The problems are accumulating for Intel and what's worse, perhaps, is that they do not provide a solution for users and investors. The promised 10 nm change was made with delays already announced and accepted by the company, not without effort, when they now face many restrictions with the latest information being dissolved.
Comet Lake-S and Z490 with LGA1200 will "support" PCIe 3.0 only
The experience in AMD with PCIe 4.0 and its outsourcing to the market have laid a strong foundation for seemingly unwilling to duplicate Intel. The durability decisions of the deployment are well-known by its users in many cases (it doesn't feel good to change the board or socket, but it is similar to the durability they provide).
In August we saw how AMD should meet the rumors (first) and evidence (in the second case) about the fact that most motherboards other than the X570 could provide PCIe 4.0 support without any problems.
UEFI and BIOS watched the Internet and were introduced immediately after new versions that removed such support at the command of Lisa Su. AMD did not want to expose itself to various problems with its GPUs and SSDs that other manufacturers in models were not ready as a support level and made Solomon's decision.
Now with Intel in the same problem, as reported, Comet Lake-S in its operating systems will have acquired and enabled PCIe 4.0 controller, so where is the problem?
Motherboards, in their entirety, do not offer size
PCIe 4.0 has come as a standard dual-speed PCIe 3.0 offering, but it also requires the highest signal integrity requirements, in short, to be robust. This is in line with the fact that Comet Lake-S will come with other prices TAU the highest of the state PL2 (56 seconds up to 250 watts, which is a double (Cofi Lake (S)), does not help standard boards under the LGA1200 socket with something called jitter.
And here's the problem exactly, dimming the jitter generated by such bandwidth and joining it to new TAU values is a problem for board manufacturers, since a series of clock generators are needed to reinforce this problem so that PCIe 4.0 interface.
In addition, it will require some power levels and MOSFETs on both CPUs and GPUs, which directly fixes the high cost that many manufacturers do not want to consider between certain motherboard ranges.
TaU, another stone in the form of PCIe 4.0
Intel's solution is excellent, as it has allowed manufacturers to convert TAU to pricing like Celt Lake-S to try that out jitter shortened and eventually adopted PCIe 4.0, but this does not seem to convince AIB, as it could mean that their models will be losing more performance compared to the higher models and their direct competitors in the market.
As if that's not enough, motherboards are in their final stages of development, so the changes may be minor in design and all suggest that Intel will choose Solomon's decision: not to allow PCIe 4.0 on the Z490 opset boards at first, then and when it arrives Rocket Lake-S, enable specific (high-end) models via UEFI and launch a new chipset with new boards that already support this interaction as usual.
So, in 2021 we will have motherboards Z490 with PCIe 4.0 powered by the emergence of new 11th Generation processes, and new boards Z590 with the support and physical development of PCIe 4.0. So, Intel intervenes with the notification system when AMD, we'll see how the war between manufacturers ends up providing this interface and if Intel finally uses it without any problems.
