In addition to Weibo, there is also WeChat
Please pay attention
WeChat public account
Shulou
2025-02-24 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > Internet Technology >
Share
Shulou(Shulou.com)06/01 Report--
This article is to share with you what are the influencing factors of power-plan, the editor thinks it is very practical, so I share it with you to learn. I hope you can get something after reading this article.
Power-Plan or PG how to play, this is a question of benevolence and wisdom, there is no standard answer, because there are various influencing factors. Here are some possible influencing factors:
Related to design
If the utilization of design is very low, you can play as close as you can. The utilization is very high, so when you hit PG, you have to consider whether it will make it difficult for the signal line to run, that is, to generate congestion or DRC.
For the same utilization and the same PG structure, the congestion of the switch module must be larger than the congestion of other modules. Therefore, for this special module, we have to carefully adjust the structure of the PG.
For design with different power requirements, such as mobile phone chips and computer chips, because of the different power requirements, the structure of PG is also very different.
Related to sign-off standard
Sign-off standards include static-IR-drop and dynamic-IR-drop, different design have different requirements for this standard, based on different standards, the requirements for PG structure are also different.
Sign-off needs to analyze the power consumption and IR according to the activity file of a work scene. The commonly used activity file formats are: VCD (Value Change Dump file), FSDB, SAIF (Switching Activity Interchange Format). Even with the same design, activity files in different work scenarios produce very different IR-Drop.
First-in experience
The general choice of PG structure is based on previous project experience. However, with the continuous progress of the process, the previous experience is not necessarily optimized or applicable to the current process.
Generally speaking, only M1 is used as rail in the 40nm process, but under the advanced technology, the resistivity of M1 is relatively high, and the IR-drop caused by M1 is relatively large, which is the main factor. So M2 may also be needed as a rail, so the first-in experience should be adjusted with the improvement of the process.
It is related to the selected process and the number of metal layers.
Different processes, pitch and width are very different, and the power consumption caused by using LVT or HVT is also very different, which can not be directly as simple as the process size of shrink.
Even with the same process, different number of metal layers can be selected. A less layer of metal saves a layer of mask and saves a lot of money. For example, in the 40nm process, if you choose 8 layers of metal, you may need 100W at a time, and if you choose 7 layers, you need 95W for one time. The difference may not be very great, but if the amount is very large, the savings will be considerable.
When different metal layers are selected, the structure of PG is naturally different.
Fine-tune PG
How to play PG, this is the secret of each company, the author can not say much, generally is to first briefly play a version, and then look at the results in the signoff tool, and then adjust based on the results and the situation of congestion. Let's talk about how to save more winding resources through fine tuning for a given PG structure, while ensuring the same IR-drop, reduce the congestion or reduce the IR-drop based on the same PG structure.
What is the principle of fine tuning?
Under the new process, the signal-routing is routed on the track, and it will not happen on the semi-track. For example, four track,2 lines are drawn below. Under the old technology, there is no problem at all. You can go as long as you want to meet min-spacing. But under the new process, you have to be well-behaved, and every alignment must be on the track (here it just means that the signal-routing must be on track).
There are several reasons why you have to be on Track:
Spacing is a discrete value. In the old process, two infinitely long and parallel lines can be arbitrary floating-point numbers as long as the spacing is greater than a certain value. But in the new process, more than a certain value is not accurate, spacing is a discrete value. For example, the table of spacing is 0.1 mine0.2, 0.3. If spacing is less than 0.3, only 0.1 and 0.2 can be selected. If spacing is greater than 0.3, it can be any floating point number.
The same goes for width, which is a discrete value.
Many DRC rules are based on this discrete value, not just this simple spacing/width, so p not only violates this one, but there are more DRC waiting for you.
The new process uses double-pattern, and the wiring is not on Track, so there will be problems when dividing different mask.
The above said that the routing must be on the track, more accurately, in order to reduce all kinds of possible DRC, the tools are on the track when they go to signal-routing, because the PR tool can't see all the DRC-rule of sign-off, so they can't and don't dare to walk around. We are still free to play PG ourselves, as long as there are no DRC violations.
Although these factors will not affect the IR-drop, but by adjusting these factors, we can reduce the congestion or enhance the PG structure based on the same congestion.
Here is an example, the data in the example are randomly taken by the author and are not representative.
one
Original structure
Assume that the original PG is shown in the following figure, with PG in the middle and signal wiring on both sides.
Let's analyze this picture:
PG is on Track. No problem.
The width takes up 3 track, no problem.
However, due to a variety of DRC, the additional 2 Track actually can not be used, in fact, PG occupies 5 track. Waste!
two
Optimization plan one
Let's do some optimization: since it takes up 5 Track, why not just PG a little wider?
And PG doesn't have to be on track, it can be in the middle of track. For example, the following figure shows that it not only widens the PG, but also saves a Track!
Optimization plan 2
As shown in the following figure, instead of using uniform-track at all, create a non-uniform-track. It still uses the same PG width, but saves a lot of winding resources.
The above are the influencing factors of power-plan, and the editor believes that there are some knowledge points that we may see or use in our daily work. I hope you can learn more from this article. For more details, please follow the industry information channel.
Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.
Views: 0
*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.
Continue with the installation of the previous hadoop.First, install zookooper1. Decompress zookoope
"Every 5-10 years, there's a rare product, a really special, very unusual product that's the most un
© 2024 shulou.com SLNews company. All rights reserved.