Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Register Log in

Is a new wave of analog techniques coming?

Status
Not open for further replies.

rfsystem

Advanced Member level 3
Joined
Feb 25, 2002
Messages
914
Helped
148
Reputation
292
Reaction score
38
Trophy points
1,308
Location
Germany
Activity points
9,550

steer

Advanced Member level 4
Joined
Jan 14, 2005
Messages
119
Helped
23
Reputation
46
Reaction score
10
Trophy points
1,298
Activity points
1,299
Re: Analog DSP

I've seen this statement quite a few times in recent 20 years. Every time this proved to be wrong. I dont held my breath this time too.
 

rfsystem

Advanced Member level 3
Joined
Feb 25, 2002
Messages
914
Helped
148
Reputation
292
Reaction score
38
Trophy points
1,308
Location
Germany
Activity points
9,550
Re: Analog DSP

If a simple NAND2 in 65nm requires some 10fQ to execute a single operation, how many 10fQ is enough to process a 4bit addition in the analog domain? From the external side you have a digital interface (sooner or later an analog, haha). So how many NAND2 you need to process the 4bit addition in the digital domain? That could make a big difference in power consumption. As I think as scaling is dead, financial earlier than technical, new design techniques have to used to proceed with Moore Law. What do you think?
 

steer

Advanced Member level 4
Joined
Jan 14, 2005
Messages
119
Helped
23
Reputation
46
Reaction score
10
Trophy points
1,298
Activity points
1,299
Re: Analog DSP

The main problem with analog DSP or CPU is complexity management.
Complexity is easily managed in software, less easily in digital hardware and extremely pourly managed in analog. So a properly designed computer platform is very complex in software (MS Windows, Unix, RTOS, etc.), less complex in digital HW (Pentium, MCU, DSP) and very simple in analog (PLL, I/O, A/D).
The current analog design techniques do no allow design anything that complex as simple programmable general 8bit MCU. Even a simple analog memory cell takes 1 week of analog designer and 1 week of layout in best case, while digital designer takes 1min to config memory compiler and thats it.
Once analog CAD gets to the level of digital one, we can think about pJ of switching energy.
 

rfsystem

Advanced Member level 3
Joined
Feb 25, 2002
Messages
914
Helped
148
Reputation
292
Reaction score
38
Trophy points
1,308
Location
Germany
Activity points
9,550
Re: Analog DSP

I think Analog DSP is not as mature to compare design efficiency. The article in IEEE Solid-State Circuits page 42 2006 show that there is accuracy limitation and programming is done by signal flow diagramm in Simulink but the results are impressive and could find application where energy supply is critical.
 

Status
Not open for further replies.
Toggle Sidebar

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top