CF Speeds .... Sandisk Extreme III

[lukeo]

This is the second posting of this information, I lost the first. I have since been guided by members of the forum to input it all into word before I cut and paste into d70users. If I fail to include all the details of the original feel free to ask. (not that anyone would know except me)

Hello all the purpose of this post is to provide information and links on compact flash speeds in the D70, and hopefully test a hypothesis. Sandisk Extreme III is quoted as having a maximum speed of 20Megabytes a second, vs Sandisk Ultra II’s 10megabytes a second. According to this bloke and others the true achieved writing speed of Ultra II is around 4.56Megabytes a second.





http://www.robgalbraith.com/bins/multi_ ... =6007-6816


It is clear that Ultra II offers a huge advantage over “generic” CF, is it worth dropping it for Ultra III? Unlikely without more testing.

As far as I can see saving as JPG results in slower speeds than writing in NEF, this is fairly easy to explain as the overhead (CPU time) required to compress JPG’s means NEF is the only format that will really see much if any improvement from Ultra III. So now all that’s left to find out is if the D70 can actually write any faster, I believe it can.

After much searching (read hours of dredging thru search engines, various engineering sites, and books) it seems Nikon’s Dynamic buffer tech is proprietary (as expected) but many digital camera buffers exist and working off nikon’s specifications, and various sites time trials with continuous shooting there are very few systems with a bandwidth of less than 40mbits (5MB/s) even with fancy memory crossbars and multiple bus access systems that could take 3fps … before falling off.

Given the example (with noise reduction off) that Ultra 2 can capture a continous 1.3fps till the card is full (after the initial 3fps) in NEF at approximately 5000kilobytes a shot. If what I think is true the Extreme III should give approximately a 8% boost in speed, or the equivalent of shooting for 8 seconds to get one extra shot. The boost would only really come in NEF and only really work if the dynamic is 40mbits in width or more. The extreme III is built on a different underlying tech the ultra II (read revamped), it has much lower access times whether this will get it to 5megabytes a second in the D70 we will need some willing members to test.

Here is the hypothesis I’d like to test (in a month or so when someone actually gets there hands on this stuff), get a D70, a friend and stopwatch. Disable noise reduction in the D70, put the shutter cap on the lens and using a 512MB or 1GB quarter fill it to avoid timings errors in continuous mode. Leave the time/model and first half of the serial number here so we can how one brand compares to another. This test should be reasonably scientific, assuming all D70’s are the same (if you can leave your D70 firmware as well).

At a minimum the cards (combined with a suitable USB2 or Firewire cardreader) are by all accounts on the web the fastest without a doubt. So if extracting those 1GB or more of photo’s from multiple cards in the minimum amount of time is your thing and you havn’t just bought a card and reader then maybe Extreme III is for your… only time and testing will tell.

p.s if your using firefox i can't explain the reason for the table below being so FAR below. (I will check in IE after this post)

<TABLE cellSpacing=1 cellPadding=2 width=70% border=1>

<TR>
<TD><B>Brand and Model(Card Identifier)1</B></TD>
<TD>
<B>Controller Source2</B></TD>
<TD >
<B>CardType3</B></TD>
<TD >
<B>Date Added or Updated</B></TD>

<TD >
<B>Write Speed -JPEG Fine4</B></TD>
<TD >
<B>Write Speed -RAW .NEF4</B></TD></TR>
<TR>
<TD>Sandisk Extreme 512MB</B>(Edge stamp: AX0310WJA CHINA)</TD>
<TD>Sandisk</TD>
<TD>I</TD>

<TD>2004/3/18</TD>
<TD>4.167MB/sec</TD>
<TD>4.561MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Ultra II 512MB</B>(Edge stamp: AX0308VZ CHINA)</TD>
<TD>Sandisk</TD>
<TD>I</TD>
<TD>2004/3/18</TD>
<TD>4.151MB/sec</TD>

<TD>4.552MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 512MB 80X Write Acceleration *second edition*</B>10</TD>
<TD>
Lexar</TD>
<TD>
I</TD>
<TD>



2004/7/29</TD>

<TD>
4.051MB/sec</TD>
<TD>
4.468MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 1GB 80X Write Acceleration *second edition*</B>10</TD>
<TD>Lexar</TD>
<TD>I</TD>
<TD>2004/7/29</TD>
<TD>4.005MB/sec</TD>

<TD>4.322MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Extreme 1GB</B>(Edge stamp: BB0307VZ CHINA)</TD>
<TD>
Sandisk</TD>
<TD>
I</TD>
<TD>
2004/3/18</TD>
<TD>
3.945MB/sec</TD>

<TD>
4.560MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Ultra II 1GB</B>(Edge stamp: BB0306VZ CHINA)</TD>
<TD>
Sandisk</TD>
<TD>
I</TD>
<TD>




2004/3/18</TD>
<TD>
3.937MB/sec</TD>
<TD>
4.560MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Extreme 2GB</B></TD>
<TD>Sandisk</TD>
<TD>I</TD>
<TD>2004/3/18</TD>

<TD>3.921MB/sec</TD>
<TD>4.407MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 2GB 80X Write Acceleration *second edition*</B>10</TD>
<TD>Lexar</TD>
<TD>I</TD>
<TD>2004/7/29</TD>
<TD>3.916MB/sec</TD>
<TD>4.314MB/sec</TD></TR>

<TR>
<TD><B>Kingston Elite Pro 512MB</B>(Edge stamp: 1GC5121MY0-2EA00)</TD>
<TD>Samsung</TD>
<TD>I</TD>
<TD>2004/7/29</TD>
<TD>3.915MB/sec</TD>
<TD>4.122MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Ultra II 2GB</B></TD>

<TD>Sandisk</TD>
<TD>I</TD>
<TD>2004/3/18</TD>
<TD>3.895MB/sec</TD>
<TD>4.405MB/sec</TD></TR>
<TR>
<TD><B>Sandisk "new" Ultra 512MB</B>(Edge stamp: AX0303TV CHINA)</TD>
<TD>
Sandisk</TD>

<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
3.332MB/sec</TD>
<TD>
3.346MB/sec</TD></TR>
<TR>
<TD><B>Transcend 1GB 45X</B>(45X on front label)</TD>

<TD>
Transcend</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
3.276MB/sec</TD>
<TD>
3.581MB/sec</TD></TR>

<TR>
<TD><B>Sandisk "new" Ultra 1GB</B>(Edge stamp: BB0303TV CHINA)</TD>
<TD>
Sandisk</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>

3.239MB/sec</TD>
<TD>
3.241MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 1GB 80X Write Acceleration *first edition*</B>10</TD>
<TD>
Lexar</TD>
<TD>
I</TD>
<TD>
2004/7/29</TD>

<TD>
3.228MB/sec</TD>
<TD>
4.009MB/sec</TD></TR>
<TR>
<TD><STRONG>Lexar Media 1GB 40X Write Acceleration*new and improved*</STRONG>9</TD>
<TD>
Lexar</TD>
<TD>
I</TD>

<TD>



2004/3/18</TD>
<TD>
3.185MB/sec</TD>
<TD>
3.803MB/sec</TD></TR>
<TR>
<TD><B>Kingston Elite Pro 1024MB</B>(Edge stamp: THNCF1G02CA)</TD>
<TD>Toshiba (SLC)5</TD>

<TD>I</TD>
<TD>2004/7/29</TD>
<TD>2.957MB/sec</TD>
<TD>3.304MB/sec</TD></TR>
<TR>
<TD><B>Microtech (Pexagon) X-treme 1GB</B>(Edge stamp: THNCF1G02CA)</TD>
<TD>Toshiba (SLC)5</TD>
<TD>I</TD>

<TD>2004/3/18</TD>
<TD>2.929MB/sec</TD>
<TD>3.237MB/sec</TD></TR>
<TR>
<TD><B>Microtech (Pexagon) X-treme 512MB</B>(Edge stamp: THNCF512MCA)</TD>
<TD>
Toshiba (SLC)5</TD>
<TD>
I</TD>

<TD>
2004/3/18</TD>
<TD>
2.928MB/sec</TD>
<TD>
3.239MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 2GB 40X Write Acceleration</B></TD>
<TD>
Lexar</TD>
<TD>
I</TD>

<TD>



2004/3/18</TD>
<TD>
2.816MB/sec</TD>
<TD>
3.391MB/sec</TD></TR>
<TR>
<TD><STRONG>Lexar Media 1GB 40X Write Acceleration</STRONG>(Edge stamp, first two digits: 38; WA on front label)</TD>
<TD>

Lexar</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
2.812MB/sec</TD>
<TD>
3.340MB/sec</TD></TR>

<TR>
<TD><B>Lexar Media 512MB 40X Write Acceleration</B>(Edge stamp, first two digits: 38; WA on front label)</TD>
<TD>
Lexar</TD>
<TD>
I</TD>
<TD>




2004/3/18</TD>

<TD>
2.709MB/sec</TD>
<TD>
3.151MB/sec</TD></TR>
<TR>
<TD><B>Ritek/Ridata 52X/PRO. 1GB</B>(Edge stamp: S4N88116)</TD>
<TD>unknown7</TD>
<TD>I</TD>
<TD>2004/3/18</TD>

<TD>2.703MB/sec</TD>
<TD>3.012MB/sec</TD></TR>
<TR>
<TD><STRONG>Lexar Media 4GB 40X Write Acceleration</STRONG>8</TD>
<TD>
Lexar</TD>
<TD>
II</TD>
<TD>




2004/3/18</TD>
<TD>
2.698MB/sec</TD>
<TD>
3.166MB/sec</TD></TR>
<TR>
<TD><B>Transcend 512MB</B>(30X label on packaging)</TD>
<TD>
Transcend</TD>
<TD>

I</TD>
<TD>



2004/3/18</TD>
<TD>
2.692MB/sec</TD>
<TD>
2.902MB/sec</TD></TR>
<TR>
<TD><B>Lexar Media 512MB 24X Write Acceleration</B>(Edge stamp, first two digits: 38; WA on front label)</TD>

<TD>
Lexar</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
2.676MB/sec</TD>
<TD>
2.913MB/sec</TD></TR>

<TR>
<TD><B>Delkin Devices PRO 640MB</B>(Label: CFPRO-640D)</TD>
<TD>
For Delkin6</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>

<TD>
2.659MB/sec</TD>
<TD>
3.006MB/sec</TD></TR>
<TR>
<TD><B>Delkin Devices PRO 512MB</B>(Label: CFPRO-512D)</TD>
<TD>
For Delkin6</TD>
<TD>
I</TD>

<TD>



2004/3/18</TD>
<TD>
2.658MB/sec</TD>
<TD>
3.015MB/sec</TD></TR>
<TR>
<TD><B>Ritek/Ridata 52X/PRO. 2GB</B>(Edge stamp: U3F06116)</TD>
<TD>

unknown7</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
2.381MB/sec</TD>
<TD>
2.884MB/sec</TD></TR>

<TR>
<TD><B>SimpleTech PRO X 512MB</B>(Internal: STI/FLASH 1T)</TD>
<TD>
SST</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>

2.160MB/sec</TD>
<TD>
2.297MB/sec</TD></TR>
<TR>
<TD><B>Sandisk "original" Ultra 512MB</B>(Edge stamp: AX0211RX CHINA)</TD>
<TD>
Sandisk</TD>
<TD>
I</TD>
<TD>




2004/3/18</TD>
<TD>
2.015MB/sec</TD>
<TD>
2.078MB/sec</TD></TR>
<TR>
<TD><B>SimpleTech PRO X 1GB</B>(Internal: STI/FLASH 1T)</TD>
<TD>
SST</TD>

<TD>
II</TD>
<TD>



2004/3/18</TD>
<TD>
1.980MB/sec</TD>
<TD>
2.124MB/sec</TD></TR>
<TR>
<TD><STRONG>Hitachi Microdrive 4GB</STRONG>8</TD>

<TD>
Hitachi</TD>
<TD>
II</TD>
<TD>
2004/3/18</TD>
<TD>
1.694MB/sec</TD>
<TD>
2.025MB/sec</TD></TR>
<TR>
<TD><B>Hitachi Microdrive 1GB</B></TD>

<TD>
Hitachi</TD>
<TD>
II</TD>
<TD>



2004/3/18</TD>
<TD>
1.466MB/sec</TD>
<TD>
1.695MB/sec</TD></TR>

<TR>
<TD><B>Viking 512MB</B>(Edge stamp: THNCF512MMA)</TD>
<TD>Toshiba (MLC)5</TD>
<TD>I</TD>
<TD>2004/3/18</TD>
<TD>1.114MB/sec</TD>
<TD>1.232MB/sec</TD></TR>
<TR>
<TD><B>Kingston Memory 1024MB</B>(Edge stamp: THNCF1G02MA)</TD>

<TD>
Toshiba (MLC)5</TD>
<TD>
I</TD>
<TD>



2004/3/18</TD>
<TD>
1012K/sec</TD>
<TD>

1.142MB/sec</TD></TR>
<TR>
<TD><B>Kingston Memory 512MB</B>(Edge stamp: THNCF512MMA)</TD>
<TD>
Toshiba (MLC)5</TD>
<TD>
I</TD>
<TD>




2004/3/18</TD>
<TD>
998K/sec</TD>
<TD>
1.140MB/sec</TD></TR>
<TR>
<TD><B>Sandisk Standard 1GB</B>(Edge stamp: BB0205NK CHINA)</TD>
<TD>Sandisk</TD>
<TD>I</TD>
<TD>2004/3/18</TD>

<TD>759K/sec</TD>
<TD>922K/sec</TD></TR></TABLE>

[atencati]

Thanks for that! If I read that correctly, there is hardly any difference between the ultraII and Extreme sandisk cards, correct? If so, thats great news! The Ultra II's will drop again in price because everyone wants the newest and greatest

Andy

[birddog114]

I still did not see any advantage in buying the:

- Sandisk Extreme III over the the Sandisk Ultra II or Transcend and Astone 45x

- the D70 with its limitation on its controller card can't read or write faster than its built.
- In the test shown the Transcend/ Astone has similar read & write speed with the Sandisk Ultra II
- Price of the Sandisk Extreme III is nearly double the Transcend/ Astone.

I rather buy 2 x Transcend 1Gb card instead of buying Sandisk Extreme III
I'll buy the Transcend or Astone 45x over the Sandisk Ultra II cos the Sandisk is more dearer than other and its performance is not or won't have a sigificant of speed over the Transcend.

To use in a card reader at home, I'm not to worry in fast or slow, leave the download job to the computer and card reader while lurking the forum and enjoy a cuppa.
I perhaps will buy the Sandisk Extreme III with the new Nikon D2x

[gstark]

Sandisk Extreme III is quoted as having a maximum speed of 20Megabytes a second, vs Sandisk Ultra II’s 10megabytes a second. According to this bloke and others the true achieved writing speed of Ultra II is around 4.56Megabytes a second.

But this is always going to be restricted by the maximum speed that the camera can achieve. In the case of the D70, the numbers you're quoting are purely for reference purposes only; the camera is somewhat slower.

It is clear that Ultra II offers a huge advantage over “generic” CF,

Is it?

Why do you say this? Taking into account that the narrowest point in the bandwith is the camera, any CF card that matches or exceeds the camera's bandwidth can (and does) provide maximum throughput that the camera is capable of.

Cards that are slower than the camera's capabilities are, of course, going to be slower. This true of generic cards, as well as of name brands like SanDisk.

Again, look to the slowest component in the package: if your card is slower than about 40x or 45x, this will be the card. Using faster cards elevates the camera to the esteemed position of bottleneck.

As far as I can see saving as JPG results in slower speeds than writing in NEF, this is fairly easy to explain as the overhead (CPU time) required to compress JPG’s means NEF is the only format that will really see much if any improvement from Ultra III. So now all that’s left to find out is if the D70 can actually write any faster, I believe it can.

Unfortunately, this seems to me to be somewhat lacking in detail.

On the D70, NEFs are also compressed. This also must take up CPU time.

But on the D70 there are several different ways that files may be saved in JPG format. There are three different quality settings, as well as a number of different resolution settings. Each of these modes will have an effect upon the amount of processing that the camera is required to perform.

Further, the subject matter that you're shooting is likely to have a significant effect upon the amount of compression being applied, and thus will also affect the amount of processing being undertaken.


[/quote]After much searching (read hours of dredging thru search engines, various engineering sites, and books) it seems Nikon’s Dynamic buffer tech is proprietary (as expected) but many digital camera buffers exist and working off nikon’s specifications, and various sites time trials with continuous shooting there are very few systems with a bandwidth of less than 40mbits (5MB/s) even with fancy memory crossbars and multiple bus access systems that could take 3fps … before falling off.

Given the example (with noise reduction off) that Ultra 2 can capture a continous 1.3fps till the card is full (after the initial 3fps) in NEF at approximately 5000kilobytes a shot. If what I think is true the Extreme III should give approximately a 8% boost in speed, or the equivalent of shooting for 8 seconds to get one extra shot. [/quote]

I'm trying to understand your thinking here. The camera is not capable of sending data to the CF card any more quickly than it is already doing; the fact that the card might be happy to accept more data is irrelevant; the camera is already at its bottleneck capability.

I'm also struggling to understand your point; even if you do achieve an 8% boost in speed, and if that does give you an extra 1 frame in 8 seconds, in NEF you're only shooting at 1 - 1.3fps at this time, and how many action sequences are likely to continue for 8 seconds?

The boost would only really come in NEF and only really work if the dynamic is 40mbits in width or more. The extreme III is built on a different underlying tech the ultra II (read revamped), it has much lower access times whether this will get it to 5megabytes a second in the D70 we will need some willing members to test.

Here is the hypothesis I’d like to test (in a month or so when someone actually gets there hands on this stuff), get a D70, a friend and stopwatch. Disable noise reduction in the D70, put the shutter cap on the lens and using a 512MB or 1GB quarter fill it to avoid timings errors in continuous mode. Leave the time/model and first half of the serial number here so we can how one brand compares to another. This test should be reasonably scientific, assuming all D70’s are the same (if you can leave your D70 firmware as well).

Actually, and with all due respect, I think it will not be all that worthwhile. Using a manual stopwatch is fraught with danger in any so-callled "scientific" test. How will you factor in reaction time at the start and end?

I'm feeling somewhat reluctant to support this, because I fail to see any value in it. The point is not to get the fastest card available, it's to get the card(s) that best match the performance of the camera.

Given that this is a D70 focussed forum, the primary point of relevance seems to me to be matching the cards to the D70, rather than the theoretical maximum speed of the card itself; getting a card that is too fast for the camera may be worthwhile if you have a D2H, D2X, or perhaps something like a 1D, but if your only (or fastest) camera is a D70, all you're doing is lining the pockets of the cards' manufacturers for no noticeable or practical benefit.

I'm open to be convinced, but I'm having trouble seeing any practical benefit from the munging of all of these numbers; wouldn't we all be better off just getting out and shooting?

[Onyx]

The dynamic buffer ability of the D70 offers a significant write speed advantage over its peers. eg. With buffers full, the D70 is faster than the 20D in terms of FPS achievable. This, I believe, is one reason you should buy fast flash memory cards for your D70 if continuous shooting ability is not to be crippled.

With regards to max. CF transfer speeds, I note on Galbraith's database the apparent differences between cards of the same brand and model but differring capacities. I believe these differences largely account for the individual sample variation, and if viewed in this light there is no significant speed differences between Sandisk Ultra II, Extreme or Lexar 80x of the first half dozen entries. Although I note the "new" Ultra and Transend 45x's numbers seem to be in a different category. Methodological issues of data testing aside...

Although I do see some benefits of buying the Extreme compared to Sandisk's Ultra or ordinary lines - dedicated customer service phone lines (at least for the US), data recovery software bundled (?), and rated to operate in ridiculous climates that the D70 probably won't.

[birddog114]

Hey!
How fast are you going with the D70?

[Grumpy]

I wouldn't be worrying about 3dp though. Too many other factors influence the speed the internal processor can clear its buffer, but the higher spec'd cards generally outperformed the lower spec'd cards.

Whether this is important to you or not will probably depend on how much sports mode shooting you do. As for transfer speed, the fastest 2GB cards might be a minute quicker. When 20GB CF cards are common, we might care more about maximum transfer speed than we do now (15 mins vs 30 mins etc).

It may be more important to you to add capacity rather than speed, which makes the cheaper cards the better option. It is like arguing a bus is not as good as a sports car. Both are good, it depends on what you want to do though.


Adam

[birddog114]

Well said Grumpy! I like that! I want more storage, not really the speed cos I understood the limitation of mysefl same as my cameras and I did not see any big difference in shooting 8fps/ 40 frames in burst mode on my D2h with both brand Transcend and Sandisk Ultra II 1Gb.
So 2Gb card of Transcend and Astone are in my inventory now and soon will be 4Gb, whatever brand but 45x is enough to serve my needs.

[lukeo]

Thanks for the feedback,

Please note that the table doesn’t include the Extreme III only the original Extreme. The III has yet to be tested in a D70. Price drops are “extremely” likely, excuse the pun, the question remains is it just good for us to grab the Ultra II a bit cheaper or is it worth the extra $$$ for the Extreme III? (if money is no objective)

Originally Posted by Birddog

I still did not see any advantage in buying the:

- Sandisk Extreme III over the the Sandisk Ultra II or Transcend and Astone 45x

- the D70 with its limitation on its controller card can't read or write faster than its built.

I am actually agreeing with you Birddog, the difference from going from “generic” CF to Sandisk Ultra II or Astone is an increase in speed of 2900Kb a second which is huge, the theoretical difference going to Extreme III would be around 400Kb a second at most. I am asking for a collection of data to test my theory that the D70 has 40mbits (5megabytes) a second worth of bandwidth and hence there is possibly a small margin for improvement. 4.56*8 = 36.48 megabits is an extremely odd bandwidth for a device to have considering it is cheaper to manufacture in the standard lot of multiples of 8megabits.

yraen69 wrote:

“Sandisk Extreme III is quoted as having a maximum speed of 20Megabytes a second, vs Sandisk Ultra II’s 10megabytes a second. According to this bloke and others the true achieved writing speed of Ultra II is around 4.56Megabytes a second.”

Originally Posted by Gstark

But this is always going to be restricted by the maximum speed that the camera can achieve. In the case of the D70, the numbers you're quoting are purely for reference purposes only; the camera is somewhat slower.

Indeed it is, just like computer standards ATA100 and SATA150 they are quoted at 100MB/s a second and 150MB/s yet a generic 80GB hard disk will struggle to hit 45MB/s a second ... much less than the rating even for ATA66 and just outdoing the rating for ATA33 from 5 years ago.

The 4.56MB/s from the sample data is the actual transfer rate of the D70 from its buffer to the Compact Flash card. This is the number I seek to improve if only slightly. Quoted Theoretical maximum speeds mean little unless the cameras buffer to card bandwidth can actually offer some more bandwidth due to the different manufacturing process and lower latency of the Extreme III, hence the 20MB/s of Extreme III is in theory double that of Ultra II so perhaps in reality we will see that elusive 400kb a second gain I am proposing we test for.

Yes generic Compact Flash is the bottleneck, from the above table the slowest card manages 759K/sec in JPG and 922K/sec in NEF, whilst the Ultra II handily out does it at4560K/sec

My argument is that we don’t actually know that the D70 is the final bottleneck, the Ultra II cards etc are the fastest cards in existence until now, they may have a bandwidth of 10megabytes a second but when driven by the D70’s chipset they only achieve 4500K/sec this is true of many computer components, driven by different chipsets they achieve different speeds. The Extreme III is new technology that hasn’t been tested, let’s test it and others.

I also understand that very few action sequences last 8 seconds, this was part of the argument against upgrading if you already own Ultra II and would fall into the minor plus’s if you don’t (and have lots of money).

The only thing I can offer in relation to the testing method is so long as the test continues for at least 3 minutes then the data error rate introduced by 2 seconds of human timing delay will offset the data by less than + or – 50K/sec which will not really impact on seeing if the D70 can make 400K/sec extra use out of Extreme III.

This also isn’t just in relation to extreme III, I am very interested to see different cards in the D70 tested for at least three minutes. This is the D70users forum so why shouldn’t we have a “database “ of our own on the top of the line Compact Flash cards, for many people here $200 dollars is pocket change compared to what was spent on lens’s some sports fanatics and nature people may enjoy any (perceptible or imperceptible) gains to had. Once you’ve taken the time to do this and compared your results to other users here you can be confident you are getting what you want to get out of D70 in regards to Compact Flash and Continuous shooting.

[kfandst]

Do faster cards download quicker to your pc/notebook?

[birddog114]

Do faster cards download quicker to your pc/notebook?

depend is what card reader you have and the connection as USB 1.1/ 2.0 or Firewire.

[gstark]

yraen69 wrote:

“Sandisk Extreme III is quoted as having a maximum speed of 20Megabytes a second, vs Sandisk Ultra II’s 10megabytes a second. According to this bloke and others the true achieved writing speed of Ultra II is around 4.56Megabytes a second.”

Originally Posted by Gstark

But this is always going to be restricted by the maximum speed that the camera can achieve. In the case of the D70, the numbers you're quoting are purely for reference purposes only; the camera is somewhat slower.

Indeed it is, just like computer standards ATA100 and SATA150 they are quoted at 100MB/s a second and 150MB/s yet a generic 80GB hard disk will struggle to hit 45MB/s a second ... much less than the rating even for ATA66 and just outdoing the rating for ATA33 from 5 years ago.

As with cameras, not all motherboards are created equal.

Compare an IBM MB's performance with, say, a noname clone. The IBM will eat it for breakfast, because of the better engineering in the circuit designs employed by the IBM.

How much of those performance differences that you're describing here can be attributed to deliberate design decisions taken by the manufacturer?

The D70 is designed down to a performance level and pricepoint. I don't have an issue with this, and happily accept that. It may well be that a firmware hack could address this, but for me the camera's performance is just fine.

Similarly, the 300D is designed down to a pricepoint. Unfortunately that designing down included a lobotomy on the camera's capabilities, turning a potential neck-snapper into the digital equalent of a horse and cart.

Having said that, The D70 was designed for optimal performance with 40x - 45x cards. Unless you can find a way to increaase the bandwith, you will not see anything significant in terms of data throughput.

It's basically like coefficient of drag - doubling the horsepower on a car with a poor coefficient of drag won't make the car go all that much faster. The car is wasting too much enrgy getting air out of its way.

The 4.56MB/s from the sample data is the actual transfer rate of the D70 from its buffer to the Compact Flash card. This is the number I seek to improve if only slightly.

How do you propose to emasure this?

And let's be generous and say you get a 10% improvement ... that'll bring you up to a bit over 5MB/s throughput.

I'm sorry, but that's really quite an underwhelming difference.

It's a bit like saying that you want to grab a 20D because it gives you 8MP instead of the D70's 6.

The practical end result of those numbers has no real meaning in a real world sense.

Putting it another way, you would need to improve throughput by something better than 75% to even begin to grab my attention. Anything less is little more than a pimple on Kim Beasley's backside.

Quoted Theoretical maximum speeds mean little unless the cameras buffer to card bandwidth can actually offer some more bandwidth due to the different manufacturing process and lower latency of the Extreme III, hence the 20MB/s of Extreme III is in theory double that of Ultra II so perhaps in reality we will see that elusive 400kb a second gain I am proposing we test for.

But only if there is surplus data being processed.

My understanding is that on the D70, there isn't any surplus data waiting to be output, and therefore any extra buffer to card bandwidth with remain exatly that - extra buffer to card bandwidth.

My argument is that we don’t actually know that the D70 is the final bottleneck,

And with respect, I believe that is where your argument is flawed. Others, with far better technical tools than we have available, have performed these tests, and they have already stated the limits of the D70.

I also understand that very few action sequences last 8 seconds, this was part of the argument against upgrading

I have yet to see an argument for upgrading ...

This is the D70users forum so why shouldn’t we have a “database “ of our own on the top of the line Compact Flash cards

Perhaps because the D70 is not a top of the line camera? It's not designed for heavy duty use. It's not designed for continuous high-speed use. It's not designed to take advantage of of the technologies or speed that the top of the line cards offer?

I'm not for a moment saying that the camera won't handle these sorts of events, btw. I'm simply being mindful of its design criteria and target markets.

If you want that ultimate performnce camera, grab the nearest D2H.

[/b]

[MattC]

Interesting...

One thing that continues to bother me is the assertion that the 45x cards will maximise the performance of the d70, when there it is in black and white on Bob Galbraiths performance chart that the Ultra II is a full 27% faster than the Transcend 45x. In the case of the Ultra II the camera or card is the bottle neck (hard to know seeing as the only faster card tested is WA enabled which the d70 does not support), in the case of the Transcend 45x the card is the bottleneck.
I would be quite comfortable with either the Astone or Transcend cards in my bag. They offer more than adequate performance for daily use, but I do not see them as in the same league as the Ultra II for continuous shooting. BTW, continuous shooting for me is compose, shoot, recompose, shoot, recompose.... Ocassionally I run into the buffer with the Ultra II and RAW.

As for those write speeds, I always regard them with a little suspicion. To me, they are burst speeds, not continuous. A maximum of 2/3 of the quoted max is the rule of thumb that I use for continuous. Not particularly accurate, and in some cases way off, but it is just a rule of thumb.

How fast is the d70? 45x, 60x, more??? These tests only ever measure averages, not maximum speeds. The cards never operate at their maximum, nor do the devices that use them.

It would be interesting to see whether the Ultra III squeezes a little more from the d70. I doubt it.

Cheers

Matt

[birddog114]

Matt,
It's the same with clock on the car, most car has its top speed is 220kph but how and which way they can achieve it.

Again and again, the Sandisk Extreme III is not faster than the Sandisk II in the D70 in whatever shooting mode and I din't see why should we buy it for that such of dollars in comparison with the Ultra II or Transcend/ Astone.

I need large storage on the CF not the speed, cos I want to shoot till the cow goes home.

[MattC]

Birddog,

Not to worry, I will not be rushing out to buy a Extreme III. As you point out, it will be no faster than the Ultra II in a D70. I already have two 1GB Ultra II's which are as good as it gets. I could be tempted to purchase a couple of Astone or Transcend cards in the future if I ever need more storage.

Cheers

Matt

[birddog114]

Birddog,

Not to worry, I will not be rushing out to buy a Extreme III. As you point out, it will be no faster than the Ultra II in a D70. I already have two 1GB Ultra II's which are as good as it gets. I could be tempted to purchase a couple of Astone or Transcend cards in the future if I ever need more storage.

Cheers

Matt

that's my message

[fozzie]

mattco6974.

I have just purchased from Birddog the ASTONE 1GB 'ULTRA' (45x) and I am more than happy with it's performance, plus the saving in cost.
Speed is not everything, remember the tortoise and the hare scenario.

Cheerio,

[birddog114]

mattco6974.

I have just purchased from Birddog the ASTONE 1GB 'ULTRA' (45x) and I am more than happy with it's performance, plus the saving in cost.
Speed is not everything, remember the tortoise and the hare scenario.

Cheerio,

Where's your new 2Gb? You've got it also.

[fozzie]

Birddog,

I have got that too, oops - short term memory loss .


Cheers,

[MattC]

Fozzie,

Like most things, I do not get to try before I buy. It is a geographical problem. When my d70 arrived I spent the first couple of weeks using a 1GB Microdrive. Aaargh! I decided that I would buy the best that I could afford. I paid AU $195 ea for the Ultra II's. That was an exceptional price at the time. This while I was still hanging out at DPR (so happy to be out of there) and well before I signed up here.

Cheers

Matt

[fozzie]

mattco6974.

I too purchased a Sandisk ULTRA II 512MB CF card in I think May/June 2004 which cost about the same as one of Birddog's 'ASTONE' 1GB CF cards. I didn't join the forum until October, 04. We have all been burnt in the past, but now have the best 'Camera Store' in Australia.


Cheers,

[MattC]

...but now have the best 'Camera Store' in Australia.

Shhhh... Lets keep it our own little secret

Cheers

Matt

[skippy]

I think there are two main benefits of the fast cards, one of which may or may not be vapour. Both benefits are only going to be of importance to some people, while others will not care.

The possible vapour one is whether the faster card will allow you to flush the buffer to card fast enough to allow you to grab another frame or two in the buffer at the full 3fps. I bought a 2nd card and was convinced to buy the standard Sandisk. It seemed slower, so I tested it. The std card gave I think 24 frames in 30 sec, where the Ultra II gave me 27. I'm a little hazy on the exact numbers (it was in October), but the difference was 3 frames in 30sec. One of those frames was at 3fps, the other two were from a faster sustained transfer rate once the buffer filled. I returned the standard card in favour of another Ultra II, mainly cause I'm fussy. Now whether a further increase in card speed over the Ultra II would translate to a corresponding increase in continuous mode shooting, we'll have to wait and see to get a definitive answer.

The second reason the Extreme III may be good for you is if you're impatient to transfer your pics to your PC. At the same time Sandisk announced the new cards they also announced a new series of USB2 card readers to support the new PIO mode 5 and 6 transfers used in the new cards. The combination according to Sandisk is supposed to give about 18MB/s CF to PC xfers. Here's a bit from the middle of the original announcement on RobGalbraith.com:

Using the computer’s built-in USB ports and the Sandisk readers, here’s what we measured:

* On a Vision3000 system from Canadian PC maker MDG, the computer’s Intel 865GBFL motherboard, P4/3.0GHz processor, 512MB of dual-channel DDR RAM and Windows XP Home allowed HDBench read speed numbers of 18.7MB/second for the Extreme III SD card, and 18.1MB/second for the Extreme III CompactFlash card. This was using the manufacturer’s stock configuration for XP Home and no other applications running besides HDBench (except for the myriad of background applications that are always loaded).

* On a Dell Inspiron 8500 laptop (Intel 845-series motherboard, P4-M/2.6GHz processor, 1GB DDR SDRAM and Windows XP Home), we managed HDBench read speed numbers of 17.6MB/second for Extreme III 1GB SD and 16.9MB/second for Extreme III 1GB CompactFlash.

In both cases, the HDBench read speed for the Extreme III cards was 7-9MB/second faster than original Extreme cards in the engineering sample Sandisk reader...

If these numbers are a significant enough difference to you it may be worth picking up the new reader and card.

Comes down to priorities I suppose. Some people want a lot of space and don't care about continuous mode shooting or xfer rates to the PC, so these new cards would be a waste of money. To others these may be important enough to justify the greater cost. Still others will just want them cause they're 'better'

Hmmm, new personal best for longest post!

[lukeo]

Thank you for the feedback,

The std card gave I think 24 frames in 30 sec, where the Ultra II gave me 27

Hence my reasoning for wanting to test the cards in first place, I understand everyone believes that the D70 itself is the limit but it is clear that from each successive generation of cards the D70 has shown incremental improvement.

The Extreme III needs to be tested, when and if someone gets some.

That is the "vapour" advantage.

The "better" advantage can only work in our favour driving prices of Ultra II down.

Transfer times might be important to some, especially those that lead business lives ... then again maybe not. Set and forget that 4GB card over night.

Well it's been a fun discussion.

regards
Luke