Commit d1c68fb1 authored by Mathieu Faverge's avatar Mathieu Faverge

Update using.org (Thanks @fpruvost)

parent d4167ee7
...@@ -165,33 +165,58 @@ ...@@ -165,33 +165,58 @@
#+end_src #+end_src
List of main options that can be used in timing: List of main options that can be used in timing:
* ~--help~: show usage * ~--help~: Show usage
* ~--threads~: Number of CPU workers (default: * Machine parameters
~_SC_NPROCESSORS_ONLN~) * ~-t x, --threads=x~: Number of CPU workers (default: automatic
* ~--gpus~: number of GPU workers (default: ~0~) detection through runtime)
* ~--n_range=R~: range of N values, with ~R=Start:Stop:Step~ * ~-g x, --gpus=x~: Number of GPU workers (default: ~0~)
(default: ~500:5000:500~) * ~-P x, --P=x~: Rows (P) in the PxQ process grid (default: ~1~)
* ~--m=X~: dimension (M) of the matrices (default: ~N~) * ~--nocpu~: All GPU kernels are exclusively executed on GPUs
* ~--k=X~: dimension (K) of the matrices (default: ~1~), useful for * Matrix parameters
GEMM algorithm (k is the shared dimension and must be defined * ~-m x, --m=X, --M=x~: Dimension (M) of the matrices (default:
>1 to consider matrices and not vectors) ~N~)
* ~--nrhs=X~: number of right-hand size (default: ~1~) * ~-n x, --n=X, --N=x~: Dimension (N) of the matrices
* ~--nb=X~: block/tile size. (default: ~128~) * ~-N R, --n_range=R~: Range of N values to time with
* ~--ib=X~: inner-blocking/IB size. (default: ~32~) ~R=Start:Stop:Step~ (default: ~500:5000:500~)
* ~--niter=X~: number of iterations performed for each test * ~-k x, --k=x, --K=x, --nrhs=x~: Dimension (K) of the matrices
(default: ~1~) or number of right-hand size (default: ~1~). This is useful for
* ~--rhblk=X~: if X > 0, enable Householder mode for QR and LQ GEMM algorithms (k is the shared dimension and must be defined
factorization. X is the size of each subdomain (default: ~0~) >1 to consider matrices and not vectors)
* ~--[no]check~: check result (default: ~nocheck~) * ~-b x, --nb=x~: NB size. (default: ~320~)
* ~--[no]profile~: print profiling informations (default: * ~-i x, --ib=x~: IB size. (default: ~32~)
~noprofile~) * Check/prints
* ~--[no]trace~: enable/disable trace generation (default: ~notrace~) * ~--niter=X~: Number of iterations performed for each test
* ~--[no]dag~: enable/disable DAG generation (default: ~nodag~) (default: ~1~)
* ~--[no]inv~: check on inverse (default: ~noinv~) * ~-W, --nowarning~: Do not show warnings
* ~--nocpu~: all GPU kernels are exclusively executed on GPUs * ~-w, --nowarmup~: Cancel the warmup run to pre-load libraries
* ~--ooc~: Enable out-of-core (available only with StarPU) * ~-c, --check~: Check result
* ~--bound~: Compare result to area bound (available only with * ~-C, --inc~: Check on inverse
StarPU) (default: ~0~) * ~--mode=x~ : Change the xLATMS matrix mode generation for
SVD/EVD (default: ~4~). It must be between 0 and 20 included.
* Profiling parameters
* ~-T, --trace~: Enable trace generation
* ~--progress~: Display progress indicator
* ~-d, --dag~: Enable DAG generation. Generates a dot_dag_file.dot.
* ~-p, --profile~: Print profiling informations
* HQR parameters
* ~-a x, --qr_a=x, --rhblk=x~: Define the size of the local TS
trees in housholder reduction trees for QR and LQ
factorization. N is the size of each subdomain (default: ~-1~)
* ~-l x, --llvl=x~: Tree used for low level reduction inside
nodes (default: ~-1~)
* ~-L x, --hlvl=x~: Tree used for high level reduction between
nodes, only if P > 1 (default: ~-1~). Possible values are -1:
Automatic, 0: Flat, 1: Greedy, 2: Fibonacci, 3: Binary, 4:
Replicated greedy.
* ~-D, --domino~: Enable the domino between upper and lower trees
* Advanced options
* ~--nobigmat~: Disable single large matrix allocation for
multiple tiled allocations
* ~-s, --sync~: Enable synchronous calls in wrapper function such
as POTRI
* ~-o, --ooc~: Enable out-of-core (available only with StarPU)
* ~-G, --gemm3m~: Use gemm3m complex method
* ~--bound~: Compare result to area bound
List of timing algorithms available: List of timing algorithms available:
* LANGE: norms of matrices * LANGE: norms of matrices
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment