mvaisakh/AOSP Clang 9.x.bench.md

Last active May 30, 2021 18:23

Star () You must be signed in to star a gist
Fork () You must be signed in to fork a gist

Learn more about clone URLs
Clone this repository at <script src="https://gist.github.com/mvaisakh/a258e3d4201aad85c380886faf54c372.js"></script>
Save mvaisakh/a258e3d4201aad85c380886faf54c372 to your computer and use it in GitHub Desktop.

Download ZIP

Different benchmarks of cannon

Raw

AOSP Clang 9.x.bench.md

Different Kernel Benchmarks of Redmi Note 9T (cannong)

AOSP Clang 9.x (Stock configuration, debugging dropped)

Perf

$ perf bench sched all

Running sched/messaging benchmark... 20 sender and receiver processes per group 10 groups == 400 processes run

 Total time: 0.243 [sec]

Running sched/pipe benchmark... Executed 1000000 pipe operations between two processes

 Total time: 8.661 [sec]

   8.661027 usecs/op
     115459 ops/sec

Dhrystone

$ dhrystone -t 8 -l 10

duration: 0 seconds
number of threads: 8
number of loops: 10000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7570940 dhrystones/second
                           4309 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7570481 dhrystones/second
                           4309 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4388167 dhrystones/second
                           2498 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4160703 dhrystones/second
                           2368 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3249150 dhrystones/second
                           1849 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3137164 dhrystones/second
                           1786 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 2883214 dhrystones/second
                           1641 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.5
This machine benchmarks at 2776063 dhrystones/second
                           1580 DMIPS

Total dhrystone run time: 0.394985 seconds.

$ dhrystone -t 8 -l 100

duration: 0 seconds
number of threads: 8
number of loops: 100000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 12500000 passes = 1.4
This machine benchmarks at 8807072 dhrystones/second
                           5013 DMIPS
Dhrystone(1.1) time for 12500000 passes = 1.4
This machine benchmarks at 8780760 dhrystones/second
                           4998 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.5
This machine benchmarks at 5060473 dhrystones/second
                           2880 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.4
This machine benchmarks at 5198567 dhrystones/second
                           2959 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.4
This machine benchmarks at 3645152 dhrystones/second
                           2075 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.5
This machine benchmarks at 3522652 dhrystones/second
                           2005 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.6
This machine benchmarks at 3449918 dhrystones/second
                           1964 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.7
This machine benchmarks at 3408979 dhrystones/second
                           1940 DMIPS

Total dhrystone run time: 3.617268 seconds.

Sched

$ schbench

Latency percentiles (usec)
	50.0th: 3684
	75.0th: 8008
	90.0th: 14000
	95.0th: 18720
	*99.0th: 29856
	99.5th: 33344
	99.9th: 40896
	min=0, max=64397

Syscalls

$ callbench

clock_gettime: ..........
    syscall:	179 ns
    libc:	68 ns

read file: ..........
    mmap:	14566 ns
    read:	5680 ns

Hackbench

$ hackbench -pTl 4000

Running in threaded mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 4000 messages of 100 bytes

Time: 12.717

Raw

AOSP-clang-9.x.riced.bench.md

Different Kernel Benchmarks of Redmi Note 9T (cannong

AOSP Clang 9.x (Riced with IRQ boosts, debugging dropped)

Perf

$ perf bench sched all

 Running sched/messaging benchmark...
 20 sender and receiver processes per group
 10 groups == 400 processes run

     Total time: 0.282 [sec]

 Running sched/pipe benchmark...
 Executed 1000000 pipe operations between two processes

     Total time: 8.364 [sec]

       8.364013 usecs/op
         119559 ops/sec

Dhrystone

$ dhrystone -t 8 -l 10

duration: 0 seconds
number of threads: 8
number of loops: 10000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7382384 dhrystones/second
                           4202 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7372022 dhrystones/second
                           4196 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4350504 dhrystones/second
                           2476 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4087198 dhrystones/second
                           2326 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3526879 dhrystones/second
                           2007 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3205630 dhrystones/second
                           1824 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3037423 dhrystones/second
                           1729 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 2871563 dhrystones/second
                           1634 DMIPS

Total dhrystone run time: 0.364035 seconds

$ dhrystone -t 8 -l 100

duration: 0 seconds
number of threads: 8
number of loops: 100000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 12500000 passes = 1.4
This machine benchmarks at 8865016 dhrystones/second
                           5046 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.3
This machine benchmarks at 5533464 dhrystones/second
                           3149 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.2
This machine benchmarks at 5653832 dhrystones/second
                           3218 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.3
This machine benchmarks at 5363113 dhrystones/second
                           3052 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.2
This machine benchmarks at 3943359 dhrystones/second
                           2244 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.1
This machine benchmarks at 3974599 dhrystones/second
                           2262 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.5
This machine benchmarks at 3526765 dhrystones/second
                           2007 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.7
This machine benchmarks at 3375170 dhrystones/second
                           1921 DMIPS

Total dhrystone run time: 1.478579 seconds.

Sched

$ schbench

Latency percentiles (usec)
	50.0th: 3716
	75.0th: 8008
	90.0th: 13328
	95.0th: 17504
	*99.0th: 27616
	99.5th: 30240
	99.9th: 37696
	min=0, max=41681

Syscalls

$ callbench

clock_gettime: ..........
    syscall:	183 ns
    libc:	68 ns

read file: ..........
    mmap:	14458 ns
    read:	5626 ns

Hackbench

$ hackbench -pTl 4000

Running in threaded mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 4000 messages of 100 bytes

Time: 13.738

Raw

Eva GCC 12.x.bench.md

Different Kernel Benchmarks of Redmi Note 9T (cannong

Eva GCC 12.x (unriced, stock configuration, no LTO+CFI, debugging dropped)

Perf

$ perf bench sched all

Running sched/messaging benchmark... 20 sender and receiver processes per group 10 groups == 400 processes run

 Total time: 0.279 [sec]

Running sched/pipe benchmark... Executed 1000000 pipe operations between two processes

 Total time: 8.171 [sec]

   8.171303 usecs/op
     122379 ops/sec

Dhrystone

$ dhrystone -t 8 -l 10

duration: 0 seconds
number of threads: 8
number of loops: 10000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7926694 dhrystones/second
                           4511 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4612342 dhrystones/second
                           2625 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4471344 dhrystones/second
                           2545 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4140087 dhrystones/second
                           2356 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 3851451 dhrystones/second
                           2192 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3392103 dhrystones/second
                           1931 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3242828 dhrystones/second
                           1846 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3002563 dhrystones/second
                           1709 DMIPS

Total dhrystone run time: 0.378159 seconds

$ dhrystone -t 8 -l 100

duration: 0 seconds
number of threads: 8
number of loops: 100000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 12500000 passes = 1.8
This machine benchmarks at 7135811 dhrystones/second
                           4061 DMIPS
Dhrystone(1.1) time for 12500000 passes = 1.7
This machine benchmarks at 7569358 dhrystones/second
                           4308 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.3
This machine benchmarks at 5426932 dhrystones/second
                           3089 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.7
This machine benchmarks at 4582644 dhrystones/second
                           2608 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.3
This machine benchmarks at 3802324 dhrystones/second
                           2164 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.4
This machine benchmarks at 3642044 dhrystones/second
                           2073 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.7
This machine benchmarks at 3406715 dhrystones/second
                           1939 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.7
This machine benchmarks at 3356737 dhrystones/second
                           1910 DMIPS

Total dhrystone run time: 3.792679 seconds.

Sched

$ schbench

Latency percentiles (usec)
	50.0th: 3500
	75.0th: 7896
	90.0th: 12976
	95.0th: 16928
	*99.0th: 26656
	99.5th: 30176
	99.9th: 37952
	min=0, max=43235

Syscalls

$ callbench

clock_gettime: ..........
    syscall:	169 ns
    libc:	51 ns

read file: ..........
    mmap:	14178 ns
    read:	9017 ns

Hackbench

$ hackbench -pTl 4000

Running in threaded mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 4000 messages of 100 bytes

Time: 12.868

Raw

Stock-MIUI-kernel.bench.md

Different Kernel Benchmarks of Redmi Note 9T (cannong

Stock MIUI Kernel

Perf

$ perf bench sched all

 Running sched/messaging benchmark...
 20 sender and receiver processes per group
 10 groups == 400 processes run

     Total time: 0.255 [sec]

 Running sched/pipe benchmark...
 Executed 1000000 pipe operations between two processes

     Total time: 8.861 [sec]

       8.861086 usecs/op
         112852 ops/sec

Dhrystone

$ dhrystone -t 8 -l 10

duration: 0 seconds
number of threads: 8
number of loops: 10000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7305584 dhrystones/second
                           4158 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.2
This machine benchmarks at 7315289 dhrystones/second
                           4164 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 4209307 dhrystones/second
                           2396 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.3
This machine benchmarks at 3939353 dhrystones/second
                           2242 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3204060 dhrystones/second
                           1824 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 3049435 dhrystones/second
                           1736 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 2808049 dhrystones/second
                           1598 DMIPS
Dhrystone(1.1) time for 1250000 passes = 0.4
This machine benchmarks at 2786720 dhrystones/second
                           1586 DMIPS

Total dhrystone run time: 0.400584 seconds.

$ dhrystone -t 8 -l 100

duration: 0 seconds
number of threads: 8
number of loops: 100000000
delay between starting threads: 0 seconds

Dhrystone(1.1) time for 12500000 passes = 1.4
This machine benchmarks at 8749023 dhrystones/second
                           4980 DMIPS
Dhrystone(1.1) time for 12500000 passes = 1.4
This machine benchmarks at 8756715 dhrystones/second
                           4984 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.5
This machine benchmarks at 5040648 dhrystones/second
                           2869 DMIPS
Dhrystone(1.1) time for 12500000 passes = 2.6
This machine benchmarks at 4802044 dhrystones/second
                           2733 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.2
This machine benchmarks at 3871060 dhrystones/second
                           2203 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.6
This machine benchmarks at 3476651 dhrystones/second
                           1979 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.6
This machine benchmarks at 3469180 dhrystones/second
                           1974 DMIPS
Dhrystone(1.1) time for 12500000 passes = 3.7
This machine benchmarks at 3390390 dhrystones/second
                           1930 DMIPS

Total dhrystone run time: 3.675858 seconds.

Sched

$ schbench

Latency percentiles (usec)
	50.0th: 3612
	75.0th: 7976
	90.0th: 13232
	95.0th: 17568
	*99.0th: 28768
	99.5th: 32736
	99.9th: 40768
	min=0, max=57748

Syscalls

$ callbench

clock_gettime: ..........
    syscall:	183 ns
    libc:	68 ns

read file: ..........
    mmap:	14468 ns
    read:	5621 ns

Hackbench

$ hackbench -pTl 4000

Running in threaded mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 4000 messages of 100 bytes

Time: 11.75