update readme

Author: WilliamZhang20

README.md

@@ -12,13 +12,25 @@ Results:
     - Fused Multiply-Add for direct FP8 -> BF16 - requires an BF16 adder which can be tough and complex circuitry
     - Simpler way - accumulate the FP8 product in an integer and then convert back to float. According to @cchan, this is also implemented in NVIDIA H100.
 - Optimized the datapath by merging the old project's `mmu_feeder` and `control_unit` into a single module `control_unit`.
-- Have come up with plans for adding a second instruction for convolutions! It takes the same number of I/O cycles and is also handy with a systolic array!
+- Have come up with plans for adding a second instruction for elementwise multiplications! It takes the same number of I/O cycles and is also handy with a systolic array!
 
 - [Read the documentation for project](docs/info.md)
 
+## Design Architecture
+
+High Level Block Diagram:
+
+![alt](docs/Block_Diagram.jpg)
+
+## Verification Infrastructure
+
+Currently two tests are available, and both pass:
+1. Testing the precision of the MAC unit of the systolic array with randomly generated inputs from -10 to 10. 
+2. Testing the functional accuracy of the matrix multiplication.
+
 ## What is Tiny Tapeout?
 
-Tiny Tapeout is a project that makes it easier and cheaper to get digital and analog designs manufactured on a real chip.
+Tiny Tapeout is a project that makes it easy and chip to get digital and analog designs manufactured on a real chip.
 
 To learn more, visit https://tinytapeout.com.