How fast is your Sandisk Extreme III card?

RJL65 said:

These results were tainted by the file system used. The CF card was
formatted as FAT32, all right - but the default FAT32 formatting used
4K clusters, which slow down access significantly. I reformatted the
card using the correct 32K clusters (still as FAT32), and performance
jumped significantly.

Here are my new results using multiple files:
Multiple files totaling 3.77 GB uploaded in 4:22 (14.7 MB/s)
Multiple files totaling 3.77 GB downloaded in 3:07 (20.6 MB/s)

I will retest the card with a single large file later today.

Click to expand...

My re-test using a single large video file results:
2.5 GB uploaded in 2:10 (19.7 MB/s)
2.5 GB downloaded in 1:48 (23.7 MB/s)

RJL65 said:

RJL65 said:

These results were tainted by the file system used. The CF card was
formatted as FAT32, all right - but the default FAT32 formatting used
4K clusters, which slow down access significantly. I reformatted the
card using the correct 32K clusters (still as FAT32), and performance
jumped significantly.

Here are my new results using multiple files:
Multiple files totaling 3.77 GB uploaded in 4:22 (14.7 MB/s)
Multiple files totaling 3.77 GB downloaded in 3:07 (20.6 MB/s)

I will retest the card with a single large file later today.

Click to expand...

My re-test using a single large video file results:
2.5 GB uploaded in 2:10 (19.7 MB/s)
2.5 GB downloaded in 1:48 (23.7 MB/s)

Click to expand...

My previous results were by rough timings.

I updated my results -- now using a neat new program called FC-Test. I also compared my results with the SanDisk Extreme III 30 MB/s Edition cards in both SDHC and CF formats (note that the SDHC results are low because the official SD Card Association specs currently limit the maximum speed to 22 MB/s, and the SanDisk ImageMate All-in-One reader does not use the proprietary timing mode needed for 30 MB/s mode operation with the SDHC card).

My new results are:

SanDisk Extreme III 30 MB/s Edition 4GB SDHC:
Single large video file:
2.45 GB uploaded in 2:49.8 (14.8 MB/s)
2.45 GB downloaded in 2:11.7 (19.1 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 5:31.1 (11.7 MB/s)
3.77 GB downloaded in 3:33.3 (18.1 MB/s)

SanDisk Extreme III 30 MB/s Edition 4GB CF:
Single large video file:
2.45 GB uploaded in 2:09.4 (19.4 MB/s)
2.45 GB downloaded in 1:46.7 (23.6 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 4:29.5 (14.3 MB/s)
3.77 GB downloaded in 2:59.5 (21.5 MB/s)

I will report additional results with these two cards in other USB 2.0 readers, and will also try out a FireWire 800 reader for the CF version of the card, in the near future.

RJL65 said:

My previous results were by rough timings.

I updated my results -- now using a neat new program called FC-Test.
I also compared my results with the SanDisk Extreme III 30 MB/s
Edition cards in both SDHC and CF formats (note that the SDHC results
are low because the official SD Card Association specs currently
limit the maximum speed to 22 MB/s, and the SanDisk ImageMate
All-in-One reader does not use the proprietary timing mode needed for
30 MB/s mode operation with the SDHC card).

My new results are:

SanDisk Extreme III 30 MB/s Edition 4GB SDHC:
Single large video file:
2.45 GB uploaded in 2:49.8 (14.8 MB/s)
2.45 GB downloaded in 2:11.7 (19.1 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 5:31.1 (11.7 MB/s)
3.77 GB downloaded in 3:33.3 (18.1 MB/s)

SanDisk Extreme III 30 MB/s Edition 4GB CF:
Single large video file:
2.45 GB uploaded in 2:09.4 (19.4 MB/s)
2.45 GB downloaded in 1:46.7 (23.6 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 4:29.5 (14.3 MB/s)
3.77 GB downloaded in 2:59.5 (21.5 MB/s)

I will report additional results with these two cards in other USB
2.0 readers, and will also try out a FireWire 800 reader for the CF
version of the card, in the near future.

Click to expand...

My results with an internal SIIG multi-slot reader are:

SanDisk Extreme III 30 MB/s Edition 4GB SDHC:
Single large video file:
2.45 GB uploaded in 2:42.6 (15.5 MB/s)
2.45 GB downloaded in 2:08.9 (19.5 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 5:27.9 (11.8 MB/s)
3.77 GB downloaded in 3:29.3 (18.4 MB/s)

SanDisk Extreme III 30 MB/s Edition 4GB CF:
Single large video file:
2.45 GB uploaded in 2:28.1 (17.0 MB/s)
2.45 GB downloaded in 1:48.9 (23.1 MB/s)
Multiple smaller image files:
3.77 GB uploaded in 4:51.5 (13.2 MB/s)
3.77 GB downloaded in 2:57.9 (21.7 MB/s)

These results go to show you that card readers which perform relatively fast with an SD card do not always perform faster (and may actually perform slightly slower) with a UDMA-enabled CF card. And note that as with the SanDisk All-in-One reader results above, the SIIG internal reader is held back by the official SD Card Association speed limit of 22 MB/s with SD/SDHC cards.

The other side of speed is how quickly your camera will write to the card when in continue shooting.

I tested continuous write speed by holding the shutter down and beginning time measurement of 40 shots beginning AFTER the buffer fills* and ending with the 40th shutter. Each card had been formatted prior to measurement.

Each camera will be different but here are the results with my Nikon D90 and several cards:
Sandisk Extreme III 1 GB, 20MBsec - 17.5MB/sec
Patriot 16GB Class 6 - 9.9 MB/sec
Patriot 4GB Class 6 - 8.9 MB/sec

Another way of looking at the results is after the buffer has filled on my D90, the Sandisk will capture 42 RAW shots in 20 seconds and the Patriots just 24. The only time this ever mattered to me was taking shots at a racetrack but it's good to know the value of the higher cost cards.

buffer fills at about 12 shots with the Sandisk and 9 shots with the Patriots

PaloAlto said:

The other side of speed is how quickly your camera will write to the
card when in continue shooting.
I tested continuous write speed by holding the shutter down and
beginning time measurement of 40 shots beginning AFTER the buffer
fills* and ending with the 40th shutter. Each card had been formatted
prior to measurement.
Each camera will be different but here are the results with my Nikon
D90 and several cards:
Sandisk Extreme III 1 GB, 20MBsec - 17.5MB/sec
Patriot 16GB Class 6 - 9.9 MB/sec
Patriot 4GB Class 6 - 8.9 MB/sec

Another way of looking at the results is after the buffer has filled
on my D90, the Sandisk will capture 42 RAW shots in 20 seconds and
the Patriots just 24. The only time this ever mattered to me was
taking shots at a racetrack but it's good to know the value of the
higher cost cards.

buffer fills at about 12 shots with the Sandisk and 9 shots with the
Patriots

Click to expand...

I did a similar test on my D90 using the two editions of the Extreme III SD/SDHC cards - a 1GB original Extreme III and a newer 8GB SDHC Extreme III 30 MB/s Edition. My tests were only 9 shots in RAW/NEF. I timed only how long it takes for the buffer to clear.

My in-camera results:
SanDisk Extreme III 1GB, 20MBsec - 17.8MB/sec
SanDisk Extreme III 8GB, 30MBsec - 21MB/sec

For comparison, the currently shipping version of the SanDisk Ultra II 8GB, 15MBsec tested out to only 10.5MB/sec in the camera.

In the near future I will retest the cards to determine how many RAW/NEF images I could write continuously before the buffer fills enough to slow down shooting, and stop the tests after 40 shots per test.

RJL65 said:

My in-camera results:
SanDisk Extreme III 1GB, 20MBsec - 17.8MB/sec
SanDisk Extreme III 8GB, 30MBsec - 21MB/sec

Click to expand...

By the way, the original SDHC 4GB and 8GB versions of the original 20MBsec Extreme III card will be a bit slower than their smaller 1GB and 2GB product-mates due to the SDHC cards' use of MLC flash chips as opposed to the SLC chips used in the 1GB and 2GB versions of the card.

RJL65 said:

By the way, the original SDHC 4GB and 8GB versions of the original
20MBsec Extreme III card will be a bit slower than their smaller 1GB
and 2GB product-mates due to the SDHC cards' use of MLC flash chips
as opposed to the SLC chips used in the 1GB and 2GB versions of the
card.

Click to expand...

How do you 'know' this, or are you going on supposition?

-- Bob Elkind
Family,in/outdoor sports, landscape, wildlife
photo galleries at http://eteam.zenfolio.com
my relationship with my camera is strictly photonic

bob elkind said:

How do you 'know' this, or are you going on supposition?

Click to expand...

I went largely with supposition. After all, most memory card companies do not specify whether a particular model of card uses SLC or MLC NAND flash chips.

OK, time for a mini-lecture.

Do YOU appreciate it when people write their opinons and suppositions in these forums as if they are fact ?

Didn't think so.

It would help to maintain your credibility and the credibility of these forums if you distinguished known supposition from known fact.

RJL65 said:

bob elkind said:

How do you 'know' this, or are you going on supposition?

Click to expand...

I went largely with supposition. After all, most memory card
companies do not specify whether a particular model of card uses SLC
or MLC NAND flash chips.

Click to expand...

The proper words to use, next time, for example: 'I think these cards use SLC rather than MLC, because ... '

My supposition (based on conversations with SanDisk reps on their VERY FASTEST memory cards and SSDs) is that your supposition is mistaken. But I don't know that for a fact.

You've made valuable contributions to these forums with your posts. Please maintain the 'quality standards' your postings deserve.

-- Bob Elkind
Family,in/outdoor sports, landscape, wildlife
photo galleries at http://eteam.zenfolio.com
my relationship with my camera is strictly photonic

RJL65 said:

RJL65 said:

My in-camera results:
SanDisk Extreme III 1GB, 20MBsec - 17.8MB/sec
SanDisk Extreme III 8GB, 30MBsec - 21MB/sec

Click to expand...

By the way, the original SDHC 4GB and 8GB versions of the original
20MBsec Extreme III card will be a bit slower than their smaller 1GB
and 2GB product-mates due to the SDHC cards' use of MLC flash chips
as opposed to the SLC chips used in the 1GB and 2GB versions of the
card.

Click to expand...

I have since acquired an 8GB SDHC Extreme III 20 MB/s version, and its write speed is definitely a bit slower than its 1GB and 2GB SD siblings when tested in a Nikon D90 DSLR shooting RAW images - only 15.8 MB/s with the 8GB card versus an average of 17.5 MB/s with the 1GB and 2GB cards in the same line. And testing all three cards using the latest version of Sandra reveals a significantly higher endurance factor score for the 8GB card than for the smaller cards. That, plus the differential between the benchmarked read and write speeds being much greater in the 8GB card compared to those in the 1GB and 2GB cards, led me to believe that SanDisk began to transition its entire Extreme III line from SLC to MLC NAND flash chips around the time that the company redesigned its packaging for all of its products in late 2007 and early 2008. My 1GB and 2GB cards, which have the old-style label design, are SLC-based while my 8GB card (which has the newer label design) is MLC-based.

RJL65 wrote:
...

RJL65 said:

I have since acquired an 8GB SDHC Extreme III 20 MB/s version, and
its write speed is definitely a bit slower than its 1GB and 2GB SD
siblings when tested in a Nikon D90 DSLR shooting RAW images - only
15.8 MB/s with the 8GB card versus an average of 17.5 MB/s with the
1GB and 2GB cards in the same line.

Click to expand...

15.8 MB/sec vs. 17.5 MB/sec. This is not a huge difference. Rob Galbraith's site also reports performance differences between 8GB and 1GB Extreme III cards, but the differences are even smaller still than what you've reported. See http://www.robgalbraith.com/bins/camera_multi_page.asp?cid=6007-9597

RJL65 said:

And testing all three cards using
the latest version of Sandra reveals a significantly higher endurance
factor score for the 8GB card than for the smaller cards. That, plus
the differential between the benchmarked read and write speeds being
much greater in the 8GB card compared to those in the 1GB and 2GB
cards, led me to believe that SanDisk began to transition its entire
Extreme III line from SLC to MLC NAND flash chips around the time
that the company redesigned its packaging for all of its products in
late 2007 and early 2008.

Click to expand...

Would you please explain your reasoning? I don't understand the connection between the Sandra scores, read/write speed differentials, and SLC/MLC technology determination. It sounds much more plausible that Sandisk redesigned their card controller between the time they manufactured your smaller cards and when they manufactured their larger and more recent cards.

RJL65 said:

My 1GB and 2GB cards, which have the old-style label design, are SLC-based
while my 8GB card (which has the newer label design) is MLC-based.

Click to expand...

Are these assertions (SLC vs. MLC) based on your supposition, or upon direct knowledge from Sandisk ? You are presenting these as fact, but I don't understand or see where you've drawn a clear line from either direct information or clear supporting data. If there is a clear indicator of 'MLC vs. SLC' from SiSoft's Sandra tool, I've not seen it in SiSoft's documentation. Perhaps you can point me to the basis for your reasoning. This would be most appreciated.

As I've said before, SanDisk's entire product line is currently based exclusively on MLC devices, according to direct contacts with SanDisk (this is my day job, so to speak). Furthermore, my SanDisk correspondents assert that MLC has been used exclusively since as far back as they can remember. This isn't prima facie evidence, but it's at least as plausible as the MLC vs. SLC explanation based on a Sandra measurement.

I'm not sure what the material significance of this issue is, in the context of these forums, but I hate to leave poorly supported or vague assertions presented as solid conclusions to remain unchallenged. If I'm completely wrong on this, please advise. I'm completely willing to be convinced that I'm mistaken.

If anyone out there can explain the meaning (intended or otherwise) of the Sandra 'endurance factor', I would greatly appreciate this. The very brief explanation in SiSoft's 'help' files does very little to explain if (or how) this measurement is a meaningful indicator (relative or absolute) of 'endurance'. A web search on this subject turned up too little of any use.

-- Bob Elkind
Family,in/outdoor sports, landscape, wildlife
photo galleries at http://eteam.zenfolio.com
my relationship with my camera is strictly photonic