This example is an approximately half-full (hx8k) FPGA:pic.twitter.com/xUlcbYmU1o
-
-
That shows (predictably) the limitations of RLE quite clearly, maybe other methods can improve with higher gate counts..
1 reply 0 retweets 0 likes -
Replying to @folknology @wavedrom
J1a is pretty much as tight as it gets (hx1k). Still 33% space savings. ("gzip -9" achieves 42% space savings.)pic.twitter.com/7rTHUssgT7
2 replies 0 retweets 3 likes -
Replying to @oe1cxw @folknology
Agree, 33% is a good start. Could you elaborate more on the algorithm your are using?
1 reply 0 retweets 1 like -
Replying to @wavedrom @folknology
I am storing the distances between ones. I'm using a simple Huffman code to use fewer bits for small distances.
4 replies 0 retweets 0 likes -
Replying to @oe1cxw @folknology
yes, this way you are winning allot on Zero and loosing on Ones. Here is 128'b1 example:pic.twitter.com/GxGyXZxqym
2 replies 0 retweets 0 likes -
Replying to @wavedrom @folknology
Improved version: http://svn.clifford.at/handicraft/2016/ice40compr … worst case is now -15%.pic.twitter.com/p9SN6ny1Hi
2 replies 0 retweets 6 likes -
Replying to @oe1cxw @folknology
Great! Can you compile it for bare-metal ARM? What is the code size?
2 replies 0 retweets 0 likes -
Replying to @wavedrom @folknology
I now wrote a pure C version of the uncompressor. Less than 200 bytes code size on ARM thumb. http://svn.clifford.at/handicraft/2016/ice40compr/ice40uncompr.c …
2 replies 0 retweets 4 likes -
Hey that's nice and simple
1 reply 0 retweets 1 like
Replying to @folknology @wavedrom
Changed the encoding slightly. Now worst case is at -14%, both example bins improved by about half a percentage point.
7:14 AM - 3 Jan 2017
0 replies
0 retweets
1 like
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.