Re: [digitalradio] Comparing data modes
Dave, The question is whether any app is being starved for CPU resources during critical operations. Is there a way to tell if that is happening? Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Friday, October 03, 2008 10:19 PM Subject: RE: [digitalradio] Comparing data modes If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Friday, October 03, 2008 6:35 PM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Dave, when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. 73, Dave, AA6YQ For what it's worth, the CPU usage hovers around 30% while running 5 different digital mode programs and the HF path simulator. Throughput is the same for each program whether they are tested individually or all at once. Not sure if the performance monitor is telling the whole story when it comes to CPU usage, but I would assume the processor is loafing along at 30%. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 3:01 PM Subject: RE: [digitalradio] Comparing data modes Stelios's note raises another point independent of sampling rate issues: when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. Given all the unknowns, I wouldn't consider a comparison between two digital mode applications to be apples and apples unless the test was accomplished on two identically configured PCs, each processing the same data stream. If the data stream can be recorded and replayed, then of course the tests can be run serially on a single PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Stelios Bounanos Sent: Tuesday, September 30, 2008 9:03 AM To: digitalradio@yahoogroups.com Cc: Tony Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different to the current modem's, fldigi will need to resample. It will also resample, using the same configured conveter, if the TX or RX ppm corrections are nonzero. Of course it's smart enough to combine both kinds of resampling into one step. I don't know enough about Vista's audio system to be able to say what happens if you run multiple programs that want different rates. Possibly some of them will see slightly increased latency. Regarding CPU usage, you should make sure that you have enough CPU cycles
Re: [digitalradio] Comparing data modes
Dave, IMO any data program will be talking to the soundcard with a nice big FIFO so CPU shouldn't be an issue. From memory I use a 10 second FIFO on receive and 2 second on transmit. I can't think of any problems that could arise as long as there are some free CPU cycles. What am I missing? Simon Brown, HB9DRV www.ham-radio-deluxe.com - Original Message - From: Dave AA6YQ If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations.
Re: [digitalradio] Comparing data modes
each processor core can only execute one instruction at a time, for one application at a time. the data bus can handle either one (32 bit) or two (64 bit) operations at a time. that you can see 5 windows open at once is the magic of the video processor and persistence of vision. Brent Gourley Dothan [EMAIL PROTECTED] - Original Message - From: Simon Brown (KNS) To: digitalradio@yahoogroups.com Sent: Saturday, October 04, 2008 3:26 AM Subject: Re: [digitalradio] Comparing data modes Dave, IMO any data program will be talking to the soundcard with a nice big FIFO so CPU shouldn't be an issue. From memory I use a 10 second FIFO on receive and 2 second on transmit. I can't think of any problems that could arise as long as there are some free CPU cycles. What am I missing? Simon Brown, HB9DRV www.ham-radio-deluxe.com - Original Message - From: Dave AA6YQ If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations.
RE: [digitalradio] Comparing data modes
You'll need a more fine-grained indicator of CPU usage; try Googling. When you have multiple applications decoding the same incoming stream, peak CPU loads are likely to occur simultaneously. To assure an apples-and-apples comparison, run each application on a dedicated PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Saturday, October 04, 2008 3:11 AM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Dave, The question is whether any app is being starved for CPU resources during critical operations. Is there a way to tell if that is happening? Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Friday, October 03, 2008 10:19 PM Subject: RE: [digitalradio] Comparing data modes If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Friday, October 03, 2008 6:35 PM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Dave, when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. 73, Dave, AA6YQ For what it's worth, the CPU usage hovers around 30% while running 5 different digital mode programs and the HF path simulator. Throughput is the same for each program whether they are tested individually or all at once. Not sure if the performance monitor is telling the whole story when it comes to CPU usage, but I would assume the processor is loafing along at 30%. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 3:01 PM Subject: RE: [digitalradio] Comparing data modes Stelios's note raises another point independent of sampling rate issues: when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. Given all the unknowns, I wouldn't consider a comparison between two digital mode applications to be apples and apples unless the test was accomplished on two identically configured PCs, each processing the same data stream. If the data stream can be recorded and replayed, then of course the tests can be run serially on a single PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Stelios Bounanos Sent: Tuesday, September 30, 2008 9:03 AM To: digitalradio@yahoogroups.com Cc: Tony Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different
RE: [digitalradio] Comparing data modes
You're making assumptions about the internal structure of the application, Simon. If you're correct and all of the applications being measured are well-buffered, then fine. But its relatively easy to overrun the buffers in PSKCORE, for example; while it informs its client when that happens, the application is not forced to report this to the user. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Simon Brown (KNS) Sent: Saturday, October 04, 2008 4:26 AM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Dave, IMO any data program will be talking to the soundcard with a nice big FIFO so CPU shouldn't be an issue. From memory I use a 10 second FIFO on receive and 2 second on transmit. I can't think of any problems that could arise as long as there are some free CPU cycles. What am I missing? Simon Brown, HB9DRV www.ham-radio-deluxe.com - Original Message - From: Dave AA6YQ If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations.
Re: [digitalradio] Comparing data modes
Dave, when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. 73, Dave, AA6YQ For what it's worth, the CPU usage hovers around 30% while running 5 different digital mode programs and the HF path simulator. Throughput is the same for each program whether they are tested individually or all at once. Not sure if the performance monitor is telling the whole story when it comes to CPU usage, but I would assume the processor is loafing along at 30%. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 3:01 PM Subject: RE: [digitalradio] Comparing data modes Stelios's note raises another point independent of sampling rate issues: when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. Given all the unknowns, I wouldn't consider a comparison between two digital mode applications to be apples and apples unless the test was accomplished on two identically configured PCs, each processing the same data stream. If the data stream can be recorded and replayed, then of course the tests can be run serially on a single PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Stelios Bounanos Sent: Tuesday, September 30, 2008 9:03 AM To: digitalradio@yahoogroups.com Cc: Tony Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different to the current modem's, fldigi will need to resample. It will also resample, using the same configured conveter, if the TX or RX ppm corrections are nonzero. Of course it's smart enough to combine both kinds of resampling into one step. I don't know enough about Vista's audio system to be able to say what happens if you run multiple programs that want different rates. Possibly some of them will see slightly increased latency. Regarding CPU usage, you should make sure that you have enough CPU cycles for everything; maybe aim to keep the system load at 50%. Fldigi can tell you if it's dropping audio because the CPU is too busy but only in debug builds; perhaps something to change in a future version. 73, Stelios, M0GLD.
RE: [digitalradio] Comparing data modes
If you're referring to CPU usage as reported by the Windows Task Manager, that's an average. The question is whether any app is being starved for CPU resources during critical operations. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Friday, October 03, 2008 6:35 PM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Dave, when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. 73, Dave, AA6YQ For what it's worth, the CPU usage hovers around 30% while running 5 different digital mode programs and the HF path simulator. Throughput is the same for each program whether they are tested individually or all at once. Not sure if the performance monitor is telling the whole story when it comes to CPU usage, but I would assume the processor is loafing along at 30%. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 3:01 PM Subject: RE: [digitalradio] Comparing data modes Stelios's note raises another point independent of sampling rate issues: when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. Given all the unknowns, I wouldn't consider a comparison between two digital mode applications to be apples and apples unless the test was accomplished on two identically configured PCs, each processing the same data stream. If the data stream can be recorded and replayed, then of course the tests can be run serially on a single PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Stelios Bounanos Sent: Tuesday, September 30, 2008 9:03 AM To: digitalradio@yahoogroups.com Cc: Tony Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different to the current modem's, fldigi will need to resample. It will also resample, using the same configured conveter, if the TX or RX ppm corrections are nonzero. Of course it's smart enough to combine both kinds of resampling into one step. I don't know enough about Vista's audio system to be able to say what happens if you run multiple programs that want different rates. Possibly some of them will see slightly increased latency. Regarding CPU usage, you should make sure that you have enough CPU cycles for everything; maybe aim to keep the system load at 50%. Fldigi can tell you if it's dropping audio because the CPU is too busy but only in debug builds; perhaps something to change in a future version. 73, Stelios, M0GLD.
Re: [digitalradio] Comparing data modes
DM780 samples at 8kHz and has a lot of encoder / decoder code in common with fldigi. MT63 / Olivia having being taken from Pavel's original code. I'll try to find the Microsoft Blog entries which explain why some rates such as 11,025Hz are so poor with W2K / XP but not with VISTA. It's not a compromise - anyway with luck I'll find this explanation this morning. FWIW I generate 8kHz from 48kHz sampling using Moe's code. Simon Brown, HB9DRV www.ham-radio-deluxe.com - Original Message - From: Dave AA6YQ Also, there is no reason to believe that the two applications would be equally penalized by a compromise sampling rate -- so the results of the comparison would be suspect unless its known that both applications use the same sampling rate. -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Dave AA6YQ Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation.
Re: [digitalradio] Comparing data modes
I'm sure fldigi uses 8kHz in general and 11,025 when needed for some Thor / DominoEx modes. Simon Brown, HB9DRV www.ham-radio-deluxe.com - Original Message - The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version.
RE: [digitalradio] Comparing data modes
Stelios's note raises another point independent of sampling rate issues: when running more than one digital mode app on the same processor, one or both apps may be starved for CPU cycles and as a result underperform. Given all the unknowns, I wouldn't consider a comparison between two digital mode applications to be apples and apples unless the test was accomplished on two identically configured PCs, each processing the same data stream. If the data stream can be recorded and replayed, then of course the tests can be run serially on a single PC. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Stelios Bounanos Sent: Tuesday, September 30, 2008 9:03 AM To: digitalradio@yahoogroups.com Cc: Tony Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different to the current modem's, fldigi will need to resample. It will also resample, using the same configured conveter, if the TX or RX ppm corrections are nonzero. Of course it's smart enough to combine both kinds of resampling into one step. I don't know enough about Vista's audio system to be able to say what happens if you run multiple programs that want different rates. Possibly some of them will see slightly increased latency. Regarding CPU usage, you should make sure that you have enough CPU cycles for everything; maybe aim to keep the system load at 50%. Fldigi can tell you if it's dropping audio because the CPU is too busy but only in debug builds; perhaps something to change in a future version. 73, Stelios, M0GLD.
Re: [digitalradio] Comparing data modes
Hello Stelios, However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample But how can you know the default native sample rate? With Multidem, I can work in full duplex in any sampling speed from 8000 to 192000 Hz. I have noticed that you can open a 44.1 KHz at 48 KHz native sample rate sound card and reversely (the driver accepts all included 192 KHz). Of course, by working in full duplex, you rapidly see what is the true native sample rate (there are no clicks if you are on the good one and you can regulate the possible small shift between RX and TX sample rates). 73 Patrick - Original Message - From: Stelios Bounanos [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Cc: Tony [EMAIL PROTECTED] Sent: Tuesday, September 30, 2008 3:03 PM Subject: Re: [digitalradio] Comparing data modes On Tue, 30 Sep 2008 00:04:19 -0400, Tony [EMAIL PROTECTED] said: What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. The MT63 modem works at 8000 Hz as Simon noted. However, fldigi's default behaviour is to open the sound card at its default (native) sample rate, usually 48 or 44.1 KHz, and then resample to/from the modem rate using one of the converters from Erik de Castro Lopo's excellent libsamplerate. The converter is chosen based on a short speed test that is done the first time fldigi is run. On all reasonably recent processors, that converter will be one of the good SINC interpolators. It can always be changed later. We basically did this to avoid the not too uncommon situation where a sound card/driver combination claims to support the modem's sample rate only to return mangled audio. To really open the sound card at the modem frequency, which is currently 11025 Hz for Thor and DominoEX 5/11/22, and 8000 Hz for everything else, use the Capture and Playback menus in the Audio settings. The default value is Native (see above). Auto will try the modem rate first and, if that fails, fall back to Native. You may want to try Auto if you know that the OS can do better/faster resampling. The other options in those menus are the rates supposedly supported by the sound card. I say supposedly because the driver may be telling some slight fibs here, i.e., if it reports the hardware native rates together with those that it can resample to in software. It's almost certainly doing that if it lists every standard rate from 8KHz up to 192KHz. Experience suggests that we can trust the default (native) sample rate, at least on the cards and platforms tested so far. The other rates give audio of uncertain quality. Anyway, if you force a rate that is different to the current modem's, fldigi will need to resample. It will also resample, using the same configured conveter, if the TX or RX ppm corrections are nonzero. Of course it's smart enough to combine both kinds of resampling into one step. I don't know enough about Vista's audio system to be able to say what happens if you run multiple programs that want different rates. Possibly some of them will see slightly increased latency. Regarding CPU usage, you should make sure that you have enough CPU cycles for everything; maybe aim to keep the system load at 50%. Fldigi can tell you if it's dropping audio because the CPU is too busy but only in debug builds; perhaps something to change in a future version. 73, Stelios, M0GLD. Announce your digital presence via our Interactive Sked Page at http://www.obriensweb.com/sked 30M digital activity at http://www.projectsandparts.com/30m Recommended software : DM780, Multipsk, FLDIGI, Winwarbler ,MMVARI. Yahoo! Groups Links
Re: [digitalradio] Comparing data modes
Hello Tony, The sample rate on Multipsk is 11025 samples/s for all modes except in MT63 where it is 8000. 73 Patrick Thanks for the info Patrick. Tony, K2MO - Original Message - From: Patrick Lindecker [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 2:59 PM Subject: Re: [digitalradio] Comparing data modes Hello Tony, The sample rate on Multipsk is 11025 samples/s for all modes except in MT63 where it is 8000. 73 Patrick - Original Message - From: Tony [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, September 30, 2008 6:04 AM Subject: Re: [digitalradio] Comparing data modes What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ Dave, The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. Tony, K2MO Announce your digital presence via our Interactive Sked Page at http://www.obriensweb.com/sked 30M digital activity at http://www.projectsandparts.com/30m Recommended software : DM780, Multipsk, FLDIGI, Winwarbler ,MMVARI. Yahoo! Groups Links
RE: [digitalradio] Comparing data modes
Also, there is no reason to believe that the two applications would be equally penalized by a compromise sampling rate -- so the results of the comparison would be suspect unless its known that both applications use the same sampling rate. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Dave AA6YQ Sent: Monday, September 29, 2008 8:14 PM To: digitalradio@yahoogroups.com Subject: RE: [digitalradio] Comparing data modes Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Rick W Sent: Monday, September 29, 2008 7:23 PM To: digitalradio@yahoogroups.com Subject: [digitalradio] Comparing data modes To do simple test comparisons of the modes, I will bring up two software programs and visually see how the print compares between the two. The main comparisons have been between Multipsk, HRD/DM780, and fldigi. For most of these tests I have been using my emachines tower with Intel 2.93 GHz running Windows XP. I also have an HP Pavilion tower with AMD 4600+ chip running Vista and have been using it primarily for tests with my SignaLinkUSB interface to my ICOM IC-7000, which also allows me to have two digital data stations in the shack to perform ARQ testing with NBEMS. I can not view both computers at the same time since I use a KVM switch to work between them. I have not been able to see any situations where one program is clearly superior to another in decoding the signals. 73, Rick, KV9U Rick previously had written: When I have done some crude comparisons with actual off air tests between different programs, there is usually not a lot of difference Tony wrote: I'm interested in your test method. Tony, K2MO
Re: [digitalradio] Comparing data modes
I COULD NOT AGREE MORE WITH THE BELOW STATEMENT EX MT 63 I RUN IZ8BLY DECODE IS GREAT BUT SLOW COMPARED TO MULTI PSK 4.9/4.10 WHICH IS 2X FASTER DECODE BUT PRONE TO MORE ERRORS TEST DONE ON SAME RECEIVER TO CPU'S AND GMFSK FOR LINUX PERSONALLY DOESN.T HOLD A CANDLE TO THE OTHER TWO AND NOT ONLY SAMPLING RATE WHAT ABOUT RESOURCES USED TO RUN TWO PROGRAMS IE RAM? FOR MORE ACCURATE DATA USE TWO CPU'S FLDIGI THOR 11 IS GREAT ON 180 M AT NIGHT AMAZING! DRM IS REALLY GOOD BUT HAVE NOT USED IT IN A WHILE GOOD LUCK IN TESTING MATT KC2PUA Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application' s performance in isolation. 73, Dave, AA6YQ -Original Message- From: digitalradio@ yahoogroups. com [mailto:digitalradi [EMAIL PROTECTED] com]On Behalf Of Rick W Sent: Monday, September 29, 2008 7:23 PM To: digitalradio@ yahoogroups. com Subject: [digitalradio] Comparing data modes To do simple test comparisons of the modes, I will bring up two software programs and visually see how the print compares between the two. The main comparisons have been between Multipsk, HRD/DM780, and fldigi. For most of these tests I have been using my emachines tower with Intel 2.93 GHz running Windows XP. I also have an HP Pavilion tower with AMD 4600+ chip running Vista and have been using it primarily for tests with my SignaLinkUSB interface to my ICOM IC-7000, which also allows me to have two digital data stations in the shack to perform ARQ testing with NBEMS. I can not view both computers at the same time since I use a KVM switch to work between them. I have not been able to see any situations where one program is clearly superior to another in decoding the signals. 73, Rick, KV9U Rick previously had written: When I have done some crude comparisons with actual off air tests between different programs, there is usually not a lot of difference Tony wrote: I'm interested in your test method. Tony, K2MO
Re: [digitalradio] Comparing data modes
Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation. 73, Dave, AA6YQ Always wondered about that Dave. I ran several programs simultaneously while testing their ability to decode MT63 using an HF path simulator. It turned out that 3 out of the 5 programs tested performed exactly the same. They decoded error-free with a minimum SNR of -8db. The other two required an SNR of -5db and -6db. The outcome was the same whether I ran the programs simultaneously or one at a time. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Monday, September 29, 2008 8:14 PM Subject: RE: [digitalradio] Comparing data modes Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Rick W Sent: Monday, September 29, 2008 7:23 PM To: digitalradio@yahoogroups.com Subject: [digitalradio] Comparing data modes To do simple test comparisons of the modes, I will bring up two software programs and visually see how the print compares between the two. The main comparisons have been between Multipsk, HRD/DM780, and fldigi. For most of these tests I have been using my emachines tower with Intel 2.93 GHz running Windows XP. I also have an HP Pavilion tower with AMD 4600+ chip running Vista and have been using it primarily for tests with my SignaLinkUSB interface to my ICOM IC-7000, which also allows me to have two digital data stations in the shack to perform ARQ testing with NBEMS. I can not view both computers at the same time since I use a KVM switch to work between them. I have not been able to see any situations where one program is clearly superior to another in decoding the signals. 73, Rick, KV9U Rick previously had written: When I have done some crude comparisons with actual off air tests between different programs, there is usually not a lot of difference Tony wrote: I'm interested in your test method. Tony, K2MO
RE: [digitalradio] Comparing data modes
What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Monday, September 29, 2008 8:55 PM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation. 73, Dave, AA6YQ Always wondered about that Dave. I ran several programs simultaneously while testing their ability to decode MT63 using an HF path simulator. It turned out that 3 out of the 5 programs tested performed exactly the same. They decoded error-free with a minimum SNR of -8db. The other two required an SNR of -5db and -6db. The outcome was the same whether I ran the programs simultaneously or one at a time. Tony, K2MO - Original Message - From: Dave AA6YQ [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Monday, September 29, 2008 8:14 PM Subject: RE: [digitalradio] Comparing data modes Its my understanding that when multiple simultaneously running applications are using the soundcard with different sampling rates, that Windows delivers a compromise sampling rate. Thus comparisons run on the same PC may not accurately reflect each application's performance in isolation. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Rick W Sent: Monday, September 29, 2008 7:23 PM To: digitalradio@yahoogroups.com Subject: [digitalradio] Comparing data modes To do simple test comparisons of the modes, I will bring up two software programs and visually see how the print compares between the two. The main comparisons have been between Multipsk, HRD/DM780, and fldigi. For most of these tests I have been using my emachines tower with Intel 2.93 GHz running Windows XP. I also have an HP Pavilion tower with AMD 4600+ chip running Vista and have been using it primarily for tests with my SignaLinkUSB interface to my ICOM IC-7000, which also allows me to have two digital data stations in the shack to perform ARQ testing with NBEMS. I can not view both computers at the same time since I use a KVM switch to work between them. I have not been able to see any situations where one program is clearly superior to another in decoding the signals. 73, Rick, KV9U Rick previously had written: When I have done some crude comparisons with actual off air tests between different programs, there is usually not a lot of difference Tony wrote: I'm interested in your test method. Tony, K2MO
Re: [digitalradio] Comparing data modes
What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ Dave, The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. Tony, K2MO
RE: [digitalradio] Comparing data modes
Thanks, Tony. Vista might be adding another variable to the equation. 73, Dave, AA6YQ -Original Message- From: digitalradio@yahoogroups.com [mailto:[EMAIL PROTECTED] Behalf Of Tony Sent: Tuesday, September 30, 2008 12:04 AM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Comparing data modes What were the sampling rates used by each of those 5 applications, Tony? 73, Dave, AA6YQ Dave, The sample rates were 11025 Hz for Mixw and IZ8BLY MT63 terminal. Looks like 8000 Hz for DM780 and Multipsk. Not sure what's going on with Fldigi. I'm using the Vista version. Tony, K2MO
