
Sfor
MemberContent Type
Profiles
Forums
Events
Everything posted by Sfor
-
So far I have confirmed the official Nvidia driver packages 71.84, 77.72 and 81.98 do have support for PCI-E GeForce 6800, 6600 and 6200. The problem is, it is necesary to skip the installer and to select the card manualy, when istalling. NV40 GeForce 6800 - I was unable to test it NV43 GeForce 6600 - Works very well with 77.72 and 71.84. The 81.98 comes with shutdown problems. NV43 GeForce 6200 - The initial release of the 6200 came from degraded 6600 chips - untested but should work. NV44 GeForce 6500 - no entry in the inf section, support unknown NV44 GeForce 6200 - I'm not sure if all the PCI-E 6200 NV44 came with the Turbo Cache function. In any case these do not seem to be working correctly. NV44a, NV44a2 GeForce 6200 - I do not know if they are available in PCI-E versions. According to my research they should be just in AGP and PCI versions. I've been testing the cards on HP DC7100 with Intel 915 chipset. But, I did not test the 3D functions. I'm curious about: - Do the official Nvidia packages provide support for the NV40 and NV43 chips with PCI-E? Or perhaps it is possible to use the NV44 as well. - Are there NV44 PCI-E 6200 cards without Turbo Cache? - Are there NV44a and NV44a2 PCI-E 6200 cards? Informations gathered: - NVidia GeForce 6200 cards with Turbo Cache are not working correctly in Windows 98 with both official and unofficial drivers. - Nvidia official driver support ends at GeForce 6800. Compatible Hardware with Windows 9x - The GeForce 6800 NV40 chip is the oldest one from the GeForce 6 series. The newest is NV44 61xx,6200,6500. The NV43 6600 is clearly supported in the official Nvidia drivers. Wkipedia - GeForce 6 Series
-
Ooops. I forgot to translate the Autostart from Polish language version. It should be Startup, I think. Both the policy rule Computer Configuration\Administrative Templates\System\Logon\ Always wait for the network at computer startup and logon and registry setting HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon "WaitForNetwork"=dword:00000001 did not change a thing, when used alone. But, if both of them are set in the same time, the Windows XP does behave the same way as Windows 2000 does. Thank you for your help. The problem has been solved.
-
I wanted to run a batch script from the Autostrart. The goal was to copy some files from a server on each system boot. The computer is set, so no user login is necesary, so there is no delay for user input during the boot procedure. As the result the Microsoft Networking services are still not operational during the Autostart procedures. The batch script complains about "network patch not available" (translated from Polish). A moment later, the same script works perfecly. There is no such problem in Windows 2000. Apparently, all the network related functions are fully operational before the user logon procedures. So, how can I force Windows XP to do the same as Windows 2000 does? I tried the idea with MMC and logon policies, but it did not changed a thing. As a workaround I did loop the network patch checking in the batch. So, the files are copied after the patch becomes available. But, I would like to find a more elegant solution.
-
The HP D530 (Intel 865) and HP D7100 (Intel 915) are both equipped with integrated Broadcom NetXtreme Gigabit Ethernet adapters. For some reason the Windows 98 does not shut down properly when the Broadcom b57w2k.sys (10.24.0.0.D) driver is loaded. I was unable to solve this problem, so far. So, I'm using some different PCI ethernet adapters with the integrated Broadcom disabled, for the moment. Does anyone with a Broadcom NetLink gigabit adapter from the BCM57xx series, have shutdown problems, as well?
-
I do have a DDS4 tape drive. I find it convenient to store rarely necesary large file sets on a tapes rather, than on multiple DVDs. In any case the Windows 98 backup tool (Seagate Backup Exec) is able to handle tapes in a quite a covenient way. I tried the backup tool from Windows 2000, and: - it does not allow to name tapes, as user see fit. (98 lets to set a tape name the way similar to a disk label). - it is necesary to use every tape from a backup chain, when restoring a single backup job (98 lets to add backup after backup creating an endless chain of tapes. Only the tapes with needed backup job are necesary when restoring). - it is not possible to see what backup jobs are on a tape without inserting the tapes preeceeding it. (98 shows all the backup jobs with a warning if the jobs are spanning to another tapes). - the backups made on Windows 98 are restored withoul an error, but the file contents are messed up. So, I can not force myself to move to Windows 2000 with the tape related operations. But, perhaps there is some software that will make the Windows 2000 work as convenient as the 98 does. Still it is hardly the reason to pay for Backup Exec for Windows 2000, when the one given by Microsoft with Windows 98 works quite well for free. (I had no opportunity to test the newer versions of the Backup Exec, so I do not know if they are as convenient as the old one).
-
I found yet another difference between Windows 98 and Windows 2000. The tape backup software is significantly different. Backups made on Windows 98 with sytem backup tool (Seagate Backup Exec) are not restored properly with Windows 2000 backup tool (from Veritas). The funny thing is, the files are actualy restored without any error message, but the data is messed up. The tape handling is also significantly different. Windows 98 backup tool gives much more freedom with handling multiple backups on the same tapes. It is possible to create an endless chain of backups one after another, by just adding another empty tape when the free space ends. In order to restore it is necesary to use only the tapes containing the particular backup job. In Windows 2000 it is also possible to add backups one after another. The problem comes with restoring. It is necesary to use every tape from the beginning of the backup chain. So, the endless backup chain is not an option. Also it is not possible to freely name tapes as user wants. So, Windows 98 let's user to treat tapes in a way more similar to standard removable media drives, while Windows 2000 forces quite a few restrictions to the way the tapes are handled. I found no good freeware program able to handle tape drives in Windows 2000, so far. Turbo Tape works, but it's no good with handling large backup sets with more than one tape.
-
I did get a 3Ghz/512kb/800Mhz P4 CPU. It works a bit better than 2.8Ghz, but a quick calculation suggests 3.6GHz CPU is necesary for smooth playback of 1920x1080 video. I do have yet another problem. How to get a full playback of Advanced Substation Alpha subtitles with ability to time shift subtitles during playback? The GOM Player does not have the full AdvSubstation support, yet it does allow to time shift subtiles. On the other hand FFDShow does fully support AdvSubstation subtitles, but the built in keyboard command functions do not work for me. I do not know if the FFDShow keyboard shorcuts were ment to be available during playback, in the first place. Are there other video players able to do the time shift with the full support for Advanced Substation Alpha subtitles?
-
I'm using FFDShow rev2322_20081114 on Windows 98 without the KernelEX. The decoder configuration settings are available through the Start Menu or tray icon during playback. It appears the FFDShow is the fastest free h264 decoder available.
-
I did some further research with the FFDShow. It does support buffering frames in advance, but this option is switched off, by default. Also, it does not do much good, when CPU is too slow. When playing with FFDShow OSD filter, I found the most useful informations is CPU load and video stream delay. The OSD filter does increase the CPU load a bit, but it can give a very good understanding of what is going on. In my particular case the video stream delay is constantly rising. When it reaches a certain configurable amount (default is 1500ms) the decoder drops video frames and starts decoding from the current audio stream time. The video decoding frame drop point is configurable, but if it exceeds 5000ms the audio stream output starts to break (that's the standard effect for the GOM Player built in codec). So, the freezing effect I was referring to in earlier posts is a situation, when video delay exceeds the default 1500ms and video decoder drops frames. Since the video delay is constantly rising with 100% CPU usage, it is obvious, the CPU can not keep up. Also, buffering in advance is useless, since the CPU is always occupied in 100%. The FFDShow does have a few interesting features related to performance. - buffering frames in advance - skipping deblocking when video delay exceeds a certain point. - configurable video frame drop point, preventing audio output interruptions Currently I'm working with a 2.8/512/800 P4, but I will get a 3.0/512/800 CPU in a few days. According to my observations, the current CPU is about 15-30% too slow.
-
Well, the Partimage utility does copy only the occupied sectors of a partition, leaving the unused space behind. It also does a copy of the whole MBR and stores everything in a single compressed archive. The supported file systems are FAT, FAT32, NTFS and EXT up to version 3. EXT4 and newer file systems are not supported, as the application is no longer worked on. It is possible to restore - the whole MBR, - just the partition data of the MBR - just the boot strap code of the MBR - partition data without the MBR By my understanding, a backup is a copy made in oder to be restored, when something bad happens. I see no meaningful difference between backing up separate files, or the whole partition. The real problem comes, when it is necesary to restore a whole partition. So, in order to use a partition level system backup it is wise to keep all the relevant user data files on other partitions. So, the system restore procedure would not have to include prior data file backup. Following the idea, data file partition level backup does not seem to have much sense. Since the thread is about Windows 2000 and 98, there is one significant difference between them when it comes to system restore. I can restore a Windows 98 using a plain DOS, while I can not do the same with Windows 2000. Windows 98 kernel is stored in just 8.3 DOS format names. The LFN support comes just after the kernel start, but before the GUI start. So, it is possible to start kernel after restoring just files with short names, then restore the long file names (LFN), then to initialize GUI. Also, it is relatively easy to move Windows 98 to other computer with very different hardware.
-
Linux in any form will not copy any Windows specific attributes. In fact it can not copy Windows FAT32 file attributes, as well. So, it is not possible to do a perfect file system backup from Linux. The best way to use Linux is a full partition backup copy, indeed. Personaly, I do prefer to use the Partimage and SystemRescue CD.
-
Well. The good thing about having a dual boot Windows 98 and Windows 2000 setup is, it is possible to make a copy of Windows 2000 system files using Windows 98 and vice versa. So, having a full system backup copy, there is no need for system reinstalation or repair.
-
I did replace Ati Radeon 9000 AGPx4 with Nvidia GeForce FX 5500 AGPx8. There is a slight improvement in the performace, but it is not clearly visible. Does anyone know the link to the Beta CoreAVC 2.0.0.0? The current version is a commercial software, apparently? Or, perhaps, the beta is the commercial software, as well.
-
I've been playing with SMPlayer on Windows 98, some time ago. It worked, but I had no possibility to controll the playback with keyboard commands, then. On the other hand it works much better on Linux. I have a Ubuntu with SMPlayer installed on the same computer Windows 98 is installed on. The difference in speed between Ubutu + SMPlayer and Windows 98 + GOM Player is huge. To put it simply, Ubuntu + SMPlayer combo does not stand a chance when facing Windows 98 + GOM Player on a computer with a single core processor. I noticed my ATI All in wonder 9000 is able to use AGPx4 only. I'm curious if the performance will be better, if I replace it with AGPx8 capable GeForce FX 5500.
-
I did an experiment with a 1920x1080 video clip with a little over 23fps. To get the best out of the computer I disabled the GOM Player internal filters, so the only available performance data were from FFDshow OSD filter. Still, the speed was changing so fast, I could not read the result. The first digit was 1 or 2, the second 0 or 6. So, the fps readouts could be anything between 10, 16, 20 or 26. On scene changes the video was often freezing with 9 or 8 fps on diplay. While using GOM internal filters the speed was changing in 15-25fps range, with video freezing from time to time. The test was performed on Windows 98 with P4 2.8Ghz/800Mhz/512kb using DDR400 in qadruple channel and ATI All in Wonder 9000. The overal experience: It is usable, yet, the frame rate is visibly lower and video freezing on scene changes effect is quiite annoying. The sound was working perfectly.
-
I've been playing with older versions with VLC player, only. As far as I remember there were no functions necesary for me. I wanted to have audio video time shift correction, video size and position correction, an a few others. It was not efficient enough, as well.
-
Well, I'm aiming much above the DVD level. The Blu ray and HD DVD resolutions are gaining market, right now. I'm also using the DVI interface. To be more specific DVI is compatible with HDMI. The difference is, the HDMI can also transfer audio, while DVI can only transfer video data. Using DVI to HDMI cable it is possible to connect to any new TV. My 42" plasma is only 1024x768, so it is not a HD screen. But, the h264 1280x720 and 1920x1080 are more and more common. In order to play them back smoothly, significant processing power is necesary. I started from a 1GHz PIII CPU as well. I noticed, it plays video much more smoothly than the PIV does. But, at 1280x720 video data resolution it is not able to keep up. The video playback is getting choppy. It looks like frame rate gets reduced. It also depends on the software used. Either the video or audio smooth play has to be sacrificed if the processing power is insufficient. In case of the P4 CPUs it seems the high resolution data stream is played smoother. Instead of a general frame reduction either video or audio gets frozen for some time, when the processing power is insufficient. It appears, the P4 is significantly faster with multimedia processing than PIII, but in case of a processing power shortage it takes much more time for it to recover the smooth play. It mostly happens on a scene changes, when the amount of data to be processed is highest. The whole thing can be related to a fact the PIV CPUs have a faster memory access in general. In case of PIII the CPU can not play smoothly high resolution video, because the amount of data to be transferred does not go through system buses.
-
For quite a time I'm using Windows 98 SE as video player. The idea was to connect a 42" plasma TV to a computer in order to play video files. In either case, the most efficient software I've found so far. GOM Player - quite a nice video player with many advantages and a few problems. - does not support a** subtitles - does not resolve correctly time conflicts in subtitles - the efficiency of the built in h264 decoder and mkv and mp4 splitters are not that great + high quality subtitles + supports many video formats with built in codecs FFDShow - it covers for almost everything what GOM Player is lacking + does support a** + the built in h264 decoder is a fast one - the non a** subtitles are poor in quality - does not seem to be able to play subtitles directly from MKV files. It is necesary to extract them to separate files, first. (It is not a problem for me, since I'm using an additional tool for subtitles timing and font enhacement.) - there are rare problems with system resources in case of some a** subtitles. Haali Matroska Splitter - a fast MKV and MP4 splitter It requires some tweaking to get the best performance, but these were the best I found. I've tried a few operating systems in order to find the one fastest on a single core CPU. Windows 98 appeared to be the fastest one. Currently I'm using a Pentium 4 2,8/800 HT CPU with DDR400 memory in dual (or rather quadruple) channel. The power is more than enough for smooth play of 1280x720 h264 coded video, but is not enough for 1920x1080. So, the question is where to go from here. It is possible to aim for multi core CPUs, but it would require to change the OS to some Windows NT core system or Linux, probably. On the other hand I've heard something about DXVA being available on Windows 98. - According to my research the newer versions of FFDShow are able to use DXVA. The newest one comes in two different builds, one is Windows 2000 compatible, the other is not. Perhaps, it could be possible to port it to Windows 98 with KernelEX. - DXVA 1.0 was released for Windows 2000 and newer. I've heard some rumors about Windows 98 compatibility, however. The DXVA 2.0 seems to available from Vista up. - It could be difficult to find a DXVA compatible graphics card able to work in Windows 98. I'm curious, if anybody was able to get the full HD with Windows 98. In my case the the playback is not efficient, enough. -------- For GOD sake. Why am I not able to write Advanced Substation Alpha file format? It always ends up as a**. The HELL with the automatic word correction on this forum.
-
Well, the PersistBrowsers registry key does work. But, I've noticed significant differencies in this feature. In case of Windows 98 all restored Explorer windows had the default size. For some reason the windows restored by Windows 2000 are resized in most cases. To be more specific, the windows just opened and left without any action are restored as they were. On the other hand windows with the default directory changed are restored in different sizes.
-
I gave up on complicated solutions. So, I used an old 20GB HDD, installed Windows 2000 on it, turned the LBA48 registry support keys on, then transferred the system to the big hdd through simple file copy. There are two things unavailable in Windows 2000. - 98 does remember the opened Explorer windows between sessions. (useful sometimes) - 98 can remember the positions and sizes of console aplications, while 2000 does not. (very useful for me)
-
Windows 98 network HTTP connection speed.
Sfor replied to Sfor's topic in Windows 9x Member Projects
I just installed Windows 2000 on another computer. It was working with just Windows 98, so far. The Windows 2000 is still faster with downloads from some servers. The hardware is completely different from the previous one, so the speed difference must be comming from OS itself. -
This plan would work perfectly with Windows 98. But, in case of Windows 2000 it does not work at all. After the first reboot, the installer ends with a message "INACCESSIBLE_BOOT_DEVICE". So, I guess, there is no easy way. I will have to add the registry patch during the installation, somehow.
-
Windows 98 network HTTP connection speed.
Sfor replied to Sfor's topic in Windows 9x Member Projects
It's a plain ethernet. To be exact, it is a city wide network with fiber optic skeleton and 10/100 ethernet inside buildings. In any case, exactly the same syptoms are visible on a cable TV connection, as well. Still, there is ethernet between computer and modem. -
Windows 98 network HTTP connection speed.
Sfor replied to Sfor's topic in Windows 9x Member Projects
The article "Windows TCP/IP Registry Entries" from Microsoft says: Since in my system there is no MaxMTU registry key, the default should be 1500, unless the media driver takes it down. I do not know how to test the media driver value, however. -
Windows 98 network HTTP connection speed.
Sfor replied to Sfor's topic in Windows 9x Member Projects
As far as I understand the TCP, MTU seems to be related to the size of outgoing packets. Since, the issue is in slow receive, the MTU setting should not have any effect in this case. In other words, the MTU does not affect the size of incoming packets, so it should not affect download speed. I had an opportunity to make yet another experiment considering file transfer through Microsoft Networking using LAN. A 2.4GHz P4 Windows 98 and a 1GHz PIII Windows 2000 were used, both having the same amount of RAM (512Mb). While sending files from 98 to 2000 it went twice as fast as the other way around. So, Windows 2000 seems to be twice as fast when working as a server. So, the 98 seems to be able to send data fast, while being slow on receiving them. On the other hand, I did some experimenting with network efficiency in Windows 98. It seemed the fastest way is to use Windows 2000 as a server, while using NetBeui on Windows 98. A 98 NetBeui client sometimes worked faster than 2000 client. The 98 TCP client was slower than the 2000 one. It seemed the 2000 does not care if NetBeui or TCP is used, the result was exactly the same. The tests were done through the time measurements of a certain network database application tasks (not SQL). The application is sensitive to the workstation CPU related power. So, it is possible to conclude the TCP is more CPU demanding on Windows 98. But, it is just a guess, since the test was not designed to prove such a theory.