takagi’s gists

takagi / read.lisp

Created March 26, 2015 11:51

Comparing efficiency of READ-BYTE with READ-SEQUENCE.

	(require :sb-sprof)

	(defun test-read-byte0 ()
	(with-open-file (in "data" :direction :input
	:element-type '(unsigned-byte 8))
	(loop repeat (* 4 1024 1024)
	do (read-byte in))))

	(defun profile-read-byte0 ()
	(sb-sprof:with-profiling (:max-samples 100

takagi / tsuru.lisp

Last active August 29, 2015 14:17

Tsuru Capital recruiting test code sample.

	;;;
	;;; Fundamental WORD/INT types and readers
	;;;

	(deftype word8 ()
	`(unsigned-byte 8))

	(deftype word16 ()
	`(unsigned-byte 16))

takagi / flexi-streams.lisp

Created September 9, 2015 14:03

flexi-streams's external format

	(with-open-file (in "/Users/mtakagi/Desktop/bin" :direction :input
	:element-type 'unsigned-byte)
	(let ((buffer (make-array 256 :element-type 'unsigned-byte)))
	(read-sequence buffer in)
	(flexi-streams:octets-to-string buffer :external-format :utf-8)))

takagi / gist:3e4f3a8678a8f74e852e

Created March 4, 2016 04:08

sph


	https://github.com/takagi/cl-cuda/tree/issue/49.symbol-macro

	・時間を食う処理は、update-density と update-force で、全体の 90% 以上

	・アルゴリズムは所与としたとき、GPU の使い方のレベルで高速化する余地はあるか？
	→メモリアクセスにあまり局所性がなさそう
	→グローバルメモリへのアクセスが律速なので、それ以上はもうやりようがない？

	・グリッドやブロックの割当てはどのようにやるもの？

takagi / dot.lisp

Last active December 13, 2016 17:43

	(defun make-dvec (input-dimension initial-element)
	(make-array input-dimension :element-type 'double-float :initial-element initial-element))

	(defmacro dovec (vec var &body body)
	`(loop for ,var fixnum from 0 to (1- (length ,vec)) do ,@body))

	(defun dot (x y)
	(declare (type (simple-array double-float) x y)
	(optimize (speed 3) (safety 0)))
	(let ((result 0.0d0))

takagi / cifar_fp16

Last active June 10, 2019 06:54

Comparison of Chainer's cifar example between in FP32 mode and in FP16 mode

	$ CHAIENR_DTYPE=float16 python train_cifar.py -d 0
	Device: @cupy:0
	# Minibatch-size: 64
	# epoch: 300

	Using CIFAR10 dataset.
	epoch main/loss validation/main/loss main/accuracy validation/main/accuracy elapsed_time
	1 2.32253 2.12119 0.175971 0.192178 23.2918
	2 1.7554 1.87858 0.304497 0.302747 49.5816
	3 1.46396 1.61379 0.450664 0.416202 75.7544

takagi / dcgan_fp16

Last active June 10, 2019 07:25

Comparison of Chainer's dcgan example between in FP32 mode and in FP16 mode

	$ CHAINER_DTYPE=float16 python train_dcgan.py -d 0
	Device: @cupy:0
	# Minibatch-size: 50
	# n_hidden: 100
	# epoch: 1000

	epoch iteration gen/loss dis/loss ................] 0.01%
	0 100 nan nan
	0 200 nan nan
	0 300 nan nan

takagi / mnist_fp16

Created June 10, 2019 08:41

Comparison of Chainer's mnist example between in FP32 mode and in FP16 mode

	$ CHAINER_DTYPE=float16 python train_mnist.py -d 0
	Device: @cupy:0
	# unit: 1000
	# Minibatch-size: 100
	# epoch: 20

	epoch main/loss validation/main/loss main/accuracy validation/main/accuracy elapsed_time
	1 nan nan 0.0994271 0.0980225 3.91818
	2 nan nan 0.0997917 0.0980225 6.22553
	3 nan nan 0.0995833 0.0980225 8.72424

takagi / memnn_fp16

Last active June 14, 2019 02:14

Comparison of Chainer's memnn example between in FP32 mode and in FP16 mode

	$ CHAINER_DTYPE=float16 python train_memnn.py tasks_1-20_v1-2/en-10k/qa1_single-supporting-fact_train.txt tasks_1-20_v1-2/en-10k/qa1_single-supporting-fact_test.txt -d 0
	Training data: tasks_1-20_v1-2/en-10k/qa1_single-supporting-fact_train.txt: 2000
	Test data: tasks_1-20_v1-2/en-10k/qa1_single-supporting-fact_test.txt: 200
	epoch main/loss validation/main/loss main/accuracy validation/main/accuracy
	1 nan nan 0.0017004 0
	2 nan nan 0 0
	3 nan nan 0 0
	4 nan nan 0 0
	5 nan nan 0 0
	6 nan nan 0 0

takagi / pos_fp16

Created June 26, 2019 07:53

	$ CHAINER_DTYPE=float16 python postagging.py -d 0
	[nltk_data] Downloading package brown to /home/ext-
	[nltk_data] mtakagi/nltk_data...
	[nltk_data] Package brown is already up-to-date!
	# of sentences: 57340
	# of words: 56057
	# of pos: 472
	epoch main/loss validation/main/loss main/accuracy validation/main/accuracy elapsed_time
	0 244.875 18.3736
	0 373.75 34.9924

Masayuki Takagi takagi