Sawys' Scramble Spirits recording
Moderator: BBH
-
- MARP Seer
- Posts: 788
- Joined: Sun May 19, 2002 9:06 am
The good time
This time it's the good one.
Sawys is Sébastien GIRAUD, admin of MAMESCORE, member of MARP, TWIN GALAXIES, RETRO UPRISING and ex-member of JVRM 

I found the answer
Ok !
I try all the options in Wolfmame and finally found what was wrong !!
The option "Use RDTSC Timing system" was not marked (don't know the exact word).
So my score is not good because in reality it was recorded at almost 50%
To have a good inp I must play in fs8 (it's harder like this).
Can you tell me if I can make a record with fs8 ?
Finally I wonder why it is possible to choose or not this option "Use RDTSC Timing system" because anybody can make a record without this option and say he records in the good speed.
The new versions of Wolfmame and Mame should be modified ! And the old banished for the record !!
The only way to be sure is to have the description of the computer used.
I tried with surf planet and my record was 100%. But in reality, the game isn't well emulated almost 45 %.
I think it's maybe the cause of certains Scores in MARP which are really very big !! Some are true but maybe some are wrong !! What do you think of it ??? How knowing if the actual inps are good or not ??
So my inp can be deleted
((
I try all the options in Wolfmame and finally found what was wrong !!
The option "Use RDTSC Timing system" was not marked (don't know the exact word).
So my score is not good because in reality it was recorded at almost 50%
To have a good inp I must play in fs8 (it's harder like this).
Can you tell me if I can make a record with fs8 ?
Finally I wonder why it is possible to choose or not this option "Use RDTSC Timing system" because anybody can make a record without this option and say he records in the good speed.
The new versions of Wolfmame and Mame should be modified ! And the old banished for the record !!
The only way to be sure is to have the description of the computer used.
I tried with surf planet and my record was 100%. But in reality, the game isn't well emulated almost 45 %.
I think it's maybe the cause of certains Scores in MARP which are really very big !! Some are true but maybe some are wrong !! What do you think of it ??? How knowing if the actual inps are good or not ??
So my inp can be deleted

Sawys is Sébastien GIRAUD, admin of MAMESCORE, member of MARP, TWIN GALAXIES, RETRO UPRISING and ex-member of JVRM 

Sometimes either one of the main timing systems may not perform terribly on some systems.
RDTSC is a CPU instruction that returns the number of CPU cycles since the system was powered on. It is only available on Pentium-class CPUs (ie: anything that's not a 386 or 486)
The default method is to use the QueryPerformanceCounter() function, which is Windows-specific. This has a lower frequency than RDTSC does, but may give better performance on some systems.
There is also a fallback if neither of the above is available, to use timeGettime() (I think) which is a Windows-specific function too.
You can check which one is used by MAME, by running it with the -log switch (referred to as a 'developers log' in the GUI), and opening error.log after exiting.
The first line will tell you which timing system is used. Something along the lines of:
RDTSC is a CPU instruction that returns the number of CPU cycles since the system was powered on. It is only available on Pentium-class CPUs (ie: anything that's not a 386 or 486)
The default method is to use the QueryPerformanceCounter() function, which is Windows-specific. This has a lower frequency than RDTSC does, but may give better performance on some systems.
There is also a fallback if neither of the above is available, to use timeGettime() (I think) which is a Windows-specific function too.
You can check which one is used by MAME, by running it with the -log switch (referred to as a 'developers log' in the GUI), and opening error.log after exiting.
The first line will tell you which timing system is used. Something along the lines of:
Code: Select all
using performance counter for timing ... cycles/second = 3579545
- Barry Rodewald
MARP Assistant Web Maintainer

MARP Assistant Web Maintainer

So how did this result in the phenomenon we were puzzled by here?mahlemiut wrote:Sometimes either one of the main timing systems may not perform terribly on some systems.
RDTSC is a CPU instruction that returns the number of CPU cycles since the system was powered on. It is only available on Pentium-class CPUs (ie: anything that's not a 386 or 486)
The default method is to use the QueryPerformanceCounter() function, which is Windows-specific. This has a lower frequency than RDTSC does, but may give better performance on some systems.
There is also a fallback if neither of the above is available, to use timeGettime() (I think) which is a Windows-specific function too.
You can check which one is used by MAME, by running it with the -log switch (referred to as a 'developers log' in the GUI), and opening error.log after exiting.
The first line will tell you which timing system is used. Something along the lines of:Code: Select all
using performance counter for timing ... cycles/second = 3579545
Can anything be done in future WolfMAMEs to eliminate the confusion?
John Cunningham (JTC)


With the Use RDTSC Timing System I have :
using RDTSC for timing ... cycles/second = 1000012928
May be that now I will have some problems to play games with this.
I try and most of them are very fast !! Is it the real speed... I don't know. I make a try with members of jvrm.net to test an inp on cairblad
using RDTSC for timing ... cycles/second = 1000012928
May be that now I will have some problems to play games with this.
I try and most of them are very fast !! Is it the real speed... I don't know. I make a try with members of jvrm.net to test an inp on cairblad
Sawys is Sébastien GIRAUD, admin of MAMESCORE, member of MARP, TWIN GALAXIES, RETRO UPRISING and ex-member of JVRM 
