Professional Documents
Culture Documents
2
Task#01
PROCEDURE TO CREATE PERSONAL LETTER:
PERSONAL LETTER
Dear irfan,
It's great to hear all your news! I was staying in my grandmother's house for a while, and couldn't respond to you
immediately. Answering your question: I used to love Saturdays. Since we are doing online university, I prefer Monday
now because I only have two classes, which is ITC and ethics.
After class, I can watch TV series without having to worry about my homework, as you've said before. In the evening, my
mom likes to cook a cake and make biscuits, which is also an activity we use to spend more time together.
At night, I play games and sometimes watch a movie. I miss hanging out with you, unfortunately, the cases of Covid-19
has been rising every day making it impossible to go out.
Anyway, how is your quarantine going? Take care of yourself and don't forget to e-mail me back.
Lots of love,
You’re Friend,
hammad
Task: 2
Create a research paper:
Page: 1
Improving Mobile Gaming Performance through
Cooperative CPU-GPU Thermal Management
Muhammad Shafique§
National University of Singapore
Author: tulika@comp.nus.edu.sg
CPU and GPU that are triggered only after the 5. CONCLUSION
temperature constraint is violated. These TMUs do
not consider the performance coupling between CPU In this paper, we performed a detailed
and4.GPU, thereby reducing
DYNAMIC THERMAL application performance. characterization of the thermal behavior of the
In [6, 7], DVFS-based power management approach CPU and GPU in a commercial mobile MPSoC and
forMANAGEMENT
such architectures haveALGORITHM
been proposed in order to showed that both CPU and GPU contribute
provide users with flexible power-performance trade- towards the rise in chip temperature and hence
The proposed control-theory based DTM
offs, however they do not focus on thermal must be managed cooperatively to achieve high
algorithm is shown in Figure 5. It strives to keep Figure 1 is showing CPU & GPU throttling
management. Similarly, several works [8, 11, 9] performance. with robust thermal management.
the MPSoC temperature at or close to a pre-
consider coordinated CPU-GPU power management We also derived models to estimate the
defined threshold by estimating and allocating the
but do not consider thermal issues. 6. REFRENCES:
available thermal headroom individually to both
CPU and GPU based on their utilization values. 1. Android Governors. http://goo.gl/8j1Eqo
Due to the absence of individual sensors, the 2. Arndale board 5250. http://goo.gl/1ZCSNX
algorithm starts by estimating the CPU and GPU 3. [3] Epic Citadel, Unreal Engine.
temperatures, T c est (t) and T g est(t), at time http://goo.gl/xdsQMr.
step (t) from the current temperature measured 4. [5] A. Bartolini et al. Thermal and Energy
by the on-chip sensor T s mesa(t) and the Management of High-Performance
reference CPU and GPU temperature T c ref (t − 1) Multicores: Distributed and
and T g ref (t − 1), set in the previous iteration. To 5. [6] A. Pathania et al. Integrated CPU-GPU
this end, we first calculate the estimated on-chip power management for 3D mobile games.
Improving Mobile
Gaming Performance
through
3
SOFTWARE ENGINEERING
Cooperative CPU-
GPU Thermal
Management
Improving Mobile
Gaming Performance
through
Cooperative CPU-
GPU Thermal
Managemen
4
SOFTWARE ENGINEERING