From Atul :Oct 1, 2024
From Matthew October 19, 2024
newgrp design_apd
cd /project/mercury
/bin/bash
. /project/mercury/bin/chip_env -proj_dir PN85.3.dft_iter -prog_desc PN85.3
export DROP_DATA=/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft
export MY_BLOCK=<your block name here - see the list below>
export MY_BLOCK_ROOT=$BLOCKPATH/$MY_BLOCK
/project/mercury/bin/mercury_block_setup
cd $GEV_PROJECT_ROOT/user/$USER/$PROJECT_DIR/impl/$MY_BLOCK
cp -rf /project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft/${MY_BLOCK}/${MY_BLOCK}.LEQ_info.tcl scripts/con/.
Blocks Dropped:
hm_ae
hm_be
hm_clipi
hm_clipo
hm_conv_n
hm_conv_s
hm_dl
hm_efr_n
hm_efr_s
hm_fh_n
hm_fh_s
hm_grad
hm_iop
hm_l1dl
hm_l1ul
hm_mac
hm_msix
hm_pss
hm_ulcap
hm_xlgx
pcie_ss_pcie_0
pss_subsys_nokmercury
Block Drop Notes:
1) All pins should be legal except MIN-PIN-SPACING which can be ignored or you can run setPinConstraint -spacing 1 to turn this stupid error off.
2) There are a number of blocks with size only information and Agnel defines how those files are provided below
3) A number of blocks have changed size but we are getting closer to a final floorplan so we can incorporate sizing feedback hopefully going into PN99.
4) The RH Context files should be available in the next week. I will also send out a picture as soon as I have a good one.
5) There is one NEW BLOCK called hm_msix It is very small with three macros and should
build fairly quickly.
6) We are planning to do a formal cubby with these block builds. Please try to have something that could be cubbied by Friday November 1.
###############
Hello Mahesh and Mercury Block Builders,
There was an error in the pin placement scripts that caused pins which are tied or floating at the top level to be placed only on M7/M8 (and in some cases spread across the entire side of a block). I've resolved that problem along with incorporating the requested pin changes as I could for the PCIE_SS and reflecting those onto the PSS_NOKMERCURY block.
You can incorporating these by using agLoadBlockPins BLOCKNAME.ptn. You will need to do this during blocktools/floorplanning.
The files can be found here:
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.pin_update
We are also working on some fixes to constraints to handle these pins better and so we may do another block drop that contains those changes in the coming days.
################################ Wednesday, October 30, 2024 2:01 PM
newgrp design_apd
cd /project/mercury
/bin/bash
. /project/mercury/bin/chip_env -proj_dir PN85.3.dft_iter_newsdc -prog_desc PN85.3
export DROP_DATA=/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.newsdc
export MY_BLOCK=<your block name here - see the list below>
export MY_BLOCK_ROOT=$BLOCKPATH/$MY_BLOCK
/project/mercury/bin/mercury_block_setup
cd $GEV_PROJECT_ROOT/user/$USER/$PROJECT_DIR/impl/$MY_BLOCK
cp -rf /project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.newsdc/${MY_BLOCK}/${MY_BLOCK}.LEQ_info.tcl scripts/con/.
We've updated the constraints based on the floating port data (and possibly some other things) and I've updated the DEF files to have the ports based on the .ptn files I sent out last week.
Any block struggling with the previous drop should probably restart using these constraints/DEF.
The new block drop area can be found here:
INFO [01:27:09] Block Drop Directory: /project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.newsdc
INFO [01:27:09] Blocks Dropped:
INFO [01:27:09] hm_ae
INFO [01:27:09] hm_be
INFO [01:27:09] hm_clipi
INFO [01:27:09] hm_clipo
INFO [01:27:09] hm_conv_n
INFO [01:27:09] hm_conv_s
INFO [01:27:09] hm_dl
INFO [01:27:09] hm_efr_n
INFO [01:27:09] hm_efr_s
INFO [01:27:09] hm_fh_n
INFO [01:27:09] hm_fh_s
INFO [01:27:09] hm_grad
INFO [01:27:09] hm_iop
INFO [01:27:09] hm_l1dl
INFO [01:27:09] hm_l1ul
INFO [01:27:09] hm_mac
INFO [01:27:09] hm_msix
INFO [01:27:09] hm_pss
INFO [01:27:09] hm_ulcap
INFO [01:27:09] hm_xlgx
INFO [01:27:09] pcie_ss_pcie_0
INFO [01:27:09] pss_subsys_nokmercury
Thursday, November 7, 2024 2:35 AM
Mahesh,
Updated timing constraints for pcie_ss_pcie_0 ready in the location below.
These new constraints should fix all the issues mentioned in your initial email except the path from input to memories.
Issues fixed:
- Feedthrough violations : added a feedthrough clock.
- input to reg and reg to out relative to test clocks : added max delay.
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.newsdc_pcie_ss_pcie_0/
Thursday, November 7, 2024 11:29 PM
Hi Mahesh,
I have over-written the clock con file in:
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.newsdc_pcie_ss_pcie_0/
The new file has these general uncertainties:
Port clock uncertainty
axi4_mst* virtual_axi_clk 0.215
axi4_slv* virtual_axi_clk 0.215
sbus2ahb_generic_* virtual_pipe_clk_core* 0.198
s2b_gang* virtual_pipe_clk_core* 0.198
s2b_pipe*P2M* virtual_pipe_clk_core* 0.198
s2b_pipe_*RxData* virtual_rx_wordclk* 0.215
Thanks for pointing this out, we are still working out the bugs on getting more accurate clock uncertainties on ported clocks.
-Woody
Hello Mahesh,
I have an updated version of the context files for PN85.3:
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.rh_context
I tried loading one of these for hm_be against the hm_be floorplan and I see connectivity down to the block. I also noticed that the AP spacing is not legal and so we will likely need to adjust the M19 spacing that we have for the additional VDDM straps. It doesn't create a short though so my hope is that this will be sufficient for us to start running redhawk.
Please let me know if you run into any issues. A block builder can do a sanity check on this by simply doing a defIn on this file on any of their runs in the invs directory and see that the connectivity makes sense.
PN85.3 pad block data drop : Fri, Nov 15, 2024 at 6:21 AM
Hi Mahesh and Tariq,
We recently received floorplan DEF for the Mercury PN85.3 pad blocks.
We've created a block drop here:
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.pad_blocks.rev2
Please use the following variables for building them:
export DROP_DATA=/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.pad_blocks.rev2
export PROJECT_DIR=PN85.3_iter0
This block drop includes data for the following blocks:
hm_io_big
hm_io_e
hm_io_w
Please let me know if anything appears missing.
Furthermore, please let us know who these blocks will be built by, and a schedule for when first-pass AVO data will be ready for them.
newgrp design_apd
cd /project/mercury
/bin/bash
. /project/mercury/bin/chip_env -proj_dir PN85.3.dft_iter0_pad_11152024 -prog_desc PN85.3
export DROP_DATA=/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.dft.pad_blocks.rev2
export MY_BLOCK=<>
export MY_BLOCK_ROOT=$BLOCKPATH/$MY_BLOCK
/project/mercury/bin/mercury_block_setup
cd $GEV_PROJECT_ROOT/user/$USER/$PROJECT_DIR/impl/$MY_BLOCK
Fri, Nov 22, 2024 at 4:07 AM
Hi Mahesh.
We reduced by using the real area of the hm_pss to from 0.252 to 0.174.
set_clock_uncertainty 0.256 -setup -from [get_clocks clk_core_a_div1 ] -to [get_clocks clk_core_a_div2 ]
set_clock_uncertainty 0.174 -hold -from [get_clocks clk_core_a_div1 ] -to [get_clocks clk_core_a_div2 ]
From:
set_clock_uncertainty 0.361 -setup -from [get_clocks clk_core_a_div1 ] -to [get_clocks clk_core_a_div2 ]
set_clock_uncertainty 0.252 -hold -from [get_clocks clk_core_a_div1 ] -to [get_clocks clk_core_a_div2 ]
You can find the new con file for hm_pss in:
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_pss.new_clock_con/hm_pss.clock_con.tcl
Let me know if you have any questions,
Thanks
Agnel
Hi Rahul, : 22 November 2024 23:33
I have updated the clock definitions and placed it in below area
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_io_e.updated_constraints/hm_io_e.sdc
Please let me know in case of any concern.
Thanks & Regards,
Richa
Hi Rahul,
Below is the delivery area for the updated pcie_ss_pcie_0.clock_cons.tcl
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.pcie_ss_pcie_0.new_clock_con
Uncertainty value has changed from 0.198 to 0.177. We can achieve an improvement of only 21ps compared to slack of 502ps.
Thanks & Regards,
Richa
Hi Mahesh,
Please find in /project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_mac_mcp_12042024/hm_mac.sdc with the new MCP with start.
I put also the files :
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_mac_mcp_12042024/hm_mac.clock_con.tcl
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_mac_mcp_12042024/hm_mac.size_ok_list
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.hm_mac_mcp_12042024/customConfigFile.tcl
In PT we cannot see any error or warning at this moment , as it seems that the "-ve cycle adjustment" can be only seen in INNOVUS.
Let me know if you have any questions,
Hi Rahul,
Please find the new SDC to fix the issues that you have reported for hm_io_e and hm_io_big.
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.pad_blocks.updated_constraints_12052024/
Let me know if you have any questions
Fri, Dec 27, 2024 at 12:47 PM
Copied the Context DEF's from my build in to the below area. Use this for the blocks which are having IVD issues due to missing bumps from IO pads. (hm_grad,hm_ae and hm_be) blocks.
/project/mercury/source/PN85.3/80_blockdrop/PN85.3.fp11.san.rh_context_27122024
Note: Before starting the run, just "defIn" this context in Innovus and see the additional bumps which were not seen earlier to confirm