Breaking News
View Comments: Oldest First | Newest First | Threaded View
bobdvb
User Rank
Author
FPGA coprocessing
bobdvb   7/3/2013 7:49:45 AM
NO RATINGS
I really want to see hybrid chips like Zynq get somewhere in the market, but the flexibility needs to be more accessible. I would like to see something along the lines of an FPGA JIT, where the OS on the fixed CPU recognises it is doing something difficult and automatically programmes a segment of the FPGA to offload that task. Think back to the transputer concepts of the 80s, well we are now doing this in software with JIT optimisation of code but we should push some of that back down to the hardware. I can foresee a JIT which reprogrammes an FPGA, I just can't make it happen.

luting
User Rank
Author
Interesting idea but not sure will fly
luting   7/3/2013 12:11:08 PM
NO RATINGS
I don't have high confidence on this type of architecture for following reasons:

1. Hard to leverage existing eco system: Compiler, Debug, library, etc

2. Hard to expect SW engineer to write a program leverage such architecture efficienctly

3. With Program SoC from Xilinx and Altera in market, there is not enough market space left for such ideal to grow

almaktabi
User Rank
Author
Re: FPGA coprocessing
almaktabi   7/6/2013 4:51:50 AM
NO RATINGS
Here is a link from Microsoft Reseach describing how to do what you want...

http://research.microsoft.com/apps/video/default.aspx?id=103616

Most Recent Comments
michigan0
 
SteveHarris0
 
realjjj
 
SteveHarris0
 
SteveHarris0
 
VicVat
 
Les_Slater
 
SSDWEM
 
witeken
Most Recent Messages
9/25/2016
4:48:30 PM
michigan0 Sang Kim First, 28nm bulk is in volume manufacturing for several years by the major semiconductor companies but not 28nm FDSOI today yet. Why not? Simply because unlike 28nm bulk the LDD(Lightly Doped Drain) to minimize hot carrier generation can't be implemented in 28nm FDSOI. Furthermore, hot carrier reliability becomes worse with scaling, That is the major reason why 28nm FDSOI is not manufacturable today and will not be. Second, how can you suppress the leakage currents from such ultra short 7nm due to the short channel effects? How thin SOI thickness is required to prevent punch-through of un-dopped 7nm FDSOI? Possibly less than 4nm. Depositing such an ultra thin film less then 4nm filum uniformly and reliably over 12" wafers at the manufacturing line is extremely difficult or not even manufacturable. If not manufacturable, the 7nm FDSOI debate is over!Third, what happens when hot carriers are generated near the drain at normal operation of 7nm FDSOI? Electrons go to the positively biased drain with no harm but where the holes to go? The holes can't go to the substrate because of the thin BOX layer. Some holes may become trapped at the BOX layer causing Vt shift. However, the vast majority of holes drift through the the un-dopped SOI channel toward the N+Source,...

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed