lu-zero

diff --git a/src/cdef.rs b/src/cdef.rs
index 6d55b195..a1f1df50 100644
--- a/src/cdef.rs
+++ b/src/cdef.rs
@@ -425,17 +425,40 @@ pub fn cdef_sb_padded_frame_copy<T: Pixel>(
 // We assume in is padded, and the area we'll write out is at least as
 // large as the unpadded area of in
 // cdef_index is taken from the block context
-pub fn cdef_filter_superblock<T: Pixel>(
-  fi: &FrameInvariants<T>, in_frame: &Frame<u16>, out_frame: &mut Frame<T>,

lu-zero

--- write_coeffs_lv_map-v0.2.0.asm	2020-02-12 17:55:40.000000000 +0100
+++ write_coeffs_lv_map-master.asm	2020-02-12 17:50:26.000000000 +0100
@@ -1,16 +1,16 @@
-file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/iter/adapters/zip.rs does not exist!
-file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/ops/arith.rs does not exist!
-file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/cmp.rs does not exist!
-file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/ptr/const_ptr.rs does not exist!
 file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/slice/mod.rs does not exist!
+file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/ops/arith.rs does not exist!
 file /rustc/58b834344fc7b9185e7a50db1ff24e5eb07dae5e/src/libcore/option.rs does not exist!

lu-zero

13cfb5b64
192d56e1f
1b0b0d533
204d825a3
34bc00179
359a67cd3
428419901
488b8e2fa
4bb0d75a9
553a7343c

lu-zero

v0.2.0
 INFO  rav1e::stats         > encoded 600 frames, 2.015 fps, 2477.58 Kb/s
 INFO  rav1e::stats         > encoded 600 frames, 2.176 fps, 2502.80 Kb/s
 INFO  rav1e::stats         > encoded 600 frames, 2.374 fps, 2535.45 Kb/s
v0.2.0 low-latency
 INFO  rav1e::stats         > encoded 600 frames, 3.111 fps, 4357.21 Kb/s
 INFO  rav1e::stats         > encoded 600 frames, 3.291 fps, 4387.68 Kb/s
 INFO  rav1e::stats         > encoded 600 frames, 3.677 fps, 4421.96 Kb/s
p20200115

lu-zero

p20200115
>  encoded 600 frames, 2.537 fps, 2617.99 Kb/s
>  encoded 600 frames, 2.688 fps, 2646.60 Kb/s
>  encoded 600 frames, 2.900 fps, 2681.09 Kb/s
p20200127
>  encoded 600 frames, 2.872 fps, 2400.45 Kb/s
>  encoded 600 frames, 3.016 fps, 2425.72 Kb/s
>  encoded 600 frames, 3.244 fps, 2466.44 Kb/s
v0.3.0

lu-zero

echidna

>  Using y4m decoder: 3840x2160p @ 30/1 fps, 4:2:0, 8-bit
>  Encoding settings: keyint_min=12 keyint_max=240 quantizer=100 bitrate=0 min_quantizer=0 low_latency=true tune=Psychovisual rdo_lookahead_frames=40 min_block_size=32x32 max_block_size=32x32 multiref=false fast_deblock=true reduced_tx_set=true tx_domain_distortion=false tx_domain_rate=false encode_bottomup=false rdo_tx_decision=false prediction_modes=Simple include_near_mvs=false no_scene_detection=false diamond_me=true cdef=true use_satd_subpel=false non_square_partition=false enable_timing_info=false
>  CPU Feature Level: AVX2
>  Using 16 tiles (4x4)
>  encoded 600 frames, 1.484 fps, 11506.51 Kb/s
>  ----------
>  Key frame:             3 | avg QP:  79.00 | avg size:  192582 B
>  Inter frame:         597 | avg QP: 100.00 | avg size:   47216 B

lu-zero

#!/bin/sh

MEM_PROF_LIB=/opt/memory_profiler/libmemory_profiler.so

LD_PRELOAD="${MEM_PROF_LIB}:${LD_PRELOAD}" "$@"

$ cargo build --release

lu-zero

$ cargo install hawktracer-converter
$ cargo run --release --features=tracing $SAMPLE --tiles $TILES -s $SPEED -o /dev/null
# produce a chrome-tracing compatible json
$ hawktracer-converter-rs -s trace.bin -o rav1e-$SAMPLE-t$TILES-s$SPEED.json

lu-zero

$ cargo build --release
$ kcov --include-path=src/ /tmp/kcov target/release/rav1e $SAMPLE -s $SPEED --tiles $TILES -o /dev/null

lu-zero

testing e6c2b8defb4471202a0d4b05f6a128be82fd480d with 8 16 32 threads and tiles.

[speed 4]
master-standard-rayon
>  encoded 600 frames, 0.449 fps, 2230.03 Kb/s
>  encoded 600 frames, 0.554 fps, 2243.06 Kb/s
>  encoded 600 frames, 0.755 fps, 2278.19 Kb/s
master-new-rayon
>  encoded 600 frames, 0.447 fps, 2230.03 Kb/s