I heard from some people they are getting 90MB/s from a disk enclosure with a single disk so maybe it is down to used HDD and enclosure? So not sure if this is a HDD/enclosure limitation or USB in R7800?
Still wondering is anyone tested it with fast HDDs idealy with a RAID5?
Got 20 MB/s using RAID1
and sadly 6.6MB/s with RAID5
Usb or esata?
USB 3 (all disks on the same port with external power supply)
Can someone give me output of this command?
cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
root@nbg6817:~# cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 37 8 11 18 14
600000: 47 0 30 25 15 16
800000: 10 26 0 67 33 2133304
1000000: 14 27 61 0 31 29
1400000: 6 17 36 31 0 46
1725000: 11 26 2133305 28 39 0
R7800, 4 days uptime, light load
-----------------------------------------------------
OpenWrt SNAPSHOT, r13978-d15b6e4895
-----------------------------------------------------
root@router1:~# cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 60281 20024 3777 0 12818
600000: 59375 0 562 534 0 2997
800000: 17989 607 0 481 0 2175
1000000: 2202 272 307 0 0 2299
1400000: 0 0 0 0 0 0
1725000: 17335 2308 358 288 0 0
root@router1:~# cat /sys/devices/system/cpu/cpufreq/policy1/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 868345 475362 30247 0 5908
600000: 977185 0 80923 2384 0 4250
800000: 374375 183861 0 938 0 2305
1000000: 22757 4583 4997 0 0 1364
1400000: 0 0 0 0 0 0
1725000: 5545 7953 197 132 0 0
Some context... I'm converting the cpufreq patch to use the cpufreq notifier as i will push an rfc patch upstream. I'm experiencing the problem that the router jump every 1-2 second from 384 to 600 and i'm trying to understand if it's same for the hacky driver.
if you are only interested in the r7800 then please ignore...
uptime command run just prior to cat trans_table
r7500v2 # cat /etc/openwrt_release | grep DESCRIPTION
DISTRIB_DESCRIPTION='OpenWrt SNAPSHOT r13673-f786de2095'
r7500v2 # uptime
14:40:59 up 3 days, 19:42, load average: 0.00, 0.04, 0.06
r7500v2 # cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1200000 1400000
384000: 0 2255 56 20 15 10
600000: 2280 0 31 6 9 18
800000: 36 48 0 12 9 18
1000000: 18 13 12 0 2 6
1200000: 8 13 8 4 0 21
1400000: 15 15 15 9 19 0
r7500v2 # uptime
14:41:34 up 3 days, 19:43, load average: 0.00, 0.03, 0.06
r7500v2 # cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1200000 1400000
384000: 0 2258 56 20 15 10
600000: 2283 0 31 6 9 18
800000: 36 48 0 12 9 18
1000000: 18 13 12 0 2 6
1200000: 8 13 8 4 0 21
1400000: 15 15 15 9 19 0
r7500v2 # uptime
14:42:55 up 3 days, 19:44, load average: 0.00, 0.02, 0.05
r7500v2 # cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1200000 1400000
384000: 0 2267 56 20 15 10
600000: 2292 0 31 6 9 18
800000: 36 48 0 12 9 18
1000000: 18 13 12 0 2 6
1200000: 8 13 8 4 0 21
1400000: 15 15 15 9 19 0
Would you like to test the new driver for me? (it doesn't cause stability problem, at worst it will scale a lot between 384 and 600 when idle)
@anon98444528 and everyone who can help...
This is the patch, just add the file to the patches folder and add
CONFIG_ARM_QCOM_KRAIT_CACHE=y
to the config-5.4 file
I can test, but it likely will be a day or two for me to get back to you. HTH
sure take your time.
If someone else want to try just make sure you don't lock the Freq to 800 min or the test is useless
OpenWrt SNAPSHOT, r14023-1deaf695ec
root@nbg6817:~# uname -a
Linux nbg6817 5.4.55 #0 SMP Mon Aug 3 10:17:08 2020 armv7l GNU/Linux
root@nbg6817:~# uptime
20:56:24 up 4 min, load average: 0.33, 0.51, 0.26
root@nbg6817:~# zcat /proc/config.gz | grep CONFIG_ARM_QCOM_KRAIT_CACHE
CONFIG_ARM_QCOM_KRAIT_CACHE=y
root@nbg6817:~# cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 313 54 34 42 173
600000: 337 0 68 49 57 387
800000: 59 76 0 401 121 315
1000000: 44 73 398 0 145 87
1400000: 103 43 123 148 0 218
1725000: 74 393 328 115 270 0
root@nbg6817:~# cat /sys/devices/system/cpu/cpufreq/policy1/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 756 101 85 49 88
600000: 803 0 58 27 36 147
800000: 98 84 0 90 112 133
1000000: 68 48 93 0 495 81
1400000: 70 33 118 479 0 82
1725000: 40 150 147 104 90 0
Just a quick test so far, but no problems seen so far.
Pls post another cat tomorrow so we can compare how Freq transition grow
root@nbg6817:~# uptime
01:07:16 up 1 day, 4:15, load average: 0.00, 0.02, 0.00
root@nbg6817:~# zcat /proc/config.gz | grep CONFIG_ARM_QCOM_KRAIT_CACHE
CONFIG_ARM_QCOM_KRAIT_CACHE=y
root@nbg6817:~# cat /sys/devices/system/cpu/cpufreq/policy0/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 145375 23920 5881 3455 23811
600000: 148021 0 3907 2718 3915 57450
800000: 25147 6187 0 6784 7000 50728
1000000: 7843 3343 5071 0 6561 7684
1400000: 12718 3428 9159 3634 0 4164
1725000: 8713 57678 53789 11485 12172 0
root@nbg6817:~# cat /sys/devices/system/cpu/cpufreq/policy1/stats/trans_table
From : To
: 384000 600000 800000 1000000 1400000 1725000
384000: 0 279221 35827 25975 13528 9996
600000: 285251 0 7027 2943 2536 15274
800000: 35331 10287 0 4101 3754 16174
1000000: 24990 4627 3334 0 4457 4725
1400000: 14816 3536 5304 3349 0 2499
1725000: 4159 15360 18155 5765 5229 0
@slh what is your use? high ? with spike ? anyway thanks a lot for the testing
@hnyman in your opinion the transition between 384 and 600 in one day are acceptable or should be smaller
I've been barely at home today, so aside from 2-3 quick speed tests immediately after rebooting yesterday, uplink and router didn't really have anything to do.
@slh made a mistake if you still want to test pls add this before the opp_table0: opp_table0 {
section
(i lost track of the recent commits tell me if we still have the ipq8064 dtsi or we patch the upstream one)
qcom-krait-cache {
compatible = "qcom,krait-cache";
clocks = <&kraitcc 4>;
clock-names = "l2";
voltage-tolerance = <5>;
l2-rates = <384000000 1000000000 1200000000>;
l2-cpufreq = <384000 600000 1200000>;
l2-volt = <1100000 1100000 1150000>;
l2-supply = <&smb208_s1a>;
};
Not really sure. I haven't monitored the stats earlier, so no real feeling on the expected changes.
But is is quite similar magnitude with mw oendata above, taken with the current implementation (where curiously the router never used the 1,4 GHz speed).