diff --git a/deepmd/utils/argcheck.py b/deepmd/utils/argcheck.py
index 51714ceaf1..684b0eacbc 100644
--- a/deepmd/utils/argcheck.py
+++ b/deepmd/utils/argcheck.py
@@ -58,7 +58,7 @@ def descrpt_local_frame_args ():
 
 
 def descrpt_se_a_args():
-    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius.'
+    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius. It is noted that the total sel value must be less than 4096 in a GPU environment.'
     doc_rcut = 'The cut-off radius.'
     doc_rcut_smth = 'Where to start smoothing. For example the 1/r term is smoothed from `rcut` to `rcut_smth`'
     doc_neuron = 'Number of neurons in each hidden layers of the embedding net. When two layers are of the same size or one layer is twice as large as the previous layer, a skip connection is built.'
@@ -90,7 +90,7 @@ def descrpt_se_a_args():
 
 
 def descrpt_se_t_args():
-    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius.'
+    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius. It is noted that the total sel value must be less than 4096 in a GPU environment.'
     doc_rcut = 'The cut-off radius.'
     doc_rcut_smth = 'Where to start smoothing. For example the 1/r term is smoothed from `rcut` to `rcut_smth`'
     doc_neuron = 'Number of neurons in each hidden layers of the embedding net. When two layers are of the same size or one layer is twice as large as the previous layer, a skip connection is built.'
@@ -129,7 +129,7 @@ def descrpt_se_a_tpe_args():
 
 
 def descrpt_se_r_args():
-    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius.'
+    doc_sel = 'A list of integers. The length of the list should be the same as the number of atom types in the system. `sel[i]` gives the selected number of type-i neighbors. `sel[i]` is recommended to be larger than the maximally possible number of type-i neighbors in the cut-off radius. It is noted that the total sel value must be less than 4096 in a GPU environment.'
     doc_rcut = 'The cut-off radius.'
     doc_rcut_smth = 'Where to start smoothing. For example the 1/r term is smoothed from `rcut` to `rcut_smth`'
     doc_neuron = 'Number of neurons in each hidden layers of the embedding net. When two layers are of the same size or one layer is twice as large as the previous layer, a skip connection is built.'
@@ -348,7 +348,7 @@ def model_compression_type_args():
 
 
 def model_args ():    
-    doc_type_map = 'A list of strings. Give the name to each type of atoms.'
+    doc_type_map = 'A list of strings. Give the name to each type of atoms. It is noted that the number of atom type of training system must be less than 128 in a GPU environment.'
     doc_data_stat_nbatch = 'The model determines the normalization from the statistics of the data. This key specifies the number of `frames` in each `system` used for statistics.'
     doc_data_stat_protect = 'Protect parameter for atomic energy regression.'
     doc_type_embedding = "The type embedding."
diff --git a/doc/getting-started.md b/doc/getting-started.md
index a1134b7fe6..fb6dedacd9 100644
--- a/doc/getting-started.md
+++ b/doc/getting-started.md
@@ -14,6 +14,7 @@ In this text, we will call the deep neural network that is used to represent the
     - [Run MD with LAMMPS](#run-md-with-lammps)
     - [Run path-integral MD with i-PI](#run-path-integral-md-with-i-pi)
     - [Use deep potential with ASE](#use-deep-potential-with-ase)
+8. [Known limitations](#known-limitations)
 
 
 ## Prepare data
@@ -368,5 +369,13 @@ dyn = BFGS(water)
 dyn.run(fmax=1e-6)
 print(water.get_positions())
 ```
+
+## Known limitations
+If you use deepmd-kit in a GPU environment, the acceptable value range of some variables are additionally restricted compared to the CPU environment due to the software's GPU implementations: 
+1. The number of atom type of a given system must be less than 128.
+2. The maximum distance between an atom and it's neighbors must be less than 128. It can be controlled by setting the rcut value of training parameters.
+3. Theoretically, the maximum number of atoms that a single GPU can accept is about 10,000,000. However, this value is actually limited by the GPU memory size currently, usually within 1000,000 atoms even at the model compression mode.
+4. The total sel value of training parameters(in model/descriptor section) must be less than 4096.
+
 [DP]:https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.143001
 [DP-SE]:https://dl.acm.org/doi/10.5555/3327345.3327356