Skip to content

Draft: inline gelu #16

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

Draft
certik wants to merge 6 commits into
base: main
Choose a base branch
from
Draft

Draft: inline gelu #16

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
1d8c856
ffn: Fuse some of the operations
certik Mar 7, 2023
2399607
Implement faster tanh(x) function
certik Mar 7, 2023
6e540ee
Implement fast_gelu
certik Mar 8, 2023
9389638
Inline fast_gelu
certik Mar 8, 2023
fdefc3e
Flip the loop order
certik Mar 8, 2023
9472f6e
Merge branch 'main' into gelu2
certik Mar 17, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 50 additions & 4 deletions gpt2.f90
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,31 @@ elemental real(sp) function fast_tanh(x) result(y)

elemental real(sp) function gelu(x) result(y)
real(sp), intent(in) :: x
y = 0.5_sp * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715_sp * x**3)))
y = 0.5_sp * x * (1 + fast_tanh(sqrt(2 / pi) * (x + 0.044715_sp * x**3)))
end function

elemental real(sp) function fast_gelu(x) result(y)
real(sp), intent(in) :: x
real(sp) :: x2, x_
x_ = sqrt(2 / pi) * (x + 0.044715_sp * x**3)
if (x_ > 5) then
y = x
elseif (x_ < -5) then
y = 0
else
x2 = x_*x_
y = x_ * (0.98569772605911309407 + x2 *(-0.2794500993392901382 &
+ x2 * (6.8280504526399188164e-2 + x2 * (-1.0972014877337651823e-2 &
+ x2 * (1.1132367134444316902e-3 + x2 * (-7.018851897305717565e-5 &
+ x2 * (2.656616768082727089e-6 + x2 * (-5.5138381821615909058e-8 &
+ x2 * 4.8162484477588665996e-10))))))))
y = 0.5_sp * x * (1 + y)

! Alternative implementation (division is expensive)
!a = x * (135135.0 + x2 * (17325.0 + x2 * (378.0 + x2)))
!b = 135135.0 + x2 * (62370.0 + x2 * (3150.0 + x2 * 28.0))
!y = a / b
end if
end function

function softmax(x) result(y)
Expand Down Expand Up @@ -73,9 +97,31 @@ function linear(x, w, b) result(y)
function ffn(x, fc_w, fc_b, proj_w, proj_b) result(y)
real(sp), intent(in) :: x(:,:), fc_w(:,:), fc_b(:), proj_w(:,:), proj_b(:)
real(sp) :: y(size(x,1),size(x,2))
!real(sp) :: a(4*size(x,1),size(x,2))
!a = gelu(linear(x, fc_w, fc_b))
y = linear(gelu(linear(x, fc_w, fc_b)), proj_w, proj_b)
real(sp) :: a(4*size(x,1),size(x,2)), x0, x_, x2
integer :: i, j
!a = linear(x, fc_w, fc_b)
call matmul_2d(fc_w, x, a)
do i = 1, size(a,2)
do j = 1, size(a,1)
!a(j,i) = fast_gelu(a(j,i) + fc_b(j))
x0 = a(j,i) + fc_b(j)
x_ = sqrt(2 / pi) * (x0 + 0.044715_sp * x0**3)
if (x_ > 5) then
a(j,i) = x0
elseif (x_ < -5) then
a(j,i) = 0
else
x2 = x_*x_
a(j,i) = x_ * (0.98569772605911309407 + x2 *(-0.2794500993392901382 &
+ x2 * (6.8280504526399188164e-2 + x2 * (-1.0972014877337651823e-2 &
+ x2 * (1.1132367134444316902e-3 + x2 * (-7.018851897305717565e-5 &
+ x2 * (2.656616768082727089e-6 + x2 * (-5.5138381821615909058e-8 &
+ x2 * 4.8162484477588665996e-10))))))))
a(j,i) = 0.5_sp * x0 * (1 + a(j,i))
end if
end do
end do
y = linear(a, proj_w, proj_b)
end function

function attention(n_embd_head,n_seq,n_seq_x, q, k, v, mask) result(y)
Expand Down