From richard at imagecraft.com Fri May 11 00:47:22 2007 From: richard at imagecraft.com (Richard) Date: Fri May 11 14:48:05 2007 Subject: [Icc-arm] Some mail messages lost due to hard drive failure Message-ID: <200705110757.l4B7uwi1095789@dragonsgate2.imagecraft.com> One of our systems suffered a hard drive failure (two less than a year old Seagate drives died within a span of 2 weeks in 2 different systems, draw your own conclusions?) Fortunately, our back up system minimizes the data loss, less minimal for the email messages because unfortunately the email clients run all the time and the backup utility cannot access the data file. So if you have emailed us in the last few weeks and are still awaiting a response from us, please email again. Sorry for the inconvenience. // richard On-line orders, support, and listservers available on web site. [ For technical support on ImageCraft products, please include all previous replies in your msgs. ] From richard at imagecraft.com Tue May 15 23:29:09 2007 From: richard at imagecraft.com (Richard) Date: Tue May 15 23:39:09 2007 Subject: [Icc-arm] 64 bits FP chip.... Message-ID: <200705160639.l4G6d7mj067739@dragonsgate2.imagecraft.com> I am considering producing something like this: 64 bits FP support will be provided by the new iFPLightning chip. The product is integrated fully into our compilers (initially AVR but later on supporting other ICC compilers as well). The approximate performance goals are 1 uSec for 32 bit FP MUL 1.5 uSec for DIV and ~2X for 64 bits. This is at least 15x-20X faster than the equivalent AVR code and of course free up code space on the AVR chip. The data transfer overheard is ~10 uSec to transfer two operands and a result. Complex expressions will use the intermediate results directly without data transfer. The API uses a stack architecture so interrupt can remain enabled except during the data transfer. The iFPLightning chip comes in a 16 or 18 pin DIP module, ~0.7" x 0.6" in size. The following pricing is very tentative, but is our best guess: 1-50 $30 100 $27 500 $25 1000 $22 It is highly likely that the chip will provide full high level math support such as sin/cos/etc. We are also open to other API such as DSP etc. What do you think? // richard From lundquist at ieee.org Wed May 16 16:01:49 2007 From: lundquist at ieee.org (David Lundquist) Date: Wed May 16 16:11:11 2007 Subject: [Icc-arm] 64 bits FP chip.... In-Reply-To: <200705160639.l4G6d7mj067739@dragonsgate2.imagecraft.com> References: <200705160639.l4G6d7mj067739@dragonsgate2.imagecraft.com> Message-ID: <464B8D5D.8030709@ieee.org> Richard, Interesting, but I'm having a tough time understanding who would find this compelling or what the potential applications might be. If you take a look at modern floating point DSP's I think they will pretty soundly beat this performance at a much lower price. Just doing a quick survey of TI's offering, the TMS320C6720-200 is a amazingly capable device running at 200MHz which can perform 400M floating point MACs per second. It natively supports both 32 bit and 64 bit floating point and 32 bit fixed point. Libraries are available to do the trig functions you mention. 1K price is $6.50. Bottom line, it is much cheaper and much faster than your proposed part. Yes, the startup effort is higher, but anybody looking at moderate volume (>1K?) will quickly justify the work from a cost standpoint. If you tolerate slower, I suspect some of the single chip ARM7 and now the ARM9 based single chip micros with good FP libraries would do nicely too. Just one man's opinion. I think it would be interesting to identify an app where your approach would be the optimal design. I for one can't think of one. One thing you have cause me to reconsider is the lack of low cost tools for these modern high performance DSP's. I haven't looked recently but my recollection is that there isn't much 3rd party stuff and the offerings from TI, Freescale, Analog Devices etc.... are VERY expensive. I might suggest that as a more lucrative avenue. Keep in mind though that the embedded ARM parts keep getting faster and cheaper. Regards, Dave Lundquist Richard wrote: > I am considering producing something like this: > > 64 bits FP support will be provided by the new iFPLightning chip. The > product is integrated fully into our compilers (initially AVR but > later on supporting other ICC compilers as well). The approximate > performance goals are > > 1 uSec for 32 bit FP MUL > 1.5 uSec for DIV > and ~2X for 64 bits. > > This is at least 15x-20X faster than the equivalent AVR code and of > course free up code space on the AVR chip. > > The data transfer overheard is ~10 uSec to transfer two operands and a > result. Complex expressions will use the intermediate results directly > without data transfer. The API uses a stack architecture so interrupt > can remain enabled except during the data transfer. > > The iFPLightning chip comes in a 16 or 18 pin DIP module, ~0.7" x 0.6" > in size. > > The following pricing is very tentative, but is our best guess: > > 1-50 $30 > 100 $27 > 500 $25 > 1000 $22 > > It is highly likely that the chip will provide full high level math > support such as sin/cos/etc. We are also open to other API such as DSP > etc. > > What do you think? > > // richard > > _______________________________________________ > Icc-arm mailing list > Icc-arm@imagecraft.com > http://dragonsgate.net/mailman/listinfo/icc-arm > From jdhharris at earthlink.net Wed May 16 16:27:12 2007 From: jdhharris at earthlink.net (John Harris) Date: Wed May 16 16:36:40 2007 Subject: [Icc-arm] 64 bits FP chip.... Message-ID: <380-220075316232712250@earthlink.net> Hi Richard, I'm with David here. The price is wrong. At $2 each per 1,000 and in a much smaller package it could be interesting Regards ____________________________________________________________ John Harris CustomStage LLC 11 Quail Hollow Drive Tijeras NM 87059 USA Phone: (505)286-8224 Cell: (505)379-3015 jdhharris@earthlink.net > [Original Message] > From: David Lundquist > To: Discussion list for ICCV7 for ARM users > Date: 5/16/2007 5:01:48 PM > Subject: Re: [Icc-arm] 64 bits FP chip.... > > Richard, > > Interesting, but I'm having a tough time understanding who would find > this compelling or what the potential applications might be. If you > take a look at modern floating point DSP's I think they will pretty > soundly beat this performance at a much lower price. Just doing a > quick survey of TI's offering, the TMS320C6720-200 is a amazingly > capable device running at 200MHz which can perform 400M floating point > MACs per second. It natively supports both 32 bit and 64 bit floating > point and 32 bit fixed point. Libraries are available to do the trig > functions you mention. 1K price is $6.50. Bottom line, it is much > cheaper and much faster than your proposed part. > > Yes, the startup effort is higher, but anybody looking at moderate > volume (>1K?) will quickly justify the work from a cost standpoint. If > you tolerate slower, I suspect some of the single chip ARM7 and now the > ARM9 based single chip micros with good FP libraries would do nicely too. > > Just one man's opinion. I think it would be interesting to identify an > app where your approach would be the optimal design. I for one can't > think of one. > > One thing you have cause me to reconsider is the lack of low cost tools > for these modern high performance DSP's. I haven't looked recently but > my recollection is that there isn't much 3rd party stuff and the > offerings from TI, Freescale, Analog Devices etc.... are VERY > expensive. I might suggest that as a more lucrative avenue. Keep in > mind though that the embedded ARM parts keep getting faster and cheaper. > > Regards, > Dave Lundquist > > > > > Richard wrote: > > I am considering producing something like this: > > > > 64 bits FP support will be provided by the new iFPLightning chip. The > > product is integrated fully into our compilers (initially AVR but > > later on supporting other ICC compilers as well). The approximate > > performance goals are > > > > 1 uSec for 32 bit FP MUL > > 1.5 uSec for DIV > > and ~2X for 64 bits. > > > > This is at least 15x-20X faster than the equivalent AVR code and of > > course free up code space on the AVR chip. > > > > The data transfer overheard is ~10 uSec to transfer two operands and a > > result. Complex expressions will use the intermediate results directly > > without data transfer. The API uses a stack architecture so interrupt > > can remain enabled except during the data transfer. > > > > The iFPLightning chip comes in a 16 or 18 pin DIP module, ~0.7" x 0.6" > > in size. > > > > The following pricing is very tentative, but is our best guess: > > > > 1-50 $30 > > 100 $27 > > 500 $25 > > 1000 $22 > > > > It is highly likely that the chip will provide full high level math > > support such as sin/cos/etc. We are also open to other API such as DSP > > etc. > > > > What do you think? > > > > // richard > > > > _______________________________________________ > > Icc-arm mailing list > > Icc-arm@imagecraft.com > > http://dragonsgate.net/mailman/listinfo/icc-arm > > > > _______________________________________________ > Icc-arm mailing list > Icc-arm@imagecraft.com > http://dragonsgate.net/mailman/listinfo/icc-arm