and Medium break 30 FPS, and none reach 60.

Q: What are the best framerate samples in the dataset?

A: The two highest averages are 119 FPS at 1080p Low and 97 FPS at 1080p Medium. The lowest are 28 FPS at 4K Ultra and 30 FPS at 4K High.

Q: How does the GPU compare to its nearest rivals by benchmark average?

A: The GTX 1070 has an average benchmark score of 9780; the AMD FirePro W5000 scores 9803 (−0.2%), NVIDIA Quadro M2000M 9832 (−0.5%), NVIDIA Tesla M10 9724 (+0.6%), and Tesla C2070 9716 (+0.7%), all within 1%, putting the card right in the middle of that pack.

Q: What is the i5-9400’s original launch status?

A: The CPU’s launch MSRP field is null, so that information is not recorded. The CPU is marked end-of-life and its successors, with its percentile among all processors at 47%.

Q: What is the characteristic of the CPU's R18 Multi-Core performance?

A: The Core i5-9400 scores 668 in Cinebench R15 Multi-Core and 2,784 in R20 Multi-Core, and its R23 Multi-Core score is 6,629. These numbers place it only a few tenths of a percent above the average of its nearest rivals—it doesn’t stand out positively or negatively.

Settings Recommendations

Looking at the measured rows across resolutions, the choice of setting depends on the target display resolution. At 1080p, the High preset returns 74 FPS and Ultra returns 70 FPS, both very close; in this context, High is a sensible choice because Ultra’s extra visual detail costs roughly four frames per second (74 vs 70) and 70 is still highly playable. If you want a higher headroom, Medium gives 97 FPS.

At 1440p, only Medium reaches the smoothness at 66 FPS (min 56, max 76). High dips to 51 FPS, Ultra to 48, would still be playable for many; Low is 88 FPS. If you consider 66 FPS acceptable, Medium is the best trade-off between visuals and frame rate, but for a more hearty margin, Low at 88 FPS is the best option.

At 4K, the average 1080p results drop off a cliff. Only Low stays above 45 FPS—Low averages 51 FPS. Medium averages 39, so only Low is reasonable for a sustained experience. Ultra and High sit at 28 and 30 FPS—likely too poor for tactical stealth where reaction time matters. Therefore, the recommendation is to use Low when running 4K, or else drop to a lower resolution. For most users, 1440p Medium or 1080p High will provide a good balance.

CPU Role

The Intel Core i5-9400 is a 6-core, 6-thread processor base speeds of 2.9 GHz and a 4.1-boost, built on the Coffee Lake architecture and released in October 2018. It fails to unlock the multiplier, cooling is specified 65W TDP, and it supports only a two-channel DDR4 bus with 42.7 GB/s buffer. Its L3 cache is a shared 9 MB, and directories are per core: 64 KB L1 per core and 256 KB L2 per core. In current benchmarks, it scores 935 in Cinebench R23 Single-Core and 6,629 in Multi-Core. The average general score across benchmarks is 2254, giving it a CPU percentile of 47%—so half of all tested CPUs are faster on aggregate.

This CPU does not appear to bottleneck the GTX 1070 in Thief. The evidence lies in the resolution scaling: when res drops from 1080p to 4K, FPS drops by roughly half at all settings, which is what you expect if the CPU remains ahead while GPU workload intensifies. The i5-9400’s single-core speed (cs). Since its cinebench single-core is 935, it has enough to handle the game’s main thread. Also, the game is from 2014, before the era of many-CPU–intensive titles, so 6 cores and 6 threads are generous. The measured FPS at 1080p across all settings stays above 70, indicating that CPU and GPU load are both in sync; if the CPU were a bottleneck, we would see a flatter framerate curve when going from Low to Ultra that resolution. The actual difference at 1080p between Ultra (70) and Low (119) is 49—which is much larger than what a weak CPU would allow.

Measured FPS Breakdown

The dataset shows 12 measurement rows across three resolutions and four settings. We list the exact averages:

• 1920x1080–Low: 119 FPS
• 1920x1080–Medium: 97 FPS (min 82, max 111)
• 1920x1080–High: 74 FPS
• 1920x1080–Ultra: 70 FPS

• 2560x1440–Low: 88 FPS
• 2560x1440–Medium: 66 FPS (min 56, max 76)
• 2560x1440–High: 51 FPS
• 2560x1440–Ultra: 48 FPS

• 3840x2160–Low: 51 FPS
• 3840x2160–Medium: 39 FPS (min 33, max 45)
• 3840x2160–High: 30 FPS
• 3840x2160–Ultra: 28 FPS

The minimum and maximum columns are only listed for the Medium setting at each resolution—the only set that captured them. For instance, at 1080p Medium the range is 82 to 111, at 1440p Medium it’s 56 to 76, and at 4K Medium from 33 to 45. These numbers show a spread of about 11–20 frames, which is typical for a GPU-bound title at this tier.

The biggest absolute difference observed is 91 frames—between the highest (119 on 1080p Low) and the lowest (28 on 4K Ultra). The combination of selecting 1080p Low versus 4K Ultra yields an improvement of 91 FPS. Within each resolution, the maximum delta between settings is: at 1080p, 49 FPS (119 vs 70); at 1440p, 40 (88 vs 48); at 4K, 23 (51 vs 28). The consistent scaling across resolution confirms that the game’s framerate is largely responsible for the pixel workload, and settings merely scale within the same pattern.

At Medium settings, the frametime is minimal (a 33–45 FPS average at 4K) but still uneven. Meanwhile, for the High and Ultra settings, the delta is trivial—typically 2–4 FPS apart, meaning Ultra is nearly as efficient as High for this mid-range card. The measured average for the row confirms 4K Ultra delivers 28 FPS, which is just under the common 30 FPS floor for a playable experience, though the average is borderline. For 1440p, the High settings gives 51 FPS, a comfortable experience for most gamers.