Multimedia Performance Solved
On Tue, 2 Jan 2001, Philipp Schulte wrote: On Tue, Jan 02, 2001 at 12:16:52AM -0800, David Steinberg wrote: A few days ago, I installed potato, and things were again not so good. I applied the IDE patch and compiled a kernel, and things improved, but they're not as good as they were on Mandrake. I still get brief clicks in my audio... From all I have heard the missing pentium optimizations shouldn't make much difference. You should check if your HD is working in UDMA-mode $hdparm /dev/hda because this is what the patch is ment for. Hi all, I just thought I'd report my conclusion on this matter, in case anyone experiences something similar. I checked the disk access speed (using hdparm) under my potato installation and compared it to that of my Mandrake installation. Under potato, it was actually slightly better (improvements in the 2.2.18 kernel compared to the 2.2.16 kernel that I'm using in my Mandrake installation, perhaps?). So, I decided the problem was more likely in user space. I was going to use strace to figure out what libraries were being used and then compare the versions of mpg123 and any supporting libraries in the two installations. Well, the first thing I noticed was that libesd was being used in Mandrake, but not in debian. I had never even realized that that was an option of mpg123! Also, Mandrake was using mpg123 0.59r, while potato was using mpg123 0.59q (although, esd support has been in mpg123 since 0.59p). Well, I quickly looked at the mpg123 packages, and there was none in potato with esd support. Before compiling my own, I checked in woody, and, low and behold: mpg123-esd (version 0.59r, no less)! I installed that version, and audio is now smooth as a baby's bottom. I'm not exactly sure why it's so much better with esd...can anyone enlighten (pardon the pun) me? Is it a faster implimentation, or does it somehow demand higher priority? Anyways, needless to say, I'm a very happy camper. And I'm ready to switch over to debian full time and reclaim some hard drive space. :) -- David Steinberg [EMAIL PROTECTED]
Multimedia Performance
Hi all, I've got a question of performance that I was hoping some with more of a feel for these things could help with. Previously, I had been using Mandrake 7.1 on my desktop machine, an Athlon 600. At first, I had performance problems with certain multimedia applications: if I was playing an mp3 with mpg123, there would often be a click when an X client popped up a new window. Playing mpeg video wasn't so good, either. I applied Andre Hedrick's IDE patch and recompiled my kernel with support for the AMD Viper, and things really improved. All of the kinds of problems I just described completely disappeared. I could even play 3 or 4 separate MPEG videos with plaympeg without noticing a performance hit. A few days ago, I installed potato, and things were again not so good. I applied the IDE patch and compiled a kernel, and things improved, but they're not as good as they were on Mandrake. I still get brief clicks in my audio, and trying to play 3 videos at a time, they all drop frames and the system response slows to a crawl. Strangely, plaympeg won't play in full screen mode any more, either. Using xawtv to watch TV on my bt878-based TV card isn't quite as good, either. My question: could the difference be the pentium optimizations in the Mandrake binaries? If so, what might be the key pieces of software that I should recompile to get similar performance? If not, any ideas what the difference might be? Any help is sincerely apprecaited. -- David Steinberg [EMAIL PROTECTED]
Re: Multimedia Performance
On Tue, Jan 02, 2001 at 12:16:52AM -0800, David Steinberg wrote: A few days ago, I installed potato, and things were again not so good. I applied the IDE patch and compiled a kernel, and things improved, but they're not as good as they were on Mandrake. I still get brief clicks in my audio, and trying to play 3 videos at a time, they all drop frames and the system response slows to a crawl. Strangely, plaympeg won't play in full screen mode any more, either. Using xawtv to watch TV on my bt878-based TV card isn't quite as good, either. My question: could the difference be the pentium optimizations in the Mandrake binaries? If so, what might be the key pieces of software that I should recompile to get similar performance? If not, any ideas what the difference might be? From all I have heard the missing pentium optimizations shouldn't make much difference. You should check if your HD is working in UDMA-mode $hdparm /dev/hda because this is what the patch is ment for. I don't need the IDE-patch on my Athlon-system, I just do a hdparm -d1 -X68 /dev/hda in my boot scripts and it works. Phil
Re: Multimedia Performance
On Tue, 2 Jan 2001, Philipp Schulte wrote: From all I have heard the missing pentium optimizations shouldn't make much difference. You should check if your HD is working in UDMA-mode $hdparm /dev/hda Hi Phil, Thanks for replying. From hdparm... using_dma= 1 (on) because this is what the patch is ment for. I don't need the IDE-patch on my Athlon-system, I just do a hdparm -d1 -X68 /dev/hda in my boot scripts and it works. As I understand it, the Viper chipset isn't supported without the patch. I found that if I tried to do hdparm -d1 without it, nothing happened. Are you using that chipset, or some other disk controller? -- David Steinberg [EMAIL PROTECTED]
Re: Multimedia Performance
David Steinberg wrote: On Tue, 2 Jan 2001, Philipp Schulte wrote: From all I have heard the missing pentium optimizations shouldn't make much difference. You should check if your HD is working in UDMA-mode $hdparm /dev/hda Hi Phil, Thanks for replying. From hdparm... using_dma= 1 (on) because this is what the patch is ment for. I don't need the IDE-patch on my Athlon-system, I just do a hdparm -d1 -X68 /dev/hda in my boot scripts and it works. As I understand it, the Viper chipset isn't supported without the patch. I found that if I tried to do hdparm -d1 without it, nothing happened. Are you using that chipset, or some other disk controller? I've not seen the Viper explicitly supported except through Hedrick's ATA patch, but using the following: $ hdparm -c1 -d1 -X67 -k1 /dev/hdX Should set you into ATA33 mode, even without the patch. It's been a while since I've used a Viper board every day, but I can look into it a bit. Note that -d1 does not always set 32bit I/0, so forcing -c1 is a good idea, and -k1 is an even better idea, since IDE resets are common, and often go without notifying root or the message log. You should not be able to use -X68 without the patch, since the Viper was never meant to do ATA66 in the first place, and Andre's ATA66 patch is a hack that just /happens/ to work on most boards. -- -=|JP|=-Why, oh, why didn't I take the blue pill? Jon Pennington| Atipa Linux Solutions -o) [EMAIL PROTECTED] | http://www.atipa.com/\\ Kansas City, MO, USA | 816-595-3000 x1550 _\_V 6D04 39E0 CAE9 9ADA 2CA3 2EBE 898A 6C37 CA1E A29C
Re: Multimedia Performance
On Tue, Jan 02, 2001 at 09:42:09AM -0800, David Steinberg wrote: On Tue, 2 Jan 2001, Philipp Schulte wrote: From all I have heard the missing pentium optimizations shouldn't make much difference. You should check if your HD is working in UDMA-mode $hdparm /dev/hda Hi Phil, Thanks for replying. From hdparm... using_dma= 1 (on) Looks good. because this is what the patch is ment for. I don't need the IDE-patch on my Athlon-system, I just do a hdparm -d1 -X68 /dev/hda in my boot scripts and it works. As I understand it, the Viper chipset isn't supported without the patch. But what do you mean by not supported? Sure your devices are working so where is the problem? I found that if I tried to do hdparm -d1 without it, nothing happened. Are you using that chipset, or some other disk controller? No, I don't use any additional controller. phil:~% /sbin/lspci 00:00.0 Host bridge: Advanced Micro Devices [AMD]: Unknown device 7006 (rev 25) 00:01.0 PCI bridge: Advanced Micro Devices [AMD]: Unknown device 7007 (rev 01) 00:07.0 ISA bridge: Advanced Micro Devices [AMD]: Unknown device 7408 (rev 01) 00:07.1 IDE interface: Advanced Micro Devices [AMD]: Unknown device 7409 (rev 07) 00:07.3 Bridge: Advanced Micro Devices [AMD]: Unknown device 740b (rev 03) My board is a MSI-K7pro and it works great. Phil
Re: Multimedia Performance
On Tue, Jan 02, 2001 at 12:30:50PM -0600, Jon Pennington wrote: and often go without notifying root or the message log. You should not be able to use -X68 without the patch, since the Viper was never meant to do ATA66 in the first place, and Andre's ATA66 patch is a hack that just /happens/ to work on most boards. I am using ATA66 without any patches and it works. From $hdparm -i: UDMA modes: mode0 mode1 mode2 mode3 *mode4 Phil