[Libva] Is there a limit on number of parallel encoders?

Torbjorn tyridal torbjorn at tyridal.no
Fri May 30 00:50:43 PDT 2014


Hi,

On 05/29/2014 06:18 AM, Zhao, Yakui wrote:
> On Wed, 2014-05-28 at 04:34 -0600, Torbjorn tyridal wrote:
>> if I run two of the above in parallel, each goes along at 230fps, 18%
>> cpu (top). ie a total of 460 fps/36% cpu
>> If I then add a third instance the frame rate drops to 72fps (total 216)
>> and the cpu load jumps to 86 / 198%.
>>
>> Is there an explanation for the above in libva? 
> How about the configuration on your machines? For example: cpuinfo
> (cat /proc/meminfo) and meminfo (cat /proc/meminfo).
proc-info attached.

> How about the test result if the encode reads the frame from the file on
> harddisk directly?
>
reading from physical disk gives comparable numbers (after heating the
cache):
1 @ 273, 2 @ 226, 3 at 73,  4 @ 51

On 05/29/2014 05:25 AM, Xiang, Haihao wrote:

> What is the result if you set to 1 slice per frame ?

single-slice performance is also comparable, and most importantly, the
same fall off at 3 concurrent encoders.
1 at 256fps, 2 at 212fps, 3 at 74fps,  4 at 51fps


the clip is a medium complex one, showing a I-110 intersection in Los
Angeles. (ie lots of moving cars)
-------------- next part --------------
~# cat /proc/cpuinfo
[...0..6...]
processor	: 7
vendor_id	: GenuineIntel
cpu family	: 6
model		: 70
model name	: Intel(R) Core(TM) i7-4770R CPU @ 3.20GHz
stepping	: 1
microcode	: 0xf
cpu MHz		: 3232.000
cache size	: 6144 KB
physical id	: 0
siblings	: 8
core id		: 3
cpu cores	: 4
apicid		: 7
initial apicid	: 7
fpu		: yes
fpu_exception	: yes
cpuid level	: 13
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid
bogomips	: 6385.64
clflush size	: 64
cache_alignment	: 64
address sizes	: 39 bits physical, 48 bits virtual
power management:


~# cat /proc/meminfo
MemTotal:       16335484 kB
MemFree:        13385140 kB
MemAvailable:   14925372 kB
Buffers:           93544 kB
Cached:          2077600 kB
SwapCached:            0 kB
Active:          2017160 kB
Inactive:         597964 kB
Active(anon):     830528 kB
Inactive(anon):   128888 kB
Active(file):    1186632 kB
Inactive(file):   469076 kB
Unevictable:           0 kB
Mlocked:               0 kB
SwapTotal:       8200188 kB
SwapFree:        8200188 kB
Dirty:                 0 kB
Writeback:             0 kB
AnonPages:        443940 kB
Mapped:            94328 kB
Shmem:            515412 kB
Slab:             152844 kB
SReclaimable:     106904 kB
SUnreclaim:        45940 kB
KernelStack:        2640 kB
PageTables:        19568 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    16367928 kB
Committed_AS:    2277328 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      403136 kB
VmallocChunk:   34359330168 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:      132652 kB
DirectMap2M:     2914304 kB
DirectMap1G:    13631488 kB

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/libva/attachments/20140530/72d67de4/attachment.sig>


More information about the Libva mailing list