Well said John.
My father, now retired was a manufacturer of radio broadcast equipment - Elan Audio.
Take a look through the Products pages and you will see quite a range of Mixers, and all manner of ancillary equipment as required by the Broadcast Industry.
The prices may seem expensive, but what you get is tested and verified gear that enables one stop shopping for any radio station installation or refit/upgrade.
What you don't see are the costs associated with bringing these items to market...ie premises and associated outgoings like utilities and insurances before one joint is soldered.
Then comes design/prototyping time, then metal work, pcb's, component store, etc.
Then comes assembly and test labour (hours plus superannuation, holiday pay, sick leave etc).
To keep current and visible, appearances at trade shows are necessary also.
So this all adds up to high embedded cost in any product that goes out the door.
In this case business success/repeat business was/is ensured by providing high quality/performance/reliability/ergonomics and easy contact with the designer (my father).
Some hi-end prices are difficult to justify, but in the main the prices reflect the materials and development costs.
Dan.
My father, now retired was a manufacturer of radio broadcast equipment - Elan Audio.
Take a look through the Products pages and you will see quite a range of Mixers, and all manner of ancillary equipment as required by the Broadcast Industry.
The prices may seem expensive, but what you get is tested and verified gear that enables one stop shopping for any radio station installation or refit/upgrade.
What you don't see are the costs associated with bringing these items to market...ie premises and associated outgoings like utilities and insurances before one joint is soldered.
Then comes design/prototyping time, then metal work, pcb's, component store, etc.
Then comes assembly and test labour (hours plus superannuation, holiday pay, sick leave etc).
To keep current and visible, appearances at trade shows are necessary also.
So this all adds up to high embedded cost in any product that goes out the door.
In this case business success/repeat business was/is ensured by providing high quality/performance/reliability/ergonomics and easy contact with the designer (my father).
Some hi-end prices are difficult to justify, but in the main the prices reflect the materials and development costs.
Dan.
Windows Start -> SearchBox -> CMD.EXE -> click on its icon
FC /?
gives you the DOS help screen for the FC (file compare) command. You probably want
FC /B file1.extension file2.extension > MaxHeadroomDiffs.txt
then open MaxHeadroomDiffs.txt and see whether it says "files are identical, no differences found"
FC /?
gives you the DOS help screen for the FC (file compare) command. You probably want
FC /B file1.extension file2.extension > MaxHeadroomDiffs.txt
then open MaxHeadroomDiffs.txt and see whether it says "files are identical, no differences found"
Last edited:
That's the thing, I suspect that the bits stored in the memory are indeed different.
From very quick perusal, as I understand it flash memory cells are not binary but multilevel and at the the geometries employed are prone to quantum and 1/f noise.
There is error detection/correction built into the system, and it may be that I am hearing differing system induced noise according to the activity of the interface/controller chip.
I will take a look at the SandyK stuff, thanks.
Dan.
Yes, but then remember that the read electronics local to those memory cells are going to give you the proper binary values, after which they're going to be going through so many logic gates that, well, digital electronics is designed the way it is for a very good reason.
Not that you can't get bit rot/corruption, but we're talking an exceptionally robust system here.
Not that you can't get bit rot/corruption, but we're talking an exceptionally robust system here.
Yup, did this in economics back in the mid-80s. Of course doesn't take into account the engineering types who know what good value actually is.
And JC is right. To make something to a certain quality costs a certain amount. But that certain amount is an order of magnitude lower than the loony end of boutique audio.
I've designed the memory cells, sense amplifiers, I/O regenerators, et al., on big ole whomper memory chips*. They're pretty robust because you're not designing to meet and exceed the datasheet, you're designing to finish #1 best-in-the-world, in the comparison testing done at huge customers who actually perform incoming test and (gasp!) vendor qualification.
It's not whether you beat the datasheet spec, it's whether you beat the datasheet spec with greater margin than Micron and Toshiba and Intel and Sandisk and Macronix and etc. He with the biggest, blackest Shmoo Plot, wins. Of course it covers the datasheet spec hypercube, that is assumed. You don't even get to compete in the final round unless you beat the shioit out of the datasheet spec. It's quite a vicious competition, but I promise you that for any given memory generation, whoever was the Preferred Vendor at IBM and Cray Research and (etc), made some damn fine, damn reliable parts. Because of the competition, everybody else did too.
*see US Patent 4,523,110 for starters. Scott Wurcer can tell you all about Shmoo Plots.
It's not whether you beat the datasheet spec, it's whether you beat the datasheet spec with greater margin than Micron and Toshiba and Intel and Sandisk and Macronix and etc. He with the biggest, blackest Shmoo Plot, wins. Of course it covers the datasheet spec hypercube, that is assumed. You don't even get to compete in the final round unless you beat the shioit out of the datasheet spec. It's quite a vicious competition, but I promise you that for any given memory generation, whoever was the Preferred Vendor at IBM and Cray Research and (etc), made some damn fine, damn reliable parts. Because of the competition, everybody else did too.
*see US Patent 4,523,110 for starters. Scott Wurcer can tell you all about Shmoo Plots.
That doesn't really fly the BER on RAM is spectacularly low, remember one flipped bit on a memory swap and blue screen. Try it some time take a JPEG or any other file and flip even one bit = trash (unless you are lucky and find some unused padding). In any case if the data is check-summed it is always trashed by even one bit flip. Quantum and 1/f noise with respect to solid state memory is meaningless, memory cells fail mainly due to ionizing radiation due to rare cosmic ray hits. There is virtually zero tolerance for bit errors in a piece of machine code, one bit can dramatically change an instruction.
BTW under Linux dd and od are useful for looking at files on the bit by bit level. I have done this and used Cooledit too, nothing to report, have never got anything but bit perfect results on any of these tests.
EDIT - You guys really have to come up with a better story, this line of reasoning is just a waste of time.
Last edited:
Is memory still a winner takes all thing? I remember TI pulling out of memory back in the early 90s as the risk:reward ratio was not as they liked.
Of course memory is not unique, but with mask sets in the millions for the mind bogglingly small feature sizes they now use it is at the extreme end. And to think back in the 80s they were convinced optical litho would crap out at 0.1micron
Of course memory is not unique, but with mask sets in the millions for the mind bogglingly small feature sizes they now use it is at the extreme end. And to think back in the 80s they were convinced optical litho would crap out at 0.1micron
Sure, no argument.
The first interestingly thing is that the filter improves/fixes playback of any in this case, mp3 file.
This could be due to altering read behaviour or altering the system decoder stage, and probably both.
I am getting further difference according to the filter being fitted during the file write process.
Having designed DRAMs at TI and observed their penchant for self-fellatiation, I suspect they pulled out of the memory market when they realized they were the 8th best memory vendor on the planet, while major customers With Three Letter Names only bought from the top 3. But that's just a guess.
It was more that first to market with bigger/faster/preferred by now moribund TLA companies reaped the benefits of higher prices, but the drop from premium to commodity pricing went from years, to 18 months to 6 months to 'premium?get knotted' .
Millions or billions to tool for a new process and its in £10 SD cards. Less risk to bet on horses from the outside!
Millions or billions to tool for a new process and its in £10 SD cards. Less risk to bet on horses from the outside!
The only plausible thing going on is the decoding software/filters are different. You'll see that in a bit-compare of the respective output streams.
That's just a waste of time, of course two different digital filters will not compare bit by bit.
- Status
- Not open for further replies.