Intel today announced the Alder Lake 12th-generation Core developer documentation, detailing the architecture configuration of the new generation of processors and the optimization of the hybrid architecture of large and small cores.
Intel confirmed,The desktop version of Alder Lake-S has two different cores, one is full of 8 large cores, 8 small cores, and 32-unit core display, and the other is 6 large cores, 0 small cores, and 32-unit core display., That is, some models will not have a small core.
The code name of the mobile version is unified as Alder Lake-P, which replaces the previous division of U series and H series. There are also two different cores, one is 6 large cores, 8 small cores, and 96 unit cores, and the other is 2 small cores and 8 large cores. Nuclear, 96-unit nuclear display.
In other words, the mobile version has at most 6 large cores, but all models will have small cores, and the nuclear display size is three times that of the desktop version.
It is worth mentioning that the instruction sets supported by the large and small cores are completely consistent.The only exception is AVX-512, which is not supported by the existing small cores, so if the small cores are turned on, the large cores will automatically disable AVX-512.
on the other hand,OEM manufacturers can decide whether to block small cores by themselves, Is not displayed in the BIOS, you can open AVX-512 freely at this time.
Intel also confirmed that the 12th-generation Core will gradually cover different product forms such as desktops, notebooks, ultrabooks, flipbooks, and 2-in-1s. The corresponding models will be released in the fourth quarter of this year.
For the optimization of the scheduling efficiency of large and small cores that everyone is most concerned about, Intel has given three levels.
One is that there is no optimization.
The software itself or the developer does not consider the issue of large and small core allocation at all,Based on the feedback and algorithm of Intel ITD (Thread Scheduler), the Windows system scheduler intelligently allocates threads and dynamically allocates load.
In most cases, they can do a good job, but sometimes they may assign key tasks to small cores, or give non-critical tasks to large cores, especially those programs that use multiple middleware components and schedule threads by themselves. .
The second is Good.
The software has certain settings and can use a hybrid architecture, but there is no targeted complete rewrite.
At this time, the priority important load will be allocated to the large cores, and the task system will be scheduled according to the number of large cores and the maximum number of threads required by the load to ensure load balance.
At the same time, there are thread priority and QoS API to ensure that the load is given to the appropriate core, such as games and rendering are assigned to the big core, background threads are handed over to the small core, and so on.
The third is the best optimization (Best).
The software can fully utilize the hybrid architecture task system and create two thread pools.
One is the priority thread pool, which is oriented towards large cores and executes tasks that require large cores or are assigned to large cores.
The other is the secondary thread pool, for small cores, performing non-critical tasks such as shader compilation, audio mixing, live streaming, and decompression.
In order to further optimize the system and enhance load balancing,Developers also need to deploy task allocation algorithms, When the load of the large core is too high, the priority thread is converted to the secondary thread and handed over to the small core.
This is especially important for notebooks because they have relatively few large cores.
– THE END –
Please indicate the source for reprinting: Fast Technology
#Intel#CPU processor#Alder Lake
Editor in charge: Q above
.