This is your reminder that we will have our first reading group today. Andy is presenting two interesting papers from ISCA that have direct bearing on our RCC Project. Please read the papers in advance and send me the answer to these two questions in direct email.
1) in the Saidi paper, how do they validate that their simulations are correct? to what degree is this scientifically sound? is the best one can do?
2) is the Rigel programming model closer to a message-passing model or a shared memory model?
For the group meeting, please reply to this announcement with your status report.
Thanks!
Ron
Last Week:
Added Interrupts to the LL_DMA Core for Shan/Bin
Built Head Node and Disk Nodes for Blast IPDPS Paper
Updated Router to handle EOF_N and SOF_N better
Fixed Wishbone Builder Script to support >8 Cores
This week:
XilKernel RedSharc System
Help Shan fix bug in B-Cast System
Test / Debug Blast IPDPS Multi-Node System
11.2 is installed on homer and marge. Everything is peachy except for the JTAG cables. But for now you can use 10.1 xmd and impact on marge with 11.2 bitstreams and elf files.
Experimenting with local user/tool installs to see if builds go faster.
REDSHARC / RAW paper hardware is behind schedule. Needs to be finished by this Friday (9/11) at the latest.
Last week:
Developed network device driver for AIREN network. The polling method works for transition and reception on two-node test. But from the efficiency point of view, the polling method is not practical enough when sending or receiving short messages.
This week:
Add interrupt to network device driver
Test “ping” & “telnet” command on two nodes
Last week:
Worked on Bcast, including hardware and software. The system is working in standalone and under Linux, but not perfectly.
This week:
Fix the bug of Bcast and clean up the design
Test the system and get some data
Start Reduce (if things go smoothly)
Completed the design for the force compute engine.
Tested the design, a few bugs that need to be fixed.
Started writing my Thesis.
To do:
Complete the chapter “Background”.
fix the bugs and test for speedup.
add the distance compute part to this core
Last week:
Completed block device driver for hwfs core and mounted a file system on it.
Currently using a basic request function for a single I/O .
To do:
Short list applications for the class projects.
Get mosaic access for the lxs machine with gpu and install cuda on it.
Last week:
Read up on some stuff for the EM project
This week:
Hope to have PRC Gate 1 ready
Last week:
Worked on the RTR system hardware, still has bugs though.
Started writing ideas discussed in last meeting.
Waiting for 11.2 PR license…
This week:
Investigate coregen for DSP type applications on the 510.
write some more…
look into the existing JHDL, and try to build up on it.
Last Week
Building some more HW components for MD App.
This Week
1. Literature Survey for MD
2. Fixing some HW bugs
Last week:
got the smith-waterman core working with standalone app.
Currently working on integrating it with a linux system.
Cross compiling the (UVa)FASTA software to work on the powerpc’s.
beginning to write thesis.
Last Week:
Tested Some CUDA applications on GPU including:
1. Monte Carlo Option Pricing
2. Matrix Multiplication
3. MersenneTwister random number generator
To do:
Make a detailed plan to map CUDA to FPGA, learn Opencl application.