Page 30 of 44
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 15:03
by JezC
:thumbup:
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 15:28
by exxos
36MHz WITH WAIT STATES
0WS_result.png
36MHZ WITHOUT WAITSTATES
NOWS_result.png
And just for the sake of comparison with the previous V1 32MHz booster
ste32mhz.jpg
So basically 36MHz current booster with waitstes on is the same speed as the previous 32MHz booster... With the exception that integer division is now at 401% from 355%. The CPU scores are showing lower because of the lower ROM access speed. Overall this makes no odds as it is still effectively a 32MHz booster.
Without the wait states we have a small boost in speed in most of the tests.. But as said before, it will be very hit and miss if any machines will run in this mode.
Now onto re-trying the 40Mhz osc with waitstates enabled...
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 15:49
by exxos
This is 40MHz with wait states
40WS_result.png
just for comparison again the 36MHz....
0WS_result.png
ROM Access has tripped over to 309% again... So I'm hoping the ROM will can okay still... But of course this is also pushing the CPU again now..
Will post back how many loops it managed to do before crashing later... :)
Though I really do hope that it will sustain 40MHz... But that still just means that this particular booster will run at that speed, it does not mean that all the CPUs will run at 40MHz....
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 15:51
by troed
I'm confused by the much higher aggregate CPU result (254%) for the 32MHz one compared to 38 or even 40MHz without waitstates.
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 15:56
by JezC
troed wrote: 09 Nov 2018 15:51
I'm confused by the much higher aggregate CPU result (254%) for the 32MHz one compared to 38 or even 40MHz without waitstates.
Yes - looks like all the individual cpu scores are higher but the overall cpu is lower...bizarre :?
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 16:14
by exxos
I said earlier... its because the ROM access drops to 267% so overall CPU scores are lower. But because its running at 38MHz, it gains back the speed loss from the slower ROM access speeds, and basically equals the 32MHz scores on individual tests.
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 16:28
by exxos
Oh no... your looking at GB6 tests from a older version before the averages was "fixed" .. you can't compare the overall scores... different calculations! Its why I introduced the version numbers in GB6 and also listed what calculation mode it was using in the title bar... Is even previous versions of GB6 may have different overall average calculations.. This actually caused a lot of confusion on my website as faster boosters were looking slower than me should be but were actually faster..
EDIT:
Made corrections to the previous benchmarks as I was actually using 36MHz osc :roll:
...Will try the 38 MHz one now...
EDIT2:
40MHz failed quickly... I'm not going to look into what is maxed out here as it is taking forever... But if this is still the ROM tripping up, it would need more weight states, so running at the high-speed is only going to increase integer division scores slightly... So I'm not going to bother even looking at 40MHz now.
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 17:07
by exxos
38MHz also locked up :(
Below are the results..
WITH WAITSTATES
GEO MODE
38WS_result.png
CLASSIC MODE
38WS2_result.png
EDIT:
Just for the hell of it I am trying 40MHz with 2 wait states.. It looks like overall scores are a couple percent slower..It locked up on the second pass loop, so I think this is the CPU maxing out now..
Overall it looks like 36MHz is just going to be the best overall compromise.
EDIT2:
Incidentally while I think about it, the CPU seems to lock up when it's running too fast, like it never sees DTACK.. No bombs or anything just freezes...
When the ROM is running too fast, it is mostly intermittent crashes, sometimes with a couple of bombs and it locks up with corrupted screen or something, but generally drops to desktop.
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 19:29
by PaulJ
Yes, back in the day of the pentium 4 everyone was over clocking.. It just got to the point where it was a supreme pain in the ass to have a processor that would randomly hang or crash. Got to the point I just gave the practice up. Random hangs and crashes are more trouble than the increased speed is worth. Had a friend that ran is motherboard in a fish tank of mineral oil and thought he has the best thing since sliced bread.. until he needed to move it.. :mrgreen: :mrgreen: :mrgreen:
Re: STE V1.5 32MHz BOOSTER - STATUS
Posted: 09 Nov 2018 19:32
by exxos
PaulJ wrote: 09 Nov 2018 19:29
Random hangs and crashes are more trouble than the increased speed is worth.
Totally agree. I did overclock my PC CPU, but I got the black edition is generally did overclocked well anyway.. In general I just don't bother any more. I just want things to be stable and work as well as possible.