Editor's note: We often see words like "28nm processor" on the parameters of mobile phones. What are the differences between different generations of factories and types of 28nm processes? Let's have a look.
Part of this article comes from Baidu , original author "Gun God" 。
This is ST( I added the names of several factories in the picture of ST
From this perspective, the positioning of various processes is clear
Some aspects of ST (ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST. ST,
In the case of TSMC, the lowest 28 LP is not HKMG, but traditional poly/Si. It is also the process used by Qualcomm and MTK6589
GF (global foundation), Samsung's 28nm family are all HKMG
In addition, GF seems to cooperate with Samsung, so GF also has 28LPH.
GF(global foundry) SLP=Super Low Power, for mobile devices
HPP=High Performance Plus, for the network, those plugged in devices have high performance, but idle power consumption will be high
LPH=Low Power, High Performance, for high-end mobile devices RK3188 is the 28 SLP of GF.
Samsung
Almost, LPH positioned high-end mobile
Octa should be 28 LPH, and Samsung's Snapdragon S4 and MDM baseband for Qualcomm OEM should be 28 LP
TSMC
The lowest end 28LP is poly/Si, which was first mass produced, representing Qualcomm S4, baseband, and now MT6589。
Evolutionary HPL uses HKMG, which is the original LP+HKMG. Its performance is improved and leakage is reduced, Tegra 4 Use this process.
HP is a high-performance version, which is the process used by the original G in combination with HKMG, AMD and NVIDIA video cards
The most high-end HPM, a performance blockhouse, has an electric leakage similar to that of LP, and its frequency is extremely high, Qualcomm Snapdragon 800, Tegra 4i Use this process (Editor: Tegra4i uses high-end HPM, and Tegra4 uses HPL, which is unscientific!!).
So, aside from the HP process that cannot be used by mobile devices, TSMC's 28LP HPL HPM performance and leakage ranking
Performance: HPM>HPL>LP
Leakage: HPL<HPM=LP=40LP
The leakage of HPL is only 60% of LP
So, NVIDIA's Why does Tegra4 A15 use HPL instead of HPM? First, the mass production of HPL is relatively fast... Second, the leakage is small. The leakage of A15 itself under low load will be improved by using A7 jack 。。。 Old Huang doesn't have A7, so I want to use HPM to leak more...
Learn more about new cool devices, stay tuned
Related links:
Research on the Internal Structure and Chip of Galaxy S4
How many "cores" is your mobile phone's GPU?
Where is "Super"? Performance comparison test of Xiaomi 2S
Galaxy S4 hands-on demonstration video
"Eight Core" Beast Samsung Galaxy S4 Hands on Evaluation