--- In l..., "Paul Curtis" wrote:
> Besides,
> there is USB overhead as Michael pointed out. Round-trip times for
> USB are fairly slow.
USB overheads in mine too as I am using FT232BM.
fastest flash prog speed for LPC2138?
Started by ●July 17, 2006
Reply by ●July 18, 20062006-07-18
Reply by ●July 18, 20062006-07-18
----- Original Message -----
From: "jayasooriah"
To:
Sent: Tuesday, July 18, 2006 1:24 PM
Subject: [lpc2000] Re: fastest flash prog speed for LPC2138?
> --- In l..., "Paul Curtis" wrote:
>>
>> Hi,
>>
>> > > Erasing completed in 2.1 s - 242,847 bytes/sec
>> >
>> > Shouldn't this be 400ms?
>>
>> Jaya, don't try to be smart. You might know that you can erase the
>> device in 400ms, but I know what CrossWorks is timing. That time is the
>> time required to reset the CPU, gain control over JTAG, to downloading a
>> program into RAM over JTAG, figuring out how much of the device needs to
>> be erased, erasing the selected sectors on the device (using a single
>> ISP command in many instances), to confirm device erasure, and to update
>> the GUI whilst all this is happening.
>>
>> Why should that be 400ms?
>>
>> --
>> Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
>> CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors
>
> Paul, no need to get defensive.
>
> To setup serial communications, download 400-byte program, run it to
> fetch some 80-byte, format and print these takes me 1.5 seconds.
>
> If I added erase and check blank to this, I reckon I would take just
> over 2s.
>
> May be I am wrong but I was under the impression JTAG allows you to do
> such things orders of magnitude faster.
The JTAG speed is dependent on how fast you can get data in and out of the
interface. Full-speed USB 2.0 only gives 12 Mbps maximum, it can be lot
lower in practise. High-speed USB is 480 Mbps, but needs a more expensive
chip.
Leon
From: "jayasooriah"
To:
Sent: Tuesday, July 18, 2006 1:24 PM
Subject: [lpc2000] Re: fastest flash prog speed for LPC2138?
> --- In l..., "Paul Curtis" wrote:
>>
>> Hi,
>>
>> > > Erasing completed in 2.1 s - 242,847 bytes/sec
>> >
>> > Shouldn't this be 400ms?
>>
>> Jaya, don't try to be smart. You might know that you can erase the
>> device in 400ms, but I know what CrossWorks is timing. That time is the
>> time required to reset the CPU, gain control over JTAG, to downloading a
>> program into RAM over JTAG, figuring out how much of the device needs to
>> be erased, erasing the selected sectors on the device (using a single
>> ISP command in many instances), to confirm device erasure, and to update
>> the GUI whilst all this is happening.
>>
>> Why should that be 400ms?
>>
>> --
>> Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
>> CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors
>
> Paul, no need to get defensive.
>
> To setup serial communications, download 400-byte program, run it to
> fetch some 80-byte, format and print these takes me 1.5 seconds.
>
> If I added erase and check blank to this, I reckon I would take just
> over 2s.
>
> May be I am wrong but I was under the impression JTAG allows you to do
> such things orders of magnitude faster.
The JTAG speed is dependent on how fast you can get data in and out of the
interface. Full-speed USB 2.0 only gives 12 Mbps maximum, it can be lot
lower in practise. High-speed USB is 480 Mbps, but needs a more expensive
chip.
Leon
Reply by ●July 18, 20062006-07-18
jayasooriah wrote:
> --- In l..., "Paul Curtis" wrote:
>
>> Besides,
>> there is USB overhead as Michael pointed out. Round-trip times for
>> USB are fairly slow.
>>
>
> USB overheads in mine too as I am using FT232BM.
>
So what download speeds do you get for 512Kbytes?
Regards
Michael
>
>
>
> --- In l..., "Paul Curtis" wrote:
>
>> Besides,
>> there is USB overhead as Michael pointed out. Round-trip times for
>> USB are fairly slow.
>>
>
> USB overheads in mine too as I am using FT232BM.
>
So what download speeds do you get for 512Kbytes?
Regards
Michael
>
>
>
Reply by ●July 18, 20062006-07-18
Wow... programming 500KB file gets even faster at about 135 KB/s:
lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
++ info: Programming using agent
++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
++ info: Programming image at absolute addres: 0x00000000
programming at 0x00000000
...
programming at 0x0007C000
++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
P.S. The test is made using adaptive jtag clock. This is the homepage
of this tool: www.ronetix.at
lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
++ info: Programming using agent
++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
++ info: Programming image at absolute addres: 0x00000000
programming at 0x00000000
...
programming at 0x0007C000
++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
P.S. The test is made using adaptive jtag clock. This is the homepage
of this tool: www.ronetix.at
Reply by ●July 18, 20062006-07-18
--- In l..., Michael Johnson wrote:
>
> jayasooriah wrote:
> > --- In l..., "Paul Curtis" wrote:
> >
> >> Besides,
> >> there is USB overhead as Michael pointed out. Round-trip times for
> >> USB are fairly slow.
> >>
> >
> > USB overheads in mine too as I am using FT232BM.
> >
> So what download speeds do you get for 512Kbytes?
>
> Regards
> Michael
Michael, I don't have raw rates but only for formatted data transfers
which from memory is about 10 seconds for 4Kbyte boot loader image at
230400 baud. I might do a raw data test to see when USB latencies cut
in as I am sure this depends on the ACK protocol.
The thing that I was surprised was that just to set up the link, and
eraseing a bunch of sectors (using a one IAP call) seems to take as
much using JTAG as it does using serial communiations.
I was aware that JTAG setup costs are high but it pays back in the
sustained transfer rates because clocking speeds are many orders of
magnititude faster than serial baud rates.
I didn't know that setup costs were that high to make serial mode
still viable for small loads, for example if loading just the boot
loaders.
Jaya
>
> jayasooriah wrote:
> > --- In l..., "Paul Curtis" wrote:
> >
> >> Besides,
> >> there is USB overhead as Michael pointed out. Round-trip times for
> >> USB are fairly slow.
> >>
> >
> > USB overheads in mine too as I am using FT232BM.
> >
> So what download speeds do you get for 512Kbytes?
>
> Regards
> Michael
Michael, I don't have raw rates but only for formatted data transfers
which from memory is about 10 seconds for 4Kbyte boot loader image at
230400 baud. I might do a raw data test to see when USB latencies cut
in as I am sure this depends on the ACK protocol.
The thing that I was surprised was that just to set up the link, and
eraseing a bunch of sectors (using a one IAP call) seems to take as
much using JTAG as it does using serial communiations.
I was aware that JTAG setup costs are high but it pays back in the
sustained transfer rates because clocking speeds are many orders of
magnititude faster than serial baud rates.
I didn't know that setup costs were that high to make serial mode
still viable for small loads, for example if loading just the boot
loaders.
Jaya
Reply by ●July 18, 20062006-07-18
Thank you for the comments. Good point that we need to look at
verfiication time as well since this in a mgmt box we cant afford stuck
bits.
-KoTex Racing , Ultimate Racing Experience
--- In l..., "kotex_m3" wrote:
> Looking at a LPC2138 as a possibility to create specialized mgmt boxes
> for racing hobbyist. With out getting into the details we need the
> ability to completely reprogram a device in the pits. Does anyone have
> a feel for how long it takes to program a LPC2138 (or even the
LPC2136)
> with a full 500k ish image over JTAG? Our eval boards are coming up
> slower than expected let alone the effort to optimize and refine.
>
> Hopefully someone has some time measurements already recorded and
could
> share..
>
> Thx a million,
>
> KoTex Racing
>
verfiication time as well since this in a mgmt box we cant afford stuck
bits.
-KoTex Racing , Ultimate Racing Experience
--- In l..., "kotex_m3" wrote:
> Looking at a LPC2138 as a possibility to create specialized mgmt boxes
> for racing hobbyist. With out getting into the details we need the
> ability to completely reprogram a device in the pits. Does anyone have
> a feel for how long it takes to program a LPC2138 (or even the
LPC2136)
> with a full 500k ish image over JTAG? Our eval boards are coming up
> slower than expected let alone the effort to optimize and refine.
>
> Hopefully someone has some time measurements already recorded and
could
> share..
>
> Thx a million,
>
> KoTex Racing
>
Reply by ●July 19, 20062006-07-19
IMHO...is this not funny or what!!!! KoTex racing???
Quoting kotex_m3 :
>
> Thank you for the comments. Good point that we need to look at
> verfiication time as well since this in a mgmt box we cant afford stuck
> bits.
>
> -KoTex Racing , Ultimate Racing Experience
> --- In l..., "kotex_m3" wrote:
> >
> >
> > Looking at a LPC2138 as a possibility to create specialized mgmt boxes
> > for racing hobbyist. With out getting into the details we need the
> > ability to completely reprogram a device in the pits. Does anyone have
> > a feel for how long it takes to program a LPC2138 (or even the
> LPC2136)
> > with a full 500k ish image over JTAG? Our eval boards are coming up
> > slower than expected let alone the effort to optimize and refine.
> >
> > Hopefully someone has some time measurements already recorded and
> could
> > share..
> >
> > Thx a million,
> >
> > KoTex Racing
>
Quoting kotex_m3 :
>
> Thank you for the comments. Good point that we need to look at
> verfiication time as well since this in a mgmt box we cant afford stuck
> bits.
>
> -KoTex Racing , Ultimate Racing Experience
> --- In l..., "kotex_m3" wrote:
> >
> >
> > Looking at a LPC2138 as a possibility to create specialized mgmt boxes
> > for racing hobbyist. With out getting into the details we need the
> > ability to completely reprogram a device in the pits. Does anyone have
> > a feel for how long it takes to program a LPC2138 (or even the
> LPC2136)
> > with a full 500k ish image over JTAG? Our eval boards are coming up
> > slower than expected let alone the effort to optimize and refine.
> >
> > Hopefully someone has some time measurements already recorded and
> could
> > share..
> >
> > Thx a million,
> >
> > KoTex Racing
>
Reply by ●July 19, 20062006-07-19
Sounds good - how much does it cost?
Regards
Michael
> Wow... programming 500KB file gets even faster at about 135 KB/s:
>
> lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
> ++ info: Programming using agent
> ++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
> ++ info: Programming image at absolute addres: 0x00000000
> programming at 0x00000000
> ...
> programming at 0x0007C000
>
> ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
> P.S. The test is made using adaptive jtag clock. This is the homepage
> of this tool: www.ronetix.at
>
>
Regards
Michael
> Wow... programming 500KB file gets even faster at about 135 KB/s:
>
> lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
> ++ info: Programming using agent
> ++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
> ++ info: Programming image at absolute addres: 0x00000000
> programming at 0x00000000
> ...
> programming at 0x0007C000
>
> ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
> P.S. The test is made using adaptive jtag clock. This is the homepage
> of this tool: www.ronetix.at
>
>
Reply by ●July 19, 20062006-07-19
I bought mine for 1590 Euro, it is not a cheap JTAG tools but it is
good and it worths its price.
--- In l..., Michael Johnson wrote:
>
> Sounds good - how much does it cost?
>
> Regards
> Michael
> > Wow... programming 500KB file gets even faster at about 135 KB/s:
> >
> > lpc>flash program ftp://user:pass@.../test.bin bin 0
> > ++ info: Programming using agent
> > ++ info: Image File: ftp://user:pass@.../test.bin
> > ++ info: Programming image at absolute addres: 0x00000000
> > programming at 0x00000000
> > ...
> > programming at 0x0007C000
> >
> > ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
> >
> >
> > P.S. The test is made using adaptive jtag clock. This is the homepage
> > of this tool: www.ronetix.at
> >
> >
> >
> >
> >
> >
> >
> >
good and it worths its price.
--- In l..., Michael Johnson wrote:
>
> Sounds good - how much does it cost?
>
> Regards
> Michael
> > Wow... programming 500KB file gets even faster at about 135 KB/s:
> >
> > lpc>flash program ftp://user:pass@.../test.bin bin 0
> > ++ info: Programming using agent
> > ++ info: Image File: ftp://user:pass@.../test.bin
> > ++ info: Programming image at absolute addres: 0x00000000
> > programming at 0x00000000
> > ...
> > programming at 0x0007C000
> >
> > ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
> >
> >
> > P.S. The test is made using adaptive jtag clock. This is the homepage
> > of this tool: www.ronetix.at
> >
> >
> >
> >
> >
> >
> >
> >
Reply by ●July 19, 20062006-07-19
Peter,
> Wow... programming 500KB file gets even faster at about 135 KB/s:
>
> lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
> ++ info: Programming using agent
> ++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
> ++ info: Programming image at absolute addres: 0x00000000
> programming at 0x00000000
> ...
> programming at 0x0007C000
>
> ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
I assume that you're programming using an Ethernet connection, given the
above, and will be using a 100Mbps connection. All this will provide a
nice fast programming speed, but at a price--$2K approx.
Watch this space. :-)
--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors
> Wow... programming 500KB file gets even faster at about 135 KB/s:
>
> lpc>flash program ftp://user:pass@192.168.3.1/test.bin bin 0
> ++ info: Programming using agent
> ++ info: Image File: ftp://user:pass@192.168.3.1/test.bin
> ++ info: Programming image at absolute addres: 0x00000000
> programming at 0x00000000
> ...
> programming at 0x0007C000
>
> ++ info: successfully programmed 500 KB (512000 bytes) in 3.7s
I assume that you're programming using an Ethernet connection, given the
above, and will be using a 100Mbps connection. All this will provide a
nice fast programming speed, but at a price--$2K approx.
Watch this space. :-)
--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors