Intel’s first Age Lake CPUs have been officially launched

After months of speculation, drips and leaks, Intel’s 12th generation processors have finally seen the light of day. The generation is codenamed Alder Lake, and here it seems that Intel has finally put itself in the saddle to take up the fight against AMD to a far greater extent than they did with the 11th generation. The latter was a bit tame, which is perhaps why Intel is launching the 12th generation just under seven months after the 11th generation was officially launched. By comparison, 11 months passed between the 10th and 11th generations.

The Age Lake family will eventually have as many as 60 members, reports Intel, but so far it is the three enthusiast models that have been announced, each of them in two editions: the K versions with Intel UHD Graphics 770 and the KF versions that do not have built-in graphics and thus need a graphics card.

Both the K and KF versions are unlocked for overclocking.

Two different types of processor cores

i9-12900K/KF

  • Parking cores: 8 pcs, 3.2-5.1 GHz
  • E-cores: 8 pcs, 2.4-3.9 GHz
  • Maks turbo boost:5.2 GHz
  • Threads: 24
  • Cache (L2/L3): (14 MB/30 MB)
  • Power consumption: 125-241 watt
  • Price for K-model: 589 USD
  • Price for KF model: 564 USD

7-12700K/KF

  • Parking cores: 8 pcs, 3.6-4.9 GHz
  • E-cores: 4 pcs, 2.7-3.8 GHz
  • Maks turbo boost:5.0 GHz
  • Threads: 20
  • Cache (L2/L3): (12 MB/25 MB)
  • Power consumption: 125-190 watt
  • Price for K-model: 409 USD
  • Price for KF model: 384 USD

i5-12600K/KF

  • Parking cores: 8 pcs, 3.7-4.9 GHz
  • E-cores: 4 pcs, 2.8-3.6 GHz
  • Threads: 16
  • Cache (L2/L3): (9.5 MB/20 MB)
  • Power consumption: 125-150 watt
  • Price for K-model: 289 USD
  • Price for KF model: 264 USD

Common to the processors is that they get two different types of cores, a bit like Apple M1 processors. The P-cores give the best performance and are intended for heavy tasks, while the E-cores use far less power and are “lightweights” who do not have enough muscles for heavy tasks. The advantage of doing it this way is that you can have a processor that uses little power – and develops correspondingly little heat – when you do not need the large processor muscles. At the same time, you have more powerful cases at your disposal should you need them.

Heat generation has also been a recurring problem for Intel in the last couple of processor generations, and it is therefore not surprising that they have found such a solution to just that problem.

On each processor is a dedicated section that Intel calls «Thread Director»And then has the task of deciding which tasks are sent to E-cores and P-cores.

Cores and threads

The P-cores support “hyperthreading”, a technique that works so that a processor core, in a simplified manner, can use a kind of surplus capacity to simulate an additional core. This is often called a logical core, as opposed to a physical core, and will usually be able to offer a performance slightly lower than a normal processor core of the physical kind. In short, the operating system will perceive each P-core as two processor cores.

The e-cores, on the other hand, do not support this technology, and thus only the P-cores are “doubled” in this way.

Where the number of P-cores varies in number by two from model to model, the E-cores are connected in groups of four. Thus, the i9 model gets eight cores of each type, briefly summarized as 8 + 8, while the i7 and i5 have to settle for 8 + 4 and 6 + 4, respectively.

Clock frequencies

The cores operate at slightly different frequencies, depending on the type and which processor model they belong to. Each of them has a “base frequency” that they run on to save power when they do not have to work particularly hard. When the workload increases, however, the frequency will increase – like the speed of a car – and the processor cores thus have an upper turbo frequency they can reach when needed. This turbo frequency will be slightly lower for the E-cores than for the P-cores, but will also vary according to need and processor model.

Two of the processor models, i9-12900K / KF and i7-12700K / KF, will also be able to find out which cores perform best and give them an extra little push in frequency. The i9 model can thus reach up to 5.2 GHz on the best cores, while the i7 model can reach 5.0 GHz. The i5 variant does not support this feature.

We have not received Norwegian prices yet, so the prices stated here are in US dollars. Intel’s stated prices also apply per processor when ordering 1000 pieces.

i9-12900K/KF


The Intel i9-12900 processors start at lower frequencies than the other models, but can in turn reach higher clock speeds when needed. In addition, they are equipped with more cache memory, both at L2 and L3 level.

Higher frequency also results in higher power consumption and heat generation, and Intel states that the i9-12900K / KF can use from 125 to 241 watts. In other words, you should have a good cooler to take away the heat when it runs at full machine.

So yes, that’s the right way. Foto: Intel Corporation

New shelf

Intel’s new processors get new shelf, LGA1700, and as the name suggests, it gets 1700 contact points. The width is the same as the previous one, LGA1200, but in the longitudinal direction Intel has added 7.5 mm.

Whether the cooling solutions intended for the LGA1200 can be used on the LGA1700 as well, seems to depend on what the motherboard manufacturer chooses to do. It may be a bit complicated by the fact that the heat spreader on the new processors is said to be a bit thinner than before, but we do not know this for sure yet.

DDR5 – and DDR4

The new processors get two memory channels, and that’s not exactly new. Intel has had it in this segment since the ‘a little later in the day’. New in the 12th generation, however, is support for DDR5 memory. This should be faster than DDR4, but also has some problems, such as dramatically higher price and potentially higher heat generation. In return, it has a kind of built-in error correction system.



Intel Alder Lake Foto: Intel Corporation

By the way, the new processors can also use “old” DDR4, but not at the same time as DDR5. Maximum specified speed for DDR4 is stated at 3200 MT / s while that for DDR5 is stated at 4800 MT / s. However, this will hardly be 50 percent faster in practice, because DDR5 also has a greater delay than DDR4. Until the DDR5 standard becomes more mature, one can probably expect better performance from a good (and expensive) set of DDR4 chips than a standard DDR5 set. DDR5 modules are not yet particularly common on store shelves.

Maximum supported memory capacity is 128 GB.

PCI-express

All new processors get 20 PCIe channels and support for PCIe 5.0. However, these are the channels that go directly from the processor. The chipset will still operate on PCIe 4.0, so the extra channels the chipset provides will not be able to provide exactly the same pace. Still, PCIe 4.0 is plenty fast enough for most people for many years yet.

Performance

Whether Intel has now managed to catch up with AMD’s lead, we can not say for sure. Of course, Intel claims that, but at the same time, they have probably carefully chosen which performance measurements and results they choose to share with the world, so exactly such claims must be taken with a pinch of salt. At least we do not say anything with certainty until we have measured ourselves, but here are at least some of Intel’s claims about what kind of performance Alder Lake can offer:

  • i9-12900K will offer 19 percent better performance in single core tasks than its predecessor, i9-11900K, at 3.3 GHz.
  • With the maximum specified watts, the i9-12900K will perform 50 percent better in BlenderMT than the i9-11900K (NB! 241 watts against 250 watts).
  • At 65 watts, the i9-12900K should perform about the same in the BlenderMT as the i9-11900K does at 250 watts.

As I said, these results must be taken with a pinch of salt, not only because they are Intel’s own, but also because there are certain differences in the measurements, as well as other considerations that can affect the result in one way or another:

  • The measurements were made with Windows 11, which could potentially favor one processor generation over another.
  • i9-12900K was measured with DDR5-4400, while i9-11900K used DDR4-3200.
  • Software support for new technology is often not fully optimized at launch. Especially the “Thread Director” and the E-cores may be affected by this.

.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.