[Letux-kernel] The new SMP driver and the new cache driver.

H. Nikolaus Schaller hns at goldelico.com
Sat Jan 16 19:10:10 CET 2021


> Am 16.01.2021 um 19:05 schrieb dave <dave at ds0.me>:
> 
> Is there a typo in the OMAP5 result? Or is it really that good :) ?

It was copy&paste and much faster than I had expected.

Well, I am not sure if I have the same debian release on both. So if that changed...

Please verify with 

	cat </dev/tcp/time.nist.gov/13 && sysbench --test=cpu --num-threads=4 run && cat </dev/tcp/time.nist.gov/13

This should also allow to verify if the internal timers and calculations are done correctly.

> 
> -------- Original message --------
> From: "H. Nikolaus Schaller" <hns at goldelico.com>
> Date: 16/01/2021 17:53 (GMT+00:00)
> To: Zhou Yanjie <zhouyanjie at wanyeetech.com>
> Cc: mips-creator-ci20-dev at googlegroups.com, Discussions about the Letux Kernel <letux-kernel at openphoenux.org>, Paul Boddie <paul at boddie.org.uk>, Riccardo Mottola <riccardo.mottola at libero.it>, Ezequiel Garcia <ezequiel at vanguardiasur.com.ar>
> Subject: Re: [Letux-kernel] The new SMP driver and the new cache driver.
> 
> Hi,
> 
> > Am 16.01.2021 um 07:55 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
> > 
> > This one booted without segfaults with the same rootfs as before. At least I haven't seen such.
> 
> I have rebuilt my tree and reinstalled and have not yet seen a segfault or hang as before.
> 
> No idea what went wrong. Could have been fetching/using the wrong defconfig. I have committed
> the letux_defconfig and there is now a new branch "letux-5.10.7+jz4780-smp" [1] which appears
> to work for me.
> 
> sysbench --cpu --num-threads=4 reports execution time (seconds) for some devices:
> 
> CI20 single threaded	139.4188
> CI20:	69.5500
> Udoo Neo (i.MX6):	287.8606	[single core Cortex A8] 
> RasPi 3B+:	103.3249	[dual core CI20 is faster than quad-core A53 with 1.4 GHz (in 32 bit mode)]
> PocketBeagle (AM3358):	359.8562	[single core Cortex A8]
> Pyra (OMAP5):	9.9983	[dual Cortex A15 with 1.5GHz]
> Pinephone:	52.9954	[quad core Cortex A53 (in 32 bit mode)]
> 
> That shows that the "old" jz4780 isn't that bad in performance if we have SMP enabled.
> And compared to the OMAP5 it stays quite cool (<40°C).
> 
> Next step is to try to integrate SMP with the CI20-HDMI stuff. Then we may get a first
> "longterm stable kernel" SMP solution, even if the patches need some polishing for
> upstreaming.
> 
> BR and thanks for all the hard work on SMP,
> Nikolaus
> 
> 
> [1] https://git.goldelico.com/?p=letux-kernel.git;a=shortlog;h=refs/heads/letux-5.10.7%2Bjz4780-smp
> 
> 
> Attached is some system info:
> 
> root at letux:~# cat /proc/cpuinfo 
> system type             : JZ4780
> machine                 : img,ci20
> processor               : 0
> cpu model               : Ingenic XBurst V4.15  FPU V0.0
> BogoMIPS                : 1196.85
> wait instruction        : yes
> microsecond timers      : no
> tlb_entries             : 32
> extra interrupt vector  : yes
> hardware watchpoint     : yes, count: 1, address/irw mask: [0x0fff]
> isa                     : mips1 mips2 mips32r1 mips32r2
> ASEs implemented        :
> shadow register sets    : 1
> kscratch registers      : 0
> package                 : 0
> core                    : 0
> VCED exceptions         : not available
> VCEI exceptions         : not available
> 
> processor               : 1
> cpu model               : Ingenic XBurst V4.15  FPU V0.0
> BogoMIPS                : 1202.58
> wait instruction        : yes
> microsecond timers      : no
> tlb_entries             : 32
> extra interrupt vector  : yes
> hardware watchpoint     : yes, count: 1, address/irw mask: [0x0fff]
> isa                     : mips1 mips2 mips32r1 mips32r2
> ASEs implemented        :
> shadow register sets    : 1
> kscratch registers      : 0
> package                 : 0
> core                    : 1
> VCED exceptions         : not available
> VCEI exceptions         : not available
> 
> root at letux:~# uname -a
> Linux letux 5.10.7-letux-ci20+ #4688 SMP PREEMPT Sat Jan 16 16:14:24 CET 2021 mips GNU/Linuxroot at letux:~# cat </dev/tcp/time.nist.gov/13 && sysbench --test=cpu run && cat </dev/tcp/time.nist.gov/13
> 
> 59230 21-01-16 16:48:00 00 0 0 651.3 UTC(NIST) * 
> sysbench 0.4.12:  multi-threaded system evaluation benchmark
> 
> Running the test with following options:
> Number of threads: 1
> 
> Doing CPU performance benchmark
> 
> Threads started!
> Done.
> 
> Maximum prime number checked in CPU test: 10000
> 
> 
> Test execution summary:
>     total time:                          139.4414s
>     total number of events:              10000
>     total time taken by event execution: 139.4188
>     per-request statistics:
>          min:                                 13.90ms
>          avg:                                 13.94ms
>          max:                                 14.46ms
>          approx.  95 percentile:              13.98ms
> 
> Threads fairness:
>     events (avg/stddev):           10000.0000/0.00
>     execution time (avg/stddev):   139.4188/0.00
> 
> 
> 59230 21-01-16 16:50:20 00 0 0 943.6 UTC(NIST) * 
> root at letux:~# 
> _______________________________________________
> https://projects.goldelico.com/p/gta04-kernel/
> Letux-kernel mailing list
> Letux-kernel at openphoenux.org
> http://lists.goldelico.com/mailman/listinfo.cgi/letux-kernel
> _______________________________________________
> https://projects.goldelico.com/p/gta04-kernel/
> Letux-kernel mailing list
> Letux-kernel at openphoenux.org
> http://lists.goldelico.com/mailman/listinfo.cgi/letux-kernel



More information about the Letux-kernel mailing list