[digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Patrick Although the amount of RAM may not be critical (provided you're not paging), the SPEED of the memory is probably the limiting factor. The summary is that in a modern cpu with multiple cores and pre-fetch pipeline processing, the processor can execute instructions faster than they can be retrieved from main memory. Access to the L1/L2/L3 caches is much faster than access to main memory, so designing a program so that as much of it and it's working memory will fit into the cache as possible can make a big difference to performance. What this means for the discussion is that when comparing PCs, don't look just at the cpu speed and amount of RAM; consider the speed of the RAM too. I'm getting ready to buy a new PC now and it having DDR3 memory is more impotent to me than the highest possible cpu speed. Sources - this presentation explains a lot about modern PC architecture and performance: http://www.infoq.com/presentations/click-crash-course-modern-hardware If you really want the nitty-gritty, take a look at this classic paper, which is LINUX oriented, but still applies to a Windows PC: What every programmer should know about memory http://people.redhat.com/drepper/cpumemory.pdf John - K6CKP --- In digitalradio@yahoogroups.com, Patrick Lindecker f6...@... wrote: Hello Tony, According to my tests, it is only the capacity to do calculations which is the key, as a lot of digital processng is done (for example for SDR or Panoramics). I don't think RAM is important. I mean either you have sufficient memory or you have not (and you will have a message error). But if you have enough, having double or four more that the minimum does not change anything. Note: with or without BPSK31 panoramic, I have about 2 % of CPU usage. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Saturday, January 30, 2010 4:35 AM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Patrick, Thanks for the information. As you may have read from my reply to Andy, my CPU usage seems to be very low with Multipsk. It's well below 10%. Is there a particular Multipsk mode or configuration that would tax the system? I'd like to try it and see how it affects CPU usage. Merci mon ami... Tony -K2MO - Original Message - From: Patrick Lindecker To: digitalradio@yahoogroups.com Sent: Friday, January 29, 2010 4:51 PM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Hello Tony, I have here two PC XP at about 2.4 GHz (single core): I have compare these two XP computers on the same file to decode (in 110A): * the first one (the oldest) which is an AMD Atlon 2500+ 1.09 GHz 768 Ko RAM takes 75 seconds to decode it, * the second one which is an AMD Atlon 2400+ 2 GHz 736 Ko RAM takes 20 seconds to decode it. On the most modern (about 3 years old) with SdR and RS ID detection on 44 KHz, the CPU load is about 35 to 40 %, but on the old one it is 100 % (the program does not work in fact). So normally with a modern PC it is OK. With an old PC, it can be problematic. Note: with my Vista laptop (dual core), the CPU load is about 25 % in the same conditions. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Friday, January 29, 2010 8:36 PM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Andy, I plan on switching to SDR in the near future. My current PC is a dual CPU 2.2GHz Dell with 3 GHz RAM. Any idea what the minimum PC requirement is to run Multipsk with SDR? Could you also tell us what processor you're running now? Thanks, Tony -K2MO - Original Message - From: Andy obrien To: digitalradio Sent: Friday, January 29, 2010 9:11 AM Subject: [digitalradio] SDR-Radio with DM780 20M Digital Band One of the things that I wanted to accomplish with an SDR receiver, is the ability to keep an eye on the whole 14065 to 14115 frequency range. If I was down on 14074 monitoring ALE 400 traffic, I would miss Olivia signals that popped up in the 14109 area. I would also miss Hell signals at 14068. Now the SDR affords the opportunity to keep an eye all all at once. My venture in to SDR from a digital mode perspective has led to a discovery that, other than Multipsk, the current state of the art does not support direct monitoring of wider I/Q data. I'm also challenged in that my PC cannot cope with the Multipsk CPU demand when I try direct monitoring. So, at the moment I am visually monitoring signals with the SDR and using traditional software methods
Re: [digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Hello John, Although the amount of RAM may not be critical (provided you're not paging), Yes I agree. the SPEED of the memory is probably the limiting factor. It could explain why my two PC at 2400 MHz have very different calculation speeds (a ratio larger than 2), the oldest being the slowest (as in real life). Thanks for the explanations and the WEB addresses. 73 Patrick - Original Message - From: jcprout jcpr...@gmail.com To: digitalradio@yahoogroups.com Sent: Monday, February 01, 2010 9:45 PM Subject: [digitalradio] Re: SDR-Radio with DM780 20M Digital Band Patrick Although the amount of RAM may not be critical (provided you're not paging), the SPEED of the memory is probably the limiting factor. The summary is that in a modern cpu with multiple cores and pre-fetch pipeline processing, the processor can execute instructions faster than they can be retrieved from main memory. Access to the L1/L2/L3 caches is much faster than access to main memory, so designing a program so that as much of it and it's working memory will fit into the cache as possible can make a big difference to performance. What this means for the discussion is that when comparing PCs, don't look just at the cpu speed and amount of RAM; consider the speed of the RAM too. I'm getting ready to buy a new PC now and it having DDR3 memory is more impotent to me than the highest possible cpu speed. Sources - this presentation explains a lot about modern PC architecture and performance: http://www.infoq.com/presentations/click-crash-course-modern-hardware If you really want the nitty-gritty, take a look at this classic paper, which is LINUX oriented, but still applies to a Windows PC: What every programmer should know about memory http://people.redhat.com/drepper/cpumemory.pdf John - K6CKP --- In digitalradio@yahoogroups.com, Patrick Lindecker f6...@... wrote: Hello Tony, According to my tests, it is only the capacity to do calculations which is the key, as a lot of digital processng is done (for example for SDR or Panoramics). I don't think RAM is important. I mean either you have sufficient memory or you have not (and you will have a message error). But if you have enough, having double or four more that the minimum does not change anything. Note: with or without BPSK31 panoramic, I have about 2 % of CPU usage. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Saturday, January 30, 2010 4:35 AM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Patrick, Thanks for the information. As you may have read from my reply to Andy, my CPU usage seems to be very low with Multipsk. It's well below 10%. Is there a particular Multipsk mode or configuration that would tax the system? I'd like to try it and see how it affects CPU usage. Merci mon ami... Tony -K2MO - Original Message - From: Patrick Lindecker To: digitalradio@yahoogroups.com Sent: Friday, January 29, 2010 4:51 PM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Hello Tony, I have here two PC XP at about 2.4 GHz (single core): I have compare these two XP computers on the same file to decode (in 110A): * the first one (the oldest) which is an AMD Atlon 2500+ 1.09 GHz 768 Ko RAM takes 75 seconds to decode it, * the second one which is an AMD Atlon 2400+ 2 GHz 736 Ko RAM takes 20 seconds to decode it. On the most modern (about 3 years old) with SdR and RS ID detection on 44 KHz, the CPU load is about 35 to 40 %, but on the old one it is 100 % (the program does not work in fact). So normally with a modern PC it is OK. With an old PC, it can be problematic. Note: with my Vista laptop (dual core), the CPU load is about 25 % in the same conditions. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Friday, January 29, 2010 8:36 PM Subject: Re: [digitalradio] SDR-Radio with DM780 20M Digital Band Andy, I plan on switching to SDR in the near future. My current PC is a dual CPU 2.2GHz Dell with 3 GHz RAM. Any idea what the minimum PC requirement is to run Multipsk with SDR? Could you also tell us what processor you're running now? Thanks, Tony -K2MO - Original Message - From: Andy obrien To: digitalradio Sent: Friday, January 29, 2010 9:11 AM Subject: [digitalradio] SDR-Radio with DM780 20M Digital Band One of the things that I wanted to accomplish with an SDR receiver, is the ability to keep an eye on the whole 14065 to 14115 frequency range. If I was down on 14074 monitoring ALE 400 traffic, I would miss Olivia signals that popped up in the 14109 area. I would also
[digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Andy, Patrick is way ahead of all of us. I forgot to mention that when I first installed the E2180 dual core cpu/motherboard it only had 500GB of memory. The digital mode and SDR applications actually ran OK and the cpu usage was quite low but the memory shortage was very obvious when starting applications and changing the windows on the screen such as when browsing the web. The SDR-14 receiver is very interesting. I am curious what can actually be done when scanning a 30MHz spectrum. It seems like scanning a 30MHz spectrum fast enough to decode RSIDs would be beyond the capabilities of current PCs. The next step in software that I would like to see is opening up the digital mode panoramic decoding display to 5KHz or more. This would help to spread the signals out a bit on 20 meters and/or open it up for more digital mode QSOs. MixW actually does have dual receive capability in addition to the simultaneous multiple mode capability. It is unfortunate that it does not have a panoramic display or RSID decoding. I am amazed at what some of the programmers like Simon and Patrick have been able to do by themselves. It seems like these programs and and especially the capabilities that SDR opens up can't possibly be one-person projects. Ed WB6YTE --- In digitalradio@yahoogroups.com, Andy obrien k3uka...@... wrote: On Sat, Jan 30, 2010 at 9:08 PM, ed_hekman ehek...@... wrote: Interesting comments, Andy. My goal is also to be able to monitor all the digital portions of the band and to be able to spot all call signs in any mode across the band. CW Skimmer is a good model for that. In addition, it would nice to be able to select a few segments of the band (~5 KHz span in each segment) and to be able to select a few specific signals in each segment for continuous decoding in various modes. As far as I know MixW is the only package that can decode multiple modes simultaneously. PSDR and SDR-Radio allow the selection of multiple segments (2 for PSDR, 3 for SDR-Radio) but the integration with digital mode decoding is not built-in with the SDR software. Ed, thanks. You are way ahead of me on this stuff. I have not tried multiple segments yet, that will be interesting to try. My maximum is 192 Khz at the moment. I expect that I will move to something like an SDR-14 in the future and have 30 Mhz capability at some point. I am glad I did not plonk down a hard earned thousand bucks to find out that the software isn't;really 'ready yet, and that my PC's will need a major upgrade. So, I am happy with the learning curve at the moment and will be better prepared when ready to move up in a serious way. PSK skimmers are essentially already within FLdigi, Multipsk, Winwarbler, and DM780. Broadening PSK callsign mining to four of 5 Khz segments should eventually be possible , and not very taxing. RTTY skimming during a contest might require several 100 kHz segments, that might be tougher than skimming the same bandwidth for CW signals. I suppose the serious digital mode skimmer would want to continually keep an eye on all PSK31 and RTTY signals just like the CW enthusiasts want all CW segments. If there were eventually PSK31 and RTTY skimmers, the remaining Olivia, Hell, MFSK16, and THOR signals would be something most would happily manually watch/listen for. Thanks for sharing your benchmarks. Andy K3UK
Re: [digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Hi Again Ed. I did play with the 3 VFO feature of Simon's software, see http://www.obriensweb.com/3vfo20.jpg http://www.obriensweb.com/3vfo.jpg (17M) It is nice to be able to 'see the CW , digital, and phone portions, and click back and forth to actually hear when needed.Thanks for pointing this out. yes, Simon and Patrick have amazing skills. My next venture is to get more adept at figuring out the transmit side of things , since I am using separates. I have managed to protect the front end of my SDR receiver, no overload when I transmit. I am manually tuning the transmitting rig to the received SDR signal, it would be nice to sync the two rigs without tapping in to the IF stage of the transceiver. I can do that with SpectraVue but like Simon's software and would like to sync SDR-Radio with the TS2000. I suspect there are ways to do this already and I just need to explore the possibilities. Having fun learning... Andy.
[digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Ed: With 500 GB of memory you should have no trouble running any program likely to be produced in the next few years. MixW can get a pretty good range of resolution. In the case of resolution, looking at 20 m on MixW at a zoom factor of 0.5, the spread on the display is 14.070 through 14.080. Looking at 20 m on MixW with a zoom factor of four, you have just about as much resolution as you could ever want. The spread goes from 14, 070.1 through 14,071.5 . That is spread out through the entire width of the page. Dick,KC4COP
[digitalradio] Re: SDR-Radio with DM780 20M Digital Band
Interesting comments, Andy. My goal is also to be able to monitor all the digital portions of the band and to be able to spot all call signs in any mode across the band. CW Skimmer is a good model for that. In addition, it would nice to be able to select a few segments of the band (~5 KHz span in each segment) and to be able to select a few specific signals in each segment for continuous decoding in various modes. As far as I know MixW is the only package that can decode multiple modes simultaneously. PSDR and SDR-Radio allow the selection of multiple segments (2 for PSDR, 3 for SDR-Radio) but the integration with digital mode decoding is not built-in with the SDR software. The hardware capability to do much of this is available in many of the computers in the consumer market. A few benchmarks that I have are: MultiPSK in 4.3KHz wide panoramic mode with I/Q direct interface: Dual core CPU, E2180, 2.0 GHz, 2.5 GB memory - 25% CPU usage Quad core CPU, Q8200, 2.33GHz, 4.0 GB memory - 12% CPU usage On the Q8200: SDR-Radio uses about 8% average. DM780 uses about 8% maximum. The total CPU usage on the E2180 was around 35% - 40% with these applications running: MultiPSK in panoramic mode the I/Q direct interface HRD/DM780/Logbook UI-View APRS Weather Display I replaced a 2.4 GHz single CPU/motherboard with the E2180 dual CPU/motherboard last year and was very pleasantly surprised at the increased capability/decreased CPU usage. About 1GB per CPU core seems to be sufficient for ham radio usage. Ed WB6YTE --- In digitalradio@yahoogroups.com, Andy obrien k3uka...@... wrote: One of the things that I wanted to accomplish with an SDR receiver, is the ability to keep an eye on the whole 14065 to 14115 frequency range. If I was down on 14074 monitoring ALE 400 traffic, I would miss Olivia signals that popped up in the 14109 area. I would also miss Hell signals at 14068. Now the SDR affords the opportunity to keep an eye all all at once. My venture in to SDR from a digital mode perspective has led to a discovery that, other than Multipsk, the current state of the art does not support direct monitoring of wider I/Q data. I'm also challenged in that my PC cannot cope with the Multipsk CPU demand when I try direct monitoring. So, at the moment I am visually monitoring signals with the SDR and using traditional software methods to decode the 3-4 kHz of audio that is fed from the SDR to applications like DM780 or Fldigi. At this screen shot http://www.obriensweb.com/sdrdm780.jpg you will see how it appears. I am simply using DM780 and SDR-Radio software together. When I need to transmit, I just use my TS2000 after dialing in the signal discovered by the SDR receiver. Simon HB9DRV will likely integrate these two applications later in 2010. I did catch a Russian on RTTY this morning that I would have otherwise missed while I was slumming it in PSK31-land.. Multisk does RS-ID over this entire 14065-14115 portion, and DM780 is likely going to include this ability in the future. If people use RS-ID often enough, it will be really cool to monitor 14065-14115 and get RS ID alerts. So, just over a week playing around with the SDR receiver... I see the potential... digital mode applications are not quite there yet. When they are there (as in Multipsk) my PC isn't. This $41.00 Ebay PC may eventually get retired for a slightly improved one with better CPU. OK, back to keeping an eye on 14065-14115. A-ha, an SV3 calling CQ RTTY, 14082. Andy K3UK
Re: [digitalradio] Re: SDR-Radio with DM780 20M Digital Band
On Sat, Jan 30, 2010 at 9:08 PM, ed_hekman ehek...@cox.net wrote: Interesting comments, Andy. My goal is also to be able to monitor all the digital portions of the band and to be able to spot all call signs in any mode across the band. CW Skimmer is a good model for that. In addition, it would nice to be able to select a few segments of the band (~5 KHz span in each segment) and to be able to select a few specific signals in each segment for continuous decoding in various modes. As far as I know MixW is the only package that can decode multiple modes simultaneously. PSDR and SDR-Radio allow the selection of multiple segments (2 for PSDR, 3 for SDR-Radio) but the integration with digital mode decoding is not built-in with the SDR software. Ed, thanks. You are way ahead of me on this stuff. I have not tried multiple segments yet, that will be interesting to try. My maximum is 192 Khz at the moment. I expect that I will move to something like an SDR-14 in the future and have 30 Mhz capability at some point. I am glad I did not plonk down a hard earned thousand bucks to find out that the software isn't;really 'ready yet, and that my PC's will need a major upgrade. So, I am happy with the learning curve at the moment and will be better prepared when ready to move up in a serious way. PSK skimmers are essentially already within FLdigi, Multipsk, Winwarbler, and DM780. Broadening PSK callsign mining to four of 5 Khz segments should eventually be possible , and not very taxing. RTTY skimming during a contest might require several 100 kHz segments, that might be tougher than skimming the same bandwidth for CW signals. I suppose the serious digital mode skimmer would want to continually keep an eye on all PSK31 and RTTY signals just like the CW enthusiasts want all CW segments. If there were eventually PSK31 and RTTY skimmers, the remaining Olivia, Hell, MFSK16, and THOR signals would be something most would happily manually watch/listen for. Thanks for sharing your benchmarks. Andy K3UK
