Compact flash transfer rate (w/microcontroller)

Thomas Magma wrote:
> According to CompactFlash Revision 3.0 the maximum transfer speed can be up 
> to 66MB/sec and with a DMA transfer mode.

Yes, but isn't that value only the speed of the interface? Isn't the 
physical medium (ie. the flash memory) going to be much slower than this?

For sure, internal buffers or cache memories are going to be included, 
but isn't the real limit the write speed of flash memories?

Bye, Aldo

"Haldo" <haldo@somewhere.net> wrote in message 
news:da368l$5ba$1@newsfeed.cineca.it...
>
>
> Thomas Magma wrote:
>> According to CompactFlash Revision 3.0 the maximum transfer speed can be 
>> up to 66MB/sec and with a DMA transfer mode.
>
> Yes, but isn't that value only the speed of the interface? Isn't the 
> physical medium (ie. the flash memory) going to be much slower than this?
>
> For sure, internal buffers or cache memories are going to be included, but 
> isn't the real limit the write speed of flash memories?
>
> Bye, Aldo
>

No, the write speed of flash memory is not the limit of the write speed of 
CompactFlash cards. Yes, each bit/byte of Flash has a write cycle limit, 
however it is possible to store incoming data to a FIFO buffer then mux it 
out to parallel channels of FIFO's that feed the multiple sections of Flash. 
Then there is no write cycle limit, and the limit then falls back to the 
port speed.

Currently, the real limits are the design costs vs. the available market.

Thomas 

Thomas Magma wrote:

> No, the write speed of flash memory is not the limit of the write speed of 
> CompactFlash cards. Yes, each bit/byte of Flash has a write cycle limit, 
> however it is possible to store incoming data to a FIFO buffer then mux it 
> out to parallel channels of FIFO's that feed the multiple sections of Flash. 
> Then there is no write cycle limit, and the limit then falls back to the 
> port speed.
> 

Sure. You can buffer.
But if you act this way you cannot "sustain" this speed, can you?
Can you write 66MB/sec continously to the flash? Unless you get a FIFO 
or a SRAM buffer of the same size.... :-)

IMHO sustained write speed are at present more or less 10MB/sec 
(according to cf makers). Maybe I'm not up to date.

I was investigating on the limit for streaming continous data to a 
compact flash (or sd/mmc) device.

Bye, have a nice day!
Aldo

Haldo wrote:
> Thomas Magma wrote:
>
> > No, the write speed of flash memory is not the limit of the write speed of
> > CompactFlash cards. Yes, each bit/byte of Flash has a write cycle limit,
> > however it is possible to store incoming data to a FIFO buffer then mux it
> > out to parallel channels of FIFO's that feed the multiple sections of Flash.
> > Then there is no write cycle limit, and the limit then falls back to the
> > port speed.
> >
>
> Sure. You can buffer.
> But if you act this way you cannot "sustain" this speed, can you?
> Can you write 66MB/sec continously to the flash? Unless you get a FIFO
> or a SRAM buffer of the same size.... :-)

What Thomas described was more interesting than just a buffer - it was
hardware capable of writing in parallel to multiply the throughput
(RAID-like?) The increased speed would seem to be sustainable.

>
> IMHO sustained write speed are at present more or less 10MB/sec
> (according to cf makers). Maybe I'm not up to date.
>
> I was investigating on the limit for streaming continous data to a
> compact flash (or sd/mmc) device.
> 
> Bye, have a nice day!
> Aldo

toby wrote:
> 
> Haldo wrote:
> 
>>Thomas Magma wrote:
>>
>>
>>>No, the write speed of flash memory is not the limit of the write speed of
>>>CompactFlash cards. Yes, each bit/byte of Flash has a write cycle limit,
>>>however it is possible to store incoming data to a FIFO buffer then mux it
>>>out to parallel channels of FIFO's that feed the multiple sections of Flash.
>>>Then there is no write cycle limit, and the limit then falls back to the
>>>port speed.
>>>> 
> 
> What Thomas described was more interesting than just a buffer - it was
> hardware capable of writing in parallel to multiply the throughput
> (RAID-like?) The increased speed would seem to be sustainable.
> 

Ok. But is this the way current compact flash memories work?

Greetz,Aldo

Haldo wrote:
> toby wrote:
> >
> > Haldo wrote:
> >
> >>Thomas Magma wrote:
> >>
> >>
> >>>No, the write speed of flash memory is not the limit of the write speed of
> >>>CompactFlash cards. Yes, each bit/byte of Flash has a write cycle limit,
> >>>however it is possible to store incoming data to a FIFO buffer then mux it
> >>>out to parallel channels of FIFO's that feed the multiple sections of Flash.

True in theory.  But think about the pin count.  This is done
internally on the flash chip, but not externally.

> >>>Then there is no write cycle limit, and the limit then falls back to the
> >>>port speed.
> >>>>
> >
> > What Thomas described was more interesting than just a buffer - it was
> > hardware capable of writing in parallel to multiply the throughput
> > (RAID-like?) The increased speed would seem to be sustainable.

The idea is already in the flash chip.  It is also the main legal issue
on the on-going case of Lexar vs. Toshiba.  The argument is whether
Toshiba came up with this simple idea (as they argued) or they learned
it from Lexar (several years ago).

> >
>
> Ok. But is this the way current compact flash memories work?

Not really.  The host is probably talking to the on-board controller
first.  The controller then feeds data to the flash chip, in 8 or 16
bits data path.  Many CFs are limited to 1.5 to 2.0 MB/s, according to
our actual experiences with them.

> 
> Greetz,Aldo

"Haldo" <haldo@somewhere.net> wrote in message 
news:dadksg$plk$1@newsfeed.cineca.it...
>
>
> Thomas Magma wrote:
>
>> No, the write speed of flash memory is not the limit of the write speed 
>> of CompactFlash cards. Yes, each bit/byte of Flash has a write cycle 
>> limit, however it is possible to store incoming data to a FIFO buffer 
>> then mux it out to parallel channels of FIFO's that feed the multiple 
>> sections of Flash. Then there is no write cycle limit, and the limit then 
>> falls back to the port speed.
>>
>
> Sure. You can buffer.
> But if you act this way you cannot "sustain" this speed, can you?
> Can you write 66MB/sec continously to the flash? Unless you get a FIFO or 
> a SRAM buffer of the same size.... :-)
>
> IMHO sustained write speed are at present more or less 10MB/sec (according 
> to cf makers). Maybe I'm not up to date.
>
> I was investigating on the limit for streaming continous data to a compact 
> flash (or sd/mmc) device.
>
> Bye, have a nice day!
> Aldo
>

Typically in a CompactFlash you have a controller and a Flash chip. Say for 
instance you replaced the Flash chip with a multiplexer, and two Flash chips 
side-by-side. The multiplexer delivers a byte of data to one Flash and then 
the next byte of data to the other Flash (a D-latch holding the data at each 
Flash). You've now doubled the write speed of the card. Now try it with 4, 
8, 16 (ect...) Flash chips in parallel. Heck, why not just design a Flash 
chip that has the multiplexer and parallel Flash sectors built right in. The 
write speed, in theory, should be able to keep up to any RAM chip. Reading 
is just the reverse using a demux.

Thomas