diff --git a/W10D1/menu.md b/W10D1/menu.md
index e69de29..ac2a499 100644
--- a/W10D1/menu.md
+++ b/W10D1/menu.md
@@ -0,0 +1 @@
+[Peng Quanquan's notes](notes-of-pqq.md)
\ No newline at end of file
diff --git a/W10D1/notes-of-pqq.md b/W10D1/notes-of-pqq.md
new file mode 100644
index 0000000..1a8cf72
--- /dev/null
+++ b/W10D1/notes-of-pqq.md
@@ -0,0 +1,54 @@
+RAID: redundant array of inexpensive disk
+
+- **Reliability:** How many disk faults can the system tolerate? 
+
+
+
+### RAID-0
+
+normal disk, can't tolerate any disk failure.
+
+- **Reliability:** 0
+
+### RAID-1(Mirroring)
+
+def: More than one copy of each block is stored in a separate disk. 
+
+Thus, every block has two (or more) copies, lying on different disks.
+
+- **Reliability:** $ N / 2$ 
+
+### RAID-3
+
+def: **(Block-Level Stripping with Dedicated Parity)** 
+
+![](https://media.geeksforgeeks.org/wp-content/uploads/raid4-1.png)
+
+parity-based approach. 
+
+can't read R0, R5 simultaneously. (P0 = D0+D1+D2+D3)
+
+increase capacity compared to RAID-1.
+
+### RAID-4
+
+small read(without parity check/only checksum)
+
+large write:
+
+### RAID-5
+
+![](https://media.geeksforgeeks.org/wp-content/uploads/raid10.png)
+
+independent writes possible because of interleaved parity.
+
+e.g. write to D0, D6 uses Disks 0, 1, 3, 4
+
+
+
+### A Little Queuing Theory 
+
+$Length=\lambda \cdot W$
+
+- $\lambda$ : long-term average effective arrival rate
+- $W$ : a customer spends in the system
\ No newline at end of file
diff --git a/W13D2/menu.md b/W13D2/menu.md
index e69de29..d1e3327 100644
--- a/W13D2/menu.md
+++ b/W13D2/menu.md
@@ -0,0 +1 @@
+[pqq's notes](pqq-W13D2.md)
\ No newline at end of file
diff --git a/W13D2/pqq-W13D2.md b/W13D2/pqq-W13D2.md
new file mode 100644
index 0000000..379bf26
--- /dev/null
+++ b/W13D2/pqq-W13D2.md
@@ -0,0 +1,73 @@
+### Hardware Support for Memory Reference Speculation
+
+When $M_x \neq M_y$:
+
+1. Store My
+2. Load Mx
+
+- Optimization: Let `Load Mx` to be executed earlier by switching the execution order.
+
+
+- Hazard: error occurs when Mx = My. 
+
+
+
+### Spinlock 
+
+A locking mechanism is called a **spinlock**(non-blocking locker)
+
+- spinlock V.S. Mutex: 
+
+  In spinlock, the task cannot sleep while waiting for the lock, whereas in mutex, the task can sleep while waiting for the lock. 
+
+### Load link & Store conditional:
+
+- definition: it's a mechanism.
+- usage: to detect whether `load/store` instruction is **atomic**.
+
+e.g.1
+
+```
+; swap operation
+; suppose R4 = y, mem[0(R1)] = x
+
+mov R3, R4		; R3 = y
+ll  R2, 0(R1) ; R2 = x
+sc  R3, 0(R1) ; mem[0(R1)] = y;
+beqz R3, try
+mov R4, R2 		; R4 = x 
+```
+
+e.g.2
+
+```
+ll 	R2, 0(R1)
+addi R2, R2 #1 ; increment
+sc	 R2, 0(R1)
+beqz R2, try
+```
+
+procedure can be described as follows:
+
+1. Load [Mx] $\rightarrow$ R2
+2. R2++
+3. store R2's value into Memory
+
+If Step 1-3 are **non-atomic**, say, [Mx] had been modified by others during Step 2.
+
+Solution:
+
+- if (LR ≠ 0) { LR = 0, R2 = 1} success // not interrupted by others
+- if (LR = 0) R2 = 0 // interrupted
+  - `beqz R2, try` will redo the procedure until success
+
+
+
+- Q: nested ll/sc ? (i.e., doing `sc Mx`, `sc My` simultaneously)
+
+  A: no (because we only have 1 single LR...)
+
+LR: linker register
+
+ll/sc V.S. CAS (compare and swap)
+