[Minor] Typo Fixes (apache#10000)

* Fix typos.

* Missed funtion -> function.

python/tvm/relay/backend/contrib/ethosu/legalize.py

@@ -230,7 +230,7 @@ def __call__(self, *args, **kwargs):
 
 def sigmoid_calc_func(x: float) -> float:
     """Function to calculate the values for sigmoid"""
-    # Thse limits are inherited from TFLite
+    # These limits are inherited from TFLite
     upper_limit = 8.0
     lower_limit = -8.0
 

python/tvm/relay/frontend/common.py

@@ -716,11 +716,11 @@ def gru_cell(
     b_inp, b_hid : relay.Expr
         bias matrices. The same order of internal parts as for weights. shape = (3 * hidden_size)
     r_act : relay.op
-        activation funtion for reset gate. it is sigmoid by default
+        activation function for reset gate. it is sigmoid by default
     z_act : relay.op
-        activation funtion for update gate. it is sigmoid by default
+        activation function for update gate. it is sigmoid by default
     n_act : relay.op
-        activation funtion for new gate. it is tanh by default
+        activation function for new gate. it is tanh by default
     backwards : bool
         Flag for reverse pass of GRU
 
@@ -812,7 +812,7 @@ def lstm_cell(
     p_i, p_f, p_o : relay.Expr
         peephole LSTM matrices. shape = (batch, hidden_size)
     f_act, g_act, h_act : relay.op
-        activation funtions
+        activation functions
     backwards : bool
         Flag for reverse pass of LSTM
 

python/tvm/runtime/ndarray.py

@@ -127,7 +127,7 @@ def __setitem__(self, in_slice, value):
             raise TypeError("type %s not supported" % str(type(value)))
 
     def copyfrom(self, source_array):
-        """Peform an synchronize copy from the array.
+        """Perform an synchronize copy from the array.
 
         Parameters
         ----------

python/tvm/topi/cuda/batch_matmul_tensorcore.py

@@ -203,7 +203,7 @@ def _schedule(cfg, s, C):
         s[BF].reorder(bs, o, i, o_ii, i_ii)
 
         # Schedule for A's(B's) shared memory load
-        def shared_shedule(stage, strides):
+        def shared_schedule(stage, strides):
             s[stage].compute_at(s[CF], ko)
             bs, xo, yo = stage.op.axis
             s[stage].storage_align(xo, strides - 1, strides)
@@ -217,8 +217,8 @@ def shared_shedule(stage, strides):
             s[stage].bind(tx, thread_x)
             s[stage].vectorize(vi)
 
-        shared_shedule(AS, AS_align)
-        shared_shedule(BS, BS_align)
+        shared_schedule(AS, AS_align)
+        shared_schedule(BS, BS_align)
 
         shape = (wmma_m, wmma_n, wmma_k)
         AL_gemm = te.placeholder((wmma_m, wmma_k), name="AL_gemm", dtype=data_dtype)

python/tvm/topi/cuda/conv2d_nhwc_winograd.py

@@ -165,7 +165,7 @@ def schedule_bgemm_tensorcore(cfg, s, bgemm, data_pack, kernel_pack):
     s[BF].reorder(i, o, i_ii, o_ii)
 
     # Schedule for A's(B's) shared memory load
-    def shared_shedule(stage, strides):
+    def shared_schedule(stage, strides):
         s[stage].compute_at(s[CF], ko)
         _, _, xo, yo = stage.op.axis
         s[stage].storage_align(xo, strides - 1, strides)
@@ -179,8 +179,8 @@ def shared_shedule(stage, strides):
         s[stage].bind(tx, thread_x)
         s[stage].vectorize(vi)
 
-    shared_shedule(AS, AS_align)
-    shared_shedule(BS, BS_align)
+    shared_schedule(AS, AS_align)
+    shared_schedule(BS, BS_align)
 
     shape = (wmma_m, wmma_n, wmma_k)
     in_dtype = "float16"

python/tvm/topi/cuda/dense_tensorcore.py

@@ -238,7 +238,7 @@ def _schedule_dense_tensorcore(cfg, s, C):
     s[BF].reorder(o, i, o_ii, i_ii)
 
     # Schedule for A's(B's) shared memory load
-    def shared_shedule(stage, strides):
+    def shared_schedule(stage, strides):
         s[stage].compute_at(s[CF], ko)
         xo, yo = stage.op.axis
         s[stage].storage_align(xo, strides - 1, strides)
@@ -252,8 +252,8 @@ def shared_shedule(stage, strides):
         s[stage].bind(tx, thread_x)
         s[stage].vectorize(vi)
 
-    shared_shedule(AS, AS_align)
-    shared_shedule(BS, BS_align)
+    shared_schedule(AS, AS_align)
+    shared_schedule(BS, BS_align)
 
     shape = (wmma_m, wmma_n, wmma_k)
     AL_gemm = te.placeholder((wmma_m, wmma_k), name="AL_gemm", dtype=data_dtype)

src/arith/int_operator.h

@@ -78,7 +78,7 @@ inline bool WillOverflow<tir::ModNode>(int64_t x, int64_t y, int64_t min_value,
 }
 
 /*!
- * \brief Peform trunc division of two integers.
+ * \brief Perform trunc division of two integers.
  * \param x The left operand.
  * \param y The right operand.
  * \return the result.
@@ -94,7 +94,7 @@ inline int64_t truncdiv(int64_t x, int64_t y) { return x / y; }
 inline int64_t truncmod(int64_t x, int64_t y) { return x % y; }
 
 /*!
- * \brief Peform floor division of two integers.
+ * \brief Perform floor division of two integers.
  * \param x The left operand.
  * \param y The right operand.
  * \return the result.

src/relay/backend/te_compiler_cache.cc

@@ -192,7 +192,7 @@ class ScheduleBuilder : public backend::MemoizedExprTranslator<Array<te::Tensor>
         }
       }
 
-      // Use TOPI schdule if user specificed, or the function has no auto_scheduler schedule.
+      // Use TOPI schedule if user specificed, or the function has no auto_scheduler schedule.
       if (!schedule.defined() && !prim_func.defined()) {
         ICHECK(anchor_implementation_.defined());
         schedule = anchor_implementation_.Schedule(anchor_attrs_, tensor_outs, target_);

src/runtime/rpc/rpc_local_session.h

@@ -60,7 +60,7 @@ class LocalSession : public RPCSession {
 
  protected:
   /*!
-   * \brief internal encode return fucntion.
+   * \brief internal encode return function.
    * \param rv The return value.
    * \param encode_return The encoding function.
    */

src/support/ring_buffer.h

@@ -87,7 +87,7 @@ class RingBuffer {
   }
 
   /*!
-   * \brief Peform a non-blocking read from buffer
+   * \brief Perform a non-blocking read from buffer
    *  size must be smaller than this->bytes_available()
    * \param data the data pointer.
    * \param size The number of bytes to read.

src/support/socket.h

@@ -516,7 +516,7 @@ class TCPSocket : public Socket {
         [&]() { return recv(sockfd, buf, static_cast<sock_size_t>(len), flags); });
   }
   /*!
-   * \brief peform block write that will attempt to send all data out
+   * \brief perform block write that will attempt to send all data out
    *    can still return smaller than request when error occurs
    * \param buf_ the pointer to the buffer
    * \param len the size of the buffer
@@ -538,7 +538,7 @@ class TCPSocket : public Socket {
     return ndone;
   }
   /*!
-   * \brief peform block read that will attempt to read all data
+   * \brief perform block read that will attempt to read all data
    *    can still return smaller than request when error occurs
    * \param buf_ the buffer pointer
    * \param len length of data to recv
@@ -654,7 +654,7 @@ struct PollHelper {
   }
 
   /*!
-   * \brief peform poll on the set defined, read, write, exception
+   * \brief perform poll on the set defined, read, write, exception
    * \param timeout specify timeout in milliseconds(ms) if negative, means poll will block
    * \return
    */

src/te/operation/create_primfunc.cc

@@ -282,7 +282,7 @@ PrimFunc CreatePrimFunc(const Array<te::Tensor>& arg_list) {
                  << "Only te.placeholder and te.compute are allowed for now.";
   }
 
-  // Infomations used in CreatePrimFunc and its sub-funtions.
+  // Infomations used in CreatePrimFunc and its sub-functions.
   CreateFuncInfo info(arg_list);
   // Root body stmts.
   Array<Stmt> root_stmts;

src/te/operation/tensorize.cc

@@ -35,7 +35,7 @@ namespace te {
 
 using namespace tir;
 
-// Detect the region of input and output to be tensrized.
+// Detect the region of input and output to be tensorized.
 // out_dom: the domain of root iter vars in output op
 // in_region: region of each input tensor.
 // return The location of the tensorized scope start.

src/tir/schedule/primitive/cache_read_write.cc

@@ -635,7 +635,7 @@ StmtSRef CacheRead(ScheduleState self, const StmtSRef& block_sref, int read_buff
                                      /*require_subtree_compact_dataflow=*/false);
   const BlockNode* scope_block = TVM_SREF_TO_BLOCK(scope_block, scope_sref);
 
-  // Step 2. Creat CacheStageInfo
+  // Step 2. Create CacheStageInfo
   CacheStageInfo info;
   info.read_buffer = read_buffer;
   // Create the corresponding buffer to be written, i.e. result of cache_read

src/tir/transforms/coproc_sync.cc

@@ -450,7 +450,7 @@ class CoProcInstDepDetector : public StmtVisitor {
     std::unordered_set<int> exit_ctx;
     // existing pop performed at enter
     std::vector<std::pair<int, int> > enter_pop;
-    // existing push peformed at exit
+    // existing push performed at exit
     std::vector<std::pair<int, int> > exit_push;
     // clear the state
     void clear() {

web/src/compact.ts

@@ -19,9 +19,9 @@
 /** NodeJS and Web compact layer */
 
 /**
- * Get performance masurement.
+ * Get performance measurement.
  */
-export function getPeformance(): Performance {
+export function getPerformance(): Performance {
   if (typeof performance == "undefined") {
     // eslint-disable-next-line @typescript-eslint/no-var-requires
     const performanceNode = require("perf_hooks");

web/src/runtime.ts

@@ -653,7 +653,7 @@ class GraphExecutor implements Disposable {
    */
   async benchmarkRuns(dev: DLDevice, number=10, repeat=4): Promise<number[]> {
     // Skip first run as it can involve GPU warmup and module loading time.
-    const perf = compact.getPeformance();
+    const perf = compact.getPerformance();
     const results = [];
     this.run();
     await dev.sync();
@@ -1049,7 +1049,7 @@ export class Instance implements Disposable {
   /** Register global packed functions needed by the backend to the env. */
   private registerEnvGlobalPackedFuncs(): void {
     // Register the timer function to enable the time_evaluator.
-    const perf = compact.getPeformance();
+    const perf = compact.getPerformance();
 
     // Helper function to time the finvoke
     const timeExecution = async (