Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

One (opencl) kernel to run them all #14

Open
wants to merge 59 commits into
base: fpga
Choose a base branch
from
Open

One (opencl) kernel to run them all #14

wants to merge 59 commits into from

Conversation

scarlehoff
Copy link
Member

There is a mccl.h which is inspired on the header from Xilinx with a few extra goodies.
I am doing all of this in a different folder from FPGA since I don't know how destructive I am being until I don't port the makefile...

@scarlehoff scarlehoff changed the title One (opencl) kernel to rule them all One (opencl) kernel to run them all Sep 26, 2019
@scarrazza
Copy link
Contributor

Looks good.

scarlehoff and others added 23 commits September 26, 2019 22:00
@scarlehoff
make the kernel file a input argument again
2731e28
@scarlehoff
now it compiles
bbd894a
@scarlehoff
now everything works, virtually same code for CPU, GPU and Xilinx
ac089a6
@scarlehoff
small changes
8574ffc
@scarlehoff
CPU and GPU now do the calculation with the random numbers in the ker…
c38fb69 
…nel, much faster. Fpga not working in this commit
@scarlehoff
small cosmetic changes
7072796
@scarlehoff
fixed makefile so it works for FPGA, now it does not forces compilati…
a79652e 
…on every time
@scarlehoff
updated gitignore;
310c2bb
@scarlehoff
go back to just one kernel (but left in the makefile the posibility f…
50465fb 
…or more)
@scarlehoff
small changes
7cde844
@scarlehoff
dropped n_events
ca980e3
@scarrazza
adding mwc64x rng
eb94bea
@scarrazza
moving seed init to events_kernel
56f8a23
@scarlehoff
Merge pull request #16 from N3PDF/mwc64x
e4214ca 
mwc64x RNG
@scarlehoff
separated random and integrand files
de460e8
@scarrazza @scarlehoff
adding RANMAR algorithm
af0fc0a
@scarrazza @scarlehoff
moving rng allocation to events_kernel
d816aa0
@scarlehoff
separate ranmar file
0c037ef
@scarlehoff
separated ranmar
4180a71
@scarlehoff
added documentation to the integrand
e1aa1df
@scarlehoff
Merge pull request #15 from N3PDF/ranmar
dfce1ab 
Ranmar RNG
@scarlehoff
ensure threads is not above number of events
6b9db32
@scarlehoff
added random_bad opencl file, small changes to makefile
7f4464a
@scarlehoff
Copy link
Member Author

I think the problems with the TARGET=hw are not due to the code because I am not able to run the old version either.

@scarrazza
Copy link
Contributor

Yeah, even xbutil validate is failing.

@scarlehoff
some small changes
226591e
@scarlehoff
added memory management
4d5bfb6
@scarlehoff
added RAM example
24d81dc
@scarlehoff
code to print the time
60ef9d1
@scarlehoff
do concurrent reading and writing of the loops
a677609
@scarlehoff
remove boilerplate code unnecesary for ocl
73252ac
@scarlehoff
simplify the makefile a bit more
cd250c1
@scarlehoff
testing dual CU
665c2ed
@scarlehoff
use due computing units
5a80a68
@scarlehoff
added a first try
c10d93e
@scarlehoff
roadend
927c49b
@scarlehoff
resetting
7645dce
@scarlehoff
created a new vegas... works in cpu... pushing to test in dom
7810b2a
@scarlehoff
make sure not to compile the xcl flags for gpu
d7b7907
@scarlehoff
xilinx is a bit more strict than i thought
a0cdd49
@scarlehoff
map each buffer to a different memory bank or the code wont compile
748688a
@scarlehoff
with one kernel 10e5 2.8sec
c2b593a
@scarlehoff
parallelize random generation and kernel running, 50% speed up
1b34ee6
@scarlehoff
hide completely the CPU generation of random points
93f5fd9
@scarlehoff
fixed the order of the loops, almost 50% extra speed up
adf1ce5
@scarlehoff
more than a factor 4 speed when fixing the number of dimensions at co…
14e0250 
…mpile time...
@scarlehoff
change everything to maxdim to let the compiler optimize
74a13f6
@scarlehoff
using 4 cu no difference is seen, impossible to know whether this is …
f7c39b8 
…due to the FPGA or the copying of the results
@scarlehoff
this version is already orders of magnitude faster than CPU
ca7f282
@scarlehoff
the fpga is definitely slower...
245ef52
@scarlehoff
compilable version of the code
ee57157
@scarlehoff
Merge pull request #19 from N3PDF/newVegasOCL
7b1c723 
Vegas: a different approach
@scarlehoff
Copy link
Member Author

We can sort of consider the opencl branch closed tbh...

@scarlehoff scarlehoff mentioned this pull request Dec 4, 2019
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@scarlehoff @scarrazza