[Spice-devel] [4/ ]remove function find_bucket_8bpc

[Christophe Fergeau]

ACK
On Mon, Sep 24, 2012 at 10:49:35AM -0500, Aric Stewart wrote:
> removing the function call overhead improves performance
> 
> Signed-off-by: Aric Stewart <aric at codeweavers.com>
> ---
> quic.js |   27 +++++++++++----------------
> 1 files changed, 11 insertions(+), 16 deletions(-)
> 
> diff --git a/quic.js b/quic.js
> index 0e33610..138b052 100644
> --- a/quic.js
> +++ b/quic.js
> @@ -436,11 +436,6 @@ function QuicChannel(model_8bpc, model_5bpc)
> 
> QuicChannel.prototype = {
> 
> -    find_bucket_8bpc : function(val)
> -    {
> -        return this.buckets_ptrs[val];
> -    },
> -
>     reste : function (bpc)
>     {
>         this.correlate_row = { zero: 0 , row: []};
> @@ -652,7 +647,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row0_seg = function (i, cur_row, end
>         c = 0;
>         do
>         {
> -            a = golomb_decoding_8bpc(this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.zero).bestcode, this.io_word);
> +            a = golomb_decoding_8bpc(this.channels[c].buckets_ptrs[this.channels[c].correlate_row.zero].bestcode, this.io_word);
>             this.channels[c].correlate_row.row[0] = a.rc;
>             cur_row[2-c] = (family_8bpc.xlatL2U[a.rc]&0xFF);
>             this.decode_eatbits(a.codewordlen);
> @@ -665,7 +660,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row0_seg = function (i, cur_row, end
>             c = 0;
>             do
>             {
> -                this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.zero).update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[0], bpc);
> +                this.channels[c].buckets_ptrs[this.channels[c].correlate_row.zero].update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[0], bpc);
>             } while (++c < n_channels);
>         }
>         stopidx = ++i + this.rgb_state.waitcnt;
> @@ -681,7 +676,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row0_seg = function (i, cur_row, end
>             c = 0;
>             do
>             {
> -                a = golomb_decoding_8bpc(this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.row[i - 1]).bestcode, this.io_word);
> +                a = golomb_decoding_8bpc(this.channels[c].buckets_ptrs[this.channels[c].correlate_row.row[i - 1]].bestcode, this.io_word);
>                 this.channels[c].correlate_row.row[i] = a.rc;
>                 cur_row[(i* rgb32_pixel_size)+(2-c)] = (family_8bpc.xlatL2U[a.rc] + cur_row[((i-1) * rgb32_pixel_size) + (2-c)]) & bpc_mask;
>                 this.decode_eatbits(a.codewordlen);
> @@ -690,7 +685,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row0_seg = function (i, cur_row, end
>         c = 0;
>         do
>         {
> -            this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.row[stopidx - 1]).update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[stopidx], bpc);
> +            this.channels[c].buckets_ptrs[this.channels[c].correlate_row.row[stopidx - 1]].update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[stopidx], bpc);
>         } while (++c < n_channels);
>         stopidx = i + (this.rgb_state.tabrand() & waitmask);
>     }
> @@ -702,7 +697,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row0_seg = function (i, cur_row, end
>         c = 0;
>         do
>         {
> -            a = golomb_decoding_8bpc(this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.row[i - 1]).bestcode, this.io_word);
> +            a = golomb_decoding_8bpc(this.channels[c].buckets_ptrs[this.channels[c].correlate_row.row[i - 1]].bestcode, this.io_word);
>             this.channels[c].correlate_row.row[i] = a.rc;
>             cur_row[(i* rgb32_pixel_size)+(2-c)] = (family_8bpc.xlatL2U[a.rc] + cur_row[((i-1) * rgb32_pixel_size) + (2-c)]) & bpc_mask;
>             this.decode_eatbits(a.codewordlen);
> @@ -759,7 +754,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
> 
>         c = 0;
>         do {
> -            a = golomb_decoding_8bpc(this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.zero).bestcode, this.io_word);
> +            a = golomb_decoding_8bpc(this.channels[c].buckets_ptrs[this.channels[c].correlate_row.zero].bestcode, this.io_word);
>             this.channels[c].correlate_row.row[0] = a.rc;
>             cur_row[2-c] = (family_8bpc.xlatL2U[this.channels[c].correlate_row.row[0]] + prev_row[2-c]) & bpc_mask;
>             this.decode_eatbits(a.codewordlen);
> @@ -771,7 +766,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
>             this.rgb_state.waitcnt = (this.rgb_state.tabrand() & waitmask);
>             c = 0;
>             do {
> -                this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.zero).update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[0], bpc);
> +                this.channels[c].buckets_ptrs[this.channels[c].correlate_row.zero].update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[0], bpc);
>             } while (++c < n_channels);
>         }
>         stopidx = ++i + this.rgb_state.waitcnt;
> @@ -779,7 +774,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
>         stopidx = i + this.rgb_state.waitcnt;
>     }
>     for (;;) {
> -        rc = 0;
> +        var rc = 0;
>         while (stopidx < end && !rc) {
>             for (; i <= stopidx && !rc; i++) {
>                 var pixel = i * rgb32_pixel_size;
> @@ -822,7 +817,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
>                     var cc = this.channels[c];
>                     var cr = cc.correlate_row;
> 
> -                    a = golomb_decoding_8bpc(cc.find_bucket_8bpc(cr.row[i-1]).bestcode, this.io_word);
> +                    a = golomb_decoding_8bpc(cc.buckets_ptrs[cr.row[i-1]].bestcode, this.io_word);
>                     cr.row[i] = a.rc;
>                 cur_row[pixel+(2-c)] = (family_8bpc.xlatL2U[a.rc] + ((cur_row[pixelm1+(2-c)] + prev_row[pixel+(2-c)]) >> 1)) & bpc_mask;
>                     this.decode_eatbits(a.codewordlen);
> @@ -833,7 +828,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
> 
>             c = 0;
>             do {
> -                this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.row[stopidx - 1]).update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[stopidx], bpc);
> +                this.channels[c].buckets_ptrs[this.channels[c].correlate_row.row[stopidx - 1]].update_model_8bpc(this.rgb_state, this.channels[c].correlate_row.row[stopidx], bpc);
>             } while (++c < n_channels);
> 
>             stopidx = i + (this.rgb_state.tabrand() & waitmask);
> @@ -878,7 +873,7 @@ QuicEncoder.prototype.quic_rgb32_uncompress_row_seg = function( prev_row, cur_ro
>             c = 0;
>             do
>             {
> -                a = golomb_decoding_8bpc(this.channels[c].find_bucket_8bpc(this.channels[c].correlate_row.row[i-1]).bestcode, this.io_word);
> +                a = golomb_decoding_8bpc(this.channels[c].buckets_ptrs[this.channels[c].correlate_row.row[i-1]].bestcode, this.io_word);
>                 this.channels[c].correlate_row.row[i] = a.rc;
>                 cur_row[pixel+(2-c)] = (family_8bpc.xlatL2U[a.rc] + ((cur_row[pixelm1+(2-c)] + prev_row[pixel+(2-c)]) >> 1)) & bpc_mask;
>                 this.decode_eatbits(a.codewordlen);
> -- 
> 1.7.7.5 (Apple Git-26)
> 
> 


> _______________________________________________
> Spice-devel mailing list
> Spice-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/spice-devel

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/spice-devel/attachments/20120928/aaf73bf4/attachment.pgp>