V4 FX12 and PLB TEMAC: no space for user logic?
Hi Mohammad,
I've just had a look at the messages you generously posted in the ml about your experience with linux on V4 FX12 FPGA. I'd like to ask your opinion about FX12 practical usability in this context (gigabit PLB TEMAC/linux). In this message
http://article.gmane.org/gmane.linux.ports.ppc.embedded/16816
you say the device is completely full. If I understand correctly your system provides just the devices required to run the bandwidth test so it seems there is no room for user logic (I think you did not even add the memory controller required to access the NOR Flash containing bootloader, kernel image and root fs that is clearly mandatory for standalone product). Is that true? If it is, this limits a lot the flexibility of this architecture in this configuration. What do you think?
Regards,
llandre
DAVE Electronics System House - R&D Department
web: http://www.dave.eu email: r&[email protected]
Hi ,
I am now developing my own IPs and when I try to synthesize normally I get an error because the router finds no place to fit everything. I have resynthesized using the option:
XIL_PAR_ENABLE_LEGALIZER=1 ( loading the program with that variable on > XIL_PAR_ENABLE_LEGALIZER=1 xps). It took so much time because it uses a much more consuming algorithm but I managed to build it correctly.
Regards,
/A
Quoted 24 lines · llandre
Hi Mohammad,
I've just had a look at the messages you generously posted in the ml about your experience with linux on V4 FX12 FPGA. I'd like to ask your opinion about FX12 practical usability in this context (gigabit PLB TEMAC/linux). In this message
http://article.gmane.org/gmane.linux.ports.ppc.embedded/16816
you say the device is completely full. If I understand correctly your system provides just the devices required to run the bandwidth test so it seems there is no room for user logic (I think you did not even add the memory controller required to access the NOR Flash containing bootloader, kernel image and root fs that is clearly mandatory for standalone product). Is that true? If it is, this limits a lot the flexibility of this architecture in this configuration. What do you think?
Regards,
llandreDAVE Electronics System House - R&D Department
web: http://www.dave.eu email: r&[email protected]_______________________________________________
Linuxppc-embedded mailing list
[email protected]
https://ozlabs.org/mailman/listinfo/linuxppc-embedded
llandre,
Our tests show that V4 FX 12 is really a small device. The FPGA was completely full with these set of modules:
1- CPU core and related circuits, 2- PLB TEMAC , 3- DDR SDRAM Controller , 4- PLB BRAM IF , 5- Bridges , 6- OPB UART and 7- OPB EMC controller (Used for interfacing to flash chips )
The synthesis tool we used during our tests was : XST
it seems that there are some solutions to have a more optimized implementation on FX-12, personally i visited one of the members here who claimed that he has two PLB TEMACs enabled on one FX-12 FPGA. I believe that this is not possible. may be if one use better synthesis tools such as Synplify he may get slightly better results.
any how, in my idea FX-12 is not suitable for serious projects, it is suitable only for beginning phases of a project and the research phase. We are now focusing on ML410, a great board by Xiling, featuring on FX-60 FPGA while keeps costs low.
any questions are welcom,
thanks,
Mohammad.
Quoted 26 lines
Date: Wed, 13 Feb 2008 10:32:58 +0100
From: r&[email protected]
To: [email protected]
CC: [email protected]
Subject: V4 FX12 and PLB TEMAC: no space for user logic?Hi Mohammad,
I've just had a look at the messages you generously posted in the ml about your experience with linux on V4 FX12 FPGA.
I'd like to ask your opinion about FX12 practical usability in this context (gigabit PLB TEMAC/linux).
In this messagehttp://article.gmane.org/gmane.linux.ports.ppc.embedded/16816
you say the device is completely full. If I understand correctly your system provides just the devices required to run the bandwidth test so it seems there is no room for user logic (I think you did not even add the memory controller required to access the NOR Flash containing bootloader, kernel image and root fs that is clearly mandatory for standalone product). Is that true? If it is, this limits a lot the flexibility of this architecture in this configuration. What do you think?
Quoted 6 lines
Regards,
llandreDAVE Electronics System House - R&D Department
web: http://www.dave.eu email: r&[email protected]
_________________________________________________________________