Intel’s flagship CPU, Raptor Lake, has been seen in leaked benchmarks that show seriously fast multi-core performance, albeit at a cost in terms of power usage that the Core i9-13900K consumes.
This comes from OneRaichu (as signaled by VideoCardz (opens in new tab)) on Twitter, a regular hardware leaker who provided benchmarks pulled from Cinebench R23, though as always we should be skeptical about any leaks and the possibility that it could be faked.
Left: Under unlimited power and current setting. (standard frequency, almost 340-350w) Right: Under limited power setting. (standard frequency, almost 250w) pic.twitter.com/rxISzs55eUAugust 7, 2022
If true, this certainly shows the 13900K processor in an impressive light for performance, as it managed to hit 2,290 in single-core and 35,693 in multi-core. This was at the default settings for clock speeds and power, with the CPU consuming 254W in the test – which is actually the rumored ‘PL2’ power cap of 13900K (the maximum power usage when running at maximum Turbo speeds).
OneRaichu notes that the 13900K was running at 5.8GHz on two performance cores and 5.5GHz on the other six. VideoCardz also points out an additional ECSM leak consisting of an Aida 64 stability test, and this shows two cores hitting 5.8GHz with the others at 5.5GHz as well.
Back to the OneRaichu leak, which gets more interesting as it also offers a Cinebench run without any CPU power cap settings. While single-core wasn’t affected here, the multi-core result jumped much higher to 40,616 even though the CPU was consuming the best part of 100W more power to achieve this (the processor peaked at 345W).
As VideoCardz points out, this suggests that compared to the current 12900K, the entry-level flagship 13900K will be around 30% faster for multi-core performance (and 48% faster with that unlimited power setting) and 48% faster. faster than the Ryzen 9 5950X (67% with unlimited power). Though it’s worth noting that Intel’s 13th Gen will take on the next-gen Ryzen 7950X.
Analysis: Underlining what we’ve already heard about Raptor Lake
That’s a pretty telling jump in performance, of course, but as we’ve already noted, we need to be very careful about leaks – and indeed, over-read the results pulled from a single benchmarking tool anyway.
In short, we should consider these percentages very carefully, especially in the case of lowering the power limits for Cinebench R23 runs. Doing so requires a seriously robust cooling solution, and while we don’t know what was used on the main spill here, in the case of the ECSM (Aida 64) spill, we were told that a 360mm all-in-one liquid cooler was employed to keep temperatures to 90 degrees with a power consumption of 314W (not also far from what was achieved in Cinebench tests with unlimited potency).
All of this seems to back up recent speculation that we’re looking at boost speeds of around 5.8GHz for the main processor, and that Raptor Lake should be a little easier to cool down than Alder Lake. The latter will be good news for overclockers, of course, as they should be able to push Intel’s next-gen CPUs further while keeping thermals relatively reasonable.
That said, enthusiasts looking to push themselves with a Raptor Lake flagship twinned with an RTX 4090 – with its rumored high power demands – may well have their work cut out in terms of keeping everything properly cooled and stable.
AMD’s Zen 4 processors also look very promising in terms of performance, with boost speeds rumored to be reaching similar levels to Raptor Lake – but in what could be a crucial twist, better power efficiency is to be expected. Ryzen 7000 silicon. Also, there may be more Zen 4 chips in stock at launch, or so the latest rumors reckon…