Press Release: HMC150, "the new DVX"

drdimento said:
BTW, on dropped frames . . we have them at least on one of 3 to 10 tapes digitized and far less frequent than when we first started using miniDV tapes when we went to RAID drives AND dual Xeon processors with nice memory. We still have frequent dropped frames like 1 out of 3 to 6 tapes on the straight SATA drive machines . . AND . . other than for outside delivery clients (shoot and ship) all our tape recordings are recycled tapes some having been at least two to three dozen trips (which is certainly scary on that comment on "tape in the camera" comment which will change the way we work henceforth . . we'll mark a first in service on the tape permanently somehow so that we can progressively move old ones out because some of our tapes are at least 4 or 5 years old even though some have set in client inventory for a year or more until the project is previewed, correction edits, and final dumped to tape.
Click to expand...

Ummm... perhaps Dimento here is just pulling our (collective) leg? Is it still April 1st?

I mean... Dimento... you're claiming to be concerned about reliability and using recycled tapes, some 4-5 years old? Nah!

You have dropouts on 1/3 of the tapes you digitize, and accept that as "normal"? But question the reliability of P2 'cause it isn't mirrored? Nah!

You're about to sell your HVX's because the internal structure isn't what you thought... but you haven't actually had any reliability problems? I mean, is your P2 experience worse than your experience with tapes (dropouts on 33%!!!)?

Nah! You're just playing with us!
 
As I understand it, flash memory does wear out eventually. Write cycles degrade the media (but read cycles don't).

I'm not sure why P2 cards aren't essentially RAID 5 arrays (maybe they are?).
 
Yes, this is true - but the "eventually" is on the order of 100,000 writes. So, if write a card once every day, you get about 270 years of use before failure. If you're reckless and use your card 10 times a day, you only get around 27 years. But by that time, you'll want to update to the new 128 TB cards anyway.

I don't know for sure about exactly what happens in the directory maintenance process, that is, if routine "housekeeping" chores are writing filesystem data more frequently - but I think one reason FAT32 is generally chosen for flash devices is that such overhead is minimal. And, there are techniques in common use to re-allocate highly used parts of the filesystem to different physical memory blocks for "wear leveling". So, I'd say, as a conservative guess, you can count on at least 10 years of pretty heavy use from a P2 card...
 
ProjX v2.0 said:
Not sure if you are aware but there are massive amounts of bootleg memory cards out on the market and if you are buying based on finding the lowest price off ebay, chances are you are buying inferior quality bootleg knockoffs and not tested, official versions. Even though they come shipped with very official looking packaging and labels, it is very hard to tell the knockoffs from the official versions and requires some effort to find the distinguishable characteristics. . . . Again, you get what you pay for. There is a reason why P2 hardware and P2 cards are not cheap.
Click to expand...

All our SD cards (other than 1 Transcend card) have been bought from either Best Buy or Staples and not from ebay or anywhere else. Surely one can trust Best Buy and Staples. The 1 Transcend was from eBay and failed right away and so we just chocked that up to the ebay junk yard experience.

All were bonafide ScanDisk or PNY bubble packed off the shelf.
 
Robert M Wright said:
I'm not sure why P2 cards aren't essentially RAID 5 arrays (maybe they are?).
Click to expand...
Indeed, maybe they are. I was told the raid was for speed, not for duplication (as the write-verification takes care of any reliability issues). But perhaps it's RAID 5, which provides both speed and duplication?
 
Robert M Wright said:
As I understand it, flash memory does wear out eventually. Write cycles degrade the media (but read cycles don't). I'm not sure why P2 cards aren't essentially RAID 5 arrays (maybe they are?).
Click to expand...

I've submitted the question to corporate before jumping off the bridge in a tangent so I should get an answer within a few days and the question was specific to the RAID arrangement. Moreover, I can always go to the guy that was at the P2 conference that was trained by the Panasonic P2 product manager at that Alaskian dog race that was over 1,000 miles long because he was there and the entire process was P2 plus they gave him all the "insider" info about P2's and I have his contact info, I just have to look it up.
 
ullanta said:
Ummm... perhaps Dimento here is just pulling our (collective) leg? Is it still April 1st? . . . I mean... Dimento... you're claiming to be concerned about reliability and using recycled tapes, some 4-5 years old? Nah! . . . Nah! You're just playing with us!
Click to expand...

I have far more important things to do than "play" with people.

Yes tapes that are 4 to 5 years old . . after all I've never had a failure. However, I never use - used - tapes on a single camera one opportunity shot - HIGH RISK - only on multiple camera low risk AND as I say never had a failure yet. For high risk, we run color bars on the tape for 20 secs and if all is well we move on. Moreover, we always have two or three tapes in our pocket IF there was an issue AND we have already had a tape restrung and only lost seconds of video on that one and only tape in over 5 years.

Lastly, while I don't like any losses, even 1 tape in 3 a few dropped frames (not 30 or 40 or 100's) don't hurt a thing. Even when on the odd occassion we've had about 6 or 8 in a row or like every other or 1,2,6,8,10 it is barely noticable. I used to puke when we got lots of dropped frames when we first started but after we got rolling with newer dual xeon machines and all SATA and SATA RAID ARRAY the dropped frames are at the very most 2 to 4 frames in a row or 2 to 6 in a path of say around 10 or 12 (this last is the worse case scenario) which at the most gives us clicks which we can mark in - mark out - quick mute and 1/30th of a second (one frame) isn't even hardly visible. Our DirectTV has 1000 times that burp in real life.

But a lost session would be death . . ours! And I'm not pulling anyones leg. I'm just trying to sort out how to assure our business that there won't be any issues. Again, a chain is no stronger than the weakest link and as a former industrial scientist I work hard to avoid potentials where we can. My experience has been miniDV tape and the reliabiliy has been quite excellent and even when there were failures (dropped frames and the one breakage I did) we weren't out completely . . just small if any interruption in workflow.

Likewise to date P2 has not cost us any losses but listing to the whole HMC150 and HPX170 thread and then comparing our SD failure rate got me to thinking about the work we've been on with national country western music stars, the hottest of the NHL, a movie star, and several projects that ended up on the history channel and national geographic and what would we have done (and what would our client have done) if these cards had failed or the HDD had failed? I would have been in a lot of trouble. For that reason I hired an extra PA on yesterday's shoot to instantly off load the cards for an 8 hour shoot and in between card exchanges duplicate the folders on a second hard drive just in case and all this running back and forth between two locations that were a 3 min drive from each other. Had I HDV tapes, I would simply have dropped them in my pocket and grabbed the next, changing tapes on the fly during relocations.

Like I say, I love the HVX200 and I want to stay with it PLUS I have worked to sell the I-frame and been loyal Panasonic but I'm starting to think that a lot of clients don't really care as long as the image is A-OK and deliverable and our tape experience has always given us that without fail but I don't have enough experience with the less than 1,000 P2 riskable off loads to know for sure. On one television show alone we do about 400 off loads a year on tape plus unsurrmountable others and thus we HAVE the experience there.

Just trying to help folks understand the "whole" picture instead of the way we just jumped when we heard what we thought was an inside 100% back up - - which may yet be, hopefully Panasonic will get back to me this week early as we have a lot of Hi Def work coming up including some movie sceens for an indy.
 
Barry_Green said:
Indeed, maybe they are. I was told the raid was for speed, not for duplication (as the write-verification takes care of any reliability issues). But perhaps it's RAID 5, which provides both speed and duplication?
Click to expand...

Barry, I didn't know the term but exactly what you say here is what I was told. This is the question that I put back to Panasonic corporate for a response although I did not have the technical term. I am hoping they respond quickly as I have a lot of up and coming HD work and I want to have what I feel totally safe with drawing upon what I know thus far although limited with P2 for most of our P2 work has been multicam stuff where there has been at least 100% back up; 30 to 60 sec commercials (gads of these) where I could always just say I want to reshoot for any number of reasons; and, scripted infomercials which also (atlhough cumbersome and painful) could be reshot for any number of explainable reasons not the least of which could have been "the light was wrong" or the "image frame was not acceptable" which would be truthful . . since it wouldn't have been present at all :o) We actually did do this a lot in the beginning with this camera because the light just simply was not good and the noise level too high so we had to go to 1080 instead of 720 AND we had to lower the LCD brightness like we did on the DVX (it also was off about 1 stop from the camera while the HVX seemed to have been almost 1.5 or 2 stops)
 
What's inside a P2 card? P2 cards are high-precision micro-computers with their own processors, firmware, a RAID controller, and gigabytes of the highest-quality zero-fault solid-state memory chips. A P2 card is an intelligent device that manages the data files — and even does a write-verification step for every byte of memory that gets written to the card (thus assuring fault-free operation). Early P2 cards were manufactured using actual SD memory cards in a striped RAID array, thus increasing the performance far beyond any individual memory chip’s speed. The newest generation of P2 cards dispenses with using individual SD memory cards and actually uses the core memory components.
Click to expand...
http://panasonic.com/business/provideo/p2-hd/faqs.asp

Notice the key words, "a write-verification step", the way to do a write verify is to READ the data you just wrote to make sure it was written correctly. While this doubles the bandwidth requirement on the card, with 600+Mbps bandwidth, who cares that we are now using 200Mbps? Well, by reading and verifying, we care because now we know that it just didn't go through the motions and nothing was written.

This write verification step is why if it's written to P2, it is recorded.
It's a safe bet that if there's a read verify error, the camera is going to warn you.

How many tape drives do a read just after the write?


What about the HMC-150?

Now, let's do the math for AVCHD: If we are recording at the maximum possible rate, 24 Mbps (Mega bits per second); 24 Mbps / 8 bits to a Byte = 3 MBps (Mega Bytes per second).

If the data rate is 3 MBps, why do I need a Class 6 card (rated for 6 MBps) for the maximum data rate? A Class 4 (4 MBps) card is fast enough for 3 MBps!!!

The reason has to be, a write verification step.

3 MBps (maximum peak for write) + 3 MBps (for a read verification) = 6 MBps


So, yes it is possible for a cheap SD flash memory to die, BUT the numbers tell me that there's a read after the write to verify the data and once it's written, it's there safe and sound. What if the read verify didn't match, the only logical answer would be for the camera to give you a card failure message.


I'm thinking that when the HMC-150 comes out, the read-verify function after a write is a very important feature. This is something Panasonic should promote to assure users that recording to SD cards is safe.




Bob Diaz
 
ullanta said:
Yes, this is true - but the "eventually" is on the order of 100,000 writes. So, if write a card once every day, you get about 270 years of use before failure. If you're reckless and use your card 10 times a day, you only get around 27 years. But by that time, you'll want to update to the new 128 TB cards anyway.

I don't know for sure about exactly what happens in the directory maintenance process, that is, if routine "housekeeping" chores are writing filesystem data more frequently - but I think one reason FAT32 is generally chosen for flash devices is that such overhead is minimal. And, there are techniques in common use to re-allocate highly used parts of the filesystem to different physical memory blocks for "wear leveling". So, I'd say, as a conservative guess, you can count on at least 10 years of pretty heavy use from a P2 card...
Click to expand...

If the figure of 100,000 write cycles to failure is anything like MTBF ratings on hard drives, there will be plenty of cards that die much earlier deaths.
 
It's an interesting question -

I tend to have more faith in flash memory failure rates than HD failure rates, just because they're easier to test. "Write cycles" rather than MTBF takes out a lot of variables, and can be tested in a fairly reasonable amount of time. I agree that HD MTBF ratings are rediculous (1 million hours? There haven't BEEN hard drives for even half that much time!) But flash is simpler, has no mechanical issues, and the measure is more reasonable. So, given my extremely conservative calculation, and even discounting that, I think 5 years is reasonable as a minimum estimate? And given that the cards are able to detect and actively correct for "dead memory cells", I think the degradation would tend to be graceful.

But, who knows, until P2 has been out a lot longer and we have decent data?
 
ullanta said:
It's an interesting question - I tend to have more faith in flash memory failure rates than HD failure rates, . . . But, who knows, until P2 has been out a lot longer and we have decent data?
Click to expand...

For us it has been exactly the opposite . . almost every hard drive failure (other than one) was giving us warnings (especially my laptop that I unfortunatly ignored) by acting up, little lock ups, slow to access, or ?? any number of examples but EVERY SD and Compact card failure has been impromtu spur of the moment - worked the last time and now failed. Two we sent to SanDisk because they were less than six months old and they just replaced them but that wasn't what we wanted we wanted the images off them (1 was a vactaion with video clips & the other was a property inspection that cost $400 to redo in the end PLUS client disruption) :o(
 
If you are paranoid by a hvx200a backup to firestore and roll tape too. If you go out and shoot your friends skating and want a nice picture in HD, not too much hdd usage and want cheap media then buy a hmc 150 If you want to agrue about lame stuff become a politician!!!

Lets focus on the camera here not on somebodys work insurance...
 
There are 2 different types of flash memory; Multi-Level Cell (MLC) Flash has a limit of 10,000 erase/write cycles and Single-Level Cell (SLC) Flash has a limit of 100,000 erase/write cycles. Now on the parts I use (PIC Micro with build in flash memory), the number is NOT a MTBF, but a minimum number of times. Thus, you will get at least that number of times. Typical is about 10x the minimum.

Also, these numbers assume that the temperature is in the range of -40C --> 85C (-40F ---> 185F). You can go to page 90 of the specification sheet and see for yourself:

http://ww1.microchip.com/downloads/en/DeviceDoc/41190E.pdf


Even with a limit of 10,000 erase/write cycles, assume 50 work weeks in a year with 5 days a week work. 50 weeks x 5 = 250 work days/year. If you did an erase/write cycle every work day, 250 work days per year x 40 years = 10,000 cycles. So, it seems VERY unlikely that anyone will even come close to the lower limit of 10,000 cycles.



This is a free country, so if anyone feels that tape is safer than flash memory, thay are free to use tape, ... but one nasty little question comes up, how do you know that the tape you purchase for your next critical job is NOT a counterfeit tape that is pure junk?????

Well, I guess you'll find out after the job is finished and you go to load it on the editing system.


Bob Diaz
 
Guys, this debate is getting just plain silly. I mean, again: the engineering departments at over 840 networks and stations (including massive organizations such as the BBC, Fox Business Channel, Fox News Channel, Australian Broadcasting Corporation, New Delhi TV, the Fox Broadcasting Networks, etc) have put this stuff through so much more testing and verification than anybody on an internet forum is going to be able to even imagine -- and they've come out and chosen to trust P2 exclusively.

That game is over. Someone can point out where an individual memory card failed, etc... fine. I can point out that you probably have a your car passengers sitting on a dozen gallons of incredibly explosive gasoline, and you may even have phenomenally explosive propane or natural gas running through your house too. So? Those systems work. The P2 system works. Look at the thread Bob started, asking about user reliability reports. Look at the people who are using it now.

As for individual SD cards or other memory cards, like it or not that's where things are going. Virtually all photography is transitioning over to it. All video manufacturers are moving to it. Hard disks and optical disks and tape have been tried and used and outgrown and rejected, solid state is the present and more and moreso it will be the future. Every major manufacturer supports it.

I understand the concerns and why it's important, but -- it's proven. No more Chicken Little, please; the sky isn't falling. Video systems designed around solid-state memory take all appropriate measures. It works. It's at least as reliable as tape, and (in the case of P2) much more reliable than tape.

If you want to debate solid state reliability, let's move it somewhere else. This whole discussion has gotten way off track.
 
I don't think any AVC-HD camera streams. It's not part of the design. Streaming has to be implemented at the OS level or as a SMPTE-certified transport protocol, and as far as I know nobody's done that or is even thinking about it.
 
It would be very disappointing if AVCHD cameras don't provide a live feed from USB port, like DV and HDV cameras do from the firewire port.
 
You must log in or register to reply here.
Back
Top