More Intel 12th Gen Core performance numbers emerge – CPU – News

Indoor Intel Core i9-12900K and Core i5-12600K

While Core nomenclature will follow an iterative path, there’s no doubt that Intel’s 12th Gen Core desktop processors, codenamed Alder Lake, are a big deal. These are the first high-performance chips built with a hybrid architecture that combines many powerful hyper-threaded cores (Golden Cove) with less efficient (Gracemont) cores.

We detailed this hybrid architecture at our August conference ArchitectureWe revealed, but as we get closer to a full desktop launch, Intel is duly offering more flesh and blood. The most prudent approach is to start with publicly available models and then work backwards to the performance data provided by Intel.

Slides courtesy of Intel, click to enlarge. Chips with the F suffix do not have integrated graphics

We now know that Intel will release six unlocked 12th Gen Core chips from the get-go. Split evenly between the Core i9, Core i7 and Core i5 series, with two each, there’s plenty to take in.

new beginning

The first thing to understand is that moving to a 12th generation core platform will require investment in a Z690 series motherboard – with other 600 series models to follow in due course – while potentially also requiring a cooler upgrade to the new LGA1700 socket. In addition, the platform is the first to offer DDR5 memory support, although the dual-format support feature of the 12th generation Core means that motherboard manufacturers also have the option of launching motherboards with traditional DDR4 memory.

Determining performance from a spec sheet is certainly more difficult than ever. The reason is the combination of high-performance and efficient CPU cores built into each chip. At the bottom is a Core i5 die with 6P and 4E cores, in the middle is a Core i7 with 8P and 4E cores, and at the top is a Core i9 with 8P and 8E cores. Knowing that the P core has hyper-threading capabilities, it is very simple to calculate the total number of threads.

Even so, the top-tier Core i9-12900K’s 24-thread complement doesn’t compare to the rival Ryzen 9 5950X’s 32-threads, but not all thread power is created equal, and Intel relies on this feature to boost performance.

By their very nature, performance cores are not only faster across threads, but they also run faster than efficient cores, which makes sense. That’s not to say E-Core is just an afterthought. Although they are derived from the low-power Atom architecture, Intel believes that their single-thread speeds are at least as fast as the 10th generation Core. This comparison is somewhat misleading, since all 10th Gen Core chips feature performance-boosting Hyper-Threading: Intel didn’t do a strict apples-to-apples comparison on the slide above.

Why hybrid?

The obvious question that needs to be answered is why did Intel venture down this dual-core path? Unlike traditional homogeneous designs, performance issues can arise as threads are assigned to cores with different functions, and Intel has gone to great lengths to build a thread director – The name is self-explanatory – used to keep performance up to par.

Intel said the decision was driven by maximizing performance per watt and overall performance. Learn how four E-cores occupy the same silicon area as a single P-core while increasing multi-thread throughput by 60%. Think of it this way: Thanks to its enhanced Golden Cove architecture and bursty, high-frequency approach, the P-core seems well-suited for single-threaded or low-threaded applications. E cores work better together to help drive peak performance for applications that utilize large numbers of threads.

So what Intel is saying is that the one-size-fits-all approach (which has been used in desktop computers for the past 30 years) is no longer the most efficient way to utilize core resources. Supporting this argument is a slide produced by Intel detailing that the MT performs 50% better than the Core i9-11900K at similar power levels, but due to a more compliant V/F curve on the hybrid design , plus moving to a smaller process now called Intel 7, the Core i9-12900K is able to produce the same peak figures at a quarter of the power. Touching.

To maximize performance, Intel no longer provides a maximum Turbo state defined by time-limited PL1 and PL2 values. Conversely, for enthusiasts, the 125W TDP specified for the top chip can be ignored. Z690 motherboards will ignore this limit and call upon the 241W power budget, allowing the chip to run at higher speeds during applications. AMD’s Ryzen chips, however, don’t do this and generally have much lower total package power.

Running in its preferred 241W state, Intel put up impressive numbers. The company thinks the Core i9-12900K is Compare Both the Core i9-11900K and the rival Ryzen 9 5950X (for gaming), and are faster than the previous mainstream champion content creation. Today’s Intel slideshow doesn’t directly mention multi-threading performance compared to the Ryzen 9 5950X; you’ll have to wait a little longer for us to reveal the secrets.

Platform potential

Rival AMD has been a pioneer in adding new features to the platform. Intel is changing this situation with its 12th generation Core. Each chip offers 16 PCIe 5.0 lanes for graphics or storage, but there are currently no devices supporting the standard. Additionally, as we talked about above, the motherboard will feature DDR4 or DDR5, with native speeds of 3,200MHz and 4,800MHz respectively. Once you start overclocking, expect both sets of frequencies to go higher.

Z690 adds 28 expansion lanes, 12 of which are PCIe 4.0 types. It’s a completely new platform and feels a bit stingy. Many Z690 motherboards are expected to feature at least three high-speed (PCIe 4.0) NVMe storage slots, matching those available on the AMD X570.

Intel knows there may be more bandwidth traffic between the chipset and CPU, so it made the DMI link twice as fast as the Z590. There are still 8 lanes, but the speed increases to 16GT/s. Intel further said that this version improves overclocking capabilities and fine-grained control over frequencies. We will examine these claims in our comprehensive review.

Aside from the CPU, not much is currently known about the built-in graphics card. Built on Xe technology, with up to 32 execution units on the desktop, performance should be similar to Rocket Lake’s ostensibly similar iGPU.

pack

Intel is carefully managing expectations for its 12th-generation Core desktops, codenamed Alder Lake, ahead of its full launch. The use of heterogeneous CPU architectures is the most radical move in more than a decade. While improvements in performance metrics are expected, many issues remain. We don’t yet know how effective Intel’s Thread Director technology is at keeping the right core working on the right thread. Ready for Windows 11, how will it perform on mass-market Windows 10 distribution? Should Intel use a different combination of P and E cores? Is this design better than a homogeneous core complex taking up the same space…with fewer P cores or more E cores? All of this hangs in the balance.

Additionally, we can’t yet tell you exactly how the 12th Gen cores compare to the Ryzen 5000 series in our set of benchmarks – the numbers we’ve run are still interesting – and pricing for CPUs, motherboards and DDR5 is still largely Official mystery. According to people familiar with the matter, the retail price of Core i9-12900K is $589, Core i9-12900KF is reduced to $564, Core i7-12700K is $409, Core i7-12700KF is $384, and Core i5-126005-12695-12695K-12600KF Priced at $264. If true, the pricing is higher than we originally expected, which will likely put AMD in trouble.

We can say without hesitation that the 12th generation Core is the most powerful mainstream CPU processor Intel has ever created. That argument makes sense, as previous generations had iteratively improved upon the venerable Skylake, while Intel was plagued by the twin plagues of lagging process technology that led to suboptimal design implementations.

Intel’s 12th generation Core breaks away from these constraints by changing the way the CPU part of the chip is constructed and operates. We’ll have a more in-depth analysis, benchmarking, and discussion of platform enhancements in our full review soon. Until then, don’t buy your next high-performance processor!