Re: [casper] Problem using the ASIAA 5Gsps chip in design
Hi Dave I tried your suggestion to reset back to the commit at hand. I tried recompiling the simple ASIAA ADC design alone. I get further along this time, but I get the following error: XPS% Loading xmp file system.xmp ERROR:EDK - This project was created with EDK version 14.6, but the installed version of EDK tools is 14.5 ERROR:EDK - while running revup So my Xilinx tool set is indeed 14.5. I even tried to create a whole new project from scratch with a XSG config object, a System generator and the ASIAA ADC. But it still produces the same error. Perhaps this is a simpler fix? Looks like the EDK version is hard-coded somewhere in the mlib-devel area. Do you have suggestions on what to do here? Thanks! Gopal On Sat, May 10, 2014 at 2:15 PM, David MacMahon dav...@astro.berkeley.eduwrote: Hi, Gopal, It doesn't look like there are any relevant changes for you since (and including) the suspect commit, so you could just reset your local mlib_devel back to the commit just prior to commit e419ce2 with this command: $ git reset --hard e419ce2^ Note the '^' at the end, that's very important. Dave On May 10, 2014, at 8:43 AM, Gopal Narayanan wrote: Dave, Is there a simple fix to xps_library/detokenize.m in the SKA mlib_devel repository? Should I just use the detokenize.m script from the SMA git branch? Should the calculation not be based on app_clk_rate but rather on 100 MHz? Thanks! Gopal On Fri, May 9, 2014 at 4:53 PM, David MacMahon dav...@astro.berkeley.edu wrote: I think Gopal's problem with the ska-sa mlib_devel is due to commit e419ce2. This commit changed the calculation of the clock factors in xps_library/detokenize.m to be based on app_clk_rate (250 MHz in Gopal's case) rather than 100 MHz. These clock factors are passed to the roach_infrastructure HDL via parameters that get defined in system.mhs. Dave -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003 -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
On May 12, 2014, at 11:34 AM, Charles Coldwell wrote: On Mon, May 12, 2014 at 2:19 PM, David MacMahon dav...@astro.berkeley.edu wrote: This version info seems to not really be used/useful at all. Agreed, and it seems to be absent from the corresponding files in the sma-wideband fork. It could/should probably be dropped, which would be a better patch. That would be great! (assuming it works, of course) Dave
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Charles, Thanks. This patch worked fine. Gopal On Mon, May 12, 2014 at 1:23 PM, Charles Coldwell coldw...@gmail.comwrote: On Mon, May 12, 2014 at 11:52 AM, Gopal Narayanan go...@astro.umass.edu wrote: Hi Dave I tried your suggestion to reset back to the commit at hand. I tried recompiling the simple ASIAA ADC design alone. I get further along this time, but I get the following error: XPS% Loading xmp file system.xmp ERROR:EDK - This project was created with EDK version 14.6, but the installed version of EDK tools is 14.5 ERROR:EDK - while running revup I had to work around this problem for version 14.4 targeting ROACH2 $ git show 39a088449a2e0261cd84740014b313f704b55ded commit 39a088449a2e0261cd84740014b313f704b55ded Author: Charles M. Coldwell coldw...@gmail.com Date: Tue Apr 29 13:08:42 2014 -0400 Change the ISE version number to the one we have. diff --git a/xps_base/XPS_ROACH2_base/system.xmp.sx475t b/xps_base/XPS_ROACH2_base/system.xmp.sx475t index 88706c5..0c6922e 100644 --- a/xps_base/XPS_ROACH2_base/system.xmp.sx475t +++ b/xps_base/XPS_ROACH2_base/system.xmp.sx475t @@ -1,5 +1,5 @@ -XmpVersion: 14.6 -VerMgmt: 14.6 +XmpVersion: 14.4 +VerMgmt: 14.4 IntStyle:default MHS File:system.mhs MSS File:system.mss Since this file is copied verbatim when building your XPS project (no variable substitution or whatnot), I couldn't think of a better way to do it. -- Charles M. Coldwell, W1CMC Belmont, Massachusetts, New England Turn on, log in, tune out -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Dave, Is there a simple fix to xps_library/detokenize.m in the SKA mlib_devel repository? Should I just use the detokenize.m script from the SMA git branch? Should the calculation not be based on app_clk_rate but rather on 100 MHz? Thanks! Gopal On Fri, May 9, 2014 at 4:53 PM, David MacMahon dav...@astro.berkeley.eduwrote: I think Gopal's problem with the ska-sa mlib_devel is due to commit e419ce2. This commit changed the calculation of the clock factors in xps_library/detokenize.m to be based on app_clk_rate (250 MHz in Gopal's case) rather than 100 MHz. These clock factors are passed to the roach_infrastructure HDL via parameters that get defined in system.mhs. Dave -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Hi Rurik I have been mostly using the SKA mlib branch until now. When I saw the problem of compiling ROACH2 and ASIAA ADC, I tried to switch to SMA mlib to see if the problem was there as well (my assumption was that since you guys are using the ASIAA ADC actively, you must not have this issue). And indeed while it solved the MMCM clock issue, the fi dependence on Fixed-Point toolbox is a show-stopper there. So I might spend some time cherry-picking the fi dependency out of the SMA branch. I'll let you know the success of this effort. Thanks Gopal On Fri, May 9, 2014 at 3:47 PM, Primiani, Rurik rprimi...@cfa.harvard.eduwrote: Hi Gopal, Your XSG core config parameters look fine to me. I do not understand how the adc0_clk clock frequency should affect the sys_clk-generating MMCM in any way. Can you send me the full output log of your compile? Also, can you send me the XPS_ROACH2_base/system.mhs file from your design directory? As for the fixed point toolbox, I see that commits 1b13e3d, 617ba13, and b3d63ce remove the fi dependence from the FFT blocks. This was after the most recent time I merged with ska-sa and so the sma-wideband FFT still depends on the fixed-point toolbox. You can try cherry-picking those commits to remove the dependence for now. I have tried this but I have not tested it. Unfortunately merging the two forks at this point would require quite a bit of careful thought since there are many conflicts that would need to be worked out. I will give this a shot since this needed to be done eventually. Best, Rurik On Fri, May 9, 2014 at 2:37 PM, Gopal Narayanan go...@astro.umass.edu wrote: Rurik For the MMCM error, here are my XSG core config mask parameters: Hardware platform: ROACH2:sx475t User IP Clock Source: adc0_clk User IP Clock Rate: 250 MHz Sample Period: 1 Synthesis Tool: XST Yes, I normally use the ska-sa branch, and I have compiled the tutorial 3 spectrometer using other boards (KatADC or iADC) successfully without needing the Fixed point Toolbox. Thanks! Gopal On Fri, May 9, 2014 at 2:24 PM, Primiani, Rurik rprimi...@cfa.harvard.edu wrote: Hi Gopal, The MMCM error you pasted above has nothing to do with the ADC5g block. It is instead coming from the roach_infrastructure pcore which generates the sys_clk signal. What are your XSG core config mask parameters set to? I'm not sure how the sma_wideband repository differs from ska-sa in terms of the FFT and fixed point toolbox. The two forks have diverged quite a bit and should probably be merged. Perhaps one of the divergences is that the fixed point blocks have been replaced in ska-sa but I am unaware of such a change. Have you tried building your FFT module using the ska-sa fork (without the ADC5g if need be)? Thanks, Rurik On Fri, May 9, 2014 at 2:09 PM, Gopal Narayanan go...@astro.umass.edu wrote: Hi Casperites, I am trying to use the ASIAA ADC chip in a simulink design. To start with all I have in my design is the chip operating in two channel (AC mode), with 1:1 demux on ZDOK 0 with a clock rate of 2 GSps. I just terminated all outputs of the ADC just for this design. I am using a ROACH2 board in my design. I get the following error when compiling: ERROR:LIT:667 - Block 'MMCM_ADV symbol physical_group_infrastructure_inst/infrastructure_inst/sys_clk_fb_int/infrastructure_inst/infrastructure_inst/MMCM_BASE_sys_clk' has its target frequency, FVCO, out of range. Valid FVCO range for speed grade -1 is 600MHz - 1200MHz. The computed FCVO is a function of the input frequency CLKIN1_PERIOD, the division factor DIVCLK_DIVIDE, and the CLKFBOUT_MULT_F attribute (FVCO = 1000*CLKFBOUT_MULT_F/(CLKIN1_PERIOD*DIVCLK_DIVIDE)). The CLKIN_PERIOD attribute may have been set by ngdbuild based on the user specified PERIOD constraint. The current calculated FVCO is 400.00 MHz. Reference the V6 architecture Users Guide or search the Xilinx Answer Records database for the error code. Errors found during logical drc. I am using Matlab R2012b and XSG 14.5. This is when I am using the SKA mlib-devel branch. Here are the details of the mlib git branch: $ git rev-parse --short HEAD b2d72d4 $ git remote -v origin https://github.com/ska-sa/mlib_devel.git (fetch) origin https://github.com/ska-sa/mlib_devel.git (push) I tried to use the SMA mlib_devel branch instead. Here are the details on the SMA mlib branch I used: $ git rev-parse --short HEAD d4954fd $ git remote -v origin https://github.com/sma-wideband/mlib_devel.git (fetch) origin https://github.com/sma-wideband/mlib_devel.git (push) With the SMA mlib devel I am able to compile the simple ASIAA chip design mentioned above. However, for more complicated designs (I'm trying to put
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Hi, Gopal, It doesn't look like there are any relevant changes for you since (and including) the suspect commit, so you could just reset your local mlib_devel back to the commit just prior to commit e419ce2 with this command: $ git reset --hard e419ce2^ Note the '^' at the end, that's very important. Dave On May 10, 2014, at 8:43 AM, Gopal Narayanan wrote: Dave, Is there a simple fix to xps_library/detokenize.m in the SKA mlib_devel repository? Should I just use the detokenize.m script from the SMA git branch? Should the calculation not be based on app_clk_rate but rather on 100 MHz? Thanks! Gopal On Fri, May 9, 2014 at 4:53 PM, David MacMahon dav...@astro.berkeley.edu wrote: I think Gopal's problem with the ska-sa mlib_devel is due to commit e419ce2. This commit changed the calculation of the clock factors in xps_library/detokenize.m to be based on app_clk_rate (250 MHz in Gopal's case) rather than 100 MHz. These clock factors are passed to the roach_infrastructure HDL via parameters that get defined in system.mhs. Dave -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
[casper] Problem using the ASIAA 5Gsps chip in design
Hi Casperites, I am trying to use the ASIAA ADC chip in a simulink design. To start with all I have in my design is the chip operating in two channel (AC mode), with 1:1 demux on ZDOK 0 with a clock rate of 2 GSps. I just terminated all outputs of the ADC just for this design. I am using a ROACH2 board in my design. I get the following error when compiling: ERROR:LIT:667 - Block 'MMCM_ADV symbol physical_group_infrastructure_inst/infrastructure_inst/sys_clk_fb_int/infrastructure_inst/infrastructure_inst/MMCM_BASE_sys_clk' has its target frequency, FVCO, out of range. Valid FVCO range for speed grade -1 is 600MHz - 1200MHz. The computed FCVO is a function of the input frequency CLKIN1_PERIOD, the division factor DIVCLK_DIVIDE, and the CLKFBOUT_MULT_F attribute (FVCO = 1000*CLKFBOUT_MULT_F/(CLKIN1_PERIOD*DIVCLK_DIVIDE)). The CLKIN_PERIOD attribute may have been set by ngdbuild based on the user specified PERIOD constraint. The current calculated FVCO is 400.00 MHz. Reference the V6 architecture Users Guide or search the Xilinx Answer Records database for the error code. Errors found during logical drc. I am using Matlab R2012b and XSG 14.5. This is when I am using the SKA mlib-devel branch. Here are the details of the mlib git branch: $ git rev-parse --short HEAD b2d72d4 $ git remote -v origin https://github.com/ska-sa/mlib_devel.git (fetch) origin https://github.com/ska-sa/mlib_devel.git (push) I tried to use the SMA mlib_devel branch instead. Here are the details on the SMA mlib branch I used: $ git rev-parse --short HEAD d4954fd $ git remote -v origin https://github.com/sma-wideband/mlib_devel.git (fetch) origin https://github.com/sma-wideband/mlib_devel.git (push) With the SMA mlib devel I am able to compile the simple ASIAA chip design mentioned above. However, for more complicated designs (I'm trying to put together a spectrometer with the ASIAA chip) when using FFT modules, the SMA mlib branch seems to need the Matlab Fixed-Point toolbox, which I don't have a license for. I came across the following thread in the casper mailing lists about the Fixed Point toolbox: https://www.mail-archive.com/casper@lists.berkeley.edu/msg04488.html Anybody have some ideas on what may be causing the MMCM_ADV error seen above or to circumvent the need for a FP toolbox license? Thanks! Gopal -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Hi Gopal, The MMCM error you pasted above has nothing to do with the ADC5g block. It is instead coming from the roach_infrastructure pcore which generates the sys_clk signal. What are your XSG core config mask parameters set to? I'm not sure how the sma_wideband repository differs from ska-sa in terms of the FFT and fixed point toolbox. The two forks have diverged quite a bit and should probably be merged. Perhaps one of the divergences is that the fixed point blocks have been replaced in ska-sa but I am unaware of such a change. Have you tried building your FFT module using the ska-sa fork (without the ADC5g if need be)? Thanks, Rurik On Fri, May 9, 2014 at 2:09 PM, Gopal Narayanan go...@astro.umass.edu wrote: Hi Casperites, I am trying to use the ASIAA ADC chip in a simulink design. To start with all I have in my design is the chip operating in two channel (AC mode), with 1:1 demux on ZDOK 0 with a clock rate of 2 GSps. I just terminated all outputs of the ADC just for this design. I am using a ROACH2 board in my design. I get the following error when compiling: ERROR:LIT:667 - Block 'MMCM_ADV symbol physical_group_infrastructure_inst/infrastructure_inst/sys_clk_fb_int/infrastructure_inst/infrastructure_inst/MMCM_BASE_sys_clk' has its target frequency, FVCO, out of range. Valid FVCO range for speed grade -1 is 600MHz - 1200MHz. The computed FCVO is a function of the input frequency CLKIN1_PERIOD, the division factor DIVCLK_DIVIDE, and the CLKFBOUT_MULT_F attribute (FVCO = 1000*CLKFBOUT_MULT_F/(CLKIN1_PERIOD*DIVCLK_DIVIDE)). The CLKIN_PERIOD attribute may have been set by ngdbuild based on the user specified PERIOD constraint. The current calculated FVCO is 400.00 MHz. Reference the V6 architecture Users Guide or search the Xilinx Answer Records database for the error code. Errors found during logical drc. I am using Matlab R2012b and XSG 14.5. This is when I am using the SKA mlib-devel branch. Here are the details of the mlib git branch: $ git rev-parse --short HEAD b2d72d4 $ git remote -v origin https://github.com/ska-sa/mlib_devel.git (fetch) origin https://github.com/ska-sa/mlib_devel.git (push) I tried to use the SMA mlib_devel branch instead. Here are the details on the SMA mlib branch I used: $ git rev-parse --short HEAD d4954fd $ git remote -v origin https://github.com/sma-wideband/mlib_devel.git (fetch) origin https://github.com/sma-wideband/mlib_devel.git (push) With the SMA mlib devel I am able to compile the simple ASIAA chip design mentioned above. However, for more complicated designs (I'm trying to put together a spectrometer with the ASIAA chip) when using FFT modules, the SMA mlib branch seems to need the Matlab Fixed-Point toolbox, which I don't have a license for. I came across the following thread in the casper mailing lists about the Fixed Point toolbox: https://www.mail-archive.com/casper@lists.berkeley.edu/msg04488.html Anybody have some ideas on what may be causing the MMCM_ADV error seen above or to circumvent the need for a FP toolbox license? Thanks! Gopal -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Rurik For the MMCM error, here are my XSG core config mask parameters: Hardware platform: ROACH2:sx475t User IP Clock Source: adc0_clk User IP Clock Rate: 250 MHz Sample Period: 1 Synthesis Tool: XST Yes, I normally use the ska-sa branch, and I have compiled the tutorial 3 spectrometer using other boards (KatADC or iADC) successfully without needing the Fixed point Toolbox. Thanks! Gopal On Fri, May 9, 2014 at 2:24 PM, Primiani, Rurik rprimi...@cfa.harvard.eduwrote: Hi Gopal, The MMCM error you pasted above has nothing to do with the ADC5g block. It is instead coming from the roach_infrastructure pcore which generates the sys_clk signal. What are your XSG core config mask parameters set to? I'm not sure how the sma_wideband repository differs from ska-sa in terms of the FFT and fixed point toolbox. The two forks have diverged quite a bit and should probably be merged. Perhaps one of the divergences is that the fixed point blocks have been replaced in ska-sa but I am unaware of such a change. Have you tried building your FFT module using the ska-sa fork (without the ADC5g if need be)? Thanks, Rurik On Fri, May 9, 2014 at 2:09 PM, Gopal Narayanan go...@astro.umass.edu wrote: Hi Casperites, I am trying to use the ASIAA ADC chip in a simulink design. To start with all I have in my design is the chip operating in two channel (AC mode), with 1:1 demux on ZDOK 0 with a clock rate of 2 GSps. I just terminated all outputs of the ADC just for this design. I am using a ROACH2 board in my design. I get the following error when compiling: ERROR:LIT:667 - Block 'MMCM_ADV symbol physical_group_infrastructure_inst/infrastructure_inst/sys_clk_fb_int/infrastructure_inst/infrastructure_inst/MMCM_BASE_sys_clk' has its target frequency, FVCO, out of range. Valid FVCO range for speed grade -1 is 600MHz - 1200MHz. The computed FCVO is a function of the input frequency CLKIN1_PERIOD, the division factor DIVCLK_DIVIDE, and the CLKFBOUT_MULT_F attribute (FVCO = 1000*CLKFBOUT_MULT_F/(CLKIN1_PERIOD*DIVCLK_DIVIDE)). The CLKIN_PERIOD attribute may have been set by ngdbuild based on the user specified PERIOD constraint. The current calculated FVCO is 400.00 MHz. Reference the V6 architecture Users Guide or search the Xilinx Answer Records database for the error code. Errors found during logical drc. I am using Matlab R2012b and XSG 14.5. This is when I am using the SKA mlib-devel branch. Here are the details of the mlib git branch: $ git rev-parse --short HEAD b2d72d4 $ git remote -v origin https://github.com/ska-sa/mlib_devel.git (fetch) origin https://github.com/ska-sa/mlib_devel.git (push) I tried to use the SMA mlib_devel branch instead. Here are the details on the SMA mlib branch I used: $ git rev-parse --short HEAD d4954fd $ git remote -v origin https://github.com/sma-wideband/mlib_devel.git (fetch) origin https://github.com/sma-wideband/mlib_devel.git (push) With the SMA mlib devel I am able to compile the simple ASIAA chip design mentioned above. However, for more complicated designs (I'm trying to put together a spectrometer with the ASIAA chip) when using FFT modules, the SMA mlib branch seems to need the Matlab Fixed-Point toolbox, which I don't have a license for. I came across the following thread in the casper mailing lists about the Fixed Point toolbox: https://www.mail-archive.com/casper@lists.berkeley.edu/msg04488.html Anybody have some ideas on what may be causing the MMCM_ADV error seen above or to circumvent the need for a FP toolbox license? Thanks! Gopal -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003 -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Hi Gopal, Your XSG core config parameters look fine to me. I do not understand how the adc0_clk clock frequency should affect the sys_clk-generating MMCM in any way. Can you send me the full output log of your compile? Also, can you send me the XPS_ROACH2_base/system.mhs file from your design directory? As for the fixed point toolbox, I see that commits 1b13e3d, 617ba13, and b3d63ce remove the fi dependence from the FFT blocks. This was after the most recent time I merged with ska-sa and so the sma-wideband FFT still depends on the fixed-point toolbox. You can try cherry-picking those commits to remove the dependence for now. I have tried this but I have not tested it. Unfortunately merging the two forks at this point would require quite a bit of careful thought since there are many conflicts that would need to be worked out. I will give this a shot since this needed to be done eventually. Best, Rurik On Fri, May 9, 2014 at 2:37 PM, Gopal Narayanan go...@astro.umass.edu wrote: Rurik For the MMCM error, here are my XSG core config mask parameters: Hardware platform: ROACH2:sx475t User IP Clock Source: adc0_clk User IP Clock Rate: 250 MHz Sample Period: 1 Synthesis Tool: XST Yes, I normally use the ska-sa branch, and I have compiled the tutorial 3 spectrometer using other boards (KatADC or iADC) successfully without needing the Fixed point Toolbox. Thanks! Gopal On Fri, May 9, 2014 at 2:24 PM, Primiani, Rurik rprimi...@cfa.harvard.edu wrote: Hi Gopal, The MMCM error you pasted above has nothing to do with the ADC5g block. It is instead coming from the roach_infrastructure pcore which generates the sys_clk signal. What are your XSG core config mask parameters set to? I'm not sure how the sma_wideband repository differs from ska-sa in terms of the FFT and fixed point toolbox. The two forks have diverged quite a bit and should probably be merged. Perhaps one of the divergences is that the fixed point blocks have been replaced in ska-sa but I am unaware of such a change. Have you tried building your FFT module using the ska-sa fork (without the ADC5g if need be)? Thanks, Rurik On Fri, May 9, 2014 at 2:09 PM, Gopal Narayanan go...@astro.umass.edu wrote: Hi Casperites, I am trying to use the ASIAA ADC chip in a simulink design. To start with all I have in my design is the chip operating in two channel (AC mode), with 1:1 demux on ZDOK 0 with a clock rate of 2 GSps. I just terminated all outputs of the ADC just for this design. I am using a ROACH2 board in my design. I get the following error when compiling: ERROR:LIT:667 - Block 'MMCM_ADV symbol physical_group_infrastructure_inst/infrastructure_inst/sys_clk_fb_int/infrastructure_inst/infrastructure_inst/MMCM_BASE_sys_clk' has its target frequency, FVCO, out of range. Valid FVCO range for speed grade -1 is 600MHz - 1200MHz. The computed FCVO is a function of the input frequency CLKIN1_PERIOD, the division factor DIVCLK_DIVIDE, and the CLKFBOUT_MULT_F attribute (FVCO = 1000*CLKFBOUT_MULT_F/(CLKIN1_PERIOD*DIVCLK_DIVIDE)). The CLKIN_PERIOD attribute may have been set by ngdbuild based on the user specified PERIOD constraint. The current calculated FVCO is 400.00 MHz. Reference the V6 architecture Users Guide or search the Xilinx Answer Records database for the error code. Errors found during logical drc. I am using Matlab R2012b and XSG 14.5. This is when I am using the SKA mlib-devel branch. Here are the details of the mlib git branch: $ git rev-parse --short HEAD b2d72d4 $ git remote -v origin https://github.com/ska-sa/mlib_devel.git (fetch) origin https://github.com/ska-sa/mlib_devel.git (push) I tried to use the SMA mlib_devel branch instead. Here are the details on the SMA mlib branch I used: $ git rev-parse --short HEAD d4954fd $ git remote -v origin https://github.com/sma-wideband/mlib_devel.git (fetch) origin https://github.com/sma-wideband/mlib_devel.git (push) With the SMA mlib devel I am able to compile the simple ASIAA chip design mentioned above. However, for more complicated designs (I'm trying to put together a spectrometer with the ASIAA chip) when using FFT modules, the SMA mlib branch seems to need the Matlab Fixed-Point toolbox, which I don't have a license for. I came across the following thread in the casper mailing lists about the Fixed Point toolbox: https://www.mail-archive.com/casper@lists.berkeley.edu/msg04488.html Anybody have some ideas on what may be causing the MMCM_ADV error seen above or to circumvent the need for a FP toolbox license? Thanks! Gopal -- Gopal Narayanan Ph #: (413) 545 0925 Department of Astronomy e-mail: go...@astro.umass.edu University of Massachusetts Amherst MA 01003 -- Gopal Narayanan
Re: [casper] Problem using the ASIAA 5Gsps chip in design
I think Gopal's problem with the ska-sa mlib_devel is due to commit e419ce2. This commit changed the calculation of the clock factors in xps_library/detokenize.m to be based on app_clk_rate (250 MHz in Gopal's case) rather than 100 MHz. These clock factors are passed to the roach_infrastructure HDL via parameters that get defined in system.mhs. Dave
Re: [casper] Problem using the ASIAA 5Gsps chip in design
Yes, Dave, I think you're right. I'm unclear why the sys_clk MMCM doesn't just have hard-coded clock factors. Its input clock never changes and it's always outputting 100 MHz and 200 MHz. On the other hand, the aux_clk MMCM does seem to have its factors hard-coded even though aux_clk can be an arbitrary frequency. Am I missing something? Side note: why are these clock factors derived in detokenize.m? Surely there's a more appropriate place to do this? Best, Rurik On Fri, May 9, 2014 at 4:53 PM, David MacMahon dav...@astro.berkeley.edu wrote: I think Gopal's problem with the ska-sa mlib_devel is due to commit e419ce2. This commit changed the calculation of the clock factors in xps_library/detokenize.m to be based on app_clk_rate (250 MHz in Gopal's case) rather than 100 MHz. These clock factors are passed to the roach_infrastructure HDL via parameters that get defined in system.mhs. Dave
Re: [casper] Problem using the ASIAA 5Gsps chip in design
On May 9, 2014, at 2:26 PM, Primiani, Rurik wrote: I'm unclear why the sys_clk MMCM doesn't just have hard-coded clock factors. Its input clock never changes and it's always outputting 100 MHz and 200 MHz. I think things are a little gray here. When the user specifies using sys_clk for the User IP Clock Source, then I think sys_clk is in fact used and its MMCM is configured to output the rate specified in User IP Clock Rate rather than a fixed 100 MHz. On the other hand, the aux_clk MMCM does seem to have its factors hard-coded even though aux_clk can be an arbitrary frequency. Am I missing something? I agree that aux_clk's MMCM usage seems suspect. Side note: why are these clock factors derived in detokenize.m? Surely there's a more appropriate place to do this? It does seem like an odd place to do this and I'm sure a more appropriate place could be found, but given that it's been there for a long time and it has worked there for a long time, I'm not sure I'd want to move it just for aesthetics sake. Dave