|
Re: AndFlmsg; attaching forms to text msgs? sent messages not always listed as sent
Hi John, Whatever the Android OS update yesterday did to my phone appears to have blocked;
- AndFlmsg settings for location (thus not allowing the app to open),
- connecting to the desktop to reload AndFlmsg and
- not allowing either of two file managers to show downloaded files (I delted AndFlmsg and downloaded latest version directly into phone)
Rats. Thus the best I can do re the image in the preceding message is to say that the header and footer sure look similar to what appeared on my phone. Apologies..... 73 Ki7MBR Ole
|
|
|
Re: AndFlmsg; attaching forms to text msgs? sent messages not always listed as sent
Hi John,
Apologies for the late reply. The above looks familiar. However, my phone updated its Android operating system and a number of things now do not work, including AndFlmsg. The phone refuses to take the location permission. I just deleted the app, will reload and report. Might be another day or two.
73, Ole
|
|
|
Hi Dave, could you pls take a look to signlal browser in CW?
It still present running QSOs on wrong frequeny. This Example a QSO on 3530, correctly shown in waterfall as 3529+1000 but in signal browser as 3529+500 73 Roland DK4RH
|
|
|
Re: Transmit audio level for FM VHF?
Dave
Every signal generated by
fldigi will have a peak-to-peak audio signal that is equal to or
less than the "Tune" signal. If you follow the advice of Harry
or John using the tune signal then all others will be OK.
David, W1HKJ
toggle quoted messageShow quoted text
On 10/26/20 12:10 PM, KL4YFD ham wrote:
Hi All,
To adjust audio-levels for FM you can listen to your
own voice-signal through a scanner and adjust your
digital-Tx-audio to be slightly quieter than voice (to
prevent overdriving).
Better to have a too-quiet signal than a too-loud
one.
The modes in Fldigi, even the 8PSK modes, are
designed to be adjusted by ear without a deviation meter
needed.
This is very handy in an emergency situation, hihi.
Since people are using Fldigi more for FM, the manual
should have some kind of FM adjustment procedure or info
added.
Here is a link for adjusting audio for 8PSK:
Can apply the same procedure for all Fldigi modes on
FM.
73,
John Phelps
KL4YFD
On Mon, Oct 26, 2020 at 9:18
AM Harry Bloomberg < hbloomb@...> wrote:
My opinion as somebody who used to conduct a ton of
NBEMS training...
Getting the modulation perfectly set on VHF/UHF FM
isn't worth the time and trouble. Adjust so that the
transmitted NBEMS audio is close to that of a spoken
voice. Sometimes "Good Enough" is good enough.
73,
Harry Bloomberg W3YJ
A few years ago, I had an IC706MK2G and also made
two installations of Fldigi with one for the HF side
and the other for VHF/UHF. The interface between the
radio and computer was (is still) a SignaLink USB
with the cable from Tiger Tronics getting the
signals to and from the radio. The settings I used
for the SL USB were to have both the TX and RX (the
transmit and receive) settings at mid-point or about
12:00 and the DLY set at full counter clockwise
(CCW). The power TO the radio (from computer to
radio) was set so that the radio would transmit a
clean signal at not over half power for whatever
band had been selected.
James Good kd5vxh
On 10/26/20 1:04 AM, Randy Allen wrote:
Oh, and forgot to mention in my previous reply,
it turns out the optimum settings on VHF/UHF for
my set up (IC-706MKIIG on low power and SignaLink)
was setting all three knobs at about the 11
O'Clock position when using FLDIGI (usually MT-63
2KL) and setting the DLY to almost minimum when
using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston W0NRP via groups.io wrote:
The operating manual explains how to set the
transmit audio for an HF transceiver, but seems
to give no guidance for setting up the audio for
VHF FM.
We have the (unverified) information that the
recommended deviation level is approximately 3
KHz.
We have been struggling with this in our
training nets, and most of us do not have
deviation measurement capability.
What methods might allow us to make some kind of
reasonable estimate of our audio level when
using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
You're right...it's been a while since I've done NBEMS on VHF/UHF FM. Slightly lower audio level than voice is indeed the best. And a deviation meter is just not needed.
73, Harry W3YJ
On Mon, Oct 26, 2020 at 1:10 PM KL4YFD ham < kl4yfd@...> wrote: Hi All,
To adjust audio-levels for FM you can listen to your own voice-signal through a scanner and adjust your digital-Tx-audio to be slightly quieter than voice (to prevent overdriving). Better to have a too-quiet signal than a too-loud one.
The modes in Fldigi, even the 8PSK modes, are designed to be adjusted by ear without a deviation meter needed. This is very handy in an emergency situation, hihi.
Since people are using Fldigi more for FM, the manual should have some kind of FM adjustment procedure or info added. Here is a link for adjusting audio for 8PSK: Can apply the same procedure for all Fldigi modes on FM.
73, John Phelps KL4YFD
On Mon, Oct 26, 2020 at 9:18 AM Harry Bloomberg < hbloomb@...> wrote: My opinion as somebody who used to conduct a ton of NBEMS training...
Getting the modulation perfectly set on VHF/UHF FM isn't worth the time and trouble. Adjust so that the transmitted NBEMS audio is close to that of a spoken voice. Sometimes "Good Enough" is good enough.
73, Harry Bloomberg W3YJ
A few years ago, I had an IC706MK2G and also made two
installations of Fldigi with one for the HF side and the other for
VHF/UHF. The interface between the radio and computer was (is
still) a SignaLink USB with the cable from Tiger Tronics getting
the signals to and from the radio. The settings I used for the SL
USB were to have both the TX and RX (the transmit and receive)
settings at mid-point or about 12:00 and the DLY set at full
counter clockwise (CCW). The power TO the radio (from computer to
radio) was set so that the radio would transmit a clean signal at
not over half power for whatever band had been selected.
James Good kd5vxh
On 10/26/20 1:04 AM, Randy Allen wrote:
Oh, and forgot to mention in my previous reply, it turns out
the optimum settings on VHF/UHF for my set up (IC-706MKIIG on
low power and SignaLink) was setting all three knobs at about
the 11 O'Clock position when using FLDIGI (usually MT-63 2KL)
and setting the DLY to almost minimum when using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston
W0NRP via groups.io wrote:
The operating manual explains how to set the transmit audio
for an HF transceiver, but seems to give no guidance for
setting up the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
Hi All,
To adjust audio-levels for FM you can listen to your own voice-signal through a scanner and adjust your digital-Tx-audio to be slightly quieter than voice (to prevent overdriving). Better to have a too-quiet signal than a too-loud one.
The modes in Fldigi, even the 8PSK modes, are designed to be adjusted by ear without a deviation meter needed. This is very handy in an emergency situation, hihi.
Since people are using Fldigi more for FM, the manual should have some kind of FM adjustment procedure or info added. Here is a link for adjusting audio for 8PSK: Can apply the same procedure for all Fldigi modes on FM.
73, John Phelps KL4YFD
On Mon, Oct 26, 2020 at 9:18 AM Harry Bloomberg < hbloomb@...> wrote: My opinion as somebody who used to conduct a ton of NBEMS training...
Getting the modulation perfectly set on VHF/UHF FM isn't worth the time and trouble. Adjust so that the transmitted NBEMS audio is close to that of a spoken voice. Sometimes "Good Enough" is good enough.
73, Harry Bloomberg W3YJ
A few years ago, I had an IC706MK2G and also made two
installations of Fldigi with one for the HF side and the other for
VHF/UHF. The interface between the radio and computer was (is
still) a SignaLink USB with the cable from Tiger Tronics getting
the signals to and from the radio. The settings I used for the SL
USB were to have both the TX and RX (the transmit and receive)
settings at mid-point or about 12:00 and the DLY set at full
counter clockwise (CCW). The power TO the radio (from computer to
radio) was set so that the radio would transmit a clean signal at
not over half power for whatever band had been selected.
James Good kd5vxh
On 10/26/20 1:04 AM, Randy Allen wrote:
Oh, and forgot to mention in my previous reply, it turns out
the optimum settings on VHF/UHF for my set up (IC-706MKIIG on
low power and SignaLink) was setting all three knobs at about
the 11 O'Clock position when using FLDIGI (usually MT-63 2KL)
and setting the DLY to almost minimum when using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston
W0NRP via groups.io wrote:
The operating manual explains how to set the transmit audio
for an HF transceiver, but seems to give no guidance for
setting up the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
My opinion as somebody who used to conduct a ton of NBEMS training...
Getting the modulation perfectly set on VHF/UHF FM isn't worth the time and trouble. Adjust so that the transmitted NBEMS audio is close to that of a spoken voice. Sometimes "Good Enough" is good enough.
73, Harry Bloomberg W3YJ
A few years ago, I had an IC706MK2G and also made two
installations of Fldigi with one for the HF side and the other for
VHF/UHF. The interface between the radio and computer was (is
still) a SignaLink USB with the cable from Tiger Tronics getting
the signals to and from the radio. The settings I used for the SL
USB were to have both the TX and RX (the transmit and receive)
settings at mid-point or about 12:00 and the DLY set at full
counter clockwise (CCW). The power TO the radio (from computer to
radio) was set so that the radio would transmit a clean signal at
not over half power for whatever band had been selected.
James Good kd5vxh
On 10/26/20 1:04 AM, Randy Allen wrote:
Oh, and forgot to mention in my previous reply, it turns out
the optimum settings on VHF/UHF for my set up (IC-706MKIIG on
low power and SignaLink) was setting all three knobs at about
the 11 O'Clock position when using FLDIGI (usually MT-63 2KL)
and setting the DLY to almost minimum when using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston
W0NRP via groups.io wrote:
The operating manual explains how to set the transmit audio
for an HF transceiver, but seems to give no guidance for
setting up the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
A few years ago, I had an IC706MK2G and also made two
installations of Fldigi with one for the HF side and the other for
VHF/UHF. The interface between the radio and computer was (is
still) a SignaLink USB with the cable from Tiger Tronics getting
the signals to and from the radio. The settings I used for the SL
USB were to have both the TX and RX (the transmit and receive)
settings at mid-point or about 12:00 and the DLY set at full
counter clockwise (CCW). The power TO the radio (from computer to
radio) was set so that the radio would transmit a clean signal at
not over half power for whatever band had been selected.
James Good kd5vxh
toggle quoted messageShow quoted text
On 10/26/20 1:04 AM, Randy Allen wrote:
Oh, and forgot to mention in my previous reply, it turns out
the optimum settings on VHF/UHF for my set up (IC-706MKIIG on
low power and SignaLink) was setting all three knobs at about
the 11 O'Clock position when using FLDIGI (usually MT-63 2KL)
and setting the DLY to almost minimum when using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston
W0NRP via groups.io wrote:
The operating manual explains how to set the transmit audio
for an HF transceiver, but seems to give no guidance for
setting up the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
Oh, and forgot to mention in my previous reply, it turns out the
optimum settings on VHF/UHF for my set up (IC-706MKIIG on low
power and SignaLink) was setting all three knobs at about the 11
O'Clock position when using FLDIGI (usually MT-63 2KL) and setting
the DLY to almost minimum when using WinLink. YMMV.
Hope this helps
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston W0NRP
via groups.io wrote:
The operating manual explains how to set the transmit audio for
an HF transceiver, but seems to give no guidance for setting up
the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: Transmit audio level for FM VHF?
A member of one of our local digital training nets did have the
proper equipment to measure on air signals. We spent one net with
him monitoring each of our signals (using FLDIGI) and providing
feed back.
Maybe check around your area and see if someone (say someone that
works in Land Mobile Radio and has access to the equipment) could
do that during a net?
One of our local clubs has a person that works in one of the
local radios shops with access to a service monitor. They held a
special club meeting where folks could bring their gear in and
have it spec'ed out.
Just a couple of ideas
73
Randy KA0AZS
On 10/26/2020 01:13, Neil Preston W0NRP
via groups.io wrote:
The operating manual explains how to set the transmit audio for
an HF transceiver, but seems to give no guidance for setting up
the audio for VHF FM.
We have the (unverified) information that the recommended
deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and
most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable
estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Transmit audio level for FM VHF?
The operating manual explains how to set the transmit audio for an HF transceiver, but seems to give no guidance for setting up the audio for VHF FM.
We have the (unverified) information that the recommended deviation level is approximately 3 KHz.
We have been struggling with this in our training nets, and most of us do not have deviation measurement capability.
What methods might allow us to make some kind of reasonable estimate of our audio level when using a SignaLink?
Thanks....
|
|
|
Re: fldigi 4.1.15 and Windows XP
Dave
Windows XP is no longer
compatible with fldigi.
David
toggle quoted messageShow quoted text
On 10/24/20 8:12 PM, Mike wrote:
I'm asking for a friend of mine. He still uses an XP computer for
radio only work.
When trying to run fldigi 4.1.15 he gets the error message:
The procedure entry point CancelIoEx could not be located in the
dynamic link library KERNEL32.dll.
Is windows XP no longer compatable with fldigi or is something
else going on?
73, Mike, K0TER
|
|
|
fldigi 4.1.15.04 alpha posted at
Dave
http://www.w1hkj.com/alpha/fldigi/
Sat Oct 24 20:38 2020 -0500 version
4.1.15.04
Add MFSK Wide Modes
* Add MFSK-32WIDE and MFSK-64WIDE
- for both Shortwave Radiogram and Amateur usage
- heavier constraint length 15 FEC (8-bit error correction)
- 3.8 second interleaver (3,840 msec)
- more robust MFSK for long transmissions (radiogram, flamp,
etc)
Phase and FEC quality display
* Add phase-accuracy display for PSK modes (non-PSK31)
* Added better FEC confidence code to THOR
Add new platform optimizations
* added new --enable-optimizations options to ./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- Use AVX extensions [avx, avx2]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Updated Doxygen Sourcecode Docs
* Now renders to SVG for graphics (doxygen 1.18)
* Fixed scripts for Cppcheck 1.9 changes
* Added auto-installer for packages on Debian-like systems
* Improved main page: index.html
fonts
* fix slow start caused by font fixed/proportional enumeration
- fixed evaluator moved to separate thread during first
instantiation of Font_Browser class. Actual time to
enumerate
2650 fonts on an i5 3.2 GHz system is approximately 25
seconds. Background process does not effect UI, but
selection
of fixed fonts in font browser dialog may not show all of
the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
fldigi 4.1.15 and Windows XP
I'm asking for a friend of mine. He still uses an XP computer for radio only work.
When trying to run fldigi 4.1.15 he gets the error message:
The procedure entry point CancelIoEx could not be located in the dynamic link library KERNEL32.dll.
Is windows XP no longer compatable with fldigi or is something else going on?
73, Mike, K0TER
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
Hi Everyone:
Wanted to point out the purpose of these new optimizations real quick, as there were no real instructions with the patch: - Compiling with "native" will get you all the optimizations that your CPU supports (unless embedded).
- For embedded, use that specific platforms optimizations.
- Compiling for "x86-64" or "i686" will give code compatible with ALL cpus of that family (portable, but not all optimizations)
John Phelps KL4YFD
Hey Dave,
Thanks for putting that table together but one thing still eludes
me on the 64bit front. If I have a 64bit distro (most common
now), I would assume I would want to use the "x86-64" option yet
your macro says this will only use up to SSE2 instructions. On my
7th gen i7 chip, /proc/cpuinfo says it has support for "sse4_1
sse4_2".
Would you recommend that I modify the macro for x86-64 to add "SSE4"?
--David
KI6ZHD
On 10/24/2020 12:08 PM, Dave wrote:
Short Form
|
Long Form
|
sse
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-mfpmath=sse" |
sse2
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -mfpmath=sse" |
sse3
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -mfpmath=sse" |
sse4
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -msse4 -mfpmath=sse" |
i486
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i486
-mtune=i486 -mno-mmx -mno-sse -mfpmath=387" |
i686
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i686
-mtune=i686 -mmmx -msse -mfpmath=sse" |
x86-64
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=x86-64
-mtune=k8 -mmmx -msse -msse2 -mfpmath=sse" |
Pi1
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=arm1176jzf-s -mfpu=vfp" |
Pi2
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
PI3
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
Pi4
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
bbb
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a8 -mfpu=neon" |
cyclone5
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a9 -mfpu=neon" |
native ARM
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=native -mfpu=auto" |
My suggestion is that you read the spec sheet on a specific
processor to know which optimizations are supported.
David
On 10/24/20 1:04 PM, David Ranch
wrote:
Adding on to Harry's question, Dave W1HKJ, can you elaborate on
some of these optimization questions?
- What is the general difference between say "RPI4" vs "bbb" vs "nativeARM"?
- What are some example X86 Intel/AMD processor families
should or shouldn't use "sse4"?
- Say if I have an Intel i7-7600U, should I pick "x86-64" or "sse4". To me, these
two are complimentary (aka "and") options not "or" options
- For building a universal x86-32 or ARM binary, what would
you recommend for the best option?
I think it would be great to put some of this guidance into the
build documentations say at https://sourceforge.net/p/fldigi/wiki/debian_howto/
and mention this in the tarball's INSTALL doc.
PS. In the fldigi-4.1.15's tarball, the INSTALL file points to:
http://www.w1hkj.com/doku/doku.php?id=howtos
I think this now is supposed to be https://sourceforge.net/p/fldigi/wiki/how_to/
--David
KI6ZHD
On 10/24/2020 10:32 AM, Harry
Bloomberg wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at
1:11 PM Dave < w1hkj@...>
wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55
2020 -0500 version 4.1.15.03
PSK phase scope
expansion
* expanded PSK phase-scope constellation markers
to 8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK
modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive
source-code docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent
confusion with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread
during first
instantiation of Font_Browser class. Actual
time to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect
UI, but selection
of fixed fonts in font browser dialog may not
show all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
Hey Dave,
Thanks for putting that table together but one thing still eludes
me on the 64bit front. If I have a 64bit distro (most common
now), I would assume I would want to use the "x86-64" option yet
your macro says this will only use up to SSE2 instructions. On my
7th gen i7 chip, /proc/cpuinfo says it has support for "sse4_1
sse4_2".
Would you recommend that I modify the macro for x86-64 to add "SSE4"?
--David
KI6ZHD
toggle quoted messageShow quoted text
On 10/24/2020 12:08 PM, Dave wrote:
Short Form
|
Long Form
|
sse
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-mfpmath=sse" |
sse2
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -mfpmath=sse" |
sse3
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -mfpmath=sse" |
sse4
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -msse4 -mfpmath=sse" |
i486
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i486
-mtune=i486 -mno-mmx -mno-sse -mfpmath=387" |
i686
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i686
-mtune=i686 -mmmx -msse -mfpmath=sse" |
x86-64
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=x86-64
-mtune=k8 -mmmx -msse -msse2 -mfpmath=sse" |
Pi1
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=arm1176jzf-s -mfpu=vfp" |
Pi2
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
PI3
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
Pi4
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
bbb
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a8 -mfpu=neon" |
cyclone5
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a9 -mfpu=neon" |
native ARM
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=native -mfpu=auto" |
My suggestion is that you read the spec sheet on a specific
processor to know which optimizations are supported.
David
On 10/24/20 1:04 PM, David Ranch
wrote:
Adding on to Harry's question, Dave W1HKJ, can you elaborate on
some of these optimization questions?
- What is the general difference between say "RPI4" vs "bbb" vs "nativeARM"?
- What are some example X86 Intel/AMD processor families
should or shouldn't use "sse4"?
- Say if I have an Intel i7-7600U, should I pick "x86-64" or "sse4". To me, these
two are complimentary (aka "and") options not "or" options
- For building a universal x86-32 or ARM binary, what would
you recommend for the best option?
I think it would be great to put some of this guidance into the
build documentations say at https://sourceforge.net/p/fldigi/wiki/debian_howto/
and mention this in the tarball's INSTALL doc.
PS. In the fldigi-4.1.15's tarball, the INSTALL file points to:
http://www.w1hkj.com/doku/doku.php?id=howtos
I think this now is supposed to be https://sourceforge.net/p/fldigi/wiki/how_to/
--David
KI6ZHD
On 10/24/2020 10:32 AM, Harry
Bloomberg wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at
1:11 PM Dave < w1hkj@...>
wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55
2020 -0500 version 4.1.15.03
PSK phase scope
expansion
* expanded PSK phase-scope constellation markers
to 8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK
modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive
source-code docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent
confusion with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread
during first
instantiation of Font_Browser class. Actual
time to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect
UI, but selection
of fixed fonts in font browser dialog may not
show all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
Dave
Short Form
|
Long Form
|
sse
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-mfpmath=sse" |
sse2
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -mfpmath=sse" |
sse3
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -mfpmath=sse" |
sse4
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -mmmx -msse
-msse2 -msse3 -msse4 -mfpmath=sse" |
i486
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i486
-mtune=i486 -mno-mmx -mno-sse -mfpmath=387" |
i686
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=i686
-mtune=i686 -mmmx -msse -mfpmath=sse" |
x86-64
|
OPT_CFLAGS="$OPT_CFLAGS_x86 -march=x86-64
-mtune=k8 -mmmx -msse -msse2 -mfpmath=sse" |
Pi1
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=arm1176jzf-s -mfpu=vfp" |
Pi2
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
PI3
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
Pi4
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a7 -mfpu=neon-vfpv4" |
bbb
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a8 -mfpu=neon" |
cyclone5
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=cortex-a9 -mfpu=neon" |
native ARM
|
OPT_CFLAGS="$OPT_CFLAGS_ARM
-mcpu=native -mfpu=auto" |
My suggestion is that you read the spec sheet on a specific
processor to know which optimizations are supported.
David
toggle quoted messageShow quoted text
On 10/24/20 1:04 PM, David Ranch wrote:
Adding on to Harry's question, Dave W1HKJ, can you elaborate on
some of these optimization questions?
- What is the general difference between say "RPI4" vs "bbb" vs "nativeARM"?
- What are some example X86 Intel/AMD processor families
should or shouldn't use "sse4"?
- Say if I have an Intel i7-7600U, should I pick "x86-64" or "sse4". To me, these two
are complimentary (aka "and") options not "or" options
- For building a universal x86-32 or ARM binary, what would you
recommend for the best option?
I think it would be great to put some of this guidance into the
build documentations say at https://sourceforge.net/p/fldigi/wiki/debian_howto/
and mention this in the tarball's INSTALL doc.
PS. In the fldigi-4.1.15's tarball, the INSTALL file points to:
http://www.w1hkj.com/doku/doku.php?id=howtos
I think this now is supposed to be https://sourceforge.net/p/fldigi/wiki/how_to/
--David
KI6ZHD
On 10/24/2020 10:32 AM, Harry
Bloomberg wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at 1:11
PM Dave < w1hkj@...>
wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55
2020 -0500 version 4.1.15.03
PSK phase scope
expansion
* expanded PSK phase-scope constellation markers to
8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive source-code
docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent confusion
with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread during
first
instantiation of Font_Browser class. Actual
time to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect UI,
but selection
of fixed fonts in font browser dialog may not
show all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
And adding even more to this...
I've got a cheap laptop running an Intel Celeron N3150 CPU. Where would I look to see which 386/486 optimization to use?
73, Harry W3YJ
Adding on to Harry's question, Dave W1HKJ, can you elaborate on some
of these optimization questions?
- What is the general difference between say "RPI4" vs " bbb" vs " nativeARM"?
- What are some example X86 Intel/AMD processor families should
or shouldn't use "sse4"?
- Say if I have an Intel i7-7600U, should I pick "x86-64" or "sse4". To me, these two
are complimentary (aka "and") options not "or" options
- For building a universal x86-32 or ARM binary, what would you
recommend for the best option?
I think it would be great to put some of this guidance into the
build documentations say at
https://sourceforge.net/p/fldigi/wiki/debian_howto/ and mention this
in the tarball's INSTALL doc.
PS. In the fldigi-4.1.15's tarball, the INSTALL file points to:
http://www.w1hkj.com/doku/doku.php?id=howtos
I think this now is supposed to be
https://sourceforge.net/p/fldigi/wiki/how_to/
--David
KI6ZHD
On 10/24/2020 10:32 AM, Harry Bloomberg
wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at 1:11
PM Dave < w1hkj@...>
wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55 2020
-0500 version 4.1.15.03
PSK phase scope expansion
* expanded PSK phase-scope constellation markers to
8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive source-code
docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent confusion
with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread during
first
instantiation of Font_Browser class. Actual time
to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect UI,
but selection
of fixed fonts in font browser dialog may not show
all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
Adding on to Harry's question, Dave W1HKJ, can you elaborate on some
of these optimization questions?
- What is the general difference between say "RPI4" vs " bbb" vs " nativeARM"?
- What are some example X86 Intel/AMD processor families should
or shouldn't use "sse4"?
- Say if I have an Intel i7-7600U, should I pick "x86-64" or "sse4". To me, these two
are complimentary (aka "and") options not "or" options
- For building a universal x86-32 or ARM binary, what would you
recommend for the best option?
I think it would be great to put some of this guidance into the
build documentations say at
https://sourceforge.net/p/fldigi/wiki/debian_howto/ and mention this
in the tarball's INSTALL doc.
PS. In the fldigi-4.1.15's tarball, the INSTALL file points to:
http://www.w1hkj.com/doku/doku.php?id=howtos
I think this now is supposed to be
https://sourceforge.net/p/fldigi/wiki/how_to/
--David
KI6ZHD
toggle quoted messageShow quoted text
On 10/24/2020 10:32 AM, Harry Bloomberg
wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at 1:11
PM Dave < w1hkj@...>
wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55 2020
-0500 version 4.1.15.03
PSK phase scope expansion
* expanded PSK phase-scope constellation markers to
8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive source-code
docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent confusion
with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread during
first
instantiation of Font_Browser class. Actual time
to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect UI,
but selection
of fixed fonts in font browser dialog may not show
all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
Re: fldigi 4.1.15.03 alpha posted at
Thank you! I'll give this a try later today.
73, Harry W3YJ
On Sat, Oct 24, 2020 at 1:41 PM Dave < w1hkj@...> wrote:
Exactly Harry. Tested here
on both a Pi4 and Pi3. Changes to flrig etal will be in future
release.
Dave
On 10/24/20 12:32 PM, Harry Bloomberg
wrote:
Wow! Thanks!
So, to optimize for Raspberry Pi 4, you need to do
something like
./configure --enable-optimizations=rpi4 ?
73,
Harry Bloomberg W3YJ
On Sat, Oct 24, 2020 at 1:11
PM Dave < w1hkj@...> wrote:
http://www.w1hkj.com/alpha/fldigi/
Removed experimental
16psk modes.
Fri Oct 23 09:20:55 2020
-0500 version 4.1.15.03
PSK phase scope expansion
* expanded PSK phase-scope constellation markers to
8PSK
* doubled display-time of phase-markers(needed for
faster modes)
Command Line Options
* Fix rx-only add tx-only commandline option
- Fixed infinite loop bug when --rx-only used
- Added new --tx-only commandline option
Modem Quality Display
* Add quality display for phase and FEC
- added FEC-confidence display to all MFSK modems
- added Phase-accuracy display to all PSK modems
(except psk31)
- updated THOR FEC-confidence display
Add new platform optimizations
* added new --enable-optimizations options to
./configure
- RaspberryPi 1,2,3,4 [rpi1, rpi2, rpi3, rpi4]
- BeagleBone Black [bbb]
- Cyclone5 [cyclone5]
- Optimize for sse4 [sse4]
- i486, i686 [i486, i686]
- x86-64 [x86-64]
- ARM CPU-native [nativeARM]
Doxygen source docs
* Fixed and upgraded Doxygen interactive source-code
docs
- Consolidated source-doc scripts to single-file
- Changed images format to SVG (doxygen 1.8.13)
- Greatly improved the main docs entry-page:
index.html
- Moved to 'srcdoc' directory to prevent confusion
with
doxygen User-Docs
- Run: /scripts/srcdoc/gen_doxygen_srcdoc.sh
fonts
* fix slow start caused by font fixed/proportional
enumeration
- fixed evaluator moved to separate thread during
first
instantiation of Font_Browser class. Actual time
to enumerate
2650 fonts on an i5 3.2 GHz system is
approximately 25
seconds. Background process does not effect UI,
but selection
of fixed fonts in font browser dialog may not show
all of the
fixed fonts until the evaluator thread exits.
PSK Reporter query
* request fails to pskreporter
- caused by hard coded request to NWS
- pskreporter site very slow in responding
to connect request; caused UI to freeze.
. change request process to separate thread
Logbook compile warning
* replace strncpy with "*comma_position ="
73, David, W1HKJ
|
|
|
