Intel 13th Gen 'Raptor Lake' Core i9-13900 ES performance looks promising, 33-50% performance uplift over 12th Gen ‘Alder Lake’ Core i9-12900

[BiG StroOnZ]

Summary

SiSoftware published a preview of what to expect from the upcoming Core i9-13900 Raptor Lake-S CPU. They present an overview of a few benchmarks. The Intel Core i9-13900 ES CPU is up to 50% faster than the Intel Core i9-12900 and AMD Ryzen 9 5900X at only 3.7GHz in select workloads.

 

 

 

 

 

Quotes

Quote

According to SiSoftware, the Core i9-13900 will have 36 megabytes of L3 cache (20 percent more than the Core i9-12900) and offer support for up to DDR5-5600 memory, PCIe 5.0, and Thunderbolt 4. The big P-cores now lack AVX-512 and feature 2 MB of L2 cache per core. We see 4 MB of L2 cache for a cluster of small E-cores. An exciting addition to E-cores is the AVX/AVX2 support.

 

When it comes to the testing, the results collected suggest Raptor Lake may deliver between 33 to 50 percent higher performance in basic arithmetic tasks over Alder Lake. In vectorized and SIMD tests, the 13th-gen part was around five to eight percent faster. This was achieved with an engineering sample that has performance cores running at 3.7GHz and efficiency cores running at 2.76GHz. It is not mentioned what exact test setup was used and if the memory was indeed running at DDR5-5600 speeds. But with that said, the results are very impressive for an ES chip with such low clock speeds. 

 

My thoughts

This is a nice showing for Raptor Lake at its ES clocks, because even if they only can muster 12900KS clocks and the 5.8GHz purported clocks are not accurate; that's still nearly a 50% uptick in clock speed left on the table compared to these current Engineering Sample clocks. In heavily vectorized workloads, the Raptor Lake part delivers less impressive performance gains, and still loses to the 5900X by a decent margin; while Zen 4 will increase this margin if AVX512 support is considered. In benchmarks like Cinebench, and gaming, Raptor Lake should perform exceptionally well, but that obviously does not tell the whole story. Zen 4 vs Raptor Lake is definitely shaping up to be the battle everyone expected. 

 

Sources

https://wccftech.com/intel-raptor-lake-core-i9-13900-es-cpu-up-to-50-percent-faster-than-alder-lake-core-i9-12900-leaked-benchmarks/

https://videocardz.com/newz/performance-preview-of-intel-core-i9-13900-with-24-cores-has-been-published-by-sisoftware 

https://www.overclock3d.net/news/cpu_mainboard/intel_s_raptor_lake_i9-13900_has_been_previewed_by_sisoftware/1

https://www.pcgamer.com/intels-next-gen-raptor-lake-doubles-cache-sizes-and-matches-amds-zen-4-thread-count/

https://www.guru3d.com/news-story/sisoftware-sandra-database-scores-for-intel-core-i9-13900-raptor-lake-s.html

https://www.techspot.com/news/94919-preview-core-i9-13900-engineering-sample-performance-looks.html

https://www.techpowerup.com/295795/intel-core-i9-13900-raptor-lake-processor-gets-a-preview

https://www.tomshardware.com/news/intel-raptor-lake-benchmark-sisoftware

https://www.pcgamesn.com/intel/raptor-lake-performance-vs-alder-lake

[Arika]

Now this is competition. About damn time, if true. 

[Salv8 (sam)]

the battle between these two right now is like a 1960's cartoon.

Tune in next time to see the conclusion to today's battle between AMD and Intel on HARDWARE WARS!

[Quackers101]

when intel does their 3D stuff too, a CPU into a block of heat radiation.

[Rauten]

Ooff...

If these numbers are a true indication of real life performance of an engineering sample, AMD should be sweating right now.

[TECHNOMANCER303]

What is ES??? is that like KS or kf or abcdefg? If it means the same as before then why does intel have to keep changing things. if its different then np.

[BabaGanuche]

3 minutes ago, TECHNOMANCER303 said:

What is ES??? is that like KS or kf or abcdefg? If it means the same as before then why does intel have to keep changing things. if its different then np.

ES = Engineering Sample

[WereCat]

Can anyone tell what these benchmarks could be compared to very roughly? Let's say physics simulation or tile rendering, encode, etc... because I have no point of reference.

[Senzelian]

All I'm seeing is the 13900k being destroyed by a 5900X in the second screenshot. Am I looking at it backwards or is it really this bad?

 

Edit: And it's not even a 5900X3D? Really? I hope these benchmarks aren't representative. 

[NF-A12x25]

6 minutes ago, Senzelian said:

All I'm seeing is the 13900k being destroyed by a 5900X in the second screenshot. Am I looking at it backwards or is it really this bad?

 

Edit: And it's not even a 5900X3D? Really? I hope these benchmarks aren't representative. 

that’s a 13900ES, which is not a 13900k by any stretch of the imagination.

[Senzelian]

1 minute ago, NF-A12x25 said:

that’s a 13900ES, which is not a 13900k by any stretch of the imagination.

Just wrote 13900k out of habit.

 

But honestly, just like this not being a 13900k, the 5900X isn't a 5900X3D and not a 7900X either. We all know that the competitor to this will be Zen4.
So all in all I'm really not too impressed by these benchmarks and considering that the 13900es already beats the older generation, I don't expect too much of a jump in performance after the release anymore.

 

Also the 12900 isn't a k-model either in the benchmark. 

[WolframaticAlpha]

27 minutes ago, Senzelian said:

All I'm seeing is the 13900k being destroyed by a 5900X in the second screenshot. Am I looking at it backwards or is it really this bad?

 

Edit: And it's not even a 5900X3D? Really? I hope these benchmarks aren't representative. 

Might be ES weirdness?

[Senzelian]

2 minutes ago, WolframaticAlpha said:

Might be ES weirdness?

Yeah maybe.
Usually engineering samples are the same CPU with lower clockspeeds and maybe non-functional parts like an iGPU or certain disabled I/O.

 

But honestly it's still just a sample. We'll see what the real deal is once it launches.

[HenrySalayne]

1 hour ago, WolframaticAlpha said:

Might be ES weirdness?

1 hour ago, Senzelian said:

Usually engineering samples are the same CPU with lower clockspeeds and maybe non-functional parts like an iGPU or certain disabled I/O.

It looks to me like the "33% to 50% performance gains" is scaled to the expected clock speed from the ES' lower clock speed. So it is not faster in these benchmarks, but waaay faster considering the lower clock speed. But take this with a grain of salt, it could also be misreported boost clocks or basically anything.