Paul
> Correct, 100Mbit is 100MHz. (Note I think you mean
100Mib/s, a capital B is
> usually Byte...)
:-), That's what I meant...
It's a pity, at least the German Wikipedia does not tell what 100MB/s
really means.
Something to learn more about ...
Cheers,
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
transfer speed of data using uipstack and free rtos
Started by ●February 22, 2009
Reply by ●February 22, 20092009-02-22
Reply by ●February 23, 20092009-02-23
>>running currently i achieved 200KB/s using a 100Mb/s
link through
>>ethernet ...im sending 1450 bytes continously ...how can i increase
>>the speed from 200KB/s to 12MB/s
12M BYTES/sec??? that is impossible. As a rough rule of thumb (to include overhead) just figure 10 bits per byte. 12MB/s would be 120Mb/s, which is faster than Enet-100 line speed.
>> can u send me that tcp/ip software of lpc2468 which supports 54Mbits/sec
I don't think the sources would like that, against license agreement. Here are some vendors I can suggest for high performance enet stacks: Interniche, CMX, Micrium. Micrium can be downloaded for free to try. You still have to do some work to get it setup to run, but that's true for any of these.
If you want something for free, uIP or lwIP is about it. uTasker has a stack also, and their price is about $500. That's relatively cheap, but you will spend time on it. I started setting it up but got pulled off to other things, so I don't know what performance it has. If you want big performance, it generally takes big dollars $$$$$, and expect a lot of fine tuning.
Regards, Chris.
>>ethernet ...im sending 1450 bytes continously ...how can i increase
>>the speed from 200KB/s to 12MB/s
12M BYTES/sec??? that is impossible. As a rough rule of thumb (to include overhead) just figure 10 bits per byte. 12MB/s would be 120Mb/s, which is faster than Enet-100 line speed.
>> can u send me that tcp/ip software of lpc2468 which supports 54Mbits/sec
I don't think the sources would like that, against license agreement. Here are some vendors I can suggest for high performance enet stacks: Interniche, CMX, Micrium. Micrium can be downloaded for free to try. You still have to do some work to get it setup to run, but that's true for any of these.
If you want something for free, uIP or lwIP is about it. uTasker has a stack also, and their price is about $500. That's relatively cheap, but you will spend time on it. I started setting it up but got pulled off to other things, so I don't know what performance it has. If you want big performance, it generally takes big dollars $$$$$, and expect a lot of fine tuning.
Regards, Chris.
Reply by ●February 23, 20092009-02-23
ur very bad in ur calculations 12 MBytes/sec is actualy 100 Mbits/sec
i have all the stacks applications and i have got output with all the stacks...... but i wanted to work without any operating systems in the stack ...like uip and lwip stacks..wer utasker and micrium all OS based stacks
________________________________
From: HM2
To: l...
Sent: Monday, 23 February, 2009 3:19:51 PM
Subject: Re: [lpc2000] transfer speed of data using uipstack and free rtos
>>running currently i achieved 200KB/s using a 100Mb/s link through
>>ethernet ...im sending 1450 bytes continously ...how can i increase
>>the speed from 200KB/s to 12MB/s
12M BYTES/sec??? that is impossible. As a rough rule of thumb (to include overhead) just figure 10 bits per byte. 12MB/s would be 120Mb/s, which is faster than Enet-100 line speed.
>> can u send me that tcp/ip software of lpc2468 which supports 54Mbits/sec
I don't think the sources would like that, against license agreement. Here are some vendors I can suggest for high performance enet stacks: Interniche, CMX, Micrium. Micrium can be downloaded for free to try. You still have to do some work to get it setup to run, but that's true for any of these.
If you want something for free, uIP or lwIP is about it. uTasker has a stack also, and their price is about $500. That's relatively cheap, but you will spend time on it. I started setting it up but got pulled off to other things, so I don't know what performance it has. If you want big performance, it generally takes big dollars $$$$$, and expect a lot of fine tuning.
Regards, Chris.
Add more friends to your messenger and enjoy! Go to http://messenger.yahoo.com/invite/
i have all the stacks applications and i have got output with all the stacks...... but i wanted to work without any operating systems in the stack ...like uip and lwip stacks..wer utasker and micrium all OS based stacks
________________________________
From: HM2
To: l...
Sent: Monday, 23 February, 2009 3:19:51 PM
Subject: Re: [lpc2000] transfer speed of data using uipstack and free rtos
>>running currently i achieved 200KB/s using a 100Mb/s link through
>>ethernet ...im sending 1450 bytes continously ...how can i increase
>>the speed from 200KB/s to 12MB/s
12M BYTES/sec??? that is impossible. As a rough rule of thumb (to include overhead) just figure 10 bits per byte. 12MB/s would be 120Mb/s, which is faster than Enet-100 line speed.
>> can u send me that tcp/ip software of lpc2468 which supports 54Mbits/sec
I don't think the sources would like that, against license agreement. Here are some vendors I can suggest for high performance enet stacks: Interniche, CMX, Micrium. Micrium can be downloaded for free to try. You still have to do some work to get it setup to run, but that's true for any of these.
If you want something for free, uIP or lwIP is about it. uTasker has a stack also, and their price is about $500. That's relatively cheap, but you will spend time on it. I started setting it up but got pulled off to other things, so I don't know what performance it has. If you want big performance, it generally takes big dollars $$$$$, and expect a lot of fine tuning.
Regards, Chris.
Add more friends to your messenger and enjoy! Go to http://messenger.yahoo.com/invite/
Reply by ●February 23, 20092009-02-23
Anand Palaniswamy schrieb:
1. Your language is bad (it is "your" not "ur").
> ur very bad in ur calculations 12 MBytes/sec is actualy 100 Mbits/sec
2. Your math as well:
12MB = 12*1024*1024*8 = 100663296 which is more than the 100Mib/s of the
ethernet
3. You have always an overhead. Even with raw-ethernet you have one of
18 bytes + 8 byte preamble, i.e. 1.7%.
4. 100Mib/s / 12208b = 8191 packet/s => 11.7MBytes/s raw-ethernet
=> 11.4MBytes/s TCP
Of course only in an ideal world where students do their homework
themselves and no one suffers from hunger.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
1. Your language is bad (it is "your" not "ur").
> ur very bad in ur calculations 12 MBytes/sec is actualy 100 Mbits/sec
2. Your math as well:
12MB = 12*1024*1024*8 = 100663296 which is more than the 100Mib/s of the
ethernet
3. You have always an overhead. Even with raw-ethernet you have one of
18 bytes + 8 byte preamble, i.e. 1.7%.
4. 100Mib/s / 12208b = 8191 packet/s => 11.7MBytes/s raw-ethernet
=> 11.4MBytes/s TCP
Of course only in an ideal world where students do their homework
themselves and no one suffers from hunger.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
Reply by ●February 23, 20092009-02-23
Hi,
> 3. You have always an overhead. Even with raw-ethernet you have one of
> 18 bytes + 8 byte preamble, i.e. 1.7%.
More than that, in fact. I calculate (preamble, 2 MAC addresses, Ethertype,
, CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
then you need to put IP headers in the payload...
--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors
> 3. You have always an overhead. Even with raw-ethernet you have one of
> 18 bytes + 8 byte preamble, i.e. 1.7%.
More than that, in fact. I calculate (preamble, 2 MAC addresses, Ethertype,
, CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
then you need to put IP headers in the payload...
--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for ARM, MSP430, AVR, MAXQ, and now Cortex-M3 processors
Reply by ●February 23, 20092009-02-23
--- In l..., "Paul Curtis" wrote:
>
> Hi,
>
> > 3. You have always an overhead. Even with raw-ethernet you have one of
> > 18 bytes + 8 byte preamble, i.e. 1.7%.
>
> More than that, in fact. I calculate (preamble, 2 MAC addresses,
Ethertype,
> , CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
> then you need to put IP headers in the payload...
Yes, the header overhead would certainly limit the data rate. Of
course, the poster may have been talking about bytes on the line, but
even then, the preamble in ENET (a string of a recognizable bit
pattern) takes overhead too.
In any case, uIP cannot possibly get anywhere near that with TCP/IP
because it does not support windowing. TCP/IP is very slow to send one
packet because you have to wait for the ACK. It can be fast to send a
huge block of data by sending out 5 packets and then sending more as
ACKs come back. But, even then, it is hard to get near to line speed.
uIP does not do this windowing, so it will send one packet and block
write until the ACK. Then, it sends the next. So, the time to send
large amounts of data is severely impacted. LwIP does allow windowing,
but how much depends on the pool size.
>
> Hi,
>
> > 3. You have always an overhead. Even with raw-ethernet you have one of
> > 18 bytes + 8 byte preamble, i.e. 1.7%.
>
> More than that, in fact. I calculate (preamble, 2 MAC addresses,
Ethertype,
> , CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
> then you need to put IP headers in the payload...
Yes, the header overhead would certainly limit the data rate. Of
course, the poster may have been talking about bytes on the line, but
even then, the preamble in ENET (a string of a recognizable bit
pattern) takes overhead too.
In any case, uIP cannot possibly get anywhere near that with TCP/IP
because it does not support windowing. TCP/IP is very slow to send one
packet because you have to wait for the ACK. It can be fast to send a
huge block of data by sending out 5 packets and then sending more as
ACKs come back. But, even then, it is hard to get near to line speed.
uIP does not do this windowing, so it will send one packet and block
write until the ACK. Then, it sends the next. So, the time to send
large amounts of data is severely impacted. LwIP does allow windowing,
but how much depends on the pool size.
Reply by ●February 23, 20092009-02-23
my language is bad because its not my mother tongue and its the way of short
form we speak
ya the calculation is correct i just said the approximate value of 11.7 MByte/sec as 12MByte/sec k va
ya i know its really impossible in the practical world atleast we can transfer the data at 80 Mbit/sec - 90 Mbits/sec
________________________________
From: 42Bastian
To: l...
Sent: Monday, 23 February, 2009 7:29:27 PM
Subject: Re: [lpc2000] transfer speed of data using uipstack and free rtos
Anand Palaniswamy schrieb:
1. Your language is bad (it is "your" not "ur").
> ur very bad in ur calculations 12 MBytes/sec is actualy 100 Mbits/sec
2. Your math as well:
12MB = 12*1024*1024* 8 = 100663296 which is more than the 100Mib/s of the
ethernet
3. You have always an overhead. Even with raw-ethernet you have one of
18 bytes + 8 byte preamble, i.e. 1.7%.
4. 100Mib/s / 12208b = 8191 packet/s => 11.7MBytes/s raw-ethernet
=> 11.4MBytes/s TCP
Of course only in an ideal world where students do their homework
themselves and no one suffers from hunger.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
Add more friends to your messenger and enjoy! Go to http://messenger.yahoo.com/invite/
ya the calculation is correct i just said the approximate value of 11.7 MByte/sec as 12MByte/sec k va
ya i know its really impossible in the practical world atleast we can transfer the data at 80 Mbit/sec - 90 Mbits/sec
________________________________
From: 42Bastian
To: l...
Sent: Monday, 23 February, 2009 7:29:27 PM
Subject: Re: [lpc2000] transfer speed of data using uipstack and free rtos
Anand Palaniswamy schrieb:
1. Your language is bad (it is "your" not "ur").
> ur very bad in ur calculations 12 MBytes/sec is actualy 100 Mbits/sec
2. Your math as well:
12MB = 12*1024*1024* 8 = 100663296 which is more than the 100Mib/s of the
ethernet
3. You have always an overhead. Even with raw-ethernet you have one of
18 bytes + 8 byte preamble, i.e. 1.7%.
4. 100Mib/s / 12208b = 8191 packet/s => 11.7MBytes/s raw-ethernet
=> 11.4MBytes/s TCP
Of course only in an ideal world where students do their homework
themselves and no one suffers from hunger.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
Add more friends to your messenger and enjoy! Go to http://messenger.yahoo.com/invite/
Reply by ●February 23, 20092009-02-23
hi the data is transfered through UDP prtocol so there is no need for ACK ..only
TCP requires ACK
________________________________
From: paulkimelman
To: l...
Sent: Monday, 23 February, 2009 9:07:15 PM
Subject: [lpc2000] Re: transfer speed of data using uipstack and free rtos
--- In lpc2000@yahoogroups .com, "Paul Curtis" wrote:
>
> Hi,
>
> > 3. You have always an overhead. Even with raw-ethernet you have one of
> > 18 bytes + 8 byte preamble, i.e. 1.7%.
>
> More than that, in fact. I calculate (preamble, 2 MAC addresses,
Ethertype,
> , CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
> then you need to put IP headers in the payload...
Yes, the header overhead would certainly limit the data rate. Of
course, the poster may have been talking about bytes on the line, but
even then, the preamble in ENET (a string of a recognizable bit
pattern) takes overhead too.
In any case, uIP cannot possibly get anywhere near that with TCP/IP
because it does not support windowing. TCP/IP is very slow to send one
packet because you have to wait for the ACK. It can be fast to send a
huge block of data by sending out 5 packets and then sending more as
ACKs come back. But, even then, it is hard to get near to line speed.
uIP does not do this windowing, so it will send one packet and block
write until the ACK. Then, it sends the next. So, the time to send
large amounts of data is severely impacted. LwIP does allow windowing,
but how much depends on the pool size.
Did you know? You can CHAT without downloading messenger. Go to http://in.webmessenger.yahoo.com/
________________________________
From: paulkimelman
To: l...
Sent: Monday, 23 February, 2009 9:07:15 PM
Subject: [lpc2000] Re: transfer speed of data using uipstack and free rtos
--- In lpc2000@yahoogroups .com, "Paul Curtis" wrote:
>
> Hi,
>
> > 3. You have always an overhead. Even with raw-ethernet you have one of
> > 18 bytes + 8 byte preamble, i.e. 1.7%.
>
> More than that, in fact. I calculate (preamble, 2 MAC addresses,
Ethertype,
> , CRC, interframe gap) = 8 + 12 + 2 + 4 + 12 = 38 bytes... And
> then you need to put IP headers in the payload...
Yes, the header overhead would certainly limit the data rate. Of
course, the poster may have been talking about bytes on the line, but
even then, the preamble in ENET (a string of a recognizable bit
pattern) takes overhead too.
In any case, uIP cannot possibly get anywhere near that with TCP/IP
because it does not support windowing. TCP/IP is very slow to send one
packet because you have to wait for the ACK. It can be fast to send a
huge block of data by sending out 5 packets and then sending more as
ACKs come back. But, even then, it is hard to get near to line speed.
uIP does not do this windowing, so it will send one packet and block
write until the ACK. Then, it sends the next. So, the time to send
large amounts of data is severely impacted. LwIP does allow windowing,
but how much depends on the pool size.
Did you know? You can CHAT without downloading messenger. Go to http://in.webmessenger.yahoo.com/
Reply by ●February 23, 20092009-02-23
> my language is bad because its not my mother tongue
and its
> the way of short form we speak
Its what's known in Blighty as WKS (Wanky Kids Speak).
(yes I'm deliberately using terms not everybody would understand)
Regards,
Richard.
+ http://www.FreeRTOS.org
Designed for Microcontrollers. More than 7000 downloads per month.
+ http://www.SafeRTOS.com
Certified by T as meeting the requirements for safety related systems.
> the way of short form we speak
Its what's known in Blighty as WKS (Wanky Kids Speak).
(yes I'm deliberately using terms not everybody would understand)
Regards,
Richard.
+ http://www.FreeRTOS.org
Designed for Microcontrollers. More than 7000 downloads per month.
+ http://www.SafeRTOS.com
Certified by T as meeting the requirements for safety related systems.
Reply by ●February 23, 20092009-02-23
Anand Palaniswamy schrieb:
> my language is bad because its not my mother tongue and its the way of short form we speak
Neither is it mine. But you should have learned in your mother language
that speaking and writing is different.
But, heck, I get int the troll-trap every time. My &&%"& filter is not
working.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
> my language is bad because its not my mother tongue and its the way of short form we speak
Neither is it mine. But you should have learned in your mother language
that speaking and writing is different.
But, heck, I get int the troll-trap every time. My &&%"& filter is not
working.
--
42Bastian
Note: SPAM-only account, direct mail to bs42@...
