@@ -6,7 +6,8 @@ This folder contains example for fmha(fused multi-head attention) using ck_tile
...
@@ -6,7 +6,8 @@ This folder contains example for fmha(fused multi-head attention) using ck_tile
```
```
# in the root of ck_tile
# in the root of ck_tile
mkdir build && cd build
mkdir build && cd build
sh ../script/cmake-ck-dev.sh ../ <arch> # you can replace this <arch> to gfx90a, gfx942...
# you can replace <arch> with the appropriate architecture (for example gfx90a or gfx942) or leave it blank
sh ../script/cmake-ck-dev.sh ../ <arch>
make tile_example_fmha_fwd -j
make tile_example_fmha_fwd -j
```
```
This will result in an executable `build/bin/tile_example_fmha_fwd`
This will result in an executable `build/bin/tile_example_fmha_fwd`
...
@@ -23,7 +24,7 @@ There are 3 template parameters for this kernel template.
...
@@ -23,7 +24,7 @@ There are 3 template parameters for this kernel template.
To speed up compile time, we instantiate the kernels into separate file. In this way we can benefit from parallel building from CMake/Make system. This is achieved by `generate.py` script. Besides, you can look into this script to learn how to instantiate a kernel instance step by step, which is described in `FMHA_FWD_KERNEL_BODY` variable.
To speed up compile time, we instantiate the kernels into separate file. In this way we can benefit from parallel building from CMake/Make system. This is achieved by `generate.py` script. Besides, you can look into this script to learn how to instantiate a kernel instance step by step, which is described in `FMHA_FWD_KERNEL_BODY` variable.
## executable
## executable
`tile_example_fmha_fwd` is the example executable, implemented in `fmha_fwd.cpp`. You can type `./bin/tile_example_fmha_fwd -?` to list all supported args. Below is an example of the output (may subject to change)
`tile_example_fmha_fwd` is the example executable, implemented in `fmha_fwd.cpp`. You can type `./bin/tile_example_fmha_fwd -?` to list all the arguments. Below is an example of the output (may subject to change)
```
```
args:
args:
-v weather do CPU validation or not (default:1)
-v weather do CPU validation or not (default:1)
...
@@ -31,48 +32,48 @@ args:
...
@@ -31,48 +32,48 @@ args:
-b batch size (default:2)
-b batch size (default:2)
-h num of head, for q (default:8)
-h num of head, for q (default:8)
-h_k num of head, for k/v, -1 means equal to h (default:-1)
-h_k num of head, for k/v, -1 means equal to h (default:-1)
if not equal to h, then this is GQA/MQA case
if not equal to h, then this is GQA/MQA case
-s seqlen_q. if group-mode, means the average value of seqlen_q (default:3328)
-s seqlen_q. if group-mode, means the average value of seqlen_q (default:3328)
total_seqlen_q = seqlen_q * batch, and seqlen_q per batch may vary
total_seqlen_q = seqlen_q * batch, and seqlen_q per batch may vary
also with "-s=s0,s1,s2..." comma seperated int to set per batch seqlen(group-mode)
also with "-s=s0,s1,s2..." comma seperated int to set per batch seqlen(group-mode)
-s_k seqlen_k, -1 means equal to s (default:-1)
-s_k seqlen_k (including new key/value), -1 means equal to s (default:-1)
-d head dim for q, k (default:128)
-d head dim for q, k (default:128)
-d_v head dim for v, -1 means equal to d (default:-1)
-d_v head dim for v, -1 means equal to d (default:-1)
-scale_s scale factor of S. 0 means equal to 1/sqrt(hdim). (default:0)
-scale_s scale factor of S. 0 means equal to 1/sqrt(hdim). (default:0)
note when squant=1, this value will be modified by range_q/k
note when squant=1, this value will be modified by range_q/k
-range_q per-tensor quantization range of q. used if squant=1. (default:16)
-range_q per-tensor quantization range of q. used if squant=1. (default:16)
-range_k per-tensor quantization range of k. used if squant=1. (default:16)
-range_k per-tensor quantization range of k. used if squant=1. (default:16)
-range_v per-tensor quantization range of v. used if squant=1. (default:16)
-range_v per-tensor quantization range of v. used if squant=1. (default:16)
-range_p per-tensor quantization range of p [e^(s-m)]. used if squant=1. (default:1)
-range_p per-tensor quantization range of p [e^(s-m)]. used if squant=1. (default:1)
-range_o per-tensor quantization range of o (p*v). used if squant=1. (default:16)
-range_o per-tensor quantization range of o (p*v). used if squant=1. (default:16)
-squant if using static quantization fusion or not. auto: fp8 will default use squant, other will not (default:auto)
-squant if using static quantization fusion or not. auto: fp8 will default use squant, other will not (default:auto)
0: no static quant(not implemented) 1: apply scale_p and scale_o with respect to P and O.
0: no static quant(not implemented) 1: apply scale_p and scale_o with respect to P and O.
calculate scale_s, scale_p, scale_o according to range_q, range_k, range_v, range_p, range_o
calculate scale_s, scale_p, scale_o according to range_q, range_k, range_v, range_p, range_o
-iperm permute input (default:1)
-iperm permute input (default:1)
if true, will be b*h*s*d, else b*s*h*d
if true, will be b*h*s*d, else b*s*h*d
-operm permute output (default:1)
-operm permute output (default:1)
-bias n or 0, no bias (default:n)
-bias n or 0, no bias (default:n)
e(lementwise) or 1, elementwise bias with 1*1*s*s. e:1, 1*h*s*s. e:2, b*h*s*s
e(lementwise) or 1, elementwise bias with 1*1*s*s. e:1, 1*h*s*s. e:2, b*h*s*s
a(libi) or 2, alibi with 1*h. a:1, b*h
a(libi) or 2, alibi with 1*h. a:1, b*h
-prec data type. fp16/bf16/fp8/bf8 (default:fp16)
-prec data type. fp16/bf16/fp8/bf8 (default:fp16)
-mask 0: no mask, 1: top-left(same as 't'), 2:bottom-right(same as 'b') (default:0)
-mask 0: no mask, 1: top-left(same as 't'), 2:bottom-right(same as 'b') (default:0)
't:l,r', top-left sliding window attn(swa) with FA style left right size
't:l,r', top-left sliding window attn(swa) with FA style left right size
'b:l,r', bottom-r sliding window attn(swa) with FA style left right size
'b:l,r', bottom-r sliding window attn(swa) with FA style left right size
'xt:window_size', xformer style masking from top-left, window_size negative is causal, positive is swa
'xt:window_size', xformer style masking from top-left, window_size negative is causal, positive is swa
'xb:window_size', xformer style masking from bottom-r, window_size negative is causal, positive is swa
'xb:window_size', xformer style masking from bottom-r, window_size negative is causal, positive is swa
'g:y,x', generic attention mask coordinate with y/x size (only debug purpose for now)
'g:y,x', generic attention mask coordinate with y/x size (only debug purpose for now)
-vlayout r for row-major(seqlen*hdim), c for col-major(hdim*seqlen) (default:r)
-vlayout r for row-major(seqlen*hdim), c for col-major(hdim*seqlen) (default:r)
-lse 0 not store lse, 1 store lse (default:0)
-lse 0 not store lse, 1 store lse (default:0)
-kname if set to 1 will print kernel name (default:0)
-kname if set to 1 will print kernel name (default:0)
-init init method. ui, uniform random int, ni, normalized random int (default:uf)
-init init method. ui, uniform random int, ni, normalized random int (default:uf)
uf, uniform random float, nf, normalized random float, tf, trig float, uf:q, quantization
uf, uniform random float, nf, normalized random float, tf, trig float, uf:q, quantization
-seed random seed used for initializing input tensors. 0 for non-deterministic seed (default:11939)
-seed random seed used for initializing input tensors. 0 for non-deterministic seed (default:11939)
-drop_seed seed for random number generator (default:1)
-drop_offset offset for random number generator (default:0)
-drop_prefs seed and offset values are present on GPU; 0 - host, 1 - device/GPU (default:0)
-warmup number of iterations before benchmark the kernel (default:5)
-warmup number of iterations before benchmark the kernel (default:5)
-repeat number of iterations to benchmark the kernel (default:20)
-repeat number of iterations to benchmark the kernel (default:20)
-drop_seed seed for the random number generator for the dropout layer, default is 1
-drop_offset offset for the dropout layer which is used during random number generation, default is 0
-drop_prefs flag to indicate `drop_seed` and `drop_offset` values if present on the GPU, default is 0, 0 - host, 1 - GPU
```
```
Example 1: `./bin/tile_example_fmha_fwd -b=1 -h=16 -s=16384 -d=128` will run a fmha case with batch=1, nhead=16, sequence length=16384, hdim=128, fp16 case.
Example 1: `./bin/tile_example_fmha_fwd -b=1 -h=16 -s=16384 -d=128` will run a fmha case with batch=1, nhead=16, sequence length=16384, hdim=128, fp16 case.
Example 2: `./bin/tile_example_fmha_fwd -b=1 -h=8 -s=16384 -d=64 -drop_prefs=1 -drop_seed=10 -drop_offset=1234` will run a fmha case with
Example 2: `./bin/tile_example_fmha_fwd -b=1 -h=8 -s=16384 -d=64 -drop_prefs=1 -drop_seed=10 -drop_offset=1234` will run a fmha case with
