Arrow Lake CPU up to 18% faster than Core i9-14900K at 250W — Ryzen 9 9950X still faster

Raptor Lake
Raptor Lake (Image credit: Intel)

Hardware leaker Jaykihn has shared various benchmarks for an Arrow Lake-S processor tested at 250W. According to the leaked results, the Arrow Lake-S chip was up to 18% faster than the Core i9-14900K, currently one of the best CPUs around.

The leaker published a benchmark table featuring two Arrow Lake processors and the Core i9-14900K. The two Arrow Lake chips included were an engineering sample and a qualification sample; however, CPU specs and model names were completely absent. That said, it is highly probable both chips are pre-production variants of the flagship Core Ultra 9 285K, which purportedly comes with 24 cores. The three chips reportedly ran at 250W. As always, we recommend treating leaked benchmarks with a pinch of salt.

For simplicity's sake, we are primarily looking at the Arrow Lake-S qualification sample since that chip is much faster than the engineering sample and is a closer representative of the final product.

Swipe to scroll horizontally
BenchmarksArrow Lake-S Qualification SampleCore i9-14900KPerformance Margin
CrossMark2,5872,432Arrow Lake-S QS 6% Faster
WebXPRT4 3.73 Chrome v114372388Core i9-14900K 4% Faster
Speedometer 2.1 Chrome v114472521Core i9-14900K 10% Faster
Geekbench 5.4.5 Single Core2,4552,432Arrow Lake-S QS 9% Faster
Geekbech 5.4.5 Multi Core27,38123,902Arrow Lake-S QS 14% Faster
Cinebench R2343,11836,681Arrow Lake-S QS 18% Faster

Benchmark results are mixed. CrossMark, Geekbench 5.4.5, and Cinebench R23 demonstrate faster performance from the Arrow Lake-S chip than the Core i9-14900K. However, WebXPRT4 and Speedometer (both running in Chrome) show more favorable results for the Core i9-14900K.

The Arrow Lake-S qualification sample was 6% faster than the Core i9-14900K in CrossMark, with a score of 2,587 versus the Core i9-14900K's 2,432. In Geekbench, the Arrow Lake chip pulls away from the Core i9-14900K, 9% faster in the single-core benchmark and 14% faster in the multi-core benchmark. Cinebench R23 showed the most favorable results for the Arrow Lake-S QS chip, with an 18% performance lead ahead of the Core i9-14900K.

Conversely, the Core i9-14900K was faster in the Chrome-based tests. WebXPRT4 shows a 4% advantage for the Raptor Lake Refresh flagship, and Speedometer also indicates a 10% lead for the Core i9-14900K.

Harukaze5719 pulled Jaykihn's Arrow Lake-S QS Cinebench R23 results and put them together with various Ryzen 9 9950X and Ryzen 7 7950X results to see where Zen 5 and Zen 4 lands against Intel's next-generation desktop architecture.

(Image credit: Harukaze5719/X)

The chart revealed that the Arrow Lake part is mainly in the same alley as the Ryzen 9 9950X in Cinebench R23, but only at specific power targets. The Arrow Lake chip's score in Cinebnech resembles a Ryzen 9 9950X result taken from the AnandTech forums operating at 160W, with the Arrow Lake chip outperforming the Ryzen 9 9950X (at 160W) by 1%.

However, bumping up the Ryzen 9 9950X to 230W purportedly nullifies the Intel chips performance lead, with the Ryzen 9 9950X at 230W outperforming the Arrow Lake qualification sample by 7%.

AMD's Ryzen 9000 processors are around the corner, while Intel's contending Arrow Lake chips are still a few months away. While we'll soon see what Zen 5 is capable of, a thorough comparison with Arrow Lake won't be possible outside of leaked benchmarks in the meantime.

Aaron Klotz
Contributing Writer

Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.

  • bit_user
    Zen 5 vs. Arrow Lake should be interesting. I think I'll need to invest in some 🍿
    Reply
  • helper800
    Admin said:
    In Geekbench, the Arrow Lake chip pulls away from the Core i9-14900K, 9% faster in the single-core benchmark...
    I am almost certain the above is incorrect based on the scores shown above this message in the chart. The article needs to be revised to reflect reality of either a typo in the chart or incorrect math to show a 9% increase.

    The single core geekbench scores don't add up. 2455 vs 2432 is a 23 point increase. That means the qualification chip only tested ~0.945% faster not 9% faster...

    Considering these are early chips that means arrowlake is going to be very competitive with AMD 9000 in single threaded at least.
    Reply
  • greymaterial
    I would not trust new intel cpu tests when the chip is fresh, degradation appears in 14th gen which are not claimed to have oxidation issue and performance are impacted to keep stability. intel can sell their improvement as advertised on arrow lake chips 9 months later when they prove themselves there is no more significant degradation.
    Reply
  • bit_user
    greymaterial said:
    I would not trust new intel cpu tests when the chip is fresh, degradation appears in 14th gen which are not claimed to have oxidation issue and performance are impacted to keep stability. intel can sell their improvement as advertised on arrow lake chips 9 months later when they prove themselves there is no more significant degradation.
    My hope is that they will have learned their lesson from this experience, and do a better job of characterizing & accommodating for degradation in future generations. Also, I think they'll be much stricter about making sure motherboard vendors color inside the lines.

    However, I understand your concern and I probably wouldn't be first in line for an Arrow Lake, either. I'm not really the early adopter type, anyhow.
    Reply
  • Amdlova
    250w how do you turn this on?
    PASS
    Reply
  • helper800
    Amdlova said:
    250w how do you turn this on?
    PASS
    With a power switch usually located on the case.
    bit_user said:
    However, I understand your concern and I probably wouldn't be first in line for an Arrow Lake, either. I'm not really the early adopter type, anyhow.
    Some saying about a six foot pole also comes to mind, in other words, GL other people.
    Reply
  • Amdlova
    @helper800 that is a cheat from age of empires 2

    "You need 250w of power to win something".
    It's crazy!
    Reply
  • helper800
    Amdlova said:
    @helper800 that is a cheat from age of empires 2

    "You need 250w of power to win something".
    It's crazy!
    Wow, that's a blast from the past. I played the OG age of empires games and do not remember that code, however, you may need to give at least a little more of a hint to get that one LOL.
    Reply
  • Main_gano
    If Arrow Lake succed Zen5 by about 2 percent with about 44 % more power, than then it is relatively clear what tech is more advanced.
    There is a good reason, why I stick to AMD (without need to change something, as my current system is build in spring 2023). A littel advantage in performance for me not justify the troubles that comes with the risk of an unstable CPU and the advantage of NVIDIA above AMD for me not justify the risk of a melting power connector.
    Reply
  • Pei-chen
    Love the laser car reference.

    And AMD needs to stick that Ryzen 9000 CPU into more laptop this time
    Reply