Does the speed of SD cards actually make a real word diff

[mattmiller]

Is there someone out there who says

Yes - I used to use card X and now I use card Y and my whole pi is significantly faster to use (in terms of things like editing files) or my specific program is significanlty faster

[fruitoftheloom]

http://www.pidramble.com/wiki/benchmarks

[mattmiller]

#lol

[DougieLawson]

Move your root filesystem to a chromium plated spinning thing with a flying head and you won't have to care about SDCard size or performance ever again.

[elkberry]

A Seagate ST4096 MFM might not exactly bode too well against even the most cheap mSD these days. But maybe someone can come up with the ST506 interface via GPIO??? Oh these there the days...!

[rpdom]

A Seagate ST4096 MFM might not exactly bode too well against even the most cheap mSD these days. But maybe someone can come up with the ST506 interface via GPIO??? Oh these there the days...!

Hmm, I have a book with details of the ST506 interface somewhere...

[mattmiller]

So we can take the answer as a no then?

No-one has noticed a real world difference between speed of SD cards

Glad we've sorted that one out

[solar3000]

I use the store brand micro SD cards. I've tried using newest blazing fast brand name sandisk cards that failed. Why flush money down the toilet. Makes no real difference. They only make benchmark differences.
The pi is not a performance machine. Why is everyone forcing it to be a 10 GHz flame thrower?

[gkreidl]

So we can take the answer as a no then?

No-one has noticed a real world difference between speed of SD cards

Glad we've sorted that one out

Simple real world test: compile a large project (something like VLC, for example).
After I had moved my first file system to USB HDD, I said "Wow!" when trying to compile a project. Don't ask me about SD card differences, I have never used them again afterwards (except for installation and the boot partition).

[dukla2000]

Is there someone out there who says

Yes - I used to use card X and now I use card Y and my whole pi is significantly faster to use (in terms of things like editing files) or my specific program is significanlty faster

Yup me. I subscribe to fruitoftheloom school of thought that an SDcard with leading random small file IO is worth the effort. For sure for me significantly improves boot speed, application loading speed and even web browsing speed.

Move your root filesystem to a chromium plated spinning thing with a flying head and you won't have to care about SDCard size or performance ever again.

IMHO I am not convinced. Haven't moved my root filesystem since I retired my PiB to the museum so don't have numbers but:
1) My Evo+ 64GB benchmarks 33.48/11.3/10.34/3.23 (as per fruitoftheloom tables and methods - in the same ballpark as his 32GB)
My WD Scorpio via SATA/USB link gets 35.03/37.0/0.52/1,68 (ditto)
OK so the magnetic option slaughters the SDcard in sequential file writes but is murdered by the SDcard in small file random reads. Which is why my / stays on my SDcard.

2) Pis are pretty restricted on IO channels. I figure using the MMC channel (which seems to me to have little apologise for on a Pi3 when done optimally compared to an HDD via USB) makes sense rather than adding to the potential USB bottleneck (either via direct connect or ETH).

[mattmiller]

For sure for me significantly improves boot speed, application loading speed and even web browsing speed.

Would you hazard an estimate of how much significantly means to those 3 categories

10? 20? 50? 100%? speedup compared to a standard official NOOBS card?

[DougieLawson]

Move your root filesystem to a chromium plated spinning thing with a flying head and you won't have to care about SDCard size or performance ever again.

IMHO I am not convinced. Haven't moved my root filesystem since I retired my PiB to the museum so don't have numbers but:

My four raspberries (B+, ZC, 3B & 3B) absolutely fly with their USB hard drives (three of them are integrated WD units). They aren't RAM constrained because you can put a swapfile on a USB HD. So you're wrong.

[dukla2000]

For sure for me significantly improves boot speed, application loading speed and even web browsing speed.

Would you hazard an estimate of how much significantly means to those 3 categories

10? 20? 50? 100%? speedup compared to a standard official NOOBS card?

I don't have a NOOBS card so can't comment directly. IIRC the NOOBS cards have benchmarked decently from other reports somewhere here.

But I'm playing with SUSE/aarch64 as well and selected my best (after my EVO+ that has Raspbian on it) card from my menagerie for openSUSE - a 16GB noname that benchmarks better than the rest of the menagerie (including a SANdisk Extreme, a SANdisk mobile Ultra and a Toshiba Exceria). First boot of openSUSE image does a whole lot of stuff like expand the root partition, create a swap partition etc and takes 9 minutes on my 16GB card. Just got a 32GB EVO+, same image took 4 minutes.

[mattmiller]

Well that's what I'd call significant Ta