Tuesday, July 2nd 2024
Intel "Arrow Lake-S" Engineering Sample Posts Over 25% 1T Perf Gain Over i9-13900K, Falls Behind in nT
An unnamed Intel Core Ultra "Arrow Lake-S" desktop processor engineering sample (ES) made it to the hands of someone willing to post its CPU-Z Bench screenshot. The processor allegedly scores a whopping 1143.2 points in the CPU-Bench single-thread benchmark; and 12922.4 points in the multithreaded benchmark. When compared with the internal Intel Core i9-13900K reference scores of CPU-Z, the single-thread benchmark score is a staggering 26.71% increase over that of the i9-13900K (902 points); while the multithreaded score is 22% lower.
Since we don't know which processor model this "Arrow Lake-S" ES is, we have no way of telling if it is the top SKU with its rumored 8P+16E core configuration, or a mid-tier Core i5 SKU with the expected 6P+8E configuration. The single-threaded test only loads one P-core, and here the IPC of one of the chip's "Lion Cove" P-cores is able to trounce one of the "Raptor Cove" P-cores of the i9-13900K reference score. You also have to understand that the Hyper-Threading plays no role in this thread. Where it could play a role is the multithreaded test. "Lion Cove" lacks HTT support unlike "Raptor Cove," and the i9-13900K is a 24-core/32-thread processor. It's important to note here, that "Arrow Lake" doesn't just have up to 8 "Lion Cove" P-cores, but also up to 16 "Skymont" E-cores that Intel claims to have achieved a massive IPC gain over its predecessor, bringing its IPC in the league of past-generation P-cores such as the "Raptor Cove" or "Golden Cove."
Source:
Wccftech
Since we don't know which processor model this "Arrow Lake-S" ES is, we have no way of telling if it is the top SKU with its rumored 8P+16E core configuration, or a mid-tier Core i5 SKU with the expected 6P+8E configuration. The single-threaded test only loads one P-core, and here the IPC of one of the chip's "Lion Cove" P-cores is able to trounce one of the "Raptor Cove" P-cores of the i9-13900K reference score. You also have to understand that the Hyper-Threading plays no role in this thread. Where it could play a role is the multithreaded test. "Lion Cove" lacks HTT support unlike "Raptor Cove," and the i9-13900K is a 24-core/32-thread processor. It's important to note here, that "Arrow Lake" doesn't just have up to 8 "Lion Cove" P-cores, but also up to 16 "Skymont" E-cores that Intel claims to have achieved a massive IPC gain over its predecessor, bringing its IPC in the league of past-generation P-cores such as the "Raptor Cove" or "Golden Cove."
72 Comments on Intel "Arrow Lake-S" Engineering Sample Posts Over 25% 1T Perf Gain Over i9-13900K, Falls Behind in nT
The 9800x3d will be a close match on nT though so will be interesting to see the pricing on arrow lake.
HT off I get about 10K MT - so roughly ~11.5 ratio. Doesn't seem like this is the 16E configuration. Would make sense if the ST is 25% faster for this to be i5. Probably an overclocked i5...
If the i5 can clock to this performance im definitely in for arrow lake.
I think also based on this chances are you can drop a P core multiplier to make more headroom to raise a E core multiplier if you really want more MT conversely. At least I would hope that works better than it does with 14th gen and really should be slightly better at that with the changes.
These will probably sort of suck in relative terms for decompression though unfortunately with changes to HT. It'll probably get a bit of performance hit.
"Arrow lake is a refresh of meteor lake" - someone who shouldn't do business in divination
which would be inline from what I've heard/read about a 50%+ improvement from the E core's.
I wasn't expecting so much from the single thread P cores though..
We will see :cool:
It is likely clocked to it's eyeballs but still -- shows great potential.
Granite Ridge vs Arrow Lake is gonna be an interesting match up.
Until we see some REAL world tests with REAL non-ES chips, it's all a bunch of confabulated hooey hogwash IMO
For example Zen 5 vs Zen 4 is supposed to be 19% faster but still at ~910 points: www.techpowerup.com/322906/amd-ryzen-9000-zen-5-single-thread-performance-at-5-80-ghz-found-19-over-zen-4?cp=2
Also other benchmarks show Zen 5 being 55% faster compared to 13900K:
www.tomshardware.com/pc-components/cpus/amds-new-zen-5-chip-up-to-55-faster-than-intels-core-i9-13900k-in-leaked-benchmark-amds-ryzen-9-9950x-purportedly-shines-in-avx-workloads
The truth is that while it's fun to speculate we wont know the full performance breakdown until end of the year. Perhaps not even until Q2 2025 if Zen 5 X3D launches that late.
There are too many variables in these leaks like ES, non-final clocks, unknown cooling, unknown memory configuration that can all affect the accuracy of the results.
If one source leaked both Zen 5 and Arrow Lake-S benches with same cooling and same RAM/Speed then it would be more representative but this rarely happens that someone gets a hold of both ES near final or even qualification chips and knows how to test properly.