[TYPO3-core] RFC #11535: Frontend rendering is slowed down by timetracking

[Rupert Germann]

Rupert Germann schrieb:
> Thanks for that hint benni,
> seems that the work is already done .... ;-)
> 
> ok, I'll test 9657.
> Let's see if instantiating a null class and calling empty methods is 
> faster than not calling them.

some updated benchmark results:
siege -c 15 -i -b -t 2M -f urls43_perf.txt

CACHED PAGES

trunk
-------------------------------------------------------------
Transactions:                  32568 hits
Availability:                 100.00 %
Elapsed time:                 120.19 secs
Data transferred:             794.69 MB
Response time:                  0.06 secs
Transaction rate:             270.97 trans/sec
Throughput:                     6.61 MB/sec
Concurrency:                   14.97
Successful transactions:       32568
Failed transactions:               0
Longest transaction:            0.42
Shortest transaction:           0.00


patched -> 11535
-------------------------------------------------------------
Transactions:                  34276 hits
Availability:                 100.00 %
Elapsed time:                 120.51 secs
Data transferred:             836.50 MB
Response time:                  0.05 secs
Transaction rate:             284.42 trans/sec
Throughput:                     6.94 MB/sec
Concurrency:                   14.96
Successful transactions:       34276
Failed transactions:               0
Longest transaction:            0.78
Shortest transaction:           0.00


patched -> 9657v1
-------------------------------------------------------------
Transactions:                  34211 hits
Availability:                 100.00 %
Elapsed time:                 120.35 secs
Data transferred:             835.26 MB
Response time:                  0.05 secs
Transaction rate:             284.26 trans/sec
Throughput:                     6.94 MB/sec
Concurrency:                   14.97
Successful transactions:       34211
Failed transactions:               0
Longest transaction:            0.52
Shortest transaction:           0.00


trunk		= 100 %
patched 9657v1	= 105 %
patched 11535	= 105 %






UNCACHED PAGES

trunk
-------------------------------------------------------------
Transactions:                   4050 hits
Availability:                 100.00 %
Elapsed time:                 120.61 secs
Data transferred:             103.33 MB
Response time:                  0.45 secs
Transaction rate:              33.58 trans/sec
Throughput:                     0.86 MB/sec
Concurrency:                   14.97
Successful transactions:        4050
Failed transactions:               0
Longest transaction:            1.80
Shortest transaction:           0.06


patched 11535
-------------------------------------------------------------
Transactions:                   4310 hits
Availability:                 100.00 %
Elapsed time:                 119.84 secs
Data transferred:             109.65 MB
Response time:                  0.42 secs
Transaction rate:              35.96 trans/sec
Throughput:                     0.91 MB/sec
Concurrency:                   14.96
Successful transactions:        4310
Failed transactions:               0
Longest transaction:            1.76
Shortest transaction:           0.07


patched 9657v1
-------------------------------------------------------------
Transactions:                   4300 hits
Availability:                 100.00 %
Elapsed time:                 120.20 secs
Data transferred:             109.85 MB
Response time:                  0.42 secs
Transaction rate:              35.77 trans/sec
Throughput:                     0.91 MB/sec
Concurrency:                   14.96
Successful transactions:        4300
Failed transactions:               0
Longest transaction:            1.73
Shortest transaction:           0.06



trunk		= 100 %
patched 9657v1	= 106,5 %
patched 11535	= 107 %


as you see the results of both patches are virtually the same.
calling empty methods seems to be as expensive as processing a lot of 
if() conditions.

but Ingo's patch is simply the more elegant solution and so I
draw back RFC 11535 in favour of 9657 (which does not seem to have been 
published in this list until now).

Today I learned:
use the search in the bugtracker - it can save you a lot of work ;-)

greets
rupert