Compiled list of card speeds

Started by Audionut, July 13, 2014, 01:58:17 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

eyeland

@ Walter
Thing is, the KB card seems to work ok and I am not new to benching.
I have also tried the different variable I could think off (eg. cluster size, format type, file-system, bench apps)
As you can see below, the KB card eats the Lexr.
Crystal Disk Mark:
Lexar 64-1066
           Sequential Read :   144.711 MB/s
          Sequential Write :    84.624 MB/s
         Random Read 512KB :   135.211 MB/s
        Random Write 512KB :    39.826 MB/s
    Random Read 4KB (QD=1) :     8.275 MB/s [  2020.3 IOPS]
   Random Write 4KB (QD=1) :     1.545 MB/s [   377.1 IOPS]
   Random Read 4KB (QD=32) :    10.513 MB/s [  2566.8 IOPS]
  Random Write 4KB (QD=32) :     1.800 MB/s [   439.5 IOPS]
  Test : 1000 MB [E: 0.0% (0.0/59.6 GB)] (x5)
Lexar 64-1066 (second copy)
           Sequential Read :   145.494 MB/s
          Sequential Write :    85.570 MB/s
         Random Read 512KB :   137.307 MB/s
        Random Write 512KB :    31.296 MB/s
    Random Read 4KB (QD=1) :     8.373 MB/s [  2044.1 IOPS]
   Random Write 4KB (QD=1) :     1.574 MB/s [   384.3 IOPS]
   Random Read 4KB (QD=32) :    10.605 MB/s [  2589.0 IOPS]
  Random Write 4KB (QD=32) :     1.821 MB/s [   444.7 IOPS]
  Test : 1000 MB [E: 0.0% (0.0/59.6 GB)] (x5)
Komputerbay
           Sequential Read :   149.989 MB/s
          Sequential Write :   131.285 MB/s
         Random Read 512KB :   142.877 MB/s
        Random Write 512KB :    37.196 MB/s
    Random Read 4KB (QD=1) :     9.305 MB/s [  2271.8 IOPS]
   Random Write 4KB (QD=1) :     2.206 MB/s [   538.6 IOPS]
   Random Read 4KB (QD=32) :    11.672 MB/s [  2849.6 IOPS]
  Random Write 4KB (QD=32) :     2.502 MB/s [   610.8 IOPS]
  Test : 1000 MB [E: 0.0% (0.0/59.6 GB)] (x5)
AS SSD:
KomputerBay 64GB-1000x =          65mb/s;95mb/s;125mb/s,125mb/s,125mb/s,125mb/s;126mb/s
Lexar 64GB-1000x=                              80mb/s;82mb/s;81mb/s;81mb/s;81mb/s;81mb/s;81mb/s
Lexar 64GB-1000x(second copy)=      80mb/s;81mb/s;81mb/s;81mb/s;81mb/s;81mb/s;81mb/s



Initial thoughts: The KB card seems to benefit from "warming up" whereas the Lexar is more constant.
I find it interesting that the READ speed on the Lexar is is on par with the KB whereas the write speed is horrible. This reminds me of my Lexar 800x which similarly benches at 140mb/s read - 47mb/s write
[All cards tested with identical settings, eg. clustersize, filesystem, camera settings ]
Daybreak broke me loose and brought me back...

Walter Schulz

Controller SM2236AB is a bit outdated.
Turn off Global Draw, activate benchmark and press PLAY pretty fast.
Post results, please.

EDIT: Or you can skip that part, return card to dealer and order one more up to date (SM2236AC or newer).
Just learned that different controller types for this card seems to be significant.


eyeland

@ Walter
Cards are still crunching, my initial observation however is that with GD off in Playback mode Sequential write is around 107mb/s for the Komputerbay card and 80-81mb/s for the Lexar.
I guess that now I know why they where cheap (91$ per card)
Thanks alot for your analysis and swift help Walter.
I guess I am sending them back and in the future I will remember to pay attention to revisions and controller versions.
EDIT: results

Daybreak broke me loose and brought me back...

eyeland

Quote from: Walter Schulz on December 15, 2014, 10:25:53 PM
Controller SM2236AB is a bit outdated. (...) order one more up to date (SM2236AC or newer).
Just learned that different controller types for this card seems to be significant.
Any advice on how to identify controller versions based on card revisions or serial numbers?
I have been talking to 7 different sellers as well as using my inferior google skills to no avail :)
Daybreak broke me loose and brought me back...

walter_schulz

I'm sorry but I can't tell for sure.
It is said H2testw is able to identify controller types if some conditions are met. USB is blocking access on this level and the CF-cardreaders I own are that type. SD-card slots on those netbooks/laptops I tested won't do either.

Once there was a site devoted to heavy-duty test routines but he stopped some years ago.
http://www.hjreggel.net/cardspeed/index.html
Seems to use EOS cam right now ...

If there is someone out there successful reading controller type by H2testw I would be happy to hear about his environment.


leon2187489341

When does the Magic Lantern firmware v.2.4 for the canon rebel t3i comes?
I don't want to install nightly builds, because im afraid that my camera breaks...

Walter Schulz

Top of page -> User Guide -> FAQ -> Last question (section Troll Questions)

DeafEyeJedi

With the recent update in nightlies for ML -- curious if I should run another BM test w a fast CF (KB 128GB 1066x) with 113 build from 3/2 for comparisons with previous BM's?
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Walter Schulz

I bet you won't see reproduceable differences. I you have time you may prove me wrong.

DeafEyeJedi

That's why I asked.

But I'm afraid you're right there's really no differentials other than pushing .3-.5 MB/s slower than last time I did this.

Guess this is to be expected from 'wear & tear' (this KB card was 5-6 months) on the CF cards.... Correct?
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Walter Schulz

Differences that low in a single sample test don't have any significance.

DeafEyeJedi

5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109


Walter Schulz

CFast is incompatible to anything found in cams supported by ML.

dd2020

Thank you, Walter Schulz..
My bad...  :D

Ali Oliya

Write speed with 5D3 with Sandisk Extreme Pro 64gb 160mb/s @ 92.7MB/s

1.1.3 (24p)

Walter Schulz

Would you mind adding a second run in Photo mode? Press Play button after benchmark starts for Playback mode.

Ali Oliya

Hi Walter,

playmode


2nd run in movie mode:


Walter Schulz


DeafEyeJedi

@Ali Oliya:

If budget limits you (or want to save $) try looking into getting one of KB's fast CF cards (1066x) since they work well for me.

They're cheap, fast and reliable!

(Includes a lifetime warranty!)
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Ali Oliya

Quote from: DeafEyeJedi on April 10, 2015, 08:44:12 PM
@Ali Oliya:

If budget limits you (or want to save $) try looking into getting one of KB's fast CF cards (1066x) since they work well for me.

They're cheap, fast and reliable!

(Includes a lifetime warranty!)

Thanks for your advice
The 95.8 MB/s record you reached with the KG 1066x rocked my socks too!! . I bought this SanDisk few weeks ago. Actually I heard lots of mixed up negative and positive thoughts about KB cards in and out of forum. some talking about the impressive speed, some mentioned it gets hot and unstable...some said it works well and some said it doesn't work at all! add lots of corruption reports to that. these completely opposite comments made me consider buying something more truthworthy and reliable like SanDisk or Lexar. (i saw this article http://www.cinema5d.com/guide-raw-on-a-5d-mark-iii-magic-lantern/) and saw zero failure reports on SanDisk cards.

I'm in "Iran" and here the fast CF cards like SanDisk are double the global price(no KB here); so i had to purchase it abroad. there is also lots of limitations in ship so if the product i receive doesn't work there will be a nightmare of sending it back and getting a new one. that's why choosing the most reliable card was the only way for me. I couldn't risk.


DeafEyeJedi

Absolutely and I totally understand. I was on the same boat for years and once I purchased several KB's man I was floored and felt like a fool for not getting them sooner.

128GB 1066x seems to be the sweet spot for the 5D3, IMHO.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Ali Oliya

How many of the purchased KB cards worked? How many failed?!

Have you ever encountered any failure report on these KB 1066x cards?

Ps. In the previous posts you have mentioned you reached 103-106 during 3.5k! Is that with card spanning?

DeafEyeJedi

I've bought 3 so far and all of them have worked as advertised. It was surprising to me because I was expecting to send one back based on mixed reviews.

Guess I got lucky?

As for 2.5k (not 3.5k) crop-mode... I think it was without card spanning but I could be wrong and will double check.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Ali Oliya

@DeafEyeJedi

would you please share your experience with 3.5K ? (3584 x 1320) ...ever been able to shoot more than 2-3 seconds? in 23 or 24fps ?