CF write speeds

[Graham00]

Can somebody please tell me what is faster, a card thats x133 or 15mb/sec write speeds, thanks.
I find the x133 is slower buffering RAW's

 

[admirable]

133x is 20MB/s or 160Mbits/s

 

[Chris Primadona]

Theres a good link on wiki that explains it pretty well, but x133 is just under 20mb/sec.

http://en.wikipedia.org/wiki/CompactFlash

 

[Graham00]

Thanks for that

 

[Snapsh0t]

Unfortunately, a card's claimed speed may not be related to its real-world performance. Have a look at Rob Galbraith's site to get some realistic info.

 

[hollis_f]

Theres a good link on wiki that explains it pretty well, but x133 is just under 20mb/sec.

Actually, it's 20MB/s which is 8000 times faster than 20mb/s

Often it doesn't really matter if one capitalises units correctly - we all know what KG means, even though it should be kg. But 'b' means 'bits', not bytes - which is 'B'. This wouldn't matter, but you'll often see transfer rates of Mb/s - where they're talking about megabits per second.

so 20mb/s is 20 millibits per second.

 

[alexam]

Hi Graham,

To cope with download speed to the card, I always use the 60MB/s CF cards with my 7D as they cope well with continued 8fps bursts and never have a buffer slow down. I have several 8GB and a couple of 16GB cards, which are the SanDisc Extreme UDMA.

Malcolm

 

[Chris Primadona]

Actually, it's 20MB/s which is 8000 times faster than 20mb/s

Often it doesn't really matter if one capitalises units correctly - we all know what KG means, even though it should be kg. But 'b' means 'bits', not bytes - which is 'B'. This wouldn't matter, but you'll often see transfer rates of Mb/s - where they're talking about megabits per second.

so 20mb/s is 20 millibits per second.

Well if we are going to nit pick!
Milibytes are not often used to mesure data. We use the the term "Byte" and since I didn't use kbyte/s it would be fair to assume what the "m" represented
1 bit = a 1 or 0
4 bits = 1 nybble
8 bits = 1 byte (B)
1024 bytes = 1 Kilobyte
1024 Kilobytes = 1 Megabyte
1024 Megabytes = 1 Gigabyte
1024 Gigabytes = 1 Terabyte
1024 Terabytes = 1 petabyte
1024 petabytes = 1 exabyte
1024 exabytes = 1 zettabyte
etc etc

 

[hollis_f]

Hi Graham,

To cope with download speed to the card, I always use the 60MB/s CF cards with my 7D as they cope well with continued 8fps bursts and never have a buffer slow down.

If you shoot raw then card speed has zero effect on how many shots before the buffer fills. It does have an effect on how quickly the buffer empties.

 

[hollis_f]

1024 Megabytes = 1 Gigabyte

Not if you're a manufacturer of hard drives. For them 1000000 Kilobytes = 1000 Megabytes = 1 Gigabyte

 

[clownfish]

So I think we are all agreed
Bytes are usually used to measure capacity and bits are usually used to measure speed.
There are 8 bits per byte therefore a MB is 8 times a Mb. The confusion comes when people use bytes for speeds. The simple capitilization of the b makes a big difference.

 

[alexam]

Hi Frank,

I do shoot in Raw and sometimes Raw & Jpeg, that's why the higher speed helps to clear the buffer, so the camera doesn't slow down. Very useful when using 8fps bursts regularly . Only the 90 Mb/s would be faster, but far more expensive.

Malcolm

 

[cam1986]

Can someone recommend a good but not costly CF card to buy?

 

[Vertigo1]

If you shoot raw then card speed has zero effect on how many shots before the buffer fills. It does have an effect on how quickly the buffer empties.

But the latter affects the former - think about it.

The buffer is flushing images to the card as you continue to shoot, so the faster it can flush images, the more shots you'll be able to take before the buffer finally becomes full and the shooting speed slows.

 

[hollis_f]

But the latter affects the former - think about it.

I did think about it. However, being a scientist I decided that wasn't good enough.

The buffer is flushing images to the card as you continue to shoot, so the faster it can flush images, the more shots you'll be able to take before the buffer finally becomes full and the shooting speed slows.

That's the way you would think it would work. And it's the way I expected it to work. But it isn't the way that it does work. I tried shooting several cards, ranging from a 60MB/s 32GB Sandisk to a 1.5MB/s 1GB non-name card. Now, a full burst on the 7D lasts 2 seconds. You'd think that the fast card would be able to have around 5 or 6 raw files writ to it in that time. But that's not what happens. The 60MB/s card fills the buffer with exactly the same number of raw files as does the 1.5MB/s card. LINK

So my little experiment suggests that the camera fills the buffer before it even thinks of writing to the card.

 

[admirable]

http://www.talkphotography.co.uk/forums/showthread.php?t=344510

 

[Canon Bob]

The buffer is flushing images to the card as you continue to shoot, so the faster it can flush images, the more shots you'll be able to take before the buffer finally becomes full and the shooting speed slows.

That's the way you would think it would work. And it's the way I expected it to work. But it isn't the way that it does work.

I actually think that you're both correct here (and both wrong too)...albeit in slightly different circumstances. I'm basing this on the 1DMkIV but have no reason to suspect that the 7D behaves any differently other than maybe having a smaller buffer.

In Hi-speed burst, there simply isn't time to achieve any meaningful buffer flushing between shots and the buffer will fill after the same number of shots (+/- a small percentage) whether you use 30MB/s regular cards or the latest and greatest Class 6 UDMA's at 90MB/s.

If you shoot in Lo-speed burst then the results are very different as the inter-shot time allows for some serious emptying...the total number of shots before the buffer is full (compared to Hi-speed) will be around 250-300% better for the Class 6 UDMA's whilst the 30MB/s will only see a measly 50% improvement.

Bob

 

[Vertigo1]

Interesting, perhaps the camera just isn't capable of both writing to and reading from the buffer simultaneously, or at least not without hampering write speeds too severely, so it only flushes during 'idle' time?

 

[Faldrax]

Re: Emptying the buffer.

I know that the Sony A700 does 'simultaneously' write to and empty the buffer - a faster CF card will get you more shots at 5fps then a slower one.

Not in any way wanting to get into an X is better than Y debate, but anyone know what the situation is on the Nikon D300, as that is what the OP is using?

 

[Canon Bob]

Re: Emptying the buffer.

I know that the Sony A700 does 'simultaneously' write to and empty the buffer - a faster CF card will get you more shots at 5fps then a slower one.

UDMA cards will certainly outstrip the body so, if what you say is correct, you'll never fill the A700's buffer!

Bob

Edit....having read through the specs and reviewers test of an A700, your "simultaneous" write and empty doesn't appear to be fact and the buffer will fill in the same way other marques buffer's fill.

 

[specialman]

Can someone recommend a good but not costly CF card to buy?

I got some sandisk extreme 8gb cards off Amazon recently - cost £35 each. Good write speed for gunning it in raw

 

[skiking]

I got some sandisk extreme 8gb cards off Amazon recently - cost £35 each. Good write speed for gunning it in raw

I was having a buffering problem on a D200 with x133 - well, I was playing and could quite easily hit the limit. I've just bought the above mentioned card and it appears to be OK so far

 

[Vertigo1]

UDMA cards will certainly outstrip the body so, if what you say is correct, you'll never fill the A700's buffer!

Depends on the body, surely? Unless you're on about the A700 specifically?

RAWs on my 7D seem to average about 23MB so at 8fps we're talking 184MB/s which no UDMA card stands any chance of keeping up with.

 

[Canon Bob]

Depends on the body, surely? Unless you're on about the A700 specifically?

RAWs on my 7D seem to average about 23MB so at 8fps we're talking 184MB/s which no UDMA card stands any chance of keeping up with.

Sorry, I wasn't clear when I wrote that. My intention was to imply that the UDMA speeds can outstrip the body's ability to transfer/allow collection of the data rather than the body's the ability to create it.

Two Nikonites seem to think that the faster cards allow bigger bursts so it appears that Nikon do it differently to Canon and Sony (unless we're talking about slower burst rates like 5 or 6 FPS)

Bob

 

[skiking]

I did think about it. However, being a scientist I decided that wasn't good enough.



That's the way you would think it would work. And it's the way I expected it to work. But it isn't the way that it does work. I tried shooting several cards, ranging from a 60MB/s 32GB Sandisk to a 1.5MB/s 1GB non-name card. Now, a full burst on the 7D lasts 2 seconds. You'd think that the fast card would be able to have around 5 or 6 raw files writ to it in that time. But that's not what happens. The 60MB/s card fills the buffer with exactly the same number of raw files as does the 1.5MB/s card. LINK

So my little experiment suggests that the camera fills the buffer before it even thinks of writing to the card.

Complete rubbish :razz: so I've done a test....x400 and against x133 card










b****r, your right :bonk: Exactly the same number of photo's.:bang:

So whats the point....I've just gone out and splashed out £35 on something that doesn't operate any quicker than a £10 card. Seems the camera manufacturers have got their s/w wrong

 

[Canon Bob]

So whats the point....I've just gone out and splashed out £35 on something that doesn't operate any quicker than a £10 card.

It does operate faster.

Seems the camera manufacturers have got their s/w wrong

No they haven't, you've bought something without understanding whether it will fulfill your expectations or requirements.

Bob

 

[skiking]

It only operates quicker when it is moving the contents of the buffer to the card i.e the faster the write speed the quicker it can empty the buffer and allow the shutter to be released again.

As for saying the camera manufacturers have got there s/w right, then I still say its questionable. The efficient way of doing it is to write to the buffer only when writing to the card isn't quick enough.

I don't think the camera instructions - and they are quite basic - are explicit enough to describe the exact flow of data or are the card manufacturers clear that the ultimate speed in writing the data to the card is impacted by the camera manufactures data flow; its this lack of clear information that lead me (and I'm sure others as well going of some of the other post) to think that spending extra dosh on a fast card is going to allow you to make use of the continuous shooting mode thus allowing you to hold your finger down on the button for longer to capture more shots in one burst.

 

[Canon Bob]

I don't think the camera instructions - and they are quite basic - are explicit enough to describe the exact flow of data or are the card manufacturers clear that the ultimate speed in writing the data to the card is impacted by the camera manufactures data flow; its this lack of clear information that lead me

I don't know who's equipment you use so you could have a valid point about the lack of information. Canon do publish this information and make it clear that faster cards will have little or no benefit at 10 FPS but will give big improvements at 5FPS.

Bob

 

[skiking]

I've gone to the dark side - Nikon

I know its late, but I must be missing the point here. The comments made previously say that the data is only transferred once the buffer is full so changes in FPS wouldn't make any difference i.e the number of pics (or data size) would stay static, just captured quicker then you wait for it to transfer to the card, which would be quicker but it still doesn't allow for a more prolong burst the faster the card.

So, if you want to shoot in bursts and then be able to recover (empty the buffer) then the card speed is irrelevant i.e ignore the cards blurb and buy the slowest but highest capacity :shake:

 

[Canon Bob]

I've gone to the dark side - Nikon

I'm sorry, I don't know how Nikon work this or whether they provide the information

I know its late, but I must be missing the point here. The comments made previously say that the data is only transferred once the buffer is full so changes in FPS wouldn't make any difference

There are comments to that effect but they're ambiguous.
My initial post (post #17) should make it clear. If the burst rate is slow enough to allow some buffer clearance between shots then the faster card will be a benefit. If the burst rate is too fast for this to occur then the only advantage of a faster card is seen when flushing the buffer after it has filled.

Bob

 

[hollis_f]

I've gone to the dark side - Nikon

I know its late, but I must be missing the point here. The comments made previously say that the data is only transferred once the buffer is full

That was my supposition, not a definite fact. And Bob's experiences suggests that is only correct if the burst rate is fast and that it will clear the buffer during a burst if the burst rate is slower. When I get the time I'll check that on the 7D.

 

[Faldrax]

UDMA cards will certainly outstrip the body so, if what you say is correct, you'll never fill the A700's buffer!

Bob

Edit....having read through the specs and reviewers test of an A700, your "simultaneous" write and empty doesn't appear to be fact and the buffer will fill in the same way other marques buffer's fill.

Well, an A700 produces Raw that are about 18 MB, so that's 90 MB/s at 5 fps, which is (as previously noted) more than my Sandisk CF are rated at, so the body will fill the buffer faster than it can empty to the card.
Jpegs, which are smaller, will fill the buffer slower than the card and hence have no limit.

According to the manual, after the listing on max number of shots in burst mode;

"These numbers depend on the writing speed of a memory card"

If I get a chance tonight I'll do a test; I have both 30 and 60 MB/s Sandisk cards, which should (If I am correct) show a difference.

 

[Canon Bob]

That was my supposition, not a definite fact. And Bob's experiences suggests that is only correct if the burst rate is fast and that it will clear the buffer during a burst if the burst rate is slower. When I get the time I'll check that on the 7D.

Frank,

The table below is from Canon's technical paper on the 1DMkIV. The three figures in the "burst" column show expected results for 30MB/s, 45MB/s and UDMA 90MB/s cards.
Looking across for the basic RAW and you see only a marginal imrovement with the fast card at Hi-speed. Add in a JPG of any type and the improvement has gone.
The Lo-speed burst gives notable increases with each level of card...buffer flushing in the inter-shot gaps.
This is MkIV data but the 7D uses the same interface and the Hi-speed isn't quite so fast and this may put the variations somewhere between the MkIV's 5fps and 10 fps figures although the numbers will be different due to the 7D's reduced buffer size.
Bob

 

[hollis_f]

Frank,

The table below is from Canon's technical paper on the 1DMkIV. The three figures in the "burst" column show expected results for 30MB/s, 45MB/s and UDMA 90MB/s cards.

Fascinating. There really is a huge difference between (H) and (L). I must try the same on the 7D.