Xilinx courses

I recently took the "Advanced FPGA Implementation (v6)" Instructor-Led Course and came out of it with a fair bit of dissappointment. I don't want to engage in Xilinx-bashing but it bothers me that the course was simply not worthy of the title it was given.

The only reason I might get something out of it will be because I will pour over the 500 page book on my own and experiment for many, many hours. The class boiled down to a bunch of slides (a very small subset of the book, maybe 20%) being read out loud with a degree of re-interpretation. The labs were based on an obscure design that was not introduced at all. So, all you could do in the alloted time was type from the book like a robot and move on. No real learning took place there.

I took the course because, after a two-year effort --starting from scratch-- to learn FPGA's, I thought that an advanced course taught by an expert in the field would be a great way to take my skills up a notch or two. I needed to get to that proverbial last few percent and, frankly, I also felt stuck with regards to timing optimization, floorplanning and other advanced areas. I thought that an "advanced" course would be taught by a peer who'd offer the sort of insight that only comes from significant experience in the field and, yes, inside information. That is certainly not what happened. I can read slides just as well as the next guy. I don't need to pay $1,000, travel and burn two days' work to endure that experience.

So, I wonder. Was this a fluke? Are the other coursed different, better, worst? Are Altera's courses better? It seems that Xilinx contracts out the trainig to a third party (a company called "Technically Speaking". I heard that Altera chooses to use insiders. Is this true? Does it make a difference?

Thanks,

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian
Loading thread data ...

That's where learning happens.

That's a pretty broad topic. Consider picking focusing on a more specific goal.

Learning that was worth the price of admission.

Hey, that's *our* trademark, "if it works, it's a fluke" :)

No. Just different.

Yes.

The problem is that vendors can't help talking about their specific architecture and special features.

There is too little coverage on design entry and simulation, where there is the most to learn.

related thread:

formatting link

-- Mike Treseler

Reply to
Mike Treseler

Martin,

I am sorry you had a bad experience.

I will ask about it. I had heard from others that this particular course was a good one (some of my own staff have taken it), so I am hoping that your experience was not the course, but perhaps the instructor (still unfortunate, and not acceptable).

Aust> I recently took the "Advanced FPGA Implementation (v6)" Instructor-Led

Reply to
Austin Lesea

To answer your question about Altera: Yes, we do have a dedicated training department with teachers who do the classes at customer sites. I understand that our distributor Arrow also leads traning events. For special workshops and things like that (such as the SOPC World events going on now or other internal training events), other Altera employees specializing in that area may present. Occasionally we have had third parties present *on their specific product*, such as those who do synthesis tools.

Here's a link to the Altera Training homepage:

formatting link

Hope this helps,

Jesse Kempa Altera Corp. jkempa at altera dot com

Reply to
Jesse Kempa

I'll have to respectfully disagree with some of what you said.

pour

If all learning could happen from books schools and universities, at all levels, wouldn't have a reason to exist. I think humankind is genetically wired to very efficiently learn through a tradition of verbal communication that cannot be ignored. Of course, a lot of real and very significant in-depth learning happens outside of that context, but one cannot state that this is the only way learning happens at the exclusion of the verbal tradition.

Then there's the issue of efficiency. I've taken a few very well taught courses over the years where, within a few days, you go from a rudimentary understanding of the subject to having a very clear and organized insight from which to build. This isn't so much due to the verbal tradition I was speaking of, but rather because someone who truly understands the subject AND is a good teacher lays out the subject right there, in front of you, to assimilate and build from. Good teachers are worth 1000 books. No doubt about it.

So, if you attend a good course, you can be on your way very quickly. It's a matter of efficiency. And, while it might be true that all in the universe could be learned from books and, these days, the Internet, there's a real imporant factor we must not forget: the business equation. If what you do is a hobby then, by all means, burn time experimenting and reading through hundreds of documents, surfing the Web or playing with dev boards. However, in the context of a business that needs to get product out the door, it is much more efficient to pay someone to show you the ropes quickly and then get on with your work.

scratch--

You misunderstood where I was coming from. Having "graduated" after two years of very hard work in the field I wanted to get an insight into techniques that would let me squeeze another 5% of perfomance out of a design. I also wanted to understand if there were better approaches to the overall subject, at a high level. That's why I went to an "ADVANCED" class, and not an intro to FPGA's. In my mind, if you say you are teaching an advanced class there are a few requirements that cannot be violated. One of them being who teaches that class and what degree of information is communicated.

Let me ask you this. Do you think that spending 45 minutes listing all of the I/O out of a DCM block has a place in an advanced class? Or how many clocks can be distributed in a Virtex II? How about getting into how to properly start-up a DCM with real-world issues and code?

Well, we didn't do the latter. We did the former. And, when a student asked about how to use the phase shifter clock input there was no answer.

$1,000 is a lot of money for a printed version of PowerPoint slides. I would gladly pay $5,000 for a class that had the right content. Money is not the issue here. If you tell me you'll teach an advanced class for $1K, then do it. If that class requires $10K, then tell me so.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

Maybe so. Hart to tell at this point. Certainly the material in the book would open the door to very interesting and useful discussion of advanced topics, none of which were explored, some were recited, others skipped over.

I think there are two groups within your company that might need a flame (not a match) lit under their chairs: web design and education/documentation.

The website can be incredibly retarded and just not up to par with what good website design folks can do today. Sure, it's expensive to hire these heavy hitters, but Xilinx can afford it.

I only have one sample of the education group's output and, as you learned, they didn't put on a good show as far as I am concerned.

I think there are huge gaping holes in the available documentation and devices are getting increasingly more complex. I think there's a need to address this --by experts, not fresh grads-- and it's not being done.

Some of these topics might include floorplanning, design optimization, timing optimization, FPGA Editor, design flow optimization and automation (XFLOW, scripting, command-line tricks, etc.). I'm not talking about being able to download a document describing the various available timing constraints, for example, but a practical, in-the-trenches set of docs treating these topics in order to support designers in both adopting and succesfully utilizing these devices in an already difficult marketplace.

Within the next few months I'll probably have a need to hire a couple of FPGA/Embedded guys, and the realization that I can't seem to rely on even sending them to a manufacturer-provided course in order to enhance their ability to generate accurate designs that perform well is what triggered some of my concern.

Still, this is not a Xilinx putdown but rather costructive criticism. I love the chips and will probably continue to use them for a long time. I have over half a dozen high-performance imaging products in the works and, at this point, all of them have Xilinx FPGA's in them.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

Well, I think there's a very important distinction that needs to be understood. The "intro" or "beginner" classes can probably be taught by just about anyone who passes the cold mirror test and is a decent teacher (although my preference would be to have an experienced individual instead). However, the minute you characterize a class as "Advanced" you better get a guy who's had some skin in the game for a while and can truly shed some light on some of the dark corners of these technologies.

I think all of you guys (meaning FPGA companies) have had it good. Engineers bust their butt's digging and experimenting and figuring things out...digging for information that you should be providing. I hope things change before we get to the 100 million gate devices, 'cause the real cost of designing with these chips is being borne by the OEMs that pay for these individuals to (through no fault of their own) put a lot more hours into a project than might be required with a higher quality of documentation.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

I agree with Martin! I can't speak for the classes, but the rest is definitely true. I hope someone at Xilinx is listening. There is no doubt (at least in my mind) that Xilinx has some great FPGAs, etc. The software seems to be fairly well done, although it could use some improvement. The area of improvement is in simple, useable documentation. If I need to check three or for different areas for a full picture of what it taakes to get a job done, can you at least create a link between the areas. Xilinx could save a bundle in tech support, if they would just improve the documentation. It might even get them a few more customers. It generally is not a good idea to PO the customer. (Usually Xilinx does not do that.)

Theron Hicks

Mart> Maybe so. Hart to tell at this point. Certainly the material in the book

Reply to
Theron Hicks (Terry)

Perhaps a part of the registration process should require the students to email the teacher a list of their most important questions, or list what they're hoping to learn. Then the teacher can say, "No, you won't get what you want from my class, so save your money and time," or "Yes, I'll be able to answer that at the class. I look forward to seeing you there." Then the teacher can better prepare the class to suit the needs of the students.

Good point about efficiency. Of course if you put 10 Virtex-II Pros in each of your products, you might have enough volume to get a dedicated FAE ;_)

--Vinh

Reply to
Vinh Pham

Theron, Martin,

We have a number of folks actively researching this issue, The class mentioned turns out to be one of the most popular (from what the surveys from the attendees say), and the one that gets very high marks from the attendees also.

The instructor has taught this (as well as other courses) for 6 years, and also gets very high marks in his reviews.

It is extrememly important to us to discover what happened in this case, as it seems very unusual, and not in keeping with the survey results, and comments from other attendees.

One bad review posted here does a lot of damage: and we are concerned especially when we have hundreds of glowing reviews for the same class/instructor!

Austin

"Ther> I agree with Martin! I can't speak for the classes, but the rest is definitely

my

if

Xilinx

Reply to
Austin Lesea

Check out publications by part:

formatting link
This was recently added this area to the web site to do just that. If you see anything missing, or have suggestions on how to improve it you can contact me directly since I do not check this newsgroup too often.

Xilinx could save a bundle in tech support, if

We have also put the ability to provide direct feedback on specific documents in our documentation collection. If you ever look at a doc and do not feel it has met your needs, you can tell us by clicking on "Helpful? Yes|No". Be as specific as possible.

Let me know if there is anything else we can do.

Thanks,

Robert

Reply to
Robert T. Binkley

That's what the newsgroup is all about.

. . .

-- I agree that instruction from a qualified and interested expert with recent industrial experience is ideal.

The practical problem is finding such an expert who is also in the business of teaching.

-- I agree that it makes good business sense to pay a consultant or take a course that teaches you something faster than you could learn in yourself. However, my experience with such seminars is that this just doesn't happen beyond the introductory tutorials.

I set aside an hour or two a day for focused self-study, and I don't believe this has adversely affected the time it takes me to get working designs out the door. In fact I think it helps.

I agree with you. This is exactly what happens. This is where I walk out and why I no longer attend.

Yes. A little more *code* please. Design entry, inference, and simulation are routinely neglected.

Maybe other designers will post some good experiences with seminars or classes. I wish I had some to report.

-- Mike Treseler

Reply to
Mike Treseler

Lest there be any confusion, I was not commenting on the classes. I was talking about documentation being unnecessarily difficult to sort through. I am sorry if I gave the indication that I was refencing the classes. I have not taken any of the classes. Documentation is my main frustration area. Even then, _if_ I make contact with someone at Xilinx, I alomost always get a solution eventually. But wouldn'y it be cheaper if I could get the answer without that human intervention at Xilinx. Theron

mentioned

attendees

also gets

as it seems

from other

especially

definitely

least in my

fairly

improvement is

different

least

support, if

more

(Usually Xilinx

book

advanced

over.

flame

what good

heavy

learned,

to

automation

being

and

marketplace.

of

even

their

triggered

I
I

and,

course

your

Instructor-Led

don't

simply

will

hours.

book,

The

So, all

and

expert

two. I

also

peer

experience in

contracts out

I
a
Reply to
Theron Hicks

especially

Please understand that I meant this to be constructive criticism, meant to help improve things, not cause damage.

Let me give you my opinion as to why you have so many glowing reviews, at least from the group I was with.

In short: Corporate students.

Pretty much everyone else at the class was from the same company. They had been sent out on what I like to call a "corporate training tour". It's a fun break away from the office and you get to learn something to boot.

These guys took a number of courses during the same trip. At least one (and probably more) of them was fresh out of school. I had a couple of good conversations with him during breaks and realized that he did not belong in that class at all. When I asked how much FPGA experience he had, he replied that he'd just done the usual labs at school, not much more.

I can't comment on every single person at the class, but you can learn a lot about them based on the questions they ask. Again, it didn't seem that they had enough time at the wheel. When the team leader is asking questions about the I/O of a DCM block in an advanced class I pretty much know that they didn't have a decent look through the VirtexII data sheet. Another couple of guys were asking about how to configure Select I/O in order to have series termination. They were also asking about good development boards, etc.

In general terms, to the uninitiated, the class was wonderful. If you are too lazy to learn the basics (ok, to be fair, maybe didn't have the time?) then the class exposed you to a lot of interesting information that, if researched further, would result in valuable learning (wheen you are coming from that context). For someone without the experience this class was rocket science and you probably left in awe of all that's possible.

Then there's the "I'm happy to be out of the office" effect. Everyone is happy to do that. At least when you are part of a large corporate entity and you are very detached from the financials. You'll probably get raving reviews out of this group as well.

Lastly, it takes caring and ... well ... balls to say what I'm saying. What does a rank-and-file guy gain by saying that the class wasn't adequate? Zilch.

As a small business owner who actually pays the bills, designs hardware, writes code, etc., etc., I'm intimately aware of the value of time and money. I came to the class with a completely different frame of reference. That's why you are not getting my stamp of approval by default, which is what I think most attendees tend to do.

Thanks for looking into it,

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

Martin,

We appreciate the feedback.

However, your comment that "you would not send anyone to the classes" really hurts. That goes beyond your opinion of this one class, and indicts the whole program (unfairly, IMHO).

Your description sounds like the folks in the class you were in were not at the right pre-requisite level, which means a great deal of frustration for the lecturer, and frustration for anyone in the class who expected the material to be useful. Again, I am only trying to see what went wrong (and give everyone the benefit of the doubt).

Like I said, we are looking into this.

Please give Xilinx educational services another look in the future, as our main interest is to enable the student to learn as fast as possible how to get their designs up, running, and in production.

Austin

Mart> "Aust>

Reply to
Austin Lesea

really

whole

I looked back through my posts to see what you are referring to. The best I could find was:

"Within the next few months I'll probably have a need to hire a couple of FPGA/Embedded guys, and the realization that I can't seem to rely on even sending them to a manufacturer-provided course in order to enhance their ability to generate accurate designs that perform well is what triggered some of my concern."

I don't think that I said I would not send anyone to a class. Particularly not in a terminal sense. I sincerly believe that there's a lot of value in sending people to good courses. I've been to a few where, in the span of a week, I went from able-but-disorganized to knowing exactly where to go and huge gains in productivity. So, clearly, nobody in their right mind can state conlusively and absolutely that any and all classes are worthless. At least that's my experience.

I have no doubt that Xilinx offers great courses. The begineer courses are probably very valuable in getting people up and running quickly. I know that when I was getting started I good kick in the right direction would have saved tons of time and money.

I think that the problem might be that there was a disconnect between what the class was supposed to be and the level of experience/knowledge of the instructor (or, at least, what came out during the class). I'd expect a class tagged as "advanced" to be taught by someone like you, or, perhaps, outside experts in the field, some of whom participate in this n.g., like Ray Andraka. That's what I thought the class would be about. A humbling, yet enlightening experience. I would have expected to dive into some juicy and difficult topics, like, how to layout a folded FIR filter to improve performance. Or, the advantages/disadvantages of using SRL's instead of FF's for the main pipe of the filter. Data path alignment techniques. You get the idea. If the course cost $5K, so be it. Money is not the issue here.

I'll repeat a few lines from a prior post:

"Still, this is not a Xilinx putdown but rather constructive criticism. I love the chips and will probably continue to use them for a long time. I have over half a dozen high-performance imaging products in the works and, at this point, all of them have Xilinx FPGA's in them."

at the

material to

Maybe so.

A few suggestions:

1- Have an online quiz of some sort as a way to grade potential students.

2- Have detailed bios on the instructors, including level of design experience, etc.

3- Make class documents (books, slides) available for purchase prior to committing to a class. Credit the cost towards taking the class within a given period of time.

4- Offer the ability to have registered students send questions to the instructor prior to attending the class. Perhaps a statement of who they are and what they want to get out of the class.

5- Have students describe current designs they may be working on and how this relates to wanting to take a class.

6- An advanced class should probably be a week-long class. Maybe that was the main problem here.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

Martin,

"Advanced" is not very definitive, I'll be the first to admit.....

Still working the issue(s). No basic disagreeements here. You found the text to which I was referring, and it was not an endorsement of the course(s) at all. That was my point, so I thank you for restating it specifically for the one course in question (rather than for "a course" which implies any or all).

Austin

Reply to
Austin Lesea

Yup. I've only taken one FPGA course ever. I'm not qualified to, nor can I offer opinion on other courses or training programs I know nothing about. I urge readers of this thread to keep this very clearly in mind.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

If I knew enough to ask the right question I could probably find the answer myself. Or ask here, or ...

The main reason I go to a class is to learn something about a topic when I don't know enough about it to ask the right questions.

Another reason to go to a formal class is to get out of the office and away from your phone/email so you can concentrate on learning for long enough to make some progress. Sometimes you really do learn things by going through "dumb" lab exercises. (Especially if there is a good instructor who can answer questions when something interesting happens.)

------

It sounds like the main part of Martin's comments was a mismatched expectation about what "advanced" meant. Was there a good description of the course? Did the course match the description?

Did a bunch of students show up who weren't ready for an "advanced" class? Maybe the instructor dropped back to their level without noticing that a few people were ready for a tougher course.

--
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
 Click to see the full signature
Reply to
Hal Murray

Not necesarily true. I would expect that anyone attending an intermediate or advanced class could generate a list of things they may have doubts about or want learn. Perhaps problems they've been having that he/she could use help with. Or maybe express a desire to concentrate on a certain topic.

While you may know enough to ask the right question, getting from there to an answer might not be all that simple. You can bounce around the online documentation from manual to manual and not get very far.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
 Click to see the full signature
Reply to
Martin Euredjian

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.