最好的NDS模拟器「No$GBA」更新到了「2.6A 」、今回主要是改进了3D的模拟速度（据测评、提高很多）…………………………
另外、诸如存档类型将进行自动检测（「AUTO」选项、不用担心不会选择存档选项了）、改进纹理贴图等等也是今回更新的主要内容……………………
………………………………………………………………………………………………………………………………
……………………………………………………
新版详细改进如下……………………………………………………
23 January 2008 - version 2.6a
(faster 3d rendering, backup detect, rtc-irq, better texture interpolation,)
(perfectly accurate edge-marking, translucent-poly-id, shadow-poly support,)
(capture in vram display mode, debug gui/warnings, cpu/ipc/div/sqrt details)
- free-download: old no$gba v2.6 gaming version now free for everybody - enjoy
- debug/symbols: resize function for Alt+L symbol list window (saved in .ini)
- debug/internal: changed computer_id handling for compatibility with win vista
- nds/debug: allows nintendo to mis-use clipmtx_result for detecting their emu
- nds/debug: allows to override mis-declared-thumb-functions by crude $t labels
- nds/rtc: triggers IRQ when SI changes HI-to-LO (only when SI-IRQ is enabled)
- hll-version: demangles strange new "_ZN3txt3txtEii" and "_Z3txtii" type labels
- nds/gba/rcnt: allows to generate SI interrupts manually by toggling RCNT bits
- nds/gba/rcnt: internally memorizes rcnt-output bits (additionally to inputs)
- nds/rtc: passes rtc 1Hz/2Hz/4Hz/8Hz/16Hz or per-minute IRQ to rcnt SI input
- nds/3d: soft-speedup: uses clean 80286 shift opcodes (instead slow 80386 shrd)
- nds/3d: soft-speedup: new scaled side_clip_x allowed to re-remove pre_add_mask
- nds/3d: soft-speedup: scaled side_clip_x coords from 0..len to 0..7FFFFFFFh
- nds/3d: soft-accuracy: texture/color interpolation with variable pre_add_mask
- nds/details: emulates all newly discovered ipcfifo/div/sqrt technical details
- nds/help: added tech specs on div/sqrt (readonly results, start/stop timings)
- nds/help: added tech specs on ipcfifo (edge triggered, underrun, fifo-disable)
- nds/help: added user settings 076h (language mask) and header 01Dh (ique flag)
- nds/3d: soft-detail: edge-mark: recurses surrounding depth values (less only)
- nds/3d/help: added more technical notes on edge marking (depth and polygon_id)
- nds/a22i: auto generates chinese crc upon .fix directive (only if version=2)
- nds/help: added info on chinese title in icon/title region (addr/version/crc)
- cpu/internal: reduced 32bit test/and to 8bit (al/bl/cl/dl instead eax/ebx/etc)
- cpu/detail: emulates mis-aligned thumb bx/blx and arm bx/blx (with warning)
- cpu/detail: emulates mis-aligned rd=r15 in arm alu opcodes (thanks jonathan)
- nds/3d/help: added double-blended-edge-glitch (edge-marking plus anti-alias)
- nds/3d/help: added translucent-edge-glitch (edge-marking plus anti-aliasing)
- nds/backup: supports re-detection (games with faulty initial initialization)
- nds/backup: added bus-width auto detection (redirecting to new general types)
- nds/backup: added new types (3x general types) (and 1x sanyo, thanks flubba)
- nds/3d: soft-detail: edge-mark: handles edges at screen border (via clear_id)
- nds/3d: soft-detail: edge-mark: applies edges in respect to surrounding pixels
- nds/3d: soft-detail: edge-mark: internally stores edge_flag for possible edges
- nds/debug: allows some games to initialize not-existing port 4001004h to zero
- nds/debug: allows nintendo to use faulty ldmib with base-inclusion-writeback
- nds/debug: allows nintendo to use invalid stmib/ldmib user bank writeback
- nds/debug: allows nintendo to write more serious nonsense to (R) baseband regs
- nds/timings: re-fixed arm7/arm9 sync (new arm9-66MHz timings vs arm7-33MHz)
- nds/video: supports capture from 2d/3d engine in vram display mode (nanostray)
- nds/help: added note on undoc nds7 port 4001080h (used by ds-lite firmware)
- nds/help: added optical mouse sensor (slider controller) (thanx daniel palmer)
- nds/help: added firmware wifi internet access point settings info (thanks cue)
- nds/help: added user settings 066h (year) and 075h (ext language) (thanks cue)
- screenshot: converts 32bpp images to 24bpp (smaller and more standard files)
- nds/3d: soft-detail: prevents rendering of translucent polys with same poly_id
- nds/3d: soft-detail: supports shadow polygons (mask/render, step 1 and step 2)
- nds/3d: soft-speedup: scaled perspective correct clp.x from 0..len to 0..7FFFh
- nds/3d: soft-speedup: mmx: faster texcoord_clipping (no_repeat,repeat,flipped)
- nds/3d: soft-speedup: pre-calc tex_clip proc, collapsed 32:32 tex_xy to 16:16
- nds/3d: soft-speedup: pre-explodes edge_color_table, optimized alpha blend
- nds/3d: soft-speedup: mmx: processes two rgba-pairs and tex_xy-pairs at once
- nds/3d: soft-speedup: mmx: nonlinear color_rgb and texcoord_xy interpolation
- nds/3d: soft-detail: stores fog bit in framebuffer (opaque/trans=replace/and)
- nds/3d: soft-speedup: mmx: linear color_rgb and texcoord_xy interpolation
- nds/3d: soft-speedup: merged texture addressing and blending into single proc
- nds/3d: soft-speedup: mmx: faster texture blending (modulate/toon/highlight)
- nds/3d: soft-speedup: pre-calculates soft3d_tex_blend_proc for blendtype/mmx
- nds/3d: soft-speedup: uses mmx (if present) (otherwise stays 80386 compatible)
- nds/3d: soft-speedup: collapsed scanline_rgba from 32:32:32:32 to 8:8:8:8 bits
- detect: added no$gmb-386/486/cpuid detection, internal: rdtsc (3d/re selftest)
……………………………………………………………………………………………………………………………………
………………………………………………………………………………………………
………………

一并附上改王殿前回的测评帖供大家参考对比…………………………………………………………………………
「NO$GBA 2.6」————http://www.acgwolf.com/discuz/viewthread.php?tid=31964&extra=page%3D5
「NO$GBA 2.5C」————http://www.acgwolf.com/discuz/viewthread.php?tid=31675&extra=page%3D4
………………………………………………………………………………………………………………
……………………………………………………………………………………

[ 本帖最后由 kanon-air 于 2008-1-29 09:21 PM 编辑 ]

DQ4模拟基本完美
改进了3D Renderer 里面 NoCrash 模式的贴图错误

个人觉得其实 2.6和2.6A在速度上差别不是太大.
2.6A只是在3D和存档设置上做了优化

期待模拟器的进一步更新，我的老电脑，目前能基本完美速度运行的有莱顿教授系列，逆转裁判4，七龙珠
  DQ4很卡
  呵呵，总的说超出我的预想了

我的模拟起来还是异常的卡````
用2.6的时候都比现在好很多``
LS的兄弟是否还需要设置什么?

[ 本帖最后由 云淡风来 于 2008-1-30 01:05 PM 编辑 ]

离完美还差点～在城镇上还是会卡～

模拟器的问题建议去这个站看看.很不错的!
http://www.wiindsbbs.com/forum-28796-1.html

3.0啊3.0，等3.0

啊，那个其实就我个人的设置其实也没动什么
一个是Emulation Speed，LCD Refresh下选择Reatime,auto
        另，也可以试试Reatime*2,auto   速度会快些，不过声音就很卡
再一个就是Sound Output Mode下我一般选择Digital Mono
如果你的电脑真的还很卡的话就只能再试试把声音设定成none了

呵呵，希望能帮到你

呵呵，有个叫紧急出口的游戏，我的电脑运行起来，菜单卡点，游戏中很流畅
跟我一样老电脑的朋友们可以找来试试，游戏也蛮有意思的

谢谢~~~
我试了试,速度果真提升了不少
特别是我把防火墙的OOXX全关掉后,已经勉强可以算是流畅了

总体觉得ds模拟器很占CPU.
紧急出口那个游戏.我的电脑配置100%完美运行.
觉得在很不错!

模拟器主要是靠CPU、其次才是显示卡和物理内存、哪怕配置较高的机子、在运行诸如SS、DC、NDS、甚至是PS2等模拟器时、也最好把某些下载工具、聊天工具、乃至杀毒工具先暂时停止为好、让CPU的资源充分体现在模拟工具上………………………………
………………………………………………………………………………………………………………………………
………………

PS2模拟器，哼哼哼，没有Q66的话跟幻灯片是一个等级的，用6320 CPU被PS2模拟器蹂躏的路过。。。

简单测试了一下，苍月的十字架更快了已接近100%了，FF3的速度也大幅提升，但是以前就比较完美的重装机兵却没啥长进，速度和以前一样。水晶编年史的第二关以后的速度依然75%左右。
对某些等3.0的同志，可以考虑玩一下新生的逆转和逆转4，这两个我完美通关了。
PS：看到kanonMM的头像很感动，对晶莹之露有爱的人终于出现了！！！

的确呢、这次对速度的提升感觉是非普遍性的、有些游戏速度又提高了许多、而也有些并没什么改善、不过、仍然感觉NDS模拟器的发展脚步真是很快呢、照这样下去、达到完美应该是不久的事吧……………………………………………………

回PS：看来改王殿也是晶莹之露的爱好者呢、握手握手！…………说真的、S社的ARPG中、晶莹之露绝对是个优秀的作品、很喜欢那种世界观架构、偶后来玩F社的双星和咕噜甚至曾怀疑F社是不是参考过这个呢…………不过、是一个被遗忘被忽视的经典啊、相对同社的圣剑、武藏、放浪等等、无论是知名度还是玩者数量都很少的说、很可惜…………不过偶对晶莹的喜爱绝不亚于那些、前不久又回顾了一遍、遂在某同人作中找了偶最喜爱的玛雅的头像作了论坛头像、嘿嘿………………………………………………………………
…………………………………………………………………………………………………………………………
…………

握手，晶莹之露这个游戏的剧情感觉和一些一流的RPG都毫不逊色，游戏的不输给ACT的手感，很少有人玩过实在太可惜了。敏特和玛雅的性格塑造都十分成功，敏特精灵搞怪，经常会做一些啼笑皆非的事情。至于玛雅印象最深的就是最后，为了救敏特魔法书失灵了，不肯抛弃这个死敌妹妹，还拼命的用脚踩书并说是跟妹妹学的，又感人又好笑，感觉她虽然外表端庄，但本质应该和妹妹差不多。。。。。。。。