Fix some content
This commit is contained in:
parent
2ea1af6ec0
commit
cd1701ddce
|
@ -58,7 +58,7 @@
|
||||||
|
|
||||||
- formula
|
- formula
|
||||||
- Speedup factor:
|
- Speedup factor:
|
||||||
$$S = T / (\alpha \times T + (1 - \alpha \times T / n)) = 1 / (\alpha + (1 - \alpha) / n)$$
|
$$S = T / (\alpha \times T + (1 - \alpha) \times T / n) = 1 / (\alpha + (1 - \alpha) / n)$$
|
||||||
- $\alpha$: fraction of serial computation
|
- $\alpha$: fraction of serial computation
|
||||||
- 1 - alpha: part that can be parallelized
|
- 1 - alpha: part that can be parallelized
|
||||||
- $n$: processors used
|
- $n$: processors used
|
||||||
|
|
|
@ -44,7 +44,7 @@
|
||||||
```
|
```
|
||||||
- Note to self: run `nvcc` with `-arch=native` otherwise it won't work.
|
- Note to self: run `nvcc` with `-arch=native` otherwise it won't work.
|
||||||
|
|
||||||
### Device code and Host code
|
### Device code and Host code (Important)
|
||||||
|
|
||||||
- Compiler `nvcc` separates source code into host and device code
|
- Compiler `nvcc` separates source code into host and device code
|
||||||
- Device code (kernels): definition marked with `__global__`, called from
|
- Device code (kernels): definition marked with `__global__`, called from
|
||||||
|
|
Binary file not shown.
Loading…
Reference in a new issue