IRIX 6.4 Development Libraries

home *** CD-ROM | disk | FTP | other *** search

/ IRIX 6.4 Development Libraries / SGI IRIX 6.4 Development Libraries.iso / docs6.4 / relnotes / gl_dev / ch4.z / ch4

Wrap

Text File | 1997-09-04 | 35.7 KB | 1,321 lines

- 1 - 4. _B_u_g__F_i_x_e_s The IRIX 6.4 release includes the functionality of IRIX 6.2 and incorporates the bug fixes and functionality of Patch 1355. 4.1 _N_e_w__I_n_f_i_n_i_t_e_R_e_a_l_i_t_y__F_e_a_t_u_r_e_s__i_n__P_a_t_c_h__S_G_0_0_0_1_3_5_5 4.1.1 _P_i_p_e_l_i_n_e__I_n_s_t_r_u_m_e_n_t_a_t_i_o_n The SGIX_instruments OpenGL extension defines a new mechanism for measuring the performance of the graphics pipeline. It can be used to determine when an application is limited by pixel fill, geometry processing load, etc. This is helpful for general performance tuning and for maintaining a guaranteed frame rate in simulation systems. 4.1.2 _F_o_r_c_i_n_g__C_o_m_p_l_e_t_i_o_n__o_f__R_a_s_t_e_r_i_z_a_t_i_o_n The SGIX_flush_raster OpenGL extension forces all rasterization operations to be completed before processing the next OpenGL command. Unlike glFinish(), it does not prevent the application from issuing new commands. This is used in conjunction with the SGIX_instruments extension to ensure that rasterization is complete before taking a measurement of the graphics pipeline. 4.1.3 _D_i_s_p_l_a_y_-_L_i_s_t__M_e_m_o_r_y__M_a_n_a_g_e_m_e_n_t One of the new hardware features of InfiniteReality is a display-list cache memory. Display lists may be transferred from this memory at roughly twice the maximum speed possible for display lists stored in main memory. The SGIX_list_priority OpenGL extension allows applications to manage the contents of the display list cache by setting residence priorities for display lists. Note that this interface is experimental and behaviour is subject to change. We are very interested in feedback wrt to its usefulness and problems. For now, applications are limited to 128M of DMA-able display lists. Beyond that, the code falls back to push- model. 4.1.4 _C_a_l_l_i_g_r_a_p_h_i_c__L_i_g_h_t__S_u_p_p_o_r_t The SGIX_calligraphic_fragment OpenGL extension allows position and coverage information for light points to be transmitted to a combination calligraphic/raster display system. This - 2 - is valuable for night-time flight simulation. 4.1.5 _L_a_r_g_e__C_o_l_o_r__T_a_b_l_e_s Luminance-format color tables may now have up to 32K entries. Non-luminance color tables may have up to 16K entries. (Performance is maximized when color tables have 4K or fewer entries, however.) 4.1.6 _V_i_r_t_u_a_l__C_l_i_p_m_a_p_s Clipmaps are an extension of mipmaps, intended to handle texture mapping for extremely large textures (such as high-resolution satellite photographs of the entire Earth). The first release of clipmaps was limited to 15 levels (implying a maximum texture size of 32Kx32K). This has been changed to allow a much larger number of levels, provided that no more than 15 adjacent levels are resident in texture memory at any one time. Even now though, there remains a limitation that over- subscribing texture memory with 2 17 level clipmaps will fail. For now, only allocate clipmaps that fit into your physical texture memory. 4.1.7 _V_i_d_e_o__P_a_n_/_Z_o_o_m libXvc now supports the ability to pan over a framebuffer area larger than the display, as well as to zoom the display up or down without re-rendering. 4.1.8 _O_l_d_-_S_t_y_l_e__S_t_e_r_e_o The first InfiniteReality release included support only for stereo-in-a-window (``new-style'' stereo). The second release also supports old-style stereo, in which the screen is split into two parts and each part is scaled up by a factor of two. 4.1.9 _D_e_p_t_h__T_e_x_t_u_r_e_s The SGIX_depth_texture extension defines the behavior of depth textures (analogous to color textures). Currently depth textures are used for real-time shadows. 4.1.10 _S_y_n_c_h_r_o_n_i_z_e_d__B_u_f_f_e_r__S_w_a_p The SGIX_swap_barrier extension allows buffer swaps to be synchronized with an external event. Normally this is used to coordinate swapping among several machines, each of which is responsible for a portion of a video wall or other sophisticated multiple-display system. - 3 - 4.1.11 _S_w_a_p__G_r_o_u_p_s The SGIX_swap_group OpenGL extension provides the ability to synchronize the buffer swaps of a group of windows. For example, a double-buffered main window and its associated double-buffered overlay window can be placed in a swap group so that they will always be buffer-swapped together. 4.1.12 _N_e_w _I_n_f_i_n_i_t_e_R_e_a_l_i_t_y _P_e_r_f_o_r_m_a_n_c_e _I_m_p_r_o_v_e_m_e_n_t_s _i_n _P_a_t_c_h _S_G_0_0_0_1_3_5_5 4.1.13 _D_i_s_p_l_a_y__L_i_s_t_s Display lists can now be transferred from main memory to the graphics pipeline by DMA. This is substantially faster than immediate mode, though not as fast as display lists in the display list cache memory. 4.1.14 _P_a_c_k_e_d__V_e_r_t_e_x__A_r_r_a_y__F_o_r_m_a_t_s Support for certain vertex array formats has been optimized with special microcode. See ``man glvertexpointerext'' for details. 4.1.15 _I_P_2_5__(_R_1_0_0_0_0_)__I_m_m_e_d_i_a_t_e__M_o_d_e The immediate-mode data transfer code for IP25 is now significantly faster. 4.1.16 _B_i_t_m_a_p_s__(_T_e_x_t_) Drawing display-listed bitmaps (OpenGL-based text) is dramatically faster. 4.1.17 _H_i_s_t_o_g_r_a_m_s Single-component histograms (a common case, for luminance-only image processing operations) and the glGetHistogramEXT() command have been tuned. 4.1.18 _S_m_a_l_l_-_A_r_e_a__P_i_x_e_l__O_p_e_r_a_t_i_o_n_s The overhead for glCopyPixels() and glDrawPixels() operations has been reduced, allowing significantly better throughput for those operations when applied to small pixel arrays. 4.1.19 _M_o_d_e__C_h_a_n_g_e_s Performance for a variety of mode- change operations has been improved. 4.1.20 _C_u_l_l_i_n_g Backface culling performance has been improved, though the penalty for using culling remains higher than it was on RealityEngine. - 4 - 4.1.21 _E_v_a_l_u_a_t_o_r_s A number of optimizations have been applied to parametric polynomial surfaces (evaluators). 4.1.22 _C_o_n_v_o_l_u_t_i_o_n Performance has been improved substantially, especially for separable convolutions and for luminance-only convolutions. 4.1.23 _T_e_x_t_u_r_e__B_i_n_d_s Binding to a texture that is resident in texture memory is now over three times faster than it was in the first release. 4.1.24 _Q_u_a_d_r_i_l_a_t_e_r_a_l_s Decomposition of quadrilaterals into triangles has been changed, yielding 30%-50% better performance for geometry-limited quadrilateral strips. 4.1.25 _I_n_f_i_n_i_t_e_R_e_a_l_i_t_y__B_u_g__F_i_x_e_s__i_n__P_a_t_c_h__S_G_0_0_0_1_3_5_5 There are a huge number of bug fixes in the second release; over 300 in all. Many are minor and won't be discussed. Others will be grouped by subject. The first batch comes from earlier patches that occurred after 6.2 was released (patch1197, patch1282, patch1301, and patch1413). PV359464 - blank video in IR textport This patch contains a fix for the InfiniteReality textport that would cause it to come up with blank video on production hardware. The bug is not fatal, but since the user can't see what's going on, it makes the system very difficult to use until the X server starts up. Because this fix was applied to all machines before they shipped, you should only need this patch if you're forced to re-install 6.2 eoe from scratch. The patch re-flashes an EEPROM on the InfiniteReality board that is used to display the textport. PV363854 - certain DG4 boards' DACs do not get initialized correctly When a machine is power-cycled while warm, a small percentage of InfiniteReality DG4 boards may not initialize the DACs correctly, so video comes up scrambled. The change is to both textport and operational ucode. PV392758 - Sometimes managed area did not match video area When 1024x768 formats were created, the managed area ended up being 1120x768 because that was how much memory was being allocated. For application (and cursor tracking) convenience, this has been fixed so the managed area - 5 - reported by Xsgi as requested. PV361565 - Sirius may not be detected (by ircombine and others) the above fix also fixes this The patch re-flashes an EEPROM on the InfiniteReality board that is used to display the textport. The version number of the textport.bin image is 104. PV356365 - InfiniteReality w/ Sirius suffers from field- dominance confusion The following worked in 6.2MR field 2 dominance in NTSC field 1 dominance in PAL Now all variations work. PV364443 - glMaterial() change outside of bgn/end, between small geometry may have caused a pipe hang. PV363681 - The quality of the multisample transparency mask was not as good as it could have been. This gets it up to 32 in the 8 multisample case. PV373067 - InfiniteReality graphics may hang during shutdown/reboot. Fixed. There is a rare graphics hang caused by strided pixel DMAs that is fixed in patch 1355. In relation to this, some of the DMA vs. PIO tuning constants have been updated so performance is maintained or increased in most cases. PV358206 - 1d convolution hangs graphics PV358926 - ilSharpenImg default kernel hangs graphics PV365808 - switching between convolution filters crashes graphics PV353894 - Texture load with convolve enable does not work, disabled, but hangs PV363481 - Redundant glXBindChannelToWindow calls results in X errors PV388118 - ircombine won't load 640x480_180q field sequential PV353401 - setmon doesn't work on InfiniteReality unless X is running - 6 - PV357335 - setmon -p <pipe> does not work PV357427 - libvidfmt validator doesn't get hw config for 2nd pipe of two pipe system PV354431 - gfxinfo does not display video info correctly PV330865 - video - 10% of time, video comes up jittering/bouncing PV375120 - IR stereo is field-swapped, cannot genlock to RE2 serrated stereo PVxxxxxx - IR stereo doesn't support newer, cheaper Crystal Eyes glasses Stereo format timing has changed in order to eliminate dark banding associated with newer Crystal Eyes stereo glasses. If you use stereo formats with projection systems, be aware that recalibration will be necessary. PV375675 - H/C sync output select doesn't work PV400363 - Light position and spot direction aren't pushed/popped correctly PV363681 - InfiniteReality sample_mask implementation only has 8 levels of transparency PV372917 - min/max blend function broken on IR PV373220 - Push/Pop attrib of blend equation is broken PV395994 - Push/Pop attrib of texgen broken PV399151 - Texture scale and bias are not set on texture object bind PV356357 - Non-2dMipmap texture binds not setting all necessary shadow parameters Various obscure graphics context switching bugs have been fixed. 4.1.26 _T_e_x_t_u_r_e__M_e_m_o_r_y__M_a_n_a_g_e_r__b_u_g__f_i_x_e_s One of the biggest areas of bug fixing was in the area of texture memory management. Here is a summary: General usage issues for best performance and reliability: - 7 - +o set all texture parameters before downloading a texture (especially the minification filter) +o enable texturing before downloading any textures +o for best performance use subtexture loads to manage an application's texture usage. These problems with texture corruption or failure have been fixed since 6.2 was released: +o If texturing is disabled, texture memory is full, and the application tries to download a new texture. we will not force a new texture down to texture memory (no textures will be swapped out). unfortunately, we may corrupt the new texture in trying to save it on the host for later use. PV #355810 +o If the application has unique unpack pixelstore modes set. the local manager swapin uses the application unpack modes. PV #372831 (Krauss-Maffei) +o If a texture as an alignment less than 4 and the local manager swaps it in and out. PV #357038 +o If the texture sizes in an application are not relatively balanced and we must do a significant amount of swapping. PV #381401 +o If a texture is very wide and short (either 1d textures or short 2d textures around 4K or wider). the fix is to use a 2d texture at least 8 texels high. in addition, if the number of bytes per row is greater than 64K, the texture will be corrupted. PV #363351 +o If a texture mode (1d/2d/3d) is disabled and the application assumes we are still enabled for a lower dimension. any disable may disable all texturing, but opengl allows us to enable/disable any combination. the fix is to enable only the desired mode after disabling any other modes. PV #355811 - 8 - +o If a 3d mipmap has an odd number of levels. the last level address is computed incorrectly, and the texture will be downloaded over good data. the fix is to use only an even number of levels in a 3d mipmap PV #372566 +o If texture attributes are pushed and popped when doing 3d texturing. we force the texture to 2d. PV #375894 +o If the application sets the min lod to a non-zero value and requires a readback (either to swap local memory or getteximage). PV #383165 +o If the application does a texture draw after a getteximage of a non- zero lod. the fix is to do a glbindtextureext after the readback. PV #383163 +o If the application tries to share filter4 textures between contexts. PV #371032 +o when running with multiple contexts that either collide in texture memory or overrun the total texture memory. corruption may occur occasionally. to minimize any problem, an application (or user) should ensure that a given context is stable before starting another context. PV #330717 #357006 (AT&T off.C) +o If we need to swap in a context that uses exactly 64MB of texture memory. PV #383350 +o 3d proxy textures do not work PV #378867 +o 3d sharpen does not work PV #356879 +o changing texture priorities on the fly may fail PV #384524 (Infobyte) +o If the system runs out of swap memory and a malloc fails in the texture manager, we could crash the pipe (dma with null pointer) PV #381089 (Schlumberger -- run out of texture memory, crash) 4.1.27 _K_n_o_w_n__P_r_o_b_l_e_m_s - 9 - 4.1.28 _C_a_v_e_a_t_:__T_e_x_t_u_r_e__B_o_r_d_e_r_s We had hoped to support texture borders in this release, but the problems involved appear to be intractable. Therefore we recommend using other techniques. (A common solution is to clamp to the texture edge, rather than to the border.) 4.1.29 _S_t_e_r_e_o__v_i_d_e_o__f_o_r_m_a_t_s__c_h_a_n_g_e_d Stereo format timing has changed in order to eliminate dark banding associated with newer Crystal Eyes stereo glasses. If you use stereo formats with projection systems, be aware that recalibration will be necessary. 4.1.30 _C_a_v_e_a_t_:__R_e_m_a_i_n_i_n_g__k_n_o_w_n__b_u_g_s Note, these are not regressions from 6.2, but were either only very recently discovered or won't be fixed. +o PV #390082 In certain rare cases, the graphics will hang and reset when multiple texturing programs are run simultaneously. If they context switch during definition/downloading of textures and the phase of the moon is correct, a deadlock will occur. You can recognize this case if the SYSLOG contains many messages from kpm similar to: "(TM hosed) RM0 tm0 not ready for TG load input" +o PV #401322 When a accumulation-buffered GL window is partially covered and rendering at high speed, slight visual artifacts (a single line and maybe some sparkling pixels) will be seen. 4.2 _B_u_g__F_i_x_e_s__B_e_t_w_e_e_n__I_R_I_X__5_._3__a_n_d__6_._2 4.2.1 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X +o Linestipple context switch can fail. (140861) +o OpenGL Drawpixel, stencil format and negative zoom do not work. (195835) +o PolygonOffset doesn't support bias term. (237150) +o OpenGL needs display list code for Sharpen and Detail. (240559) - 10 - +o XCopyArea fails. (242766) +o OpenGL context destroy/recreate problems. (242847) +o Overlay planes were not being cleared on system startup. (243299) +o Read of null ALPHA buffer not functioning. (244161) +o glDrawPixel with multisampled visuals not functioning. (245384) +o Detail texturing breaks in some cases and interferes with 2D. (245538) +o glTexParameter* routines fail in dlist mode for DETAIL pnames. (253982) +o OpenGL GLX extensions need to be documented better. (254680) +o stopgfx and startgfx zero /etc/inittab if root is full. (255197) +o Polygon offset enable does not get pushed/popped with GL_POLYGON_BIT. (255266) +o OpenGL Histogram & Minmax have wrong default values. (256562) +o DrawPixels(...,STENCIL_INDEX,..) with negative X zoom hangs graphics. (259550) +o OpenGL convolution filter values are clamped to 0..1. (260661) +o Histogram and minmax don't work on read of rgba from index visual. (260999) +o Color values need to be clamped before histogram and minmax. (261001) +o glPixelZoom can causes skipped pixels in rendering. (263195) +o EXT_abgr mixes up components when intlfmt=RGB. (263461) +o OpenGL INTENSITY* formats treated just like LUMINANCE. (263708) - 11 - +o Zoom function bugs. (264568) +o OpenGL overlapped pixel copies do not work in all cases.. (265484) +o Fractional zoom bug on latest salpha. (266099) +o OpenGL low res pixel read back of signed LUMINANCE_ALPHA broken. (266100) +o Odd size glTexSubImage2DEXT can halt graphics pipe. (270303) +o OpenGL triangle fan/material change bug in IRIX 5.3 + Patch 154. (286122) +o Xor'ed X rendering to OpenGL double buffered window unreliable. (293933) +o ReadLUMINANCE_ALPHA can hang. (308184) +o glXMakeCurrent fpes. (334231) +o Picking namestack problems. (280928) +o OpenGL applying glScale to texture matrix jumbles texels. (271551) +o OpenGL blending bug when the glColorMask is on for more than one color. (272593) +o OpenGL zmax pick code doesn't work. (278560) +o The SETMON(1G) man page doesn't specify where the setmon command lives. (287865) +o OpenGL does not properly register BeginMode. (333997) 4.2.2 _M_a_x__I_m_p_a_c_t_,__H_i_g_h_t__I_m_p_a_c_t_,__a_n_d__S_o_l_i_d__I_m_p_a_c_t +o fast useracc for improved dma performance +o fixes for opengl dispatch table bugs +o fixes for stereo +o useracc fixes +o 3D textures in display lists may be incorrect (351008) - 12 - +o various pixel path bug fixes +o fix for hinv on solid/solid dual head +o fix for lines on overlay depth tested (336292) +o fix for impact copypixels ignores scissor (318279) +o USE_NEW_PIPE works +o fixes for more tex bugs - hansen +o fix for pbuffer ctx sw timeout +o texmgr bug fixes (341965,340098,336615,332498) +o mg10 and soft path for wide AA lines/polys +o glPixelZoom crashes system (334835) +o Subtex loads slow over time (337357) +o XOR ovly drawing fix (324853,339596,331375,343714) +o video texture support (342875) +o libsk fixes for banyan merge & presenter (340294) +o X/kernel changes for presenter (341020) +o proxy texture calls in display lists (342365) +o Two-sided lighting (339222) +o ogtst nextgen selectbuf failed (330422) +o fog problems (335059) +o GL_SGI_texture_color_table not advertised (336651) +o Evaluator fixes (no bug #) +o some pixel texture modes not working (310291) +o fast copypixels w/ new pixel pipe (344646) +o LUT optimization regression in 6.2 (345063) +o color index bitmap glDrawPixels hangs Xsgi (345148) - 13 - +o irix panics caused by screen savers (328927,328929,332671,332683) +o Presenter probe code screws up video (345027) +o hq3 idle wait timeout (327035) +o bfifo overflow (339247) +o characters are not fogged (344399) +o pullable BindTexture (343055) +o n32 opengl (345195) +o GRAPHICS CRASH in certain impact (345349) +o hang in makecurentread (344554) +o Multiple simultaneous pbuffer visuals (341938) +o performance issues (packed pix) (332436) +o New timing tables (incl 1600x1200) (346980) +o image format conversion bug (347124) +o glPixelZoom broken for fractional val > 1 (342034) +o gtstf -t image causes a CTXSW timeout (346093) +o N32 GL busted (347379) +o overlay crud fix (324853) +o segv in glMgrClipBox (342034) +o CADCENTRE replicated texture smears (330726) +o Using fast_useracc() causes munpin failures (347731) +o fbcheck fix (347957) +o makecurrent trashes viewport/proj matrix (348718) +o texcolortable ignored (343545) +o X server loads presenter table on video bd (343819) +o mg10 diags (347955) - 14 - +o monpanel changes (348702) +o MakeCurrentRead can't copy 12,12,12->8,8,8,8 (341773) +o 5x5 convole fails in chainp (333475) +o ilchain shows garbage (298333) +o Certain Impact pixel garbage on left side (350281) +o GL ReadPixels range checks (349177) +o Pbuffers visual artifacts (341788) +o presenter probe gives bad result sometimes (353549) +o convolve + pixel zoom draws incorrectly, then hangs system (345903) partial fix +o Single buffered video textures hang system (351384) +o bus error or panic just after CTXSW timeout (344908) +o Vulcan death grip causes panic (348467) +o gfxinfo reports bogus chip revisions (326155) +o draw/readpixels gets bad data for luminance visuals (354158) +o sub tex / copy tex / copy sub tex bugs (338490) +o ISG: texture loads don't use pixel tiles (353107) +o volren: texture restore speed not as fast as forced download (354139) +o loading paged RGBA texture with color table enabled can fail (354869) +o decouple swap from vertical retrace (343783) +o glReadPixels returns corrupt data in GL_INT mode (352836) +o 1024x768_pbuf timing tables w/o db32 added (344554) +o ctxsw crash during dma of depth/stencil (350308) +o pbuffer zbuffer causes scrn corruption on max 1024x768 (351168) - 15 - +o kernel panic w/color matrix (351525) +o heap corruption following get histogram (351377) +o pipe crash when redefining histogram (351368) +o imgview hangs (349211) +o pipe crashes during imaging operations (349177) +o Shared contexts confuse buffer swaps (350611) +o cfifo timeout moving opaque windows 0n 1024x768 screen (353372) +o kernel hang in ogtst stress_writepixel (353509) +o AA wide lines drop to 1 pixel width (341696) +o MakeCurrent with Null context is slow (334880) +o SDRC - CTXSW timeout takes user back to login screen(350398) +o System Hangs during XcopyArea (350209) +o MakeCurrent with Null context is slow (334880) 4.2.3 _O_t_h_e_r__H_i_g_h_-_E_n_d__C_o_n_f_i_g_u_r_a_t_i_o_n_s +o On Crimson LG1, glxinfo doesn't work. (295947) +o Lack of support for more than four IO4 boards. (317376) 4.2.4 _V_G_X__a_n_d__V_G_X_T +o Concave polygon drawn as convex. (243909) 4.3 _B_u_g__F_i_x_e_s__B_e_t_w_e_e_n__I_R_I_X__6_._1__a_n_d__6_._2 4.3.1 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X +o OpenGL texture performance bug, fixed by the texture_object extension. (253132) +o Multiple graphics contexts have performance problems. (256131) +o On PowerOnyx, bug in the O32 vertex array assembly code. (283018) - 16 - +o Color index clamping broken when drawing polygons, problem with double buffered overlay readback. (284245) +o Graphics may hang during "init". (290694) +o On PowerOnyx, GE10prom.u not installed. (303564) +o Gfx context switching degradation. (322614) +o Color matrix is not applied in single component read. (302450) +o Scale and bias API of OpenGL texture color table needs updating. (309742) +o MakeCurrentRead Accum buf may not use the proper color buffer. (313966) +o Pushattrib causes stereobuffer to fail on colorindex. (323684) 4.3.2 _O_t_h_e_r__H_i_g_h_-_E_n_d__C_o_n_f_i_g_u_r_a_t_i_o_n_s +o On Onyx Extreme, DANG error interrupt can occur. (301890) +o On PowerOnyx Extreme, lrectwrite w/ rectzoom y!=1 bug. (316482) 4.4 _B_u_g__F_i_x_e_s__B_e_t_w_e_e_n__I_R_I_X__6_._0_._1__a_n_d__6_._1 4.4.1 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X +o On PowerOnyx, OpenGL stereo-in-a-window can have visual problems. (252555) +o On PowerOnyx 90Mhz graphics performance degradation. (284209) +o On PowerOnyx, single processor can fail "init 0" from terminal with graphics running. (285223) 4.5 _B_u_g__F_i_x_e_s__B_e_t_w_e_e_n__I_R_I_X__5_._2__a_n_d__5_._3 4.5.1 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X +o Bad data on lrectread after switching output format. (203736). - 17 - +o gdiff fails: client on 5.2, display on 4.0.5. (203734) +o TX_SELECT broken for 3D textures. (212748). +o Kernel Panic after exiting multi-threaded application (211712) +o chost hogs unnecessary graphics resources preventing use of z buffer and aux buffers. (231777) +o readdisplay does not read overlay planes. (204945). +o photoshop 2.5 crashes graphics. (223409) +o Parts of a polygon will occasionally lose some of its texture. (221645) +o lmdef after lmbind with only new color causes lights to move. (225477). +o OpenGL push/pop GL_DRAW_BUFFER busted. (202414). +o FG_PIX_LIN fog sometimes wrong. (223323) +o Multiple pick/select hits not recorded unless name stack changes. (237549). +o polymode (PYM_HOLLOW) does not work. (238848). +o Race condition between swtch and rrmSuspend/rmResume causes kernel panic. (208514) +o Picking does not return all hits. (207529). +o On a multi-pipe RealityEngine2, kernel panics using vout on :0.1 from :0.0 with :0.1 as 1280x960_180qi and :0.0 as 1280x1024_60. (219702) +o POLYGON_SMOOTH enable is backwards. (240719) +o OpenGL default texture filters (min and max) are wrong. (243604) +o OpenGL sample select doesn't work correctly. (204636) +o Strange effects when using nurbs curves and fog. (207531) +o XDrawPoint(s) fails to draw when the window is partially obscured by another window. (209041) - 18 - +o scrnselect/winopen problems on multi-head. (214802) +o Push/Pop attributes has the side effect of turning off two-sided lighting. (225775) +o When GL_BLEND and GL_COLOR_MATERIAL are enabled, so is backface culling. (237877) +o OpenGL SHADE_MODEL is not pushed/popped. (241917) 4.5.2 _O_t_h_e_r__H_i_g_h_-_E_n_d__C_o_n_f_i_g_u_r_a_t_i_o_n_s +o On GT, snapshot confused about snap coordinates. (223210) 4.5.3 _X_S_,__X_Z_,__E_l_a_n_,__a_n_d__E_x_t_r_e_m_e +o A serious graphics problem in 5.2 reduces performance of some applications by up to 30%. (202239,219242) +o IrisGL getviewport performance extremely slow (30 times slower than it should be). (200478) +o Polygons with more than 4 vertices are drawn slower than in 4.0.5. +o glDrawPixels sometimes draws in wrong position. (208211) +o Severe performance slowdown on overlay drawing. (209449) +o On power-up, line stipple sometimes not initialized properly. (209715) +o Overlay planes not always cleared. (209783) +o Colormaps improperly overwritten. (209980) +o Textures don't draw in rightmost pixel of viewport. (210666) +o OpenGL color and depth clear is slower than IrisGL equivalents. (210874) +o IrisGL arcf sometimes draws incomplete arcs. (214681) +o Backfaced polygons of less than 1 pixel width sometimes fail to draw. (215735) - 19 - +o User defined clip planes generate bad polygons and can crash graphics. (216191) +o Blending and flat fill modes do not work correctly together. (217940, 218024, 225773) +o Specular highlights can occur when light normal is directly opposite the eye vector. (226049) +o Textures not sized to a power of 2 display inproper blending at edges. (226321) +o Problems with accumulation buffer rendering. (227985) +o IrisGL linewidthf fails to set linewidth to closest integer value. (230299) +o pixmode(PM_OFFSET,16) doesn't work with pixmode(PM_SIZE,16). Problem can cause incorrect drawing and graphics crashes. (234060) +o IrisGL setmon can crash system. (244248) 4.5.4 _I_n_d_y__a_n_d__I_n_d_i_g_o_2__X_L +o IrisGL gflush() could cause core dump. (238324) +o OpenGL accumulation buffer may not always work. (232024) +o IrisGL problem using afunction with texturing. (237905) +o OpenGL bug rendering tri strips to pixmaps. (235948) +o OpenGL glColorMaterial not always updated when glEnable(GL_COLOR_MATERIAL) called. (229948) +o IrisGL problem using accumulation buffer in mixed-mode (glx) environment. (225992) +o IrisGL had problems with wide, patterned lines. (225231) +o IrisGL underlays did not work (acted as overlays). (225003) +o IrisGL could core dump when rendering z-buffered anti- aliased lines (224031). +o OpenGL failed to update material property changes in a display list. (223591, 224151) - 20 - +o IrisGL fogging lines incorrectly when clipping planes used. (223419) +o IrisGL push/popattributes() wasn't restoring color writemask. (230393) +o IrisGL line patterns were sometimes incorrect. (225546)