[NTG-context] bottlenecks

[Hans Hagen]

On 12/17/2022 5:02 PM, Henning Hraban Ramm via ntg-context wrote:
> Am 16.12.22 um 21:02 schrieb Hans Hagen via ntg-context:
>>> (2)
>>> This is a 720 page novel (first run with this LMTX version, some 
>>> fonts were still uncached):
>>>
>>
>>> mkiv lua stats  > runtime: 19.275 seconds, 720 processed pages, 720 
>>> shipped pages, 37.355 pages/second
>> so what if you add to the top
>>
>> \enableexperiments[fonts.compact]
> 
> ok, updated to latest LMTX, first run, all fonts still uncached:
> 
> mkiv lua stats  > runtime: 19.025 seconds, 720 processed pages, 720 
> shipped pages, 37.846 pages/second
> system          | total runtime: 73.873 seconds of 73.964 seconds
> 
> real    1m13,971s
> user    1m10,581s
> sys    0m2,312s
> 
> next run, with fonts.compact:
> 
> mkiv lua stats  > runtime: 19.327 seconds, 720 processed pages, 720 
> shipped pages, 37.254 pages/second
> system          | total runtime: 19.353 seconds of 19.453 seconds
> 
> real    0m19,461s
> user    0m18,966s
> sys    0m0,386s
> 
> without fonts.compact:
> 
> mkiv lua stats  > runtime: 19.252 seconds, 720 processed pages, 720 
> shipped pages, 37.398 pages/second
> system          | total runtime: 19.283 seconds of 19.379 seconds
> 
> 
> real    0m19,385s
> user    0m18,943s
> sys    0m0,380s
> 
> minimal difference.

the differences are mostly in loading times (say 10 times some cjk font 
or math font)

the best way to check if compact mode was active is to check if the 
number of font instances changed

for the record: compact mode actually has more overhead in the engine 
but it is compensated by other aspects of it

Hans


-----------------------------------------------------------------
                                           Hans Hagen | PRAGMA ADE
               Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
        tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl
-----------------------------------------------------------------