PDA

View Full Version : CF cards - does UDMA make a difference with older cameras?



MattNQ
30-07-2015, 10:32am
Looking at getting a couple more cards for the D3.
My current cards are older Sandisk Extreme pro that came with my cameras. Some 30mb/s some 60mb/s
Can't say I've really noticed the difference between the cards in speed when shooting .

Now the interweb illuminati seem to say that the write speed of a D3 is 35mb/s. It does support UDMA.

I do shoot sport in raw, primarily because in bright sunlight, it leaves me much more control over blown highlights (which often happens in the background when exposing for players faces)
Netball I'm shooting bursts of typically 3-5 shots then a break until the next play, and in athletics, jumps are 4-5 to get the takeoff, mid-air & landing. But I have found that shooting 8 lanes of hurdles, I'm hitting the buffer limits regularly.

My question is - will I really notice the difference between a Sandisk Ultra (non-UDMA) 50mb/s card and a Sandisk Extreme (UDMA) at 60mb/s on my D3?

ameerat42
30-07-2015, 11:05am
Matt. Any write speeds associated with UDMA? If so, how do they compare?
Am.

MattNQ
30-07-2015, 11:20am
Matt. Any write speeds associated with UDMA? If so, how do they compare?
Am.

It varies. There are different versions. UDMA7 seems to be the latest at 120mb/s or 160mb/s. UDMA cards seem to be 60mb/s.
But my older 30mb/s card is still UDMA. Might have to do some more research.

ameerat42
30-07-2015, 11:29am
Well, from my non-Nikon experience, the write speed of the camera caps what the card does.
Only other thing is read speeds, and if different for camera (as usually are for cards) then "faster" might be "better".
Also, what speeds do the cards cite, read, or write? I get benefit from faster cards when reading from a USB3 card reader.
Am.
PS: as camera is USB2.

MattNQ
30-07-2015, 11:39am
Thanks Am,
The read speed in a card reader doesn't bother me so much - I just set it copying & go & do something else
The reason why I ask is Cheap Chips currently have great CF specials;

16Gb/s Sandisk Ultra 50mb/s for $34.80
16Gb/s Sandisk Extreme 120mb/s (UDMA7) for $66.00

32Gb/s Sandisk Ultra 50mb/s for $52.10
32Gb/s Sandisk Extreme 120mb/s (UDMA7) for $98.00

So I'm thinking if my camera only writes to card at 35mb/s, anything over that is overkill, so maybe I should buy two lower spec cards instead of one higher spec card if it makes no real world difference :D

ameerat42
30-07-2015, 12:05pm
Wow! I nearly have the first of these. Haven't seen a price like that since 2013.
I paid that for a 60MB/S back then. You sure it's not 60? - Though even 50 is OK.

MattNQ
30-07-2015, 12:48pm
Wow! I nearly have the first of these. Haven't seen a price like that since 2013.
I paid that for a 60MB/S back then. You sure it's not 60? - Though even 50 is OK.

Yes, 50 it is, strangely enough. Unless that is the limit of non-UDMA cards perhaps?

JJM
30-07-2015, 1:14pm
This is a small quote from a website that might help you

"Burst depth varies depending on whether you're using a standard or UDMA CompactFlash card. And since the 5D Mark II is compatible with UDMA through mode 6, the buffer clearing speeds could get faster. For UDMA cards, the 5D Mark II allows unlimited burst depth when shooting large / fine JPEGs, or as many as 14 Raw files; slightly shorter than the 17 Raw file burst possible in the original 5D, but then that's perhaps understandable given the higher burst speed and vastly higher image resolutions and resulting file sizes. Non-UDMA CompactFlash burst depth in Raw mode is only just slightly shorter, at 13 frames."

Edit: Further reading suggests that upgrading the firmware on your camera can upgrade the UDMA level that the camera is compatible with.

arthurking83
30-07-2015, 2:21pm
Rob Galbraith used to have a database of what cards work faster with which camera(brand and model, etc)

I remember that the Lexar 1000x CF cards worked faster than all other cards, by a significant margin.
The 1000x Lexars all worked at over 40Mb/s on a D700 and D3s, whereas the Sandisk 60Mb/s cards seemed to operate in the mid 30's or so.

I got a 800x Lexar card based on this info(plus a USB3 reader for it too)

The problem is that Rob Galbraith stopped updating the site(and hence database) and I don't know if it's still available for viewing.

The other thing to note too with this, if you do use the D3 for sports, and the most likely situation will be that you'd probably do lots of burst shooting, of lots of images .. all of which adds up to lots of images on each card.
So even tho a card may be capable of only 35(or 40) Mb/s write times on the camera due to the limitations of the cameras hardware, it's always handy to have nice fast read times for when it comes time to get the images off the card and onto the computer.

Note that while there may be many brands claiming the same data transfer speeds(eg. 60Mb/s or 90Mb/s) they don't all operate at the same speeds, and from what I've noticed(and also gleaned from Rob's database) is that particular brands of cards seem to work better with particular brands of camera.
It could be due to the way specific hardware interoperate with each other.

Also note that Lexar use the multiplication factor nomenclature for rating their cards.
So on a CF card, the 1x speed rating is for 150KB/s, so 600x speeds are supposedly(150 x600=90,000) 90MB/s .. and the 1000x cards are supposed to be 150MB/s.

Have you compared the prices of the 'equivalent' Lexar cards at Cheap Chips?
If not it may be worth a look.

I don't have any fast Sandisk CF cards to compare with(only older slower types now), but I do have a single 45MB/s SD card, and it is a lot slower than either of my fast CF cards(a lot more than I'd have thought it'd be).

Another card I have is a Patriot 600x CF card, and I use that to compare the real performance of my Lexar 800x Cf card.
There is an anomaly with this card tho in that I can't get the D800E to recognise it, but on the D300, the two cards feel similarly speedy.
On the computer tho(Win7) via the Lexar USB3 reader, both for read and write times the much older Patriot CF card is faster.
A lot faster in write times tho(just transferring a bunch of normal files and stuff) where I've seen 10-15MB/s faster speeds and usually faster for read times too, but not enough of a difference to call it much of a difference.

(on the PC) Read times are generally in the 80-ish MB/s range .. and when your raw files are 80Mb, it helps immensely.
(again on the PC)Write times are in the 45-65MB/s range. Like I said tho, I can't get the D800E to recognise the Patriot card to do a more strenuous test.

So back on the topic of card brands and camera brands .. I reckon have a look at the Lexar models and if they compare favourably on price to the Sandisk cards, go for them.
Lexar seem to work nicer with Nikon cameras.

MattNQ
30-07-2015, 4:05pm
Thanks Arthur,

The Lexars are fairly on par with the Sandisk Extremes price-wise.

Lexar 800x 16Gb 120mb/s $74.10
Lexar 1066x 16Gb 160mb/s $81.90

Sandisk Extreme 16Gb 120mb/s $66.00
Sandisk Extreme Pro 16Gb 160mb/s $80.60


My other concern is that on Nikon's current recommended memory card list for the D3, they've tested up to UDMA6. (updated 2013)
I'd assume UDMA7 would also work, but sometimes as you know, slight differences can result in the camera getting confused.

I get a little over 600 images on a 16Gb card (14 bit raw). That takes about 20mins to copy to my external drive attached to the pc.

arthurking83
01-08-2015, 2:18pm
AH!
Back when I got my cards(from Cheap Chips too) similarly paced Lexars were always a little cheaper than Sandisks(ie. you paid a small premium for the name Sandisk) .. and Lexars one rung faster(ie. 60MB/s Sandisks vs 90MB/s Lexars) were at the same price point.

So in your price list .. I'd say the 1066x Lexar is good value( but that is from my POV).

Going by the time taken to transfer 16G .. at about 20mins .. sounds like USB2 speeds.
About 15MB/s or something like that.

USB3 is so much faster.

(assuming you have or haven't got assess to USB3).

if you do, then upgrading to a USB3 reader could be an advantage.

if you don't, then think about an updated PC, or a USB3 PCI addon card for access to USB3.

I'm fairly sure that most computers post about 2009 have some form of USB3 connection(even if it's not the most ideal USB3 chip used).

For a comparison.
with the Lexar 800x 32G CF I have via this Lexar USB3 reader, I can transfer 10G in about 2mins(actually 1:50!)
Data starts to transfer at about 130MB/s then very slowly drops to about 100MB/s over the course of the first half of the transfer.
And this is to a quite speedy mechanical HDD, capable of 120 or more MB/s write speeds.

If I transfer the 10G of data to my slowest internal drive, it much much slower.
It initially starts at about 130 or so MB/s then drops quickly to about 80MB/s after about 30sec or so. That's the caching system coming into play(ie. not real transfer to the HDD).
After about 30sec Windows copy past tool reports 2mins 30sec to go for the transfer, but increases slowly.
The overall transfer(once data has actually stopped) takes more like 4mins or so .. or about 30-40MB/s.
This is simply because the old drive is old(tech) and it's maximum write speed is about 30-40MB/s.

From your POV, if your computer hardware isn't capable of taking advantage of the speed of faster hardware, or you don't have plans to update/upgrade any hardware to suit .. then saving the money on faster than usable hardware may be the better choice.
That is, why spend an extra $10 if you're not going to see any benefit .. ever.
But if you reckon you'll end with with a spanking new computer at some point in the future, then I reckon .. for an extra tenner .. the faster card may be the better option.

I'll post some actual transfer rates later if you want to see actual data(easy to do, but time limited tonight).
I do it anyhow as I'm constantly trying to test new/updated/different drivers somewhere along the line.

USB3 has been driving me bonkers recently with some odd random disconnects, and I'm twiddling with stuff before I install Win10 soon.

OH! and if you want to test how fast some of your hardware is(HDDs, SSDs CF cards all memory of some type) .. try Crystal DiskMark.
If you do, get the portable version .. not the install version!
it just runs some tests to see how fast your data transfers can be.

MattNQ
03-08-2015, 10:48pm
Thanks for all that info Arthur.
I think my dodgy laptop's only USB2. My work laptop is USB3, but it has some pretty invasive encryption software on it which can sometimes encrypt portable device attached as well. So I'm not always game to use it.

I tried downloading straight to my internal HDD instead of going straight to my external HDD.16Gb is now only 11 mins as opposed to 20mins.:)
Over the weekend I shot a heap of netball. I didn't find any glaring difference between the 30mb/s & 60mb/s.... but I neglected to make a note of when I did hit the buffer (sadly only about 15-17 shots on the D3 - I assume mine didn't get the buffer upgrade) whether it cleared quicker on the faster card.

Warbler
04-08-2015, 6:59am
You ADF, Matt? The faster cards will serve you well with the replacement for your D3 anyway. I've been intrigued by the CFast standard. These seem to be for the newest as best models.

http://www.sandisk.com/products/memory-cards/cfastpro/

MattNQ
04-08-2015, 9:05am
You ADF, Matt? The faster cards will serve you well with the replacement for your D3 anyway. I've been intrigued by the CFast standard. These seem to be for the newest as best models.

http://www.sandisk.com/products/memory-cards/cfastpro/

Not ADF, but a certain big Telco that everyone loves to hate :D
They have really stepped up their cyber security recently. Even our mobiles are encrypted now...which means you can't then plug the micro SD card into a reader and dowload photos onto a computer.

Replacement for D3? I only just upgraded to the 7yo D3 this year (I am a slooow upgrade!) Unless I win lotto, it will be unlikely for a while :(
If I accidentally found a spare thousand or three I'd probably grab a longer lens first -need something to get me out to 300-400mm...preferably at f4 or faster :D

The CFast standard is interesting - seems to be only video cameras using it at present. With the likes of Arri and BlackMagic using it, it must be good. It is not compatable in any way with CF though. I read somewhere the standard could be capable of 1gb/s transfer

MattNQ
13-08-2015, 11:48pm
Quick card update.
I cheaped out & bought the 50mb/s Sandisk Ultra (non-UDMA) for $34.80.

Shooting rowing last weekend, it was noticeably slower to write from buffer to card than the 60mb/s UDMA card I have, and also seemed slower than my old 30mb/s extreme III (which is non-UDMA)
Not a scientific test I know :D

So will be fine for landscapes, but sport I'll pick up another UDMA card I think.

Warbler
14-08-2015, 6:13am
These are a very good deal, even with the Aussie $ so low. I have the earlier 1000X and it is the fastest card I've ever used.

http://www.bhphotovideo.com/c/product/1031508-REG/lexar_lcf32gcrbna10662_32gb_pro_compact_flash.html

arthurking83
14-08-2015, 9:18am
I just did a quick test with the Lexar 800x on my D800E

just two tests by hand. 15 shots in continuous mode till buffer full.
total size of the 15 files came to 1.08Gb(~ 72Mb per file)

I timed(by hand) the time from the final image to when the small green light switched off to give a rough indication of write performance with this 800x CF card inside the camera.

1.08 Gb of data took between 31.8 and 32.4 sec(with a separate test taking 31.3s to tho) .. so lets call it 32sec.

so the write rate of the card in the camera is approx: 31.5MB/s
This may actually be inaccurate as the buffer writes to the card near the start of the 15shot sequence anyhow. So the speed isn't really an indication of the write time of the camera's internals plus card's capability .. it's simply describes how quickly the buffer clears once full.
Because it makes no difference to us in a real sense .. it can be considered 'the same thing' tho.

The other test I did(where the buffer clear time was 31.3s) also involved another burst session during the write time.
At about the half way point(abt 15s) I shot another sequence of images.
It allowed be 6 more shots .. and I then timed the clearance rate after the buffer filled again .. that was the 31.3s

Alternatively:

On the PC(with Win7) the card's specs are more like:


-----------------------------------------------------------------------
CrystalDiskMark 4.1.0 x64 (C) 2007-2015 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

Sequential Read (Q= 32,T= 1) : 123.109 MB/s
Sequential Write (Q= 32,T= 1) : 51.283 MB/s
Random Read 4KiB (Q= 32,T= 1) : 12.494 MB/s [ 3050.3 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 1.313 MB/s [ 320.6 IOPS]
Sequential Read (T= 1) : 120.587 MB/s
Sequential Write (T= 1) : 23.909 MB/s
Random Read 4KiB (Q= 1,T= 1) : 9.563 MB/s [ 2334.7 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 1.216 MB/s [ 296.9 IOPS]

Test : 1024 MiB [E: 0.0% (0.0/29.8 GiB)] (x3) [Interval=5 sec]
Date : 2015/07/09 13:51:54
OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

The two important figures being the sequential write times.

I dunno what "Q=32, T=1" means and why the "T=1" write times differ down to half the other "Q=32" speed.

I just transfer a few hundred MB of data to it via various methods and it's more like the 50MB/s speed on the PC.

basically what this highlights is that the write times on the PC are going to be different to a certain extent.
While I see more like the 50MB/s using the computer to rate the card, the "T=1" time via Crystal Disk Mark seems to be more reflective of the actual write times that the camera can achieve.
That is, taking into account that the camera writes the files to buffer during the shooting sequence anyhow .. the 24MB/s seems to be a more accurate indicator as to the write time this card can achieve in the D800E.
In another camera it may well be different.

I still reckon it's all to easy to 'overcapitalise' on an accessory such as a memory card. That is, a faster card may not work that much faster in a particular camera brand or model.

MattNQ
15-04-2016, 8:47pm
b




My other concern is that on Nikon's current recommended memory card list for the D3, they've tested up to UDMA6. (updated 2013)
I'd assume UDMA7 would also work, but sometimes as you know, slight differences can result in the camera getting confused.



For completeness, I'm following this old thread up with some anecdotal trivia to bore you all with...
After filling almost all of my cards in one day at a netball carnival last weekend, I figured it was time to get a new one. 4 out of my 5 cards I bought second hand with my cameras, so I am not sure of their age.

Cheap Chips had a 32Gb UDMA7 120mb/s Sandisk Extreme for $73 - seemed to be one of the cheapest oz-based deals around without risking getting a non-genuine one.

So despite a little trepidation, I have had no dramas using the UDMA7 card on the D3, which was last officially tested as compatible by Nikon with UDMA6 cards.

Some internet comments indicated it may be a problem, but as I suspected, the camera seems to simply work to its UDMA6 limits and no further.

Anyway, here's some test buffer clearing times to compare,
14 RAW images fill the D3's buffer.
Single hand-timed test like Arthurs :D

Sandisk Extreme UDMA7 120Mb/s - 22.3s
Sandisk Extreme UDMA (UDMA6 I assume) 60mb/s - 22.6s
Sandisk Ultra (non-UDMA) 50mb/s - 28.3s
Sandisk Extreme III (non-UDMA) 30mb/s - 31.5s

In reality, I tend to shoot 2-5 frames each action sequence for netball, with a few seconds to the next sequence, so it is rare I have to wait for the buffer to clear.
Athletics is different though - when you are trying to cover 8 lanes of hurdlers - I smash the buffer big time :D