X-Virus-Scanned: clean according to Sophos on Logan.com Return-Path: Sender: To: lml@lancaironline.net Date: Fri, 24 Aug 2007 16:52:50 -0400 Message-ID: X-Original-Return-Path: Received: from wind.imbris.com ([216.18.130.7] verified) by logan.com (CommuniGate Pro SMTP 5.1.11) with ESMTPS id 2282680 for lml@lancaironline.net; Fri, 24 Aug 2007 13:18:15 -0400 Received-SPF: none receiver=logan.com; client-ip=216.18.130.7; envelope-from=brent@regandesigns.com Received: from [192.168.1.100] (cbl-238-80.conceptcable.com [207.170.238.80] (may be forged)) (authenticated bits=0) by wind.imbris.com (8.12.11/8.12.11.S) with ESMTP id l7OHHa7N078584 for ; Fri, 24 Aug 2007 10:17:36 -0700 (PDT) (envelope-from brent@regandesigns.com) X-Original-Message-ID: <46CF12B1.1040808@regandesigns.com> X-Original-Date: Fri, 24 Aug 2007 10:17:37 -0700 From: Brent Regan User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax) X-Accept-Language: en-us, en MIME-Version: 1.0 X-Original-To: Lancair Mailing List Subject: Re: fallability in the digital age Content-Type: multipart/alternative; boundary="------------030500060806090305040301" This is a multi-part message in MIME format. --------------030500060806090305040301 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit Colyn writes: <<>> You are assuming that it is possible to implement and CERTIFY a nVidia graphics processor. While it may be theoretically possible, it is highly unlikely. Graphic processors are indeed processors and processors run software and that software MUST be certified under DO178B. So, nVidia must release the crown jewels of their corporate IP to a third party for in depth scrutiny where EVERY conditional state must be "exorcised". From what I have seen of the driver code (NOT pretty) of one of your competitors, certifying anything graphics processor related would take an effort measured in geologic time. The ice caps have melted, sea levels have risen 200 feet, people evolved gills and a sixth finger on each hand and, oh look, the FAA certified a graphics chip for EFIS. What is an EFIS again? The next problem is that if you do get the chip software through certification and have spent the seven figures to do so, the chip manufacturer will not guarantee chip availability for 10 years so you are forced to buy enough chips now to satisfy your product lifetime requirements. At ~$200 a copy that is another six figure expense. Next. the frame buffers are not ECC. Next. How am I going to drive the graphics processor with my OS free system? Will nVidia support a one of a kind, low volume driver? Next. Will nVidea expose themselves to product liability and product support for the sake of a day's worth of sales? "Your honor, my Billy-bob would never have flown his airplane into that mountain unless the graphics processor told him it wer OK." Next. The FAA has recently decided that graphics processors are evil and have ordered the scribes to chisel this in stone. If you want to try to change their minds you are welcome to try. I like nVidia products. I have a high end nVidea card in the workstation that I am using to design the IDU-IV for CFS. Unfortunately that is the closest a graphics processor will get to the next generation of certified EFIS products. Again, making something that looks good on the ground is trivially easy. You can breadboard hardware together for a few thousand dollars and a few weeks time. Making something that performs reliably in flight is the hard part. The vast majority of the experimental EFIS systems out there would not survive or pass the most benign Safety of Flight (SOF) test levels in DO160E. Never mind the higher levels required for certification. The vibration test alone would, as my son observed, convert most experimental EFIS systems into "Electronic Piņatas". John is accurate in saying that innovation frequently comes from a small talented group that works for long hours with little pay to revolutionize and industry. I have been fortunate to be part of one of those small groups that have succeeded in showing the way to revolutionize an industry. Our group, now designing the next generation EFIS, is still small and they still work 14 hour days 7 days a week, and have been doing it for a DECADE. Promises come far cheaper than experience. Regards Brent Regan --------------030500060806090305040301 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Colyn writes:

<<<I have to disagree that you would emit detailed terrain for lack of processing resources.  At this point, the amount of 3d processing power available in a single chip is huge.  (disclaimer: I'm a chip architect for nVidia)
>>>
You are assuming that it is possible to implement and CERTIFY a nVidia graphics processor. While it may be theoretically possible, it is highly unlikely. Graphic processors are indeed processors and processors run software and that software MUST be certified under DO178B. So, nVidia must release the crown jewels of their corporate IP to a third party for in depth scrutiny where EVERY conditional state must be "exorcised". From what I have seen of the driver code (NOT pretty) of one of  your competitors, certifying anything graphics processor related would take an effort measured in geologic time. The ice caps have melted, sea levels have risen 200 feet, people evolved gills and a sixth finger on each hand and, oh look,  the FAA certified a graphics chip for EFIS. What is an EFIS again?

The next problem is that if you do get the chip software through certification and have spent the seven figures to do so,  the chip manufacturer will not guarantee chip availability for 10 years so you are forced to buy enough chips now to satisfy your product lifetime requirements.  At ~$200 a copy that is another six figure expense.

Next. the frame buffers are not ECC.

Next. How am I going to drive the graphics processor with my OS free system? Will nVidia support a one of a kind, low volume driver?

Next. Will nVidea expose themselves to product liability and product support for the sake of a day's worth of sales?  "Your honor, my Billy-bob would never have flown his airplane into that mountain unless the graphics processor told him it wer OK."

Next. The FAA has recently decided that graphics processors are evil and have ordered the scribes to chisel this in stone. If you want to try to change their minds you are welcome to try.

I like nVidia products. I have a high end nVidea card in the workstation that I am using to design the IDU-IV for CFS. Unfortunately that is the closest a graphics processor will get to the next generation of certified EFIS products.

Again, making something that looks good on the ground is trivially easy. You can breadboard hardware together for a few thousand dollars and a few weeks time. Making something that performs reliably in flight is the hard part. The vast majority of the experimental EFIS systems out there would not survive or pass the most benign Safety of Flight (SOF) test levels in DO160E. Never mind the higher levels required for certification.  The vibration test alone would, as my son observed,  convert most experimental EFIS systems into "Electronic Piñatas".

John is accurate in saying that innovation frequently comes from a small talented group that works for long hours with little pay to revolutionize and industry.  I have been fortunate to be part of one of those small groups that have succeeded in showing the way to revolutionize an industry. Our group, now designing the next generation EFIS, is still small and they still work 14 hour days 7 days a week, and have been doing it for a DECADE.

Promises come far cheaper than experience.

Regards
Brent Regan


 --------------030500060806090305040301--