Staying power of 8051- A closer look at the 80C251 Soft Core

In my last post on 8015 couple of days back, I promised to revisit the soft core descendant of the 80C251 processor. Here we take a closer look at what it has to offer.

The advantage of micro-controllers is that it can offers many peripherals on chip. The more it can offer, more diverse systems can be implemented with it. So, let's take a look at the peripherals the DQ 80251, the soft core has to offer. One unique peripheral is a block that is JTAG compatible hardware. Other units on-chip include a power management unit, an interrupt controller, 4 x 8 bit I/O units, 2 x timer/controllers, and  a full duplex serial port. We shall take a look at the peripherals in another post.

Meanwhile let us look at the debug facilities available through the access provided in the core. The interface follows the JTAG and other standards. The soft core bundle comes with a set of debug software too. So when you have created a device and implemented a system, you would be able to get a detailed look at what is happening when your application code is running. Major source of problems at development time is how the program logic was implemented. The logic used in developing algorithm(s) could be wrong. next thing is the problem in semantics. You used source statements meant to implement some logic but the exact semantics (that is rigid in programming languages) could be different. The same thing can extend to object code level. The compiler used may have used some machine instructions to implement the source statements but the actual machine code may do things differently, based on what it is defined to do.

To understand what exactly has happened inside the target machine can only be decoded by looking at the processor status (the ALU, registers, Stack pointer, any special registers, the flags, etc.), values of relevant variables, etc. Any good debug system should let you control execution and find these information bits.

The debug process breaks the chain of execution at some suitable point (chosen by the developer/tester) and allowing you to look at the relevant information as mentioned already. You should be able to do this after a block of code has been executed or even after every statement is executed. The debug hardware module and the software provided gives quite a sophisticated set up. You should be able to debug most kinds of systems implemented using the DQ80251.

10 steps to selecting a microcontroller

In an article in EDN, 10 things to be looked at for selecting a micro-controller were discussed. The items discussed were 

1. Looking at the interfaces you are going to require for connecting to external devices
2. Look at the software architecture
3. Select an architecture
4. Assess memory needs
5. Evaluate microcontroller and select one
6. Cost and power constraints study
7. Check availability
8. Make sure development software support is available
9. Take up testing of actual devices

This is a good check-list. But, do remember that it is an iterative process, not a sequential one. You will need to go back and forth between the steps. For example, at step 6 you find that the controller you selected at step 5 does not meet your power budget, you will need to go back and look at other alternatives as in step 5.

At step 1, the problem may look like a simple hardware issue. Not so with embedded systems. Hardware and  the external connections influence what amount of software you might need. If the connections are for USB or may be even Ethernet (communications in general) will need software support of the communications protocol stack.

Software architecture will determine if it will be able to support the processing you require, how often you need interrupts to be serviced, how complex those services are, if you need floating point support. In fat you will need to consider here if floating requirements are heavy enough to make sure floating point support is available on-chip.

You should have a fair idea of what kind of micro-controller you are going to need. Current needs, likely future needs, costs involved, software development tool available are some factors to be considered to arrive at a couple of likely candidates.Likely memory requirements and what addressable memory is available with the controller are some other considerations too. If you are trying to make a minimal hardware system, you need to look for adequate amount of on-chip memory. This assessment should include some extra resources as features always creep up and additional memory is required eventually. If planning on external memory, the hardware needed should be minimal and address logic should be simple. Often the external memory access is slower. You will need to make sure the access time will be acceptable.

Rest of the steps are quite self evident. Of these, the software support point is very crucial though.Read the EDN article here.
Full credentials of Debasis

8051 micro Has Tremendous Staying Power, Evolves into Quad-Pipelined Soft Core

The latest avatar is a soft core and much evolved version of the 8 bit microcontroller family. The designing company the Digital Core Design has been working with the 8051 for a long time, for past 11 years! This core is a 16/32 bit version of the industry standard and much popular 80C51 and 80C251 devices. While the 80C51 was, as we all know, was a preferred 8 bit device, the 80C251 was an evolve 16 bit version. This core being offered has a 32 bit option too.

The devices you could create out of this core could be more than 50 time faster that the 80C51 and about 5 times faster than the 80C251. One feature that contributes to the speed up is the heavily pipeline processor core. What is more interesting is that it will consume much less power to get that performance. The on-chip configurable debug unit (compatible with JTAG) is interesting. The PMU or the power management unit is useful for managing the power use and optimizing the power to performance. Rest of the peripherals are predictable and compatible with what you have come to expect in the 8051 devices. The designers provide all the support needed for compiling specific devices including the VHDL and Verilog source code.

This sure is an interesting device and we plan to spend more time on the same in future posts.

For complete details about Debasis please look up the this site.

Apple's A6 Processor

There has been huge interest generated about Apple's A6 processor being used in the latest iPhone 5. That is largely because Apple had not tried processor design at all so far. Its last processor A5 was a Samsung design and was based on ARM designs. Speculations continued as Apple acquired processor design teams. Now it is confirmed that this team has delivered and delivered quite a punch apparently. This new A6 is equivalent of a ARM Cortex A15 core but is an entirely different design. It is compatible externally to ARM core alright[2].

It has delivered a great performance for the iPhone 5. A recent tear-down report says the CPU core has been manually laid out rather than some EDA tool doing it, for the best possible performance[1]. A CNET report confirms that the performance is faster than expected[3].

It will be interesting to see what else comes out of this design group. Also interesting will be if innovation continues into the time when Jobs is not around. In one of my post on providing air-cover for a team to innovate, I had found that Jobs was capable of it. Will it continue in the new dispensation!

References:

1. MacRumor, Teardown of Apple's A6 Chip Reveals Manual Layout of Custom Dual-Core CPU, http://www.macrumors.com/2012/09/25/teardown-of-apples-a6-chip-reveals-manual-layout-of-custom-dual-core-cpu/
2. International Business Times, Apple Inside: Engineers Tweaked A6 Chip, Expert Says,http://www.ibtimes.com/apple-inside-engineers-tweaked-a6-chip-expert-says-843109
3. CNET, Apple's A6 processor appears faster than previously thought, http://news.cnet.com/8301-13579_3-57521088-37/apples-a6-processor-appears-faster-than-previously-thought/

Intel to bundle Liquid Cooling with Retail Sandy Bridge-E Processors?

First news of Intel bundling a liquid cooler with its high performance processors appeared back in Apr this year. Intel made a reference to it during it's developer's forum at that time.

 With processors being built with dynamic clocking, they can be used in both directions. One is to slow down to clock to slow down and thus reduce power consumption. On the other hand you could "overclock" to get the last ounce of performance out of them. Thermal dissipation will obviously increase. Overclocking has been the tricks used by enthusiasts, particularly gaming guys, to squeeze out the last bit of performance. Liquid cooling has been a trick used by this crowd. Now by providing this option Intel is letting board designers and hence anyone needing it, the option of getting enhanced performance through this. Last checked, it is actually being offered now. This is how it is being offered according to Tom's Hardware.

Top engineering schools ranking

Top universities and engineering schools rankings came out as in every other year. The rankings were taken by the usual suspects. These rankings give rise to the usual debates too. Main issue, that resonate with what I speculated on a couple posts back, is if the teachers are the best engineering teachers(for engineering schools, in particular). Some of the comments are significant and I quote from them in the following. Some of the points relevant have been highlighted. As I had mentioned in my post earlier, where is the emphasis on "teaching" engineering!

"December 29, 2011

In response to: Top engineering schools ranking
Glen Bishop commented:

Allegiance to one's alma mater seems to be clouding the idea of "best" engineering school. As a Princeton freshman, I lived the feeling of "woe unto me for being the dumbest member of the entering class". With high school grades and test scores in the top one percent, I probably WAS the dumbest in the class. Other students breezed through material that I found challenging, but then again the undergrad curricula in junior year featured compilers, quantum mechanics, finite element analysis, and digital signal processing. Princeton offers undergrad classes with 30 students taught by a full professor, and labs stuffed with cutting edge equipment. The "Harry Potter" style dining halls offer a clue, but the conversations with students and faculty make it clear that this is one of the "best" schools. Beware though: the "best" school is definitely NOT the "best" choice for the average or even the above average engineering student. Students are advised to select schools that can teach no more and no less than what the student can absorb in each 12 week term. Hence the "best" school is a "bad" choice for the majority of students. We are all equal before god and the law, but final exam scores at those "best" schools will reveal something else."

"December 29, 2011

In response to: Top engineering schools ranking
MA Engr commented:

Lots of great comments here. I received my BSEE/MSEE in the early '80s. Things have changed much since then. Now I have been helping my sons select universities. Having been in the business awhile, I am appalled at some of new grads who can't even figure out basic circuits. Are the "best" schools weighted by research grant money, famous faculty, etc? Which is all great, but perhaps what should get more weighting for non-academic career people is how good is the TEACHING. Are all classes taught by grad students? Are they really learning fundamentals well, or are they jumping into programming ASICs and FPGAs before they even know what a resistor is? Look at things like class size, curriculum, faculty focus on teaching."

"September 30, 2011

In response to: Top engineering schools ranking
Disgruntled commented:

For my undergraduate BSME, I went to an ABET-accredited college ranked slightly below the middle. It was mostly a friendly school, and I enjoyed learning and studying there. It was in some ways the best time of my life. I’m not bragging, but just to establish the situation: I did very well there, being first in the Mechanical Engineering Department out of 92 people.
For my graduate MSME, I went to Stanford. I was rather naïve, and didn’t even realize that it had such a good reputation. I was pushed into applying there by several of my professors; without my even knowing much about Stanford itself.
There is no doubt that attending this institution has caused people to pay more attention to my resume, and opened some doors to me that would have remained closed if I’d stayed at my 3rd tier undergraduate school.
Nearly all of the students at Stanford were very bright, some true geniuses. But also, some of them were literally high-functioning autistics - with no social skills at all, often very rude or non-communicative. At my previous lower-ranked undergraduate school, the upper quarter of the class was also fairly bright. The main difference from Stanford, was that the bottom half of the class wasn’t so bright; and probably stood little chance of getting a real engineering job, unless they had connections. 
In Stanford, I was under the most stress I have ever been in my life. It was very competitive. Everyone was out for themselves, and no one helped anyone else. The course load was very hectic, and I came close to having a nervous breakdown. Even though most of the students were supposedly selected based on academic achievement and merit; there was a very elitist, snobby, moneyed attitude present. Coming from the middle class, I felt patronized by the rich students. The popular topic of conversation around the dinner table was where everyone had gone on their third trip to Europe, and what wonderful art museums they had seen there. (I had never even been outside of the US.)
And some of the teachers at Stanford were really not all that good. Because the school selected only the highest-ranked students, they could have probably just taught basket-weaving to the students for the whole time they were there. Once graduated, because of their innate talent, most the students would have probably done all right anyway, and adapted to the real-world needs of their jobs.
I had always intended to get my Phd. However, all of the stress was too much for me. After getting my MSME; I fled to a situation where I could go home at 5 in the evening and relax; instead of studying 80 or 90 hours a week. I became too busy with work and home life to ever pursue a Phd again. Stanford, and its elitist attitude, broke my self-confidence. I now regard attending this so-called top-ranked school, as being one of my life’s greatest mistakes.
Looking back over 35 years now, the main moral is: A school’s ranking isn’t everything. Select a school that you’re comfortable in. It may not necessarily be the top-ranked school, but rather a school that fits your personality and lifestyle."

Visicalc-Grandmother of all spreadsheets, Excel included!!

Last Oct 22 VisiCalc was launched. I am fortunate to have seen an early version on a HP machine. The link provided here is by the developers (Dan Bricklin and Bob Frankston) of the spreadsheet and there is a lot of information about the program. 


This is one solution I have always admired, nice, elegant and simple! The qualities any elegant solution should have. This is great.