Raspberry Pi 4 Thread - general discussion

[jbeale]

/dev/mmcblk0: Timing buffered disk reads: 96 MB in 3.03 seconds = 31.66 MB/sec

Thanks for the reply. 31.66 MB/sec seems pretty fast for a SD Card, at least on the Pi.
I see https://www.pidramble.com/wiki/benchmarks/microsd-cards reports about the same rate as the fastest result for the Pi2B using a SSD via USB.

[morphious66]

/dev/mmcblk0: Timing buffered disk reads: 96 MB in 3.03 seconds = 31.66 MB/sec

Thanks for the reply. 31.66 MB/sec seems pretty fast for a SD Card, at least on the Pi.
I see https://www.pidramble.com/wiki/benchmarks/microsd-cards reports about the same rate as the fastest result for the Pi2B using a SSD via USB.

Very true! but 2/3rds down it shows the overclocked ones, some hitting 40. I hope we have this abilty on the PI 4 to crank it up....

About previous security concern on the Curl link posted. As someone with 27+ years of working experience in the tech sector, I would personally never post a link to anything that is not a previously posted and tested CLI command / script. A quick search on Google will show this curl command and URL / script has been used by dozens of groups to test speeds of SD cards.

[elatllat]

Someone lucky enough to already have a pi-4, please post the output of

Code: Select all

cryptsetup benchmark

[Edit]
Looks like crypto will slow the pi-4 down to USB2 speed.

https://forum.odroid.com/viewtopic.php? ... 3&p=215037

[jdb]

I've removed references to commands that do "download this random shellscript and execute it as root". This is universally bad practice - don't do it.
There are any number of benchmarking applications out there - bonnie++ and iozone are two examples.

[morphious66]

I've removed references to commands that do "download this random shellscript and execute it as root". This is universally bad practice - don't do it.
There are any number of benchmarking applications out there - bonnie++ and iozone are two examples.

ok, Is this a better solution? Can someone here please go to https://www.pidramble.com/wiki/benchmarks/microsd-cards or here https://www.raspberrypi.org/forums/view ... 3&t=140461 and use one of the methods listed on this page to get a good reading from their Pi 4 for the SD card speed and clock speed of it.

[HiassofT]

ok, Is this a better solution? Can someone here please go to https://www.pidramble.com/wiki/benchmarks/microsd-cards or here https://www.raspberrypi.org/forums/view ... 3&t=140461 and use one of the methods listed on this page to get a good reading from their Pi 4 for the SD card speed and clock speed of it.

Testing with a Sandisk Extreme 32GB class A1 on RPi4, current Raspbian Buster Lite (stock settings / config.txt, no overclock etc):

Code: Select all

root@raspberrypi:~# hdparm -t /dev/mmcblk0

/dev/mmcblk0:
 HDIO_DRIVE_CMD(identify) failed: Invalid argument
 Timing buffered disk reads: 132 MB in  3.00 seconds =  43.95 MB/sec

Code: Select all

root@raspberrypi:~# dd if=/dev/zero of=/home/pi/test bs=8k count=50k conv=fsync
51200+0 records in
51200+0 records out
419430400 bytes (419 MB, 400 MiB) copied, 11.4197 s, 36.7 MB/s

Code: Select all

root@raspberrypi:/home/pi/install/iozone/iozone3_434/src/current# ./iozone -e -I -a -s 100M -r 4k -i 0 -i 1 -i 2
	Iozone: Performance Test of File I/O
	        Version $Revision: 3.434 $
		Compiled for 32 bit mode.
		Build: linux-arm 
		
	Contributors: ...

	Run began: Tue Jul  2 23:09:48 2019

	Include fsync in write timing
	O_DIRECT feature enabled
	Auto Mode
	File size set to 102400 kB
	Record Size 4 kB
	Command line used: ./iozone -e -I -a -s 100M -r 4k -i 0 -i 1 -i 2
	Output is in kBytes/sec
	Time Resolution = 0.000001 seconds.
	Processor cache size set to 1024 kBytes.
	Processor cache line size set to 32 bytes.
	File stride size set to 17 * record size.

                                                             random    random     bkwd    record    stride                                    
              kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
          102400       4     3034     3151    10099    10100     9269     4343                                                                

iozone test complete.

Edit: iozone results seem to vary a bit, so take these with a grain of salt.

I did some more tests with performance governor, that bumped random read to about 12MB/sec but random write was between 4.4 and 4.9 (did 3 runs):

Code: Select all

                                                              random    random     bkwd    record    stride                                    
              kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
          102400       4     3401     3435    13154    13151    11905     4368                                                                
          102400       4     3365     3456    13198    13194    12027     4962                                                                
          102400       4     3326     3414    13181    13179    12009     4405                                                                

so long,

Hias

[morphious66]

Timing buffered disk reads: 132 MB in 3.00 seconds = 43.95 MB/sec

That looks to be a little more the double the 3 B+

I wonder if this is due to a DDR50 increase or a clock speed increase.

can you test for this one?

Raspberry Pi Dramble microSD benchmarks

Should see something like this:

microSD clock: 100.000 MHz

Also anyone with a SanDisk or other A2 micro SD card that can test this as well?

[Gomoto]

Code: Select all

root@raspberrypi:/home/pi# hdparm -t /dev/sda3

/dev/sda3:
 Timing buffered disk reads: 644 MB in  3.00 seconds = 214.51 MB/sec

OCZ Vertex 2 60GB & Sabrent 2.5" SATA USB 3.0 Adapter
Booting from sd card & root on ssd

[HiassofT]

Sorry, I'm not going to run random bash scripts from the internet. I used the commands as on the Raspberry Pi Dramble microSD benchmarks site.

Most benchmarkers though recommend to set the performance governor as ondemand can skew the results and make them impossible to compare - see the update to my previous post.

The performance is about what you'd expect from DDR50 clocking.

Here's the output from dmesg:

Code: Select all

root@raspberrypi:~# dmesg | grep mmc0
[    1.129772] mmc0: SDHCI controller on fe340000.emmc2 [fe340000.emmc2] using ADMA
[    1.234256] mmc0: new ultra high speed DDR50 SDHC card at address aaaa
[    1.237037] mmcblk0: mmc0:aaaa SE32G 29.7 GiB

so long,

Hias

[morphious66]

The performance is about what you'd expect from DDR50 clocking.

Here's the output from dmesg:

Code: Select all

root@raspberrypi:~# dmesg | grep mmc0
[    1.129772] mmc0: SDHCI controller on fe340000.emmc2 [fe340000.emmc2] using ADMA
[    1.234256] mmc0: new ultra high speed DDR50 SDHC card at address aaaa
[    1.237037] mmcblk0: mmc0:aaaa SE32G 29.7 GiB

Strange its not showing the clock speed such as mine from Pi 3 B+ Overclocked:

~# dmesg | grep mmc0
[ 0.797229] mmc0: sdhost-bcm2835 loaded - DMA enabled (>1)
[ 0.927537] mmc0: host does not support reading read-only switch, assuming write-enable
[ 0.932638] mmc0: overclocking to 100000000Hz
[ 0.935758] mmc0: new high speed SDHC card at address 59b4
[ 0.939475] mmcblk0: mmc0:59b4 00000 14.9 GiB

[Heater]

morphious66,

About previous security concern on the Curl link posted. As someone with 27+ years of working experience in the tech sector, I would personally never post a link to anything that is not a previously posted and tested CLI command / script. A quick search on Google will show this curl command and URL / script has been used by dozens of groups to test speeds of SD cards.

With all due respect to your years of experience, that is exactly what you did.

Your suggested command was (suitably obfuscated):

Code: Select all

curl https://raw.githubusercontent.com/randomeUser/randomProject/master/setup/benchmarks/randomScript.sh | sudo bash

Quite likely that is fine. It may well have been used by many and work well.

BUT DON'T DO THAT. Because you do not know if that random script does not get changed in the future.

By all means link to any script or other software. But we should not be encouraging people to habitually download and run random stuff off the net as root.

(Sorry for shouting. This is important)

[jbeale]

Very true! but 2/3rds down it shows the overclocked ones, some hitting 40. I hope we have this ability on the PI 4 to crank it up....

Clearly I haven't been paying attention, I didn't realize you could get such a huge boost by overclocking the SD memory on the Pi3. I'm amazed that 40 MB/s is possible. Maybe I am too old and conservative but it's been my experience those SD cards are by far the least reliable part of a Pi system. If you do a lot of data writing (heavy data logging etc) they can wear out in a year or so, and sometimes sooner if you have a power failure at a bad time. So I don't know about the Pi4 situation but I worry about the reliability implications of overclocking what seems to me already the weakest link in the system.

[bensimmo]

I posted the results of the pidramble benchmarks some days ago (pidramble have been used in MagPi and link the RPi Blog many times iirc)

It is seen at 50MHz and the 'overclock' does nothing. Tested with SanDisk A1 card so far
It nothing earth shattering, if not mostly slower than an overclocked SD on a Pi3.


Edit link back to it.
https://www.raspberrypi.org/forums/view ... k#p1486278

Not one is stretch, Pi4 is buster so not directly comparable for benchmarking.

[Imperf3kt]

Very true! but 2/3rds down it shows the overclocked ones, some hitting 40. I hope we have this ability on the PI 4 to crank it up....

Clearly I haven't been paying attention, I didn't realize you could get such a huge boost by overclocking the SD memory on the Pi3. I'm amazed that 40 MB/s is possible. Maybe I am too old and conservative but it's been my experience those SD cards are by far the least reliable part of a Pi system. If you do a lot of data writing (heavy data logging etc) they can wear out in a year or so, and sometimes sooner if you have a power failure at a bad time. So I don't know about the Pi4 situation but I worry about the reliability implications of overclocking what seems to me already the weakest link in the system.

The overclock applies to the bus speed of the SD card. The cards themselves are able to handle faster than the overclock, but the Pi itself bottlenecked that speed (hence the overclock)

[a22zsolutions]

Does rpi4 has multiple/additional uart, spi, i2c etc peripherals?

As per rpi4 schematic it doesnt?

[gordon77]

Is the CPU Usage Monitor supposed to show a figure on the Pi4 with Buster ?

[jamesh]

Does rpi4 has multiple/additional uart, spi, i2c etc peripherals?

As per rpi4 schematic it doesnt?

Yes, it does, they are muxed on to the GPIO's.

[jamesh]

Is the CPU Usage Monitor supposed to show a figure on the Pi4 with Buster ?

Right click the icon, select CPU Usage monitor settings, select Show usage as a percentage.

[gordon77]

Is the CPU Usage Monitor supposed to show a figure on the Pi4 with Buster ?

Right click the icon, select CPU Usage monitor settings, select Show usage as a percentage.

Thanks

[ColonelDare]

Does anyone know what happened to Clementine in the Buster repos?
I use it lots, on several devices, and will miss it if I have to move up from Stretch

It looks like it got caught up in the QT opengl es mess.

Debian builds (and raspbian doesn't change this) QT5 against opengl ES on armel and armhf because more arm boards have (or at least had) drivers for opengl ES than for desktop opengl. Unfortunately this breaks apps that rely on both desktop opengl and QT. The result seems to be that when clementine moved from QT4 to QT5, support for armel and armhf was dropped

Thanks for the reply Pugwash - am I getting seriously off topic or is there likely a work around the QT thing?

Although I fearlessly battle aliens across the galaxy [cf. my handle] I am not very smart at moving across repos or side-loading things I don't understand (but I am smart enough not to load stuff I don't know about). Would moaning at Clementine help or just make me feel better?

Kind regards, Doug

[djazz]

Something wrong with OpenGL and textures on the RPi 4?

Pi 3:
Pi 4 with 4 GB:

Code: Select all

Mesa: User error: GL_INVALID_ENUM in glTexParameter(pname=GL_TEXTURE_MAX_ANISOTROPY)

is the only debug message Mesa gives me. Game runs otherwise, just wrong texture mapping, or wrong texture completely. Memory issue? It''s the 4 GB model.

It runs slower (by 1-2 fps) than Pi 3B at stock clocks.

[thejck]

Hello, I am new to the rasberry pi world but have used other boards.
I have been looking for a device that has dual hdmi outputs for a while.
My use case senario is that I want to send 1 hdmi signal to my 4k hdr tv and the second to my AVR.
I do not want to upgrade to a new 4k receiver. I also like to run my setup with my tv off for music sometimes. so this would be perfect.
Does anyone know if this would be possible witth the new board?
I would prefer to use the libreelec OS if possible but am open to other suggestions. My use case is for a entertainment center so something with KODI is preferred.

[jamesh]

Hello, I am new to the rasberry pi world but have used other boards.
I have been looking for a device that has dual hdmi outputs for a while.
My use case senario is that I want to send 1 hdmi signal to my 4k hdr tv and the second to my AVR.
I do not want to upgrade to a new 4k receiver. I also like to run my setup with my tv off for music sometimes. so this would be perfect.
Does anyone know if this would be possible witth the new board?
I would prefer to use the libreelec OS if possible but am open to other suggestions. My use case is for a entertainment center so something with KODI is preferred.

I am currently looking at directing HDMI sound to a different port to the video.

[thejck]

I am currently looking at directing HDMI sound to a different port to the video.

Thank You. what OS and media application are you using?

[6by9]

I am currently looking at directing HDMI sound to a different port to the video.

Thank You. what OS and media application are you using?

It's a function of the firmware, so OS and application independent, however our default OS is always Raspbian.