VOGONS

Search

Common searches

Search results

Display options

Display options

Re: 486DX-33 vs 386DX-33

Posted on 2009-08-14, 03:13

1. Load times should still be mostly a function of disk, program execution will be faster on the 486. 2 doom ii, 486 faster. 3. the early wc series is non speed governed, in which a 486 can be too fast. I recall a discussion around here about the sweet spot, around 386-33 or 386-40/486-20. fyi, the …

CGA Compatibility Tester

Posted on 2009-08-12, 01:09

I figured this has long been known but a google site search of vogons didn't come up with any hits. Trixter wrote a CGA test program: http://www.oldskool.org/pc/cgacomp Running cvs set to cga, errors I've seen are: Under color select register, border/overscan color select doesn't show a border. …

Re: Vintage Win9X OS vs. practical audio choices

Posted on 2009-08-08, 16:43

http://sound-drivers.org/?id=6614&comp=Advanced%20Gravis&model=Gravis%20Ultrasound%20Classic%2C%20Ace%2C%20Pnp%20%28Interwave%29&fname=gus-bin-0.10.zip "WDM driver for Gravis UltraSound soundcards. It works under Windows 2000 and Windows 98 Second Edition and supports Gravis UltraSound Classic, Ace, …

"include 4DOS"

Posted on 2009-07-29, 11:23

This inspired an interesting thought, a dosbox distribution. Next time there's another 4DOS guy, can suggest they start a distro. I know I wouldn't mind seeing a pre collected set of free DOS utils and games.

Re: Got gcov to work in msys.

Posted on 2009-07-28, 18:13

Even though I run dynamic core all the time, lots of normal core running. These small changes to the inlined fetches give a big drop, 6574 bytes. core_normal.cpp: 501065914: 110:static INLINE Bit8u Fetchb() { 250532957: 111: Bit8u temp=LoadMb(core.cseip); 250532957: 112: core.cseip+=1; -: 113: …

Re: Got gcov to work in msys.

Posted on 2009-07-28, 15:54

Here's a size optimized finddynreg. Besides being 10 bytes smaller overall, the big nut is no replication of the for loop, which should improve cache thrashing(gcov shows both loops being liberally used.) I haven't measured the cycles usage difference yet though. //194 KB (198,811 bytes) //194 KB ( …

Re: Got gcov to work in msys.

Posted on 2009-07-28, 13:42

If you're interested in threading, this guy is a candidate, though a little fine grained: function _ZN15CodePageHandler14FindCacheBlockEj called 408253959 returned 100% blocks executed 100% 408253959: 313: CacheBlock * FindCacheBlock(Bitu start) { 408253959: 314: CacheBlock * block=hash_map[1+(start …

Re: Got gcov to work in msys.

Posted on 2009-07-28, 10:50

Meh, I'd leave (enabled && needed>done) if it's going to muck up behavior bad enough, it's just at first glance it doesn't look like that would be. @wd yeah, mix is pretty far down the list, it was just the first thing I ran into and my penchant for optimizing just because, though gprof shows …

Re: Got gcov to work in msys.

Posted on 2009-07-27, 21:14

I misspoke. Yes, it doesn't matter here. 331812562: 451:bool PIC_RunQueue(void) { -: 452: /* Check to see if a new milisecond needs to be started */ 331812562: 453: CPU_CycleLeft+=CPU_Cycles; 331812562: 454: CPU_Cycles=0; 331812562: 455: if (CPU_CycleLeft<=0) { 333454: 456: return false; -: 457: } …

Re: Got gcov to work in msys.

Posted on 2009-07-27, 17:33

Enabled is checked, and it's an &&, no short circuit involved. Can get rid of the inside if else too, which gcov shows oscillates between both conditions, and also save half a cache line in the process. // 784 KB (803,055 bytes) before // 784 KB (803,023 bytes) after void MixerChannel::Mix(Bitu _ …

Got gcov to work in msys.

Posted on 2009-07-27, 17:08

Browsing the output, noticed in the while below, enabled doesn't appear to be changed unless it's possible by handler(todo)?, so the check could be pulled out of the loop with "if (enabled) while (needed>done) {.." function _ZN12MixerChannel3MixEj called 782296 returned 100% blocks executed 100%^M …

Page 11 of 41

Go to top of page Go to top of page
Close window
Close window