As we get closer to the looming launch of Intel's 13th Generation Core 'Raptor Lake' processors, we learn more about them and sometimes even get a sneak peek at their performance numbers in synthetic benchmarks. This time around someone has posted benchmark results of Intel's 65W Core i9-13900 processor to the Geekbench 5 database (via @BenchLeaks), allegedly revealing its specifications and performance numbers. For now, take the numbers with a pinch of salt.
Intel's Core i9-13900 is fully-fledged Raptor Lake silicon with eight high-performance Raptor Cove cores with Hyper-Threading (optimized Golden Cove cores) and 16 energy-efficient Gracemont cores without simultaneous multithreading that together can process up to 32 threads concurrently. By contrast, Intel's Core i9-12900 has eight Golden Cove and eight Gracemont cores and therefore can process up to 24 threads simultaneously.
As it turns out, Intel's forthcoming Core i9-13900 CPU seemingly has more energy-efficient cores than its processor, the Core i9-12900. This greatly affects its performance in multi-threaded workloads, but also features significantly higher turbo clocks for the performance cores 5576 MHz vs 5081 MHz. This would imply a noticeable edge in single-threaded workloads.
|Header Cell - Column 0||Core i9-13900||Core i9-12900|
|General specifications||8P, 16E ? ~ 5.58 GHz, 36MB||8P, 8E, 2.40 ~ 5.10 GHz, 30MB|
|Single-Core | Integer||1904||1627|
|Single-Core | Float||2278||2035|
|Single-Core | Crypto||4185||5313|
|Single-Core | Score||2130||1934|
|Row 5 - Cell 0||Row 5 - Cell 1||Row 5 - Cell 2|
|Multi-Core | Integer||19664||17933|
|Multi-Core | Float||22875||19882|
|Multi-Core | Crypto||9729||13261|
|Multi-Core | Score||20131||18284|
Due to its high turbo clocks and a larger L3 cache, the new 65W Raptor Lake looks set to beat 65W Alder Lake processor single-core integer and single-core floating point workloads in Geekbench 5 by 17% and 12%, respectively. We are yet to discover how this will impact performance in real-world applications, but according to the Geekbench 5 report, performance uplift is noticeable.
When it comes to multi-thread workloads, the new CPU appears to be 9.6% faster in integer and 15% faster in floating point workloads, which is a quite substantial advantage, at least in a synthetic benchmark.
Is that 5.58Ghz all-P-core? Single P-core? What about the E-core clocks? If they moved the balance for more P-core speed (which would make sense) to keep a similar power draw and thermal envelope (kinda?), then the E-cores are probably slower in heavily constrainded TDP/Power environments? What is that 65W? I'd imagine TDP since I'd doubt a 13900 would pull just 65W with an all-core workload and 5.58Ghz.
This doesn't really tell me much?
If not otherwise stated you can be pretty sure that it's the Intel® Turbo Boost Max Technology 3.0 Frequency. (So single thread max clock on the best available p-core)
And the 65w is only for base TDP, while this was, again almost certainly, made with max power,
if not outright unlimited PL2 and unlocked clocks because intel would not use an uneven number for their turbo clocks.
I doubt this is the 13900, probably the 13900K.
It tells enthusiasts something.
True at default power limits the chip will only boost a core or two to 5.8Ghz. However, if you have sufficient cooling and enough VRM power it should be trivial to boost all cores to 5.8Ghz by power unlocking it and have a 100% success rate.
The reason being, it is rated for 5.8, it is just a matter of power and cooling to get there on all cores - it's not even really overclocking, just power unlock.
I doubt this is a 13900 though, as the retail leaks I noted above say the 13900K is rated for a 5.8Ghz boost clock. 13900 I would expect to be at least 200Mhz less.
The reason being that the cores are rated and only the two best ones are rated to reach the highest clock.
I figured someone would argue the semantics.
I'm sure someone, somewhere, has failed on a power unlock getting max boost. I haven't seen it.
I've done it many times and as I said, it is trivial and the success rate is near 100%. It's no different than AMDs PBO.
It does. TDP isn't power draw, it's size of the CPUs required thermal dissipation in watts. TDP has never been power draw (though they are related of course). If you have a cooler that can dissipate 65 watts then you can use it with this processor. That is all TDP means and all it ever meant. This is true for AMD as well.
Us - Clever girl...
It's a minimum guaranteed all core clock (base frequency) under heavy load.
And they also state the maximum turbo power which is the max power draw of the CPU that they allow, not the cooling but the actual power. (At steady state they are basically the same anyway)
Any ark page of any CPU when you click on it:
For AMD it doesn't mean anything, it's the temp at the cpu case subtracting the temp at the heat sink intake divided by the quality rating of the heat sink material...
The only reason the 3700x is 65w and the 3800x is 105W ,according to AMD, is because the 3700x is considered to run a bit hotter and have a crappier cooler (HSF θca) ...