-
Notifications
You must be signed in to change notification settings - Fork 26
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.
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
Expand the size of the matrix #1
Comments
Did you re-compile the overlay?
… On Dec 10, 2018, at 10:30 PM, leankemski ***@***.***> wrote:
I tried to change the dimensions for matrix A and B into 1000*1000, and I just changed the size of matrix in mmilt.hpp and sharedmem.py. It did not work, the result of the matrix was wrong. Here are the codes I changed, the board I used is PYNQ-Z1. Can you tell me what else should I change, thx.
in mmult.hpp
// Dimensions for A and B (which determine C)
#define A_ROWS 1000
#define A_COLS 1000
#define B_ROWS A_COLS
#define B_COLS 1000
in sharedmem.py
__MMULT_A_SHAPE = (1000, 1000)
__MMULT_BT_SHAPE = (1000, 1000)
__MMULT_C_SHAPE = (1000, 1000)
(https://user-images.githubusercontent.com/34908328/49782331-deaeaa80-fd50-11e8-8177-13a968a5ca57.png <https://user-images.githubusercontent.com/34908328/49782331-deaeaa80-fd50-11e8-8177-13a968a5ca57.png>)
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub <#1>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGFUdF3j5mw7pdtileBH5F4PPL2HFd-Pks5u31EBgaJpZM4ZMwES>.
|
Yes, I re-compiled the overlay according to the tutorial. |
Increase the delay in sharedmem.py. I think your matrix is taking longer than 1 second to multiply
… On Dec 12, 2018, at 4:47 PM, leankemski ***@***.***> wrote:
Did you re-compile the overlay?
…
On Dec 10, 2018, at 10:30 PM, leankemski ***@***.**> wrote: I tried to change the dimensions for matrix A and B into 10001000, and I just changed the size of matrix in mmilt.hpp and sharedmem.py. It did not work, the result of the matrix was wrong. Here are the codes I changed, the board I used is PYNQ-Z1. Can you tell me what else should I change, thx. in mmult.hpp // Dimensions for A and B (which determine C) #define A_ROWS 1000 #define A_COLS 1000 #define B_ROWS A_COLS #define B_COLS 1000 in sharedmem.py __MMULT_A_SHAPE = (1000, 1000) __MMULT_BT_SHAPE = (1000, 1000) __MMULT_C_SHAPE = (1000, 1000) (https://user-images.githubusercontent.com/34908328/49782331-deaeaa80-fd50-11e8-8177-13a968a5ca57.png https://user-images.githubusercontent.com/34908328/49782331-deaeaa80-fd50-11e8-8177-13a968a5ca57.png) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1>, or mute the thread https://github.com/notifications/unsubscribe-auth/AGFUdF3j5mw7pdtileBH5F4PPL2HFd-Pks5u31EBgaJpZM4ZMwES.
Yes, I re-compiled the overlay according to the tutorial.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
|
Thank you. I changed delay to 100, it worked for a 300 matrix. But for a matrix of 1000, it should be more longer. I think if I change the type into float, it will be much worse. Will the mmult perform better if I unroll the loop? There is no accerlerlated computation now. |
Hi, I tried to change the dimensions for matrix A and B into 1000*1000, and I just changed the size of matrix in mmilt.hpp and sharedmem.py. It did not work, the result of the matrix was wrong. Here are the codes I changed, the board I used is PYNQ-Z1. Can you tell me what else should I change, thx.
in mmult.hpp
// Dimensions for A and B (which determine C)
#define A_ROWS 1000
#define A_COLS 1000
#define B_ROWS A_COLS
#define B_COLS 1000
in sharedmem.py
__MMULT_A_SHAPE = (1000, 1000)
__MMULT_BT_SHAPE = (1000, 1000)
__MMULT_C_SHAPE = (1000, 1000)
(https://user-images.githubusercontent.com/34908328/49782331-deaeaa80-fd50-11e8-8177-13a968a5ca57.png)
I also want to know if i change the type to float would the ip work correctly?
The text was updated successfully, but these errors were encountered: