01-24-2012 10:09 AM
First of all, I have gone through pretty significant steps in trying to debug what is going on. However, my main issue is that if my test bench becomes to complicated...either I get an unexpected error or a gui not responding error. Worse yet, is that the logic internally will beging to act ranodmly (e.g. a singal will randomly change state). In a test bench, when a process kicks off, it prevents another process from continuing to run (it halts it for a pseudo random period of time).
I have submitted a webcase, but i'm trying to see what is going on here has happened to other people. I have had this issue with 12.3, 13.3, and now 13.4 on two different machines. Both are running Win7 Enterprise 64-bit. The largest difference between the two machines was the video card (AMD vs nVidia).
A good thing is that if I run a simple test bench, it appears to be okay. However, I'm in a spartan 6. This is not a small or trivial device and thus for long term verificaiton, a small or trivial test bench will not suffice. I really want to keep a single workflow and not have to have a seperate simulator (like Aldec) and pay addtionally for it. I've been okay not being able to display variables and some of the other limits, but getting different results (like 'stable not working at all 'quiet is working) on the same signal is driving me up a wall. Doing verificaiton by bench testing on 40k lines of VHDL will be the death of me if I can't simulate properly.
02-09-2012 03:54 PM
The issues I had with my ISIM simulation crashing in windows 7 appears to be fixed in the 13.4 release. I was able to run the simulation for quite a period of time in 13.4 after the period that I would get a windows crash in 13.3 or earlier ISE versions.
01-24-2012 11:18 AM
I have also had ISIM crash when using windows 7 with a large simulation test bench. I can run about 8us of simulation before windows 7 crashes with a GUI not responding. I have put up with this ever since going to windows 7 with several versions of ISE. At this point I have setup a dual boot machine with 32bit XP and windows 7. Anytime I wish to run ISIM I reboot to 32bit XP. The exact same test bench will run without crashes in 32bit XP. I will note that 64bit XP will also crash ISIM.
For several releases Xilinx said they would get an analog trace format into ISIM like Modelsim had. They never have and in fact took the note away in the user guide for ISIM that said they would do so in a future release. Many users have asked for this function which Xilinx has never gotten into ISIM.
Because of these I expect that Xilinx will never fix the problems in ISIM like it crashing in windows 7. I will add the just last week I was doing some code editing in ISE 13.4 with windows 7and all of a sudden the ISE editor stopped responding just like does for ISIM.
01-25-2012 02:00 AM
02-09-2012 03:54 PM
The issues I had with my ISIM simulation crashing in windows 7 appears to be fixed in the 13.4 release. I was able to run the simulation for quite a period of time in 13.4 after the period that I would get a windows crash in 13.3 or earlier ISE versions.
03-22-2012 02:44 PM
Glad you're up and running Jay. Afraid to report that I'm having the same issues as you they won't go away with the update to 13.4. Like yourself, I'm simulating a Spartan 6 device. My simulation is cut right down to contain only the block of code I'm currently working on, but it's still giving the isimgui.exe has stopped working error from windows. The Xilinx error is "A Xilinx Application has encountered an unexpected error. It is recommended that you save any unsaved work in the event that this condition persist. For further assistance, please consult the Answers Database and other online resources at http://support.xilinx.com."
So afraid it's still broken. I'm on the illfated 64-bit windows 7 machine. Any help please Xilinx? I can't design this without a simulator. Incidentally, why did Xilinx move away from Modelsim towards ISim? I wouldn't mind but there are so many features that are missing. No signal spy. No variables... Broken attribute handling... It's a real shame. It's forcing users to spend money twice.
03-23-2012 10:32 AM
Sometimes it helps to look at Windows Event Log (Control Panel -> Administartive Tools -> Event Viewer) messages and check for (error) events at time you hit the application crash.
Hopefully you can see some pattern of events that you can connect with application chrash.
04-02-2012 09:28 PM
Interesting idea. Here's the event log for the failure:
Faulting application name: isimgui.exe, version: 0.0.0.0, time stamp: 0x4f082a27
Faulting module name: libSimBridgeData.dll, version: 0.0.0.0, time stamp: 0x4f07b60e
Exception code: 0xc0000005
Fault offset: 0x000000000000a5cf
Faulting process id: 0x1988
Faulting application start time: 0x01cd114e723b6863
Faulting application path: C:\Xilinx\13.4\ISE_DS\ISE\bin\nt64\unwrapped\isimgui.exe
Faulting module path: C:\Xilinx\13.4\ISE_DS\ISE\lib\nt64\libSimBridgeData.dll
Report Id: 90ac342d-7d43-11e1-9a4f-782bcbae99a0
Xilinx? Does this mean anything to you. ISim is still broken... I haven't moved back to Modelsim yet, but seriously, it's looking more and more like a good idea...
04-03-2012 05:57 AM
05-04-2012 03:26 AM
By the same if I am simulating simple design no problem, but with growing number of loged signals and also complexity of design the time is shorter before crash. One of designs I have with lot of signals si crashing after 1-2us of simulation. The only way to simulate is to remove most of signals and look only on few. However it is hard to find bug in comlicated design.
05-24-2012 09:25 AM
I know, it's a bit unconstructive, but:
It's a pitty that ISim is still so unstable eventhough it is none of the first versions (v13...) . Sometimes it's a real challange just to simulate parts of a virtex 6 project.
05-27-2012 10:50 PM
I have this kind of crash happens on Win7 64-bit machine too. The version I'm using is latest one ISE 14.1. Does anyone try ISE14.1 already ?
06-19-2012 02:41 PM
Yes, I am using the latest 14.1 version with the illfated Win7 64-bit configuration. Why hasn't Xilinx rectified this issue even though they planned to resolve it in 13.4 itself ?!
12-04-2012 06:58 AM
I have the same problem in 14.3.
My crash log looks very similar to the one previously posted in this thread.
It seems to somehow be related to the number of signals, or signal groups, in the waveform viewer (at least in my case). I've found that by reducing the number of signals, it can, sometimes, work better. But it's very frustrating when you're trying to debug something. You kick off a simulation, that can take many minutes before it finally is doing something (for example, finally get out of reset), you finally start to see some signals toggling, and then......ISIM crashes! So, let's see, delete some signals, restart, hope it doesn't crash. If it doesn't, oh joy good, then add more signals that you needed, but wait, don't add too many signals, better delete some of the ones you already have, or that nasty crash will come back. Restart the whole thing, wait, wait, repeat, crash, repeat, wait, wait, pull your hair out and curse.
After years, I'm getting sick of this, and I'm not a webpack user, my company is paying for several licenses, using the "real" ISIM. This is not acceptable.
Dan
12-04-2012 06:18 PM
12-05-2012 04:01 AM
12-14-2012 06:12 AM
I just had the same problem in Xilinx ISE 12.3 on a Win7 64 bit computer. SOLUTİON: I solved the problem by changing the properties of the isimgui.exe. I set the properties of isimgui.exe as : run in administrator mode and run in WinXP SP3 mode. It solved my problem, now it does not crash with large simulation sets.
12-14-2012 06:30 AM
That makes sense because the problem seems to be something to do with Windows 7 Professional 64 bit. Unfortunately, my simulation won't run with only 4 gig of memory so I can't use Win XP.
I opened a case about this and basically, they couldn't reproduce it under Windows 7 Enterprise 64 bit, however, they were only running with 8gig and my simulation needs 10-12gig, so basically I believe it was self-limiting such that the problem didn't happen. Also, they got a warning about insufficient memory whereas I did not.
In the end, I reproduced this on 3 differenet Windows 7 Pro 64 bit machines here, with 24 to 32gig of memory. The memory is not a problem since it never 'exploded' or anything like that. Isimgui.exe just hangs.
I also tried it under Linux (Centos) 64 bit and it doesn't happen there.
Xilinx filed a CR, which is what they've done for the last 2-3 years with this problem, and then nothing happens. When looking through my case history, I realized that I've opened a case for this type of problem at least 4-6 times previously, starting with ISE 12.4.
Sometimes the problem will "go away" for a while, but then comes back. Some things seem to help, some of the time. Such as, start the simulation with the wcfg file open, with the signals you want to see, stop the simulation, close the wcfg window and then continue the simulation to the end, and open the wcfg file to see the signals. But that doesn't work much of the time. ISIM just crashes or turns the signals to garbage, when you open the wcfg file.
However, I do get to hear repeatedly how much better the simulation under Vivado will be. Of course that's useless for all designs, except Series-7 FPGAs, which are barely available for a year.
It seems to me that the resources are shifting to Vivado, leaving Virtex-6/Spartan-6 and earlier customers in the cold. I've already heard that only minimal changes will be made to the existing tools, or only what is really necessary. All of that makes the competition look more attractive....
Dan
12-14-2012 11:55 AM
@dbarrowman wrote:
It seems to me that the resources are shifting to Vivado, leaving Virtex-6/Spartan-6 and earlier customers in the cold. I've already heard that only minimal changes will be made to the existing tools, or only what is really necessary. All of that makes the competition look more attractive....
Preaching to the choir.
12-26-2012 07:12 PM
I have also encountered the same problem too. My PC configuration is Windows 7 Professional 64 bit with 8GB of RAM. The ISIM simply crashed after a short simulation and. the signals turned into unknow state. However, in some rare occasions, the simulation managed to complete with no issue. I had reflected this to Xilinx Field Application Engineer and he could not replicated it under Windows 7 Enterprise 64 bit.
May I ask.
1) How do I know how much memory is needed for my simulation?
2) Is there any work-around solution for this? (e.g switch to Window 7 Enterprise or Linux)
Thanks.
Melvin
12-28-2012 04:49 AM
01-14-2013 05:38 PM
Hi Dan
Thank you for reply. Recently, I filed a webcase with Xilinx and got a reply from them. They acknowledged this issue and claimed that the latest Vivado Simulation (XSIM) has a permanent solution to it. Hope it helps. Will update again if I manage to get the latest ISE 14.4 and run it in XSIM. Thanks.
Regards
Melvin
01-16-2013 06:31 AM
02-04-2013 04:59 AM
I had the same problem with ISE 14.3. So I tried ISE 14.4 and it worked for about 3 weeks and then iSim started to crash again. I did a reinstall and now it is working again, but it will probably start crashing in a few weeks again. Can it be some temporary or buffert file(s) that is growing over time and gets too big and that are not cleaned or removed properly?
Regards,
Gustav
05-31-2013 07:42 AM
Hello,
I have the same problem with ISE 14.4, Windows 7, Lenovo T430s laptop - 64bit machine. I will reinstall ISE to see if it's resolved.
Thanks,
Quoc
06-03-2013 12:43 AM
A work around that has worked for me is to load iSim add all the files to the wave window. Then hit the "Re-launch" button in iSim. If I do this every time I start iSim it dosen't crash.
Regards,
Gustav
06-19-2013 03:54 PM
Im having issues with 14.5. Wasnt having problems untill today i started a system level testbench which are bigger, dont have time to debug issues with isim, but find modelsim or even riviera much better.
from the isim.log file :
ISim log file
Running: C:\\WORK\omniblast\xbuild\xil_ise\omniblast_top_tb_isim_beh.exe -intstyle ise -gui -tclbatch isim.cmd -view C://WORK/omniblast/sim/system_top.wcfg -wdb C:/SCOLE/WORK/omniblast/xbuild/xil_ise/omniblast_top_tb_isim_beh.wdb
ISim P.58f (signature 0x7708f090)
This is a Full version of ISim.
Time resolution is 1 ps
# onerror resume
# run 10 us
Simulator is doing circuit initialization process.
at 0 ps, Instance /omniblast_top_tb/omniblast_top_dut/pkt_system/incoming_buf/ : Warning: There is an 'U'|'X'|'W'|'Z'|'-' in an arithmetic operand, the result will be 'X'(es).
Finished circuit initialization process.
# run 1 ms
The simulator has terminated in an unexpected manner. Please review the ISim log (isim.log) for details.
priceless.... this is the isim.log file. (what a helpful message lol )
10-04-2013 12:02 PM
Thank you!! Finally a workaround that seems to work for me!!. I have been dealing with this issue for a long time.
I am running ISE 14.1 on a Windows 7 Ultimate PC with 32 GB of memory.
01-20-2015 05:19 PM
I'm also experiencing pseudo-random crashes. I'm runing Version 14.2 and it crashes in both 32 and 64-bit versions. And if it doesn't crash I get stuff like this
Or this
The signals causing this issue are 6 by 8 by 84 (4032 STD_LOGIC signals)
01-21-2015 01:28 AM
I also still see these problems, even with ISE 14.7, in a known good working simulation. Maybe it's a memory issue but the PC has a quite large memory.
Just a tip....I have seen cases with ISIM where the way something was written (incorrectly) in VHDL or Verilog resulted in a explosion of memory use. This can happen with any simulator. The code was actually incorrect, however, still valid code for simulation (would never synthesize, at least not into something useful). Unfortunately those kinds of things can be hard to find without some kind of "lint" tool for checking your HDL.
Speaking of lint tools, I would appreciate it if anyone has advice as what's available and good.
Dan
04-16-2015 12:27 AM
there is some other problem too. If you add too many signal to ISIM wavwform and cause cerash and close, in the future run if you delete extra signal, it don't have effect and it will crash with every amount of signal!!