Merge branch 'kernels'

11 years ago · cae071b837
14 changed files with 3579 additions and 878 deletions
--- a/AUTHORS.md
+++ b/AUTHORS.md
@ -1,6 +1,6 @@
				@@ -1,6 +1,6 @@
 # Authors

-## Current
+## Core

 * Scrypt-only refactor and maintenance: Noel Maersk <veox@wemakethings.net> LfxRFgXzA13TSTxgFGUFEtumv5ShGzAmLw
 * Core: Martin Danielsen <kalroth@gmail.com> 1DNBcSEENBwDKrcTyTW61ezWhzsPy5imkn
@ -9,6 +9,13 @@
				@@ -9,6 +9,13 @@
 * API: Andrew Smith <kan0i {at} kano-kun [dot] net> 1Jjk2LmktEQKnv8r2cZ9MvLiZwZ9gxabKm


+## OpenCL kernels
+
+* scrypt: Colin Percival
+* zuikkis: Zuikkis LeXck7EYgxyjw13zNDxZFmmgmWffFvhmSh
+* alexkarnew/alexkarold: Alexey Karimov LMqRcHdwnZtTMH6c2kWoxSoKM5KySfaP5C
+
+
 ## Bug fixes

 * Gabriel Devenyi
--- a/Makefile.am
+++ b/Makefile.am
@ -37,7 +37,7 @@ sgminer_SOURCES += ocl.c ocl.h
				@@ -37,7 +37,7 @@ sgminer_SOURCES += ocl.c ocl.h
 sgminer_SOURCES += findnonce.c findnonce.h
 sgminer_SOURCES += adl.c adl.h adl_functions.h
 sgminer_SOURCES += scrypt.c scrypt.h
-sgminer_SOURCES += *.cl
+sgminer_SOURCES += kernel/*.cl

-bin_SCRIPTS	= $(top_srcdir)/*.cl
+bin_SCRIPTS	= $(top_srcdir)/kernel/*.cl

--- a/README.md
+++ b/README.md
@ -18,9 +18,10 @@ License: GPLv3.  See `COPYING` for details.
				@@ -18,9 +18,10 @@ License: GPLv3.  See `COPYING` for details.
 Documentation is available in directory `doc`. For details on several topics, see:

 * `API` for the RPC API specification;
-* `SCRYPT` for how to find the right balance in GPU configuration to mine Scrypt-based coins effectively;
 * `FAQ` for frequently asked questions;
 * `GPU` for semi-obsolete information on GPU configuration options and mining SHA256d-based coins;
+* `KERNEL.md` for OpenCL kernel-related information;
+* `MINING` for how to find the right balance in GPU configuration to mine Scrypt-based coins effectively;
 * `windows-build.txt` for information on how to build on Windows.

 Note that **most of the documentation is outdated**. If you want to contribute, fork this repository, update as needed, and submit a pull request.
--- a/configure.ac
+++ b/configure.ac
@ -317,8 +317,10 @@ fi
				@@ -317,8 +317,10 @@ fi

 AC_DEFINE_UNQUOTED([SGMINER_PREFIX], ["$prefix/bin"], [Path to sgminer install])

-AC_DEFINE_UNQUOTED([SCRYPT_KERNNAME], ["scrypt130511"], [Filename for scrypt kernel])
-
+AC_DEFINE_UNQUOTED([ALEXKARNEW_KERNNAME], ["alexkarnew"], [Filename for Alexey Karimov's optimised kernel for Catalyst >=13.4])
+AC_DEFINE_UNQUOTED([ALEXKAROLD_KERNNAME], ["alexkarold"], [Filename for Alexey Karimov's optimised kernel for Catalyst <13.4])
+AC_DEFINE_UNQUOTED([CKOLIVAS_KERNNAME], ["ckolivas"], [Filename for original scrypt kernel])
+AC_DEFINE_UNQUOTED([ZUIKKIS_KERNNAME], ["zuikkis"], [Filename for Zuikkis' optimised kernel])

 AC_SUBST(OPENCL_LIBS)
 AC_SUBST(OPENCL_FLAGS)
--- a/doc/KERNEL.md
+++ b/doc/KERNEL.md
@ -0,0 +1,86 @@
				@@ -0,0 +1,86 @@
+# Kernels
+
+## Available OpenCL kernels
+
+See directory `kernel`.
+
+
+## Parameter configuration
+
+### Common
+
+In general, switching kernels requires reconfiguring mining parameters,
+such as (but not necessarily limited to) `thread-concurrency`, `intensity`,
+`gpu-engine` and `gpu-memclock`.
+
+A description of how to do this is available in `doc/MINING`.
+
+
+### alexkarnew
+
+Alaxey Karimov's optimised kernel, based on `ckolivas`. For Catalyst >=13.4.
+
+Only supports `vectors=1`.
+
+[Announcement](https://litecointalk.org/index.php?topic=4082.0).
+
+
+### alexkarold
+
+Alaxey Karimov's optimised kernel, based on `ckolivas`. For Catalyst <13.4.
+
+Only supports `vectors=1`.
+
+[Announcement](https://litecointalk.org/index.php?topic=4082.0).
+
+
+### ckolivas
+
+The original Colin Percival `scrypt` kernel, maintained for a long time by
+Con Kolivas in `cgminer` and renamed to reflect the fact.
+
+Only supports `vectors=1`.
+
+
+### zuikkis
+
+Zuikkis' optimised kernel, based on `ckolivas`.
+
+Only supports `vectors=1`, `lookup-gap=2` and `worksize=256`.
+
+[Announcement](https://litecointalk.org/index.php?topic=6058.msg90873#msg90873).
+
+
+## Submitting new kernels
+
+### Requirements
+
+* OpenCL source code only, licenced under GPLv3 (or later).
+* Not hard-coded for a specific GPU model or manufacturer.
+* Known limitations and any specific configuration quirks must be mentioned.
+
+
+### Procedure
+
+1. Copy the kernel you wish to modify and commit it verbatim.
+
+This way, it is easy to verify that there are no hidden changes. Note in
+the commit message which kernel is used as a base.
+
+2. Make changes to the kernel. Commit them.
+
+This allows to produce a diff that makes sense.
+
+3. Search for KL_CKOLIVAS and CKOLIVAS_KERNNAME in the top-level source
+directory and make additions to the listed files in order to integrate
+the new kernel.
+
+Now it can be selected when starting via the `--kernel` argument or
+`kernel` configuration option.
+
+4. Add yourself to the "kernels" section in `AUTHORS.md`. Keep it short.
+
+5. Submit a pull request on GitHub, or file it at the issue tracker,
+outlining the changes made, known limitations, and tested GPUs. List
+your git repository and branch name. The current repository and issue
+tracker links should be in `README.md`.
--- a/doc/MINING
+++ b/doc/MINING
--- a/driver-opencl.c
+++ b/driver-opencl.c
@ -196,8 +196,14 @@ char *set_thread_concurrency(char *arg)
				@@ -196,8 +196,14 @@ char *set_thread_concurrency(char *arg)

 static enum cl_kernels select_kernel(char *arg)
 {
-	if (!strcmp(arg, "scrypt"))
-		return KL_SCRYPT;
+	if (!strcmp(arg, ALEXKARNEW_KERNNAME))
+		return KL_ALEXKARNEW;
+	if (!strcmp(arg, ALEXKAROLD_KERNNAME))
+		return KL_ALEXKAROLD;
+	if (!strcmp(arg, CKOLIVAS_KERNNAME))
+		return KL_CKOLIVAS;
+	if (!strcmp(arg, ZUIKKIS_KERNNAME))
+		return KL_ZUIKKIS;

 	return KL_NONE;
 }
@ -1291,8 +1297,17 @@ static bool opencl_thread_prepare(struct thr_info *thr)
				@@ -1291,8 +1297,17 @@ static bool opencl_thread_prepare(struct thr_info *thr)
 	if (!cgpu->kname)
 	{
 		switch (clStates[i]->chosen_kernel) {
-			case KL_SCRYPT:
-				cgpu->kname = "scrypt";
+			case KL_ALEXKARNEW:
+				cgpu->kname = ALEXKARNEW_KERNNAME;
+				break;
+			case KL_ALEXKAROLD:
+				cgpu->kname = ALEXKAROLD_KERNNAME;
+				break;
+			case KL_CKOLIVAS:
+				cgpu->kname = CKOLIVAS_KERNNAME;
+				break;
+			case KL_ZUIKKIS:
+				cgpu->kname = ZUIKKIS_KERNNAME;
 				break;
 			default:
 				break;
@ -1322,11 +1337,15 @@ static bool opencl_thread_init(struct thr_info *thr)
				@@ -1322,11 +1337,15 @@ static bool opencl_thread_init(struct thr_info *thr)
 	}

 	switch (clState->chosen_kernel) {
-		case KL_SCRYPT:
-			thrdata->queue_kernel_parameters = &queue_scrypt_kernel;
-			break;
-		default:
-			break;
+	case KL_ALEXKARNEW:
+	case KL_ALEXKAROLD:
+	case KL_CKOLIVAS:
+	case KL_ZUIKKIS:
+		thrdata->queue_kernel_parameters = &queue_scrypt_kernel;
+		break;
+	default:
+		applog(LOG_ERR, "Failed to choose kernel in opencl_thread_init");
+		break;
 	}

 	thrdata->res = calloc(buffersize, 1);
--- a/kernel/alexkarnew.cl
+++ b/kernel/alexkarnew.cl
@ -0,0 +1,858 @@
				@@ -0,0 +1,858 @@
+/*-
+ * Copyright 2009 Colin Percival, 2011 ArtForz, 2011 pooler, 2012 mtrlt,
+ * 2012-2013 Con Kolivas, 2013 Alexey Karimov.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * This file was originally written by Colin Percival as part of the Tarsnap
+ * online backup system.
+ */
+
+__constant uint ES[2] = { 0x00FF00FF, 0xFF00FF00 };
+__constant uint K[] = {
+	0x428a2f98U,
+	0x71374491U,
+	0xb5c0fbcfU,
+	0xe9b5dba5U,
+	0x3956c25bU,
+	0x59f111f1U,
+	0x923f82a4U,
+	0xab1c5ed5U,
+	0xd807aa98U,
+	0x12835b01U,
+	0x243185beU, // 10
+	0x550c7dc3U,
+	0x72be5d74U,
+	0x80deb1feU,
+	0x9bdc06a7U,
+	0xe49b69c1U,
+	0xefbe4786U,
+	0x0fc19dc6U,
+	0x240ca1ccU,
+	0x2de92c6fU,
+	0x4a7484aaU, // 20
+	0x5cb0a9dcU,
+	0x76f988daU,
+	0x983e5152U,
+	0xa831c66dU,
+	0xb00327c8U,
+	0xbf597fc7U,
+	0xc6e00bf3U,
+	0xd5a79147U,
+	0x06ca6351U,
+	0x14292967U, // 30
+	0x27b70a85U,
+	0x2e1b2138U,
+	0x4d2c6dfcU,
+	0x53380d13U,
+	0x650a7354U,
+	0x766a0abbU,
+	0x81c2c92eU,
+	0x92722c85U,
+	0xa2bfe8a1U,
+	0xa81a664bU, // 40
+	0xc24b8b70U,
+	0xc76c51a3U,
+	0xd192e819U,
+	0xd6990624U,
+	0xf40e3585U,
+	0x106aa070U,
+	0x19a4c116U,
+	0x1e376c08U,
+	0x2748774cU,
+	0x34b0bcb5U, // 50
+	0x391c0cb3U,
+	0x4ed8aa4aU,
+	0x5b9cca4fU,
+	0x682e6ff3U,
+	0x748f82eeU,
+	0x78a5636fU,
+	0x84c87814U,
+	0x8cc70208U,
+	0x90befffaU,
+	0xa4506cebU, // 60
+	0xbef9a3f7U,
+	0xc67178f2U,
+	0x98c7e2a2U,
+	0xfc08884dU,
+	0xcd2a11aeU,
+	0x510e527fU,
+	0x9b05688cU,
+	0xC3910C8EU,
+	0xfb6feee7U,
+	0x2a01a605U, // 70
+	0x0c2e12e0U,
+	0x4498517BU,
+	0x6a09e667U,
+	0xa4ce148bU,
+	0x95F61999U,
+	0xc19bf174U,
+	0xBB67AE85U,
+	0x3C6EF372U,
+	0xA54FF53AU,
+	0x1F83D9ABU, // 80
+	0x5BE0CD19U,
+	0x5C5C5C5CU,
+	0x36363636U,
+	0x80000000U,
+	0x000003FFU,
+	0x00000280U,
+	0x000004a0U,
+	0x00000300U
+};
+
+#define rotl(x,y) rotate(x,y)
+#define Ch(x,y,z) bitselect(z,y,x)
+#define Maj(x,y,z) Ch((x^z),y,z)
+
+#define EndianSwap(n) (rotl(n & ES[0], 24U)|rotl(n & ES[1], 8U))
+
+#define Tr2(x)		(rotl(x, 30U) ^ rotl(x, 19U) ^ rotl(x, 10U))
+#define Tr1(x)		(rotl(x, 26U) ^ rotl(x, 21U) ^ rotl(x, 7U))
+#define Wr2(x)		(rotl(x, 25U) ^ rotl(x, 14U) ^ (x>>3U))
+#define Wr1(x)		(rotl(x, 15U) ^ rotl(x, 13U) ^ (x>>10U))
+
+#define RND(a, b, c, d, e, f, g, h, k)	\
+	h += Tr1(e); 			\
+	h += Ch(e, f, g); 		\
+	h += k;				\
+	d += h;				\
+	h += Tr2(a); 			\
+	h += Maj(a, b, c);
+
+void SHA256(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+	
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	uint4 W[4];
+
+	W[ 0].x = block0.x;
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[0]);
+	W[ 0].y = block0.y;
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[1]);
+	W[ 0].z = block0.z;
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[2]);
+	W[ 0].w = block0.w;
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[3]);
+
+	W[ 1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[ 1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[ 1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[ 1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+
+	W[ 2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[ 2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[ 2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[ 2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+
+	W[ 3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[ 3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[ 3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[ 3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void SHA256_fresh(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+#define A (*state0).x
+#define B (*state0).y
+#define C (*state0).z
+#define D (*state0).w
+#define E (*state1).x
+#define F (*state1).y
+#define G (*state1).z
+#define H (*state1).w
+
+	uint4 W[4];
+
+	W[0].x = block0.x;
+	D= K[63] +W[0].x;
+	H= K[64] +W[0].x;
+
+	W[0].y = block0.y;
+	C= K[65] +Tr1(D)+Ch(D, K[66], K[67])+W[0].y;
+	G= K[68] +C+Tr2(H)+Ch(H, K[69] ,K[70]);
+
+	W[0].z = block0.z;
+	B= K[71] +Tr1(C)+Ch(C,D,K[66])+W[0].z;
+	F= K[72] +B+Tr2(G)+Maj(G,H, K[73]);
+
+	W[0].w = block0.w;
+	A= K[74] +Tr1(B)+Ch(B,C,D)+W[0].w;
+	E= K[75] +A+Tr2(F)+Maj(F,G,H);
+
+	W[1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+	
+	W[2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+	
+	W[3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += (uint4)(K[73], K[77], K[78], K[79]);
+	*state1 += (uint4)(K[66], K[67], K[80], K[81]);
+}
+
+__constant uint fixedW[64] =
+{
+	0x428a2f99,0xf1374491,0xb5c0fbcf,0xe9b5dba5,0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5,
+	0xd807aa98,0x12835b01,0x243185be,0x550c7dc3,0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf794,
+	0xf59b89c2,0x73924787,0x23c6886e,0xa42ca65c,0x15ed3627,0x4d6edcbf,0xe28217fc,0xef02488f,
+	0xb707775c,0x0468c23f,0xe7e72b4c,0x49e1f1a2,0x4b99c816,0x926d1570,0xaa0fc072,0xadb36e2c,
+	0xad87a3ea,0xbcb1d3a3,0x7b993186,0x562b9420,0xbff3ca0c,0xda4b0c23,0x6cd8711a,0x8f337caa,
+	0xc91b1417,0xc359dce1,0xa83253a7,0x3b13c12d,0x9d3d725d,0xd9031a84,0xb1a03340,0x16f58012,
+	0xe64fb6a2,0xe84d923a,0xe93a5730,0x09837686,0x078ff753,0x29833341,0xd5de0b7e,0x6948ccf4,
+	0xe0a1adbe,0x7c728e11,0x511c78e4,0x315b45bd,0xfca71413,0xea28f96a,0x79703128,0x4e1ef848,
+};
+
+void SHA256_fixed(uint4*restrict state0,uint4*restrict state1)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	RND(A,B,C,D,E,F,G,H, fixedW[0]);
+	RND(H,A,B,C,D,E,F,G, fixedW[1]);
+	RND(G,H,A,B,C,D,E,F, fixedW[2]);
+	RND(F,G,H,A,B,C,D,E, fixedW[3]);
+	RND(E,F,G,H,A,B,C,D, fixedW[4]);
+	RND(D,E,F,G,H,A,B,C, fixedW[5]);
+	RND(C,D,E,F,G,H,A,B, fixedW[6]);
+	RND(B,C,D,E,F,G,H,A, fixedW[7]);
+	RND(A,B,C,D,E,F,G,H, fixedW[8]);
+	RND(H,A,B,C,D,E,F,G, fixedW[9]);
+	RND(G,H,A,B,C,D,E,F, fixedW[10]);
+	RND(F,G,H,A,B,C,D,E, fixedW[11]);
+	RND(E,F,G,H,A,B,C,D, fixedW[12]);
+	RND(D,E,F,G,H,A,B,C, fixedW[13]);
+	RND(C,D,E,F,G,H,A,B, fixedW[14]);
+	RND(B,C,D,E,F,G,H,A, fixedW[15]);
+	RND(A,B,C,D,E,F,G,H, fixedW[16]);
+	RND(H,A,B,C,D,E,F,G, fixedW[17]);
+	RND(G,H,A,B,C,D,E,F, fixedW[18]);
+	RND(F,G,H,A,B,C,D,E, fixedW[19]);
+	RND(E,F,G,H,A,B,C,D, fixedW[20]);
+	RND(D,E,F,G,H,A,B,C, fixedW[21]);
+	RND(C,D,E,F,G,H,A,B, fixedW[22]);
+	RND(B,C,D,E,F,G,H,A, fixedW[23]);
+	RND(A,B,C,D,E,F,G,H, fixedW[24]);
+	RND(H,A,B,C,D,E,F,G, fixedW[25]);
+	RND(G,H,A,B,C,D,E,F, fixedW[26]);
+	RND(F,G,H,A,B,C,D,E, fixedW[27]);
+	RND(E,F,G,H,A,B,C,D, fixedW[28]);
+	RND(D,E,F,G,H,A,B,C, fixedW[29]);
+	RND(C,D,E,F,G,H,A,B, fixedW[30]);
+	RND(B,C,D,E,F,G,H,A, fixedW[31]);
+	RND(A,B,C,D,E,F,G,H, fixedW[32]);
+	RND(H,A,B,C,D,E,F,G, fixedW[33]);
+	RND(G,H,A,B,C,D,E,F, fixedW[34]);
+	RND(F,G,H,A,B,C,D,E, fixedW[35]);
+	RND(E,F,G,H,A,B,C,D, fixedW[36]);
+	RND(D,E,F,G,H,A,B,C, fixedW[37]);
+	RND(C,D,E,F,G,H,A,B, fixedW[38]);
+	RND(B,C,D,E,F,G,H,A, fixedW[39]);
+	RND(A,B,C,D,E,F,G,H, fixedW[40]);
+	RND(H,A,B,C,D,E,F,G, fixedW[41]);
+	RND(G,H,A,B,C,D,E,F, fixedW[42]);
+	RND(F,G,H,A,B,C,D,E, fixedW[43]);
+	RND(E,F,G,H,A,B,C,D, fixedW[44]);
+	RND(D,E,F,G,H,A,B,C, fixedW[45]);
+	RND(C,D,E,F,G,H,A,B, fixedW[46]);
+	RND(B,C,D,E,F,G,H,A, fixedW[47]);
+	RND(A,B,C,D,E,F,G,H, fixedW[48]);
+	RND(H,A,B,C,D,E,F,G, fixedW[49]);
+	RND(G,H,A,B,C,D,E,F, fixedW[50]);
+	RND(F,G,H,A,B,C,D,E, fixedW[51]);
+	RND(E,F,G,H,A,B,C,D, fixedW[52]);
+	RND(D,E,F,G,H,A,B,C, fixedW[53]);
+	RND(C,D,E,F,G,H,A,B, fixedW[54]);
+	RND(B,C,D,E,F,G,H,A, fixedW[55]);
+	RND(A,B,C,D,E,F,G,H, fixedW[56]);
+	RND(H,A,B,C,D,E,F,G, fixedW[57]);
+	RND(G,H,A,B,C,D,E,F, fixedW[58]);
+	RND(F,G,H,A,B,C,D,E, fixedW[59]);
+	RND(E,F,G,H,A,B,C,D, fixedW[60]);
+	RND(D,E,F,G,H,A,B,C, fixedW[61]);
+	RND(C,D,E,F,G,H,A,B, fixedW[62]);
+	RND(B,C,D,E,F,G,H,A, fixedW[63]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void shittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[1].x,B[2].y,B[3].z,B[0].w);
+	tmp[1] = (uint4)(B[2].x,B[3].y,B[0].z,B[1].w);
+	tmp[2] = (uint4)(B[3].x,B[0].y,B[1].z,B[2].w);
+	tmp[3] = (uint4)(B[0].x,B[1].y,B[2].z,B[3].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[5].x,B[6].y,B[7].z,B[4].w);
+	tmp[1] = (uint4)(B[6].x,B[7].y,B[4].z,B[5].w);
+	tmp[2] = (uint4)(B[7].x,B[4].y,B[5].z,B[6].w);
+	tmp[3] = (uint4)(B[4].x,B[5].y,B[6].z,B[7].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void unshittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[3].x,B[2].y,B[1].z,B[0].w);
+	tmp[1] = (uint4)(B[0].x,B[3].y,B[2].z,B[1].w);
+	tmp[2] = (uint4)(B[1].x,B[0].y,B[3].z,B[2].w);
+	tmp[3] = (uint4)(B[2].x,B[1].y,B[0].z,B[3].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[7].x,B[6].y,B[5].z,B[4].w);
+	tmp[1] = (uint4)(B[4].x,B[7].y,B[6].z,B[5].w);
+	tmp[2] = (uint4)(B[5].x,B[4].y,B[7].z,B[6].w);
+	tmp[3] = (uint4)(B[6].x,B[5].y,B[4].z,B[7].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void salsa(uint4 B[8])
+{
+	uint4 w[4];
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i]^=B[i+4]);
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i+4]^=(B[i]+=w[i]));
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] += w[i];
+}
+
+void scrypt_core(uint4 X[8], __global uint4*restrict lookup)
+{
+	shittify(X);
+	const uint zSIZE = 8;
+	const uint xSIZE = CONCURRENT_THREADS;
+	uint x = get_global_id(0)%xSIZE;
+	uint CO=rotl(x,3U);
+	uint CO_tmp=rotl(xSIZE,3U);
+
+	for(uint y=0; y<1024/LOOKUP_GAP; ++y, CO+=CO_tmp)
+	{
+		uint CO_reg=CO;
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z, ++CO_reg)
+			lookup[CO_reg] = X[z];
+		for(uint i=0; i<LOOKUP_GAP; ++i) 
+			salsa(X);
+	}
+
+	CO_tmp=rotl(x,3U);
+
+#if (LOOKUP_GAP != 1) && (LOOKUP_GAP != 2) && (LOOKUP_GAP != 4) && (LOOKUP_GAP != 8)
+	{
+		uint y = (1024/LOOKUP_GAP);
+		CO=CO_tmp+rotl(y*xSIZE,3U);
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z, ++CO)
+			lookup[CO] = X[z];
+		for(uint i=0; i<1024%LOOKUP_GAP; ++i)
+			salsa(X); 
+	}
+#endif
+
+	for (uint i=0; i<1024; ++i) 
+	{
+		uint4 V[8];
+		uint j = X[7].x & K[85];
+		uint y = (j/LOOKUP_GAP);
+		uint CO_reg=CO_tmp+rotl(xSIZE*y,3U);
+
+		for(uint z=0; z<zSIZE; ++z, ++CO_reg)
+			V[z] = lookup[CO_reg];
+
+#if (LOOKUP_GAP == 1)
+#elif (LOOKUP_GAP == 2)
+		if (j&1)
+			salsa(V);
+#else
+		uint val = j%LOOKUP_GAP;
+		for (uint z=0; z<val; ++z) 
+			salsa(V);
+#endif
+
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z)
+			X[z] ^= V[z];
+		salsa(X);
+	}
+	unshittify(X);
+}
+
+#define FOUND (0xFF)
+#define SETFOUND(Xnonce) output[output[FOUND]++] = Xnonce
+
+__attribute__((reqd_work_group_size(WORKSIZE, 1, 1)))
+__kernel void search(__global const uint4 * restrict input,
+volatile __global uint*restrict output, __global uint4*restrict padcache,
+const uint4 midstate0, const uint4 midstate16, const uint target)
+{
+	uint gid = get_global_id(0);
+	uint4 X[8];
+	uint4 tstate0, tstate1, ostate0, ostate1, tmp0, tmp1;
+	uint4 data = (uint4)(input[4].x,input[4].y,input[4].z,gid);
+	uint4 pad0 = midstate0, pad1 = midstate16;
+
+	SHA256(&pad0,&pad1, data, (uint4)(K[84],0,0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[86]));
+	SHA256_fresh(&ostate0,&ostate1, pad0^ K[82], pad1^ K[82], K[82], K[82]);
+	SHA256_fresh(&tstate0,&tstate1, pad0^ K[83], pad1^ K[83], K[83], K[83]);
+
+	tmp0 = tstate0;
+	tmp1 = tstate1;
+	SHA256(&tstate0, &tstate1, input[0],input[1],input[2],input[3]);
+
+#pragma unroll
+	for (uint i=0; i<4; i++) 
+	{
+		pad0 = tstate0;
+		pad1 = tstate1;
+		X[rotl(i,1U) ] = ostate0;
+		X[rotl(i,1U)+1] = ostate1;
+
+		SHA256(&pad0,&pad1, data, (uint4)(i+1,K[84],0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[87]));
+		SHA256(X+rotl(i,1U),X+rotl(i,1U)+1, pad0, pad1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+	}
+	scrypt_core(X,padcache);
+	SHA256(&tmp0,&tmp1, X[0], X[1], X[2], X[3]);
+	SHA256(&tmp0,&tmp1, X[4], X[5], X[6], X[7]);
+	SHA256_fixed(&tmp0,&tmp1);
+	SHA256(&ostate0,&ostate1, tmp0, tmp1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+
+	bool result = (EndianSwap(ostate1.w) <= target);
+	if (result)
+		SETFOUND(gid);
+}
--- a/kernel/alexkarold.cl
+++ b/kernel/alexkarold.cl
@ -0,0 +1,855 @@
				@@ -0,0 +1,855 @@
+/*-
+ * Copyright 2009 Colin Percival, 2011 ArtForz, 2011 pooler, 2012 mtrlt,
+ * 2012-2013 Con Kolivas, 2013 Alexey Karimov.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * This file was originally written by Colin Percival as part of the Tarsnap
+ * online backup system.
+ */
+
+__constant uint ES[2] = { 0x00FF00FF, 0xFF00FF00 };
+__constant uint K[] = {
+	0x428a2f98U,
+	0x71374491U,
+	0xb5c0fbcfU,
+	0xe9b5dba5U,
+	0x3956c25bU,
+	0x59f111f1U,
+	0x923f82a4U,
+	0xab1c5ed5U,
+	0xd807aa98U,
+	0x12835b01U,
+	0x243185beU, // 10
+	0x550c7dc3U,
+	0x72be5d74U,
+	0x80deb1feU,
+	0x9bdc06a7U,
+	0xe49b69c1U,
+	0xefbe4786U,
+	0x0fc19dc6U,
+	0x240ca1ccU,
+	0x2de92c6fU,
+	0x4a7484aaU, // 20
+	0x5cb0a9dcU,
+	0x76f988daU,
+	0x983e5152U,
+	0xa831c66dU,
+	0xb00327c8U,
+	0xbf597fc7U,
+	0xc6e00bf3U,
+	0xd5a79147U,
+	0x06ca6351U,
+	0x14292967U, // 30
+	0x27b70a85U,
+	0x2e1b2138U,
+	0x4d2c6dfcU,
+	0x53380d13U,
+	0x650a7354U,
+	0x766a0abbU,
+	0x81c2c92eU,
+	0x92722c85U,
+	0xa2bfe8a1U,
+	0xa81a664bU, // 40
+	0xc24b8b70U,
+	0xc76c51a3U,
+	0xd192e819U,
+	0xd6990624U,
+	0xf40e3585U,
+	0x106aa070U,
+	0x19a4c116U,
+	0x1e376c08U,
+	0x2748774cU,
+	0x34b0bcb5U, // 50
+	0x391c0cb3U,
+	0x4ed8aa4aU,
+	0x5b9cca4fU,
+	0x682e6ff3U,
+	0x748f82eeU,
+	0x78a5636fU,
+	0x84c87814U,
+	0x8cc70208U,
+	0x90befffaU,
+	0xa4506cebU, // 60
+	0xbef9a3f7U,
+	0xc67178f2U,
+	0x98c7e2a2U,
+	0xfc08884dU,
+	0xcd2a11aeU,
+	0x510e527fU,
+	0x9b05688cU,
+	0xC3910C8EU,
+	0xfb6feee7U,
+	0x2a01a605U, // 70
+	0x0c2e12e0U,
+	0x4498517BU,
+	0x6a09e667U,
+	0xa4ce148bU,
+	0x95F61999U,
+	0xc19bf174U,
+	0xBB67AE85U,
+	0x3C6EF372U,
+	0xA54FF53AU,
+	0x1F83D9ABU, // 80
+	0x5BE0CD19U,
+	0x5C5C5C5CU,
+	0x36363636U,
+	0x80000000U,
+	0x000003FFU,
+	0x00000280U,
+	0x000004a0U,
+	0x00000300U
+};
+
+#define rotl(x,y) rotate(x,y)
+#define Ch(x,y,z) bitselect(z,y,x)
+#define Maj(x,y,z) Ch((x^z),y,z)
+
+#define EndianSwap(n) (rotl(n & ES[0], 24U)|rotl(n & ES[1], 8U))
+
+#define Tr2(x)		(rotl(x, 30U) ^ rotl(x, 19U) ^ rotl(x, 10U))
+#define Tr1(x)		(rotl(x, 26U) ^ rotl(x, 21U) ^ rotl(x, 7U))
+#define Wr2(x)		(rotl(x, 25U) ^ rotl(x, 14U) ^ (x>>3U))
+#define Wr1(x)		(rotl(x, 15U) ^ rotl(x, 13U) ^ (x>>10U))
+
+#define RND(a, b, c, d, e, f, g, h, k)	\
+	h += Tr1(e); 			\
+	h += Ch(e, f, g); 		\
+	h += k;				\
+	d += h;				\
+	h += Tr2(a); 			\
+	h += Maj(a, b, c);
+
+void SHA256(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+	
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	uint4 W[4];
+
+	W[ 0].x = block0.x;
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[0]);
+	W[ 0].y = block0.y;
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[1]);
+	W[ 0].z = block0.z;
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[2]);
+	W[ 0].w = block0.w;
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[3]);
+
+	W[ 1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[ 1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[ 1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[ 1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+
+	W[ 2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[ 2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[ 2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[ 2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+
+	W[ 3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[ 3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[ 3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[ 3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void SHA256_fresh(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+#define A (*state0).x
+#define B (*state0).y
+#define C (*state0).z
+#define D (*state0).w
+#define E (*state1).x
+#define F (*state1).y
+#define G (*state1).z
+#define H (*state1).w
+
+	uint4 W[4];
+
+	W[0].x = block0.x;
+	D= K[63] +W[0].x;
+	H= K[64] +W[0].x;
+
+	W[0].y = block0.y;
+	C= K[65] +Tr1(D)+Ch(D, K[66], K[67])+W[0].y;
+	G= K[68] +C+Tr2(H)+Ch(H, K[69] ,K[70]);
+
+	W[0].z = block0.z;
+	B= K[71] +Tr1(C)+Ch(C,D,K[66])+W[0].z;
+	F= K[72] +B+Tr2(G)+Maj(G,H, K[73]);
+
+	W[0].w = block0.w;
+	A= K[74] +Tr1(B)+Ch(B,C,D)+W[0].w;
+	E= K[75] +A+Tr2(F)+Maj(F,G,H);
+
+	W[1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+	
+	W[2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+	
+	W[3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += (uint4)(K[73], K[77], K[78], K[79]);
+	*state1 += (uint4)(K[66], K[67], K[80], K[81]);
+}
+
+__constant uint fixedW[64] =
+{
+	0x428a2f99,0xf1374491,0xb5c0fbcf,0xe9b5dba5,0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5,
+	0xd807aa98,0x12835b01,0x243185be,0x550c7dc3,0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf794,
+	0xf59b89c2,0x73924787,0x23c6886e,0xa42ca65c,0x15ed3627,0x4d6edcbf,0xe28217fc,0xef02488f,
+	0xb707775c,0x0468c23f,0xe7e72b4c,0x49e1f1a2,0x4b99c816,0x926d1570,0xaa0fc072,0xadb36e2c,
+	0xad87a3ea,0xbcb1d3a3,0x7b993186,0x562b9420,0xbff3ca0c,0xda4b0c23,0x6cd8711a,0x8f337caa,
+	0xc91b1417,0xc359dce1,0xa83253a7,0x3b13c12d,0x9d3d725d,0xd9031a84,0xb1a03340,0x16f58012,
+	0xe64fb6a2,0xe84d923a,0xe93a5730,0x09837686,0x078ff753,0x29833341,0xd5de0b7e,0x6948ccf4,
+	0xe0a1adbe,0x7c728e11,0x511c78e4,0x315b45bd,0xfca71413,0xea28f96a,0x79703128,0x4e1ef848,
+};
+
+void SHA256_fixed(uint4*restrict state0,uint4*restrict state1)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	RND(A,B,C,D,E,F,G,H, fixedW[0]);
+	RND(H,A,B,C,D,E,F,G, fixedW[1]);
+	RND(G,H,A,B,C,D,E,F, fixedW[2]);
+	RND(F,G,H,A,B,C,D,E, fixedW[3]);
+	RND(E,F,G,H,A,B,C,D, fixedW[4]);
+	RND(D,E,F,G,H,A,B,C, fixedW[5]);
+	RND(C,D,E,F,G,H,A,B, fixedW[6]);
+	RND(B,C,D,E,F,G,H,A, fixedW[7]);
+	RND(A,B,C,D,E,F,G,H, fixedW[8]);
+	RND(H,A,B,C,D,E,F,G, fixedW[9]);
+	RND(G,H,A,B,C,D,E,F, fixedW[10]);
+	RND(F,G,H,A,B,C,D,E, fixedW[11]);
+	RND(E,F,G,H,A,B,C,D, fixedW[12]);
+	RND(D,E,F,G,H,A,B,C, fixedW[13]);
+	RND(C,D,E,F,G,H,A,B, fixedW[14]);
+	RND(B,C,D,E,F,G,H,A, fixedW[15]);
+	RND(A,B,C,D,E,F,G,H, fixedW[16]);
+	RND(H,A,B,C,D,E,F,G, fixedW[17]);
+	RND(G,H,A,B,C,D,E,F, fixedW[18]);
+	RND(F,G,H,A,B,C,D,E, fixedW[19]);
+	RND(E,F,G,H,A,B,C,D, fixedW[20]);
+	RND(D,E,F,G,H,A,B,C, fixedW[21]);
+	RND(C,D,E,F,G,H,A,B, fixedW[22]);
+	RND(B,C,D,E,F,G,H,A, fixedW[23]);
+	RND(A,B,C,D,E,F,G,H, fixedW[24]);
+	RND(H,A,B,C,D,E,F,G, fixedW[25]);
+	RND(G,H,A,B,C,D,E,F, fixedW[26]);
+	RND(F,G,H,A,B,C,D,E, fixedW[27]);
+	RND(E,F,G,H,A,B,C,D, fixedW[28]);
+	RND(D,E,F,G,H,A,B,C, fixedW[29]);
+	RND(C,D,E,F,G,H,A,B, fixedW[30]);
+	RND(B,C,D,E,F,G,H,A, fixedW[31]);
+	RND(A,B,C,D,E,F,G,H, fixedW[32]);
+	RND(H,A,B,C,D,E,F,G, fixedW[33]);
+	RND(G,H,A,B,C,D,E,F, fixedW[34]);
+	RND(F,G,H,A,B,C,D,E, fixedW[35]);
+	RND(E,F,G,H,A,B,C,D, fixedW[36]);
+	RND(D,E,F,G,H,A,B,C, fixedW[37]);
+	RND(C,D,E,F,G,H,A,B, fixedW[38]);
+	RND(B,C,D,E,F,G,H,A, fixedW[39]);
+	RND(A,B,C,D,E,F,G,H, fixedW[40]);
+	RND(H,A,B,C,D,E,F,G, fixedW[41]);
+	RND(G,H,A,B,C,D,E,F, fixedW[42]);
+	RND(F,G,H,A,B,C,D,E, fixedW[43]);
+	RND(E,F,G,H,A,B,C,D, fixedW[44]);
+	RND(D,E,F,G,H,A,B,C, fixedW[45]);
+	RND(C,D,E,F,G,H,A,B, fixedW[46]);
+	RND(B,C,D,E,F,G,H,A, fixedW[47]);
+	RND(A,B,C,D,E,F,G,H, fixedW[48]);
+	RND(H,A,B,C,D,E,F,G, fixedW[49]);
+	RND(G,H,A,B,C,D,E,F, fixedW[50]);
+	RND(F,G,H,A,B,C,D,E, fixedW[51]);
+	RND(E,F,G,H,A,B,C,D, fixedW[52]);
+	RND(D,E,F,G,H,A,B,C, fixedW[53]);
+	RND(C,D,E,F,G,H,A,B, fixedW[54]);
+	RND(B,C,D,E,F,G,H,A, fixedW[55]);
+	RND(A,B,C,D,E,F,G,H, fixedW[56]);
+	RND(H,A,B,C,D,E,F,G, fixedW[57]);
+	RND(G,H,A,B,C,D,E,F, fixedW[58]);
+	RND(F,G,H,A,B,C,D,E, fixedW[59]);
+	RND(E,F,G,H,A,B,C,D, fixedW[60]);
+	RND(D,E,F,G,H,A,B,C, fixedW[61]);
+	RND(C,D,E,F,G,H,A,B, fixedW[62]);
+	RND(B,C,D,E,F,G,H,A, fixedW[63]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void shittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[1].x,B[2].y,B[3].z,B[0].w);
+	tmp[1] = (uint4)(B[2].x,B[3].y,B[0].z,B[1].w);
+	tmp[2] = (uint4)(B[3].x,B[0].y,B[1].z,B[2].w);
+	tmp[3] = (uint4)(B[0].x,B[1].y,B[2].z,B[3].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[5].x,B[6].y,B[7].z,B[4].w);
+	tmp[1] = (uint4)(B[6].x,B[7].y,B[4].z,B[5].w);
+	tmp[2] = (uint4)(B[7].x,B[4].y,B[5].z,B[6].w);
+	tmp[3] = (uint4)(B[4].x,B[5].y,B[6].z,B[7].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void unshittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[3].x,B[2].y,B[1].z,B[0].w);
+	tmp[1] = (uint4)(B[0].x,B[3].y,B[2].z,B[1].w);
+	tmp[2] = (uint4)(B[1].x,B[0].y,B[3].z,B[2].w);
+	tmp[3] = (uint4)(B[2].x,B[1].y,B[0].z,B[3].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[7].x,B[6].y,B[5].z,B[4].w);
+	tmp[1] = (uint4)(B[4].x,B[7].y,B[6].z,B[5].w);
+	tmp[2] = (uint4)(B[5].x,B[4].y,B[7].z,B[6].w);
+	tmp[3] = (uint4)(B[6].x,B[5].y,B[4].z,B[7].w);
+	
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void salsa(uint4 B[8])
+{
+	uint4 w[4];
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i]^=B[i+4]);
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i+4]^=(B[i]+=w[i]));
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+#pragma unroll
+	for(uint i=0; i<4; ++i)
+		B[i+4] += w[i];
+}
+
+void scrypt_core(uint4 X[8], __global uint4*restrict lookup)
+{
+	shittify(X);
+	const uint zSIZE = 8;
+	const uint xSIZE = CONCURRENT_THREADS;
+	uint x = get_global_id(0)%xSIZE;
+	uint CO_tmp=xSIZE<<3U;
+	uint CO_tmp2=x<<3U;
+
+	for(uint y=0; y<1024/LOOKUP_GAP; ++y)
+	{
+		uint CO=y*CO_tmp+CO_tmp2;
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z,++CO)
+			lookup[CO] = X[z];
+		for(uint i=0; i<LOOKUP_GAP; ++i) 
+			salsa(X);
+	}
+	
+#if (LOOKUP_GAP != 1) && (LOOKUP_GAP != 2) && (LOOKUP_GAP != 4) && (LOOKUP_GAP != 8)
+	{
+		uint y = (1024/LOOKUP_GAP);
+		uint CO=y*CO_tmp+CO_tmp2;
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z)
+			lookup[CO] = X[z];
+		for(uint i=0; i<1024%LOOKUP_GAP; ++i)
+			salsa(X); 
+	}
+#endif
+	for (uint i=0; i<1024; ++i) 
+	{
+		uint4 V[8];
+		uint j = X[7].x & K[85];
+		uint y = (j/LOOKUP_GAP);
+		uint CO=y*CO_tmp+CO_tmp2;
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z)
+			V[z] = lookup[CO+z];
+
+#if (LOOKUP_GAP == 1)
+#elif (LOOKUP_GAP == 2)
+		if (j&1)
+			salsa(V);
+#else
+		uint val = j%LOOKUP_GAP;
+		for (uint z=0; z<val; ++z) 
+			salsa(V);
+#endif
+
+#pragma unroll
+		for(uint z=0; z<zSIZE; ++z)
+			X[z] ^= V[z];
+		salsa(X);
+	}
+	unshittify(X);
+}
+
+#define FOUND (0xFF)
+#define SETFOUND(Xnonce) output[output[FOUND]++] = Xnonce
+
+__attribute__((reqd_work_group_size(WORKSIZE, 1, 1)))
+__kernel void search(__global const uint4 * restrict input,
+volatile __global uint*restrict output, __global uint4*restrict padcache,
+const uint4 midstate0, const uint4 midstate16, const uint target)
+{
+	uint gid = get_global_id(0);
+	uint4 X[8];
+	uint4 tstate0, tstate1, ostate0, ostate1, tmp0, tmp1;
+	uint4 data = (uint4)(input[4].x,input[4].y,input[4].z,gid);
+	uint4 pad0 = midstate0, pad1 = midstate16;
+
+	SHA256(&pad0,&pad1, data, (uint4)(K[84],0,0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[86]));
+	SHA256_fresh(&ostate0,&ostate1, pad0^ K[82], pad1^ K[82], K[82], K[82]);
+	SHA256_fresh(&tstate0,&tstate1, pad0^ K[83], pad1^ K[83], K[83], K[83]);
+
+	tmp0 = tstate0;
+	tmp1 = tstate1;
+	SHA256(&tstate0, &tstate1, input[0],input[1],input[2],input[3]);
+
+#pragma unroll
+	for (uint i=0; i<4; i++) 
+	{
+		pad0 = tstate0;
+		pad1 = tstate1;
+		X[i*2 ] = ostate0;
+		X[i*2+1] = ostate1;
+
+		SHA256(&pad0,&pad1, data, (uint4)(i+1,K[84],0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[87]));
+		SHA256(X+i*2,X+i*2+1, pad0, pad1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+	}
+	scrypt_core(X,padcache);
+	SHA256(&tmp0,&tmp1, X[0], X[1], X[2], X[3]);
+	SHA256(&tmp0,&tmp1, X[4], X[5], X[6], X[7]);
+	SHA256_fixed(&tmp0,&tmp1);
+	SHA256(&ostate0,&ostate1, tmp0, tmp1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+
+	bool result = (EndianSwap(ostate1.w) <= target);
+	if (result)
+		SETFOUND(gid);
+}
--- a/kernel/ckolivas.cl
+++ b/kernel/ckolivas.cl
--- a/kernel/zuikkis.cl
+++ b/kernel/zuikkis.cl
@ -0,0 +1,841 @@
				@@ -0,0 +1,841 @@
+/*-
+ * Copyright 2009 Colin Percival, 2011 ArtForz, 2011 pooler, 2012 mtrlt,
+ * 2012-2013 Con Kolivas.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * This file was originally written by Colin Percival as part of the Tarsnap
+ * online backup system.
+ */
+
+__constant uint ES[2] = { 0x00FF00FF, 0xFF00FF00 };
+__constant uint K[] = {
+	0x428a2f98U,
+	0x71374491U,
+	0xb5c0fbcfU,
+	0xe9b5dba5U,
+	0x3956c25bU,
+	0x59f111f1U,
+	0x923f82a4U,
+	0xab1c5ed5U,
+	0xd807aa98U,
+	0x12835b01U,
+	0x243185beU, // 10
+	0x550c7dc3U,
+	0x72be5d74U,
+	0x80deb1feU,
+	0x9bdc06a7U,
+	0xe49b69c1U,
+	0xefbe4786U,
+	0x0fc19dc6U,
+	0x240ca1ccU,
+	0x2de92c6fU,
+	0x4a7484aaU, // 20
+	0x5cb0a9dcU,
+	0x76f988daU,
+	0x983e5152U,
+	0xa831c66dU,
+	0xb00327c8U,
+	0xbf597fc7U,
+	0xc6e00bf3U,
+	0xd5a79147U,
+	0x06ca6351U,
+	0x14292967U, // 30
+	0x27b70a85U,
+	0x2e1b2138U,
+	0x4d2c6dfcU,
+	0x53380d13U,
+	0x650a7354U,
+	0x766a0abbU,
+	0x81c2c92eU,
+	0x92722c85U,
+	0xa2bfe8a1U,
+	0xa81a664bU, // 40
+	0xc24b8b70U,
+	0xc76c51a3U,
+	0xd192e819U,
+	0xd6990624U,
+	0xf40e3585U,
+	0x106aa070U,
+	0x19a4c116U,
+	0x1e376c08U,
+	0x2748774cU,
+	0x34b0bcb5U, // 50
+	0x391c0cb3U,
+	0x4ed8aa4aU,
+	0x5b9cca4fU,
+	0x682e6ff3U,
+	0x748f82eeU,
+	0x78a5636fU,
+	0x84c87814U,
+	0x8cc70208U,
+	0x90befffaU,
+	0xa4506cebU, // 60
+	0xbef9a3f7U,
+	0xc67178f2U,
+	0x98c7e2a2U,
+	0xfc08884dU,
+	0xcd2a11aeU,
+	0x510e527fU,
+	0x9b05688cU,
+	0xC3910C8EU,
+	0xfb6feee7U,
+	0x2a01a605U, // 70
+	0x0c2e12e0U,
+	0x4498517BU,
+	0x6a09e667U,
+	0xa4ce148bU,
+	0x95F61999U,
+	0xc19bf174U,
+	0xBB67AE85U,
+	0x3C6EF372U,
+	0xA54FF53AU,
+	0x1F83D9ABU, // 80
+	0x5BE0CD19U,
+	0x5C5C5C5CU,
+	0x36363636U,
+	0x80000000U,
+	0x000003FFU,
+	0x00000280U,
+	0x000004a0U,
+	0x00000300U
+};
+
+#define rotl(x,y) rotate(x,y)
+#define Ch(x,y,z) bitselect(z,y,x)
+#define Maj(x,y,z) Ch((x^z),y,z)
+
+#define EndianSwap(n) (rotl(n & ES[0], 24U)|rotl(n & ES[1], 8U))
+
+#define Tr2(x)		(rotl(x, 30U) ^ rotl(x, 19U) ^ rotl(x, 10U))
+#define Tr1(x)		(rotl(x, 26U) ^ rotl(x, 21U) ^ rotl(x, 7U))
+#define Wr2(x)		(rotl(x, 25U) ^ rotl(x, 14U) ^ (x>>3U))
+#define Wr1(x)		(rotl(x, 15U) ^ rotl(x, 13U) ^ (x>>10U))
+
+#define RND(a, b, c, d, e, f, g, h, k)	\
+	h += Tr1(e); 			\
+	h += Ch(e, f, g); 		\
+	h += k;				\
+	d += h;				\
+	h += Tr2(a); 			\
+	h += Maj(a, b, c);
+
+void SHA256(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+	
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	uint4 W[4];
+
+	W[ 0].x = block0.x;
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[0]);
+	W[ 0].y = block0.y;
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[1]);
+	W[ 0].z = block0.z;
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[2]);
+	W[ 0].w = block0.w;
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[3]);
+
+	W[ 1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[ 1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[ 1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[ 1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+
+	W[ 2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[ 2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[ 2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[ 2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+
+	W[ 3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[ 3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[ 3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[ 3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[ 0].x += Wr1(W[ 3].z) + W[ 2].y + Wr2(W[ 0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[ 0].y += Wr1(W[ 3].w) + W[ 2].z + Wr2(W[ 0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[ 0].z += Wr1(W[ 0].x) + W[ 2].w + Wr2(W[ 0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[ 0].w += Wr1(W[ 0].y) + W[ 3].x + Wr2(W[ 1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[ 1].x += Wr1(W[ 0].z) + W[ 3].y + Wr2(W[ 1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[ 1].y += Wr1(W[ 0].w) + W[ 3].z + Wr2(W[ 1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[ 1].z += Wr1(W[ 1].x) + W[ 3].w + Wr2(W[ 1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[ 1].w += Wr1(W[ 1].y) + W[ 0].x + Wr2(W[ 2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[ 2].x += Wr1(W[ 1].z) + W[ 0].y + Wr2(W[ 2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[ 2].y += Wr1(W[ 1].w) + W[ 0].z + Wr2(W[ 2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[ 2].z += Wr1(W[ 2].x) + W[ 0].w + Wr2(W[ 2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[ 2].w += Wr1(W[ 2].y) + W[ 1].x + Wr2(W[ 3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[ 3].x += Wr1(W[ 2].z) + W[ 1].y + Wr2(W[ 3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[ 3].y += Wr1(W[ 2].w) + W[ 1].z + Wr2(W[ 3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[ 3].z += Wr1(W[ 3].x) + W[ 1].w + Wr2(W[ 3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[ 3].w += Wr1(W[ 3].y) + W[ 2].x + Wr2(W[ 0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void SHA256_fresh(uint4*restrict state0,uint4*restrict state1, const uint4 block0, const uint4 block1, const uint4 block2, const uint4 block3)
+{
+#define A (*state0).x
+#define B (*state0).y
+#define C (*state0).z
+#define D (*state0).w
+#define E (*state1).x
+#define F (*state1).y
+#define G (*state1).z
+#define H (*state1).w
+
+	uint4 W[4];
+
+	W[0].x = block0.x;
+	D= K[63] +W[0].x;
+	H= K[64] +W[0].x;
+
+	W[0].y = block0.y;
+	C= K[65] +Tr1(D)+Ch(D, K[66], K[67])+W[0].y;
+	G= K[68] +C+Tr2(H)+Ch(H, K[69] ,K[70]);
+
+	W[0].z = block0.z;
+	B= K[71] +Tr1(C)+Ch(C,D,K[66])+W[0].z;
+	F= K[72] +B+Tr2(G)+Maj(G,H, K[73]);
+
+	W[0].w = block0.w;
+	A= K[74] +Tr1(B)+Ch(B,C,D)+W[0].w;
+	E= K[75] +A+Tr2(F)+Maj(F,G,H);
+
+	W[1].x = block1.x;
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[4]);
+	W[1].y = block1.y;
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[5]);
+	W[1].z = block1.z;
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[6]);
+	W[1].w = block1.w;
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[7]);
+	
+	W[2].x = block2.x;
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[8]);
+	W[2].y = block2.y;
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[9]);
+	W[2].z = block2.z;
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[10]);
+	W[2].w = block2.w;
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[11]);
+	
+	W[3].x = block3.x;
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[12]);
+	W[3].y = block3.y;
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[13]);
+	W[3].z = block3.z;
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[14]);
+	W[3].w = block3.w;
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[76]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[15]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[16]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[17]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[18]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[19]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[20]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[21]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[22]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[23]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[24]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[25]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[26]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[27]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[28]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[29]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[30]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[31]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[32]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[33]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[34]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[35]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[36]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[37]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[38]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[39]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[40]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[41]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[42]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[43]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[44]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[45]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[46]);
+
+	W[0].x += Wr1(W[3].z) + W[2].y + Wr2(W[0].y);
+	RND(A,B,C,D,E,F,G,H, W[0].x+ K[47]);
+
+	W[0].y += Wr1(W[3].w) + W[2].z + Wr2(W[0].z);
+	RND(H,A,B,C,D,E,F,G, W[0].y+ K[48]);
+
+	W[0].z += Wr1(W[0].x) + W[2].w + Wr2(W[0].w);
+	RND(G,H,A,B,C,D,E,F, W[0].z+ K[49]);
+
+	W[0].w += Wr1(W[0].y) + W[3].x + Wr2(W[1].x);
+	RND(F,G,H,A,B,C,D,E, W[0].w+ K[50]);
+
+	W[1].x += Wr1(W[0].z) + W[3].y + Wr2(W[1].y);
+	RND(E,F,G,H,A,B,C,D, W[1].x+ K[51]);
+
+	W[1].y += Wr1(W[0].w) + W[3].z + Wr2(W[1].z);
+	RND(D,E,F,G,H,A,B,C, W[1].y+ K[52]);
+
+	W[1].z += Wr1(W[1].x) + W[3].w + Wr2(W[1].w);
+	RND(C,D,E,F,G,H,A,B, W[1].z+ K[53]);
+
+	W[1].w += Wr1(W[1].y) + W[0].x + Wr2(W[2].x);
+	RND(B,C,D,E,F,G,H,A, W[1].w+ K[54]);
+
+	W[2].x += Wr1(W[1].z) + W[0].y + Wr2(W[2].y);
+	RND(A,B,C,D,E,F,G,H, W[2].x+ K[55]);
+
+	W[2].y += Wr1(W[1].w) + W[0].z + Wr2(W[2].z);
+	RND(H,A,B,C,D,E,F,G, W[2].y+ K[56]);
+
+	W[2].z += Wr1(W[2].x) + W[0].w + Wr2(W[2].w);
+	RND(G,H,A,B,C,D,E,F, W[2].z+ K[57]);
+
+	W[2].w += Wr1(W[2].y) + W[1].x + Wr2(W[3].x);
+	RND(F,G,H,A,B,C,D,E, W[2].w+ K[58]);
+
+	W[3].x += Wr1(W[2].z) + W[1].y + Wr2(W[3].y);
+	RND(E,F,G,H,A,B,C,D, W[3].x+ K[59]);
+
+	W[3].y += Wr1(W[2].w) + W[1].z + Wr2(W[3].z);
+	RND(D,E,F,G,H,A,B,C, W[3].y+ K[60]);
+
+	W[3].z += Wr1(W[3].x) + W[1].w + Wr2(W[3].w);
+	RND(C,D,E,F,G,H,A,B, W[3].z+ K[61]);
+
+	W[3].w += Wr1(W[3].y) + W[2].x + Wr2(W[0].x);
+	RND(B,C,D,E,F,G,H,A, W[3].w+ K[62]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+
+	*state0 += (uint4)(K[73], K[77], K[78], K[79]);
+	*state1 += (uint4)(K[66], K[67], K[80], K[81]);
+}
+
+__constant uint fixedW[64] =
+{
+	0x428a2f99,0xf1374491,0xb5c0fbcf,0xe9b5dba5,0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5,
+	0xd807aa98,0x12835b01,0x243185be,0x550c7dc3,0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf794,
+	0xf59b89c2,0x73924787,0x23c6886e,0xa42ca65c,0x15ed3627,0x4d6edcbf,0xe28217fc,0xef02488f,
+	0xb707775c,0x0468c23f,0xe7e72b4c,0x49e1f1a2,0x4b99c816,0x926d1570,0xaa0fc072,0xadb36e2c,
+	0xad87a3ea,0xbcb1d3a3,0x7b993186,0x562b9420,0xbff3ca0c,0xda4b0c23,0x6cd8711a,0x8f337caa,
+	0xc91b1417,0xc359dce1,0xa83253a7,0x3b13c12d,0x9d3d725d,0xd9031a84,0xb1a03340,0x16f58012,
+	0xe64fb6a2,0xe84d923a,0xe93a5730,0x09837686,0x078ff753,0x29833341,0xd5de0b7e,0x6948ccf4,
+	0xe0a1adbe,0x7c728e11,0x511c78e4,0x315b45bd,0xfca71413,0xea28f96a,0x79703128,0x4e1ef848,
+};
+
+void SHA256_fixed(uint4*restrict state0,uint4*restrict state1)
+{
+	uint4 S0 = *state0;
+	uint4 S1 = *state1;
+
+#define A S0.x
+#define B S0.y
+#define C S0.z
+#define D S0.w
+#define E S1.x
+#define F S1.y
+#define G S1.z
+#define H S1.w
+
+	RND(A,B,C,D,E,F,G,H, fixedW[0]);
+	RND(H,A,B,C,D,E,F,G, fixedW[1]);
+	RND(G,H,A,B,C,D,E,F, fixedW[2]);
+	RND(F,G,H,A,B,C,D,E, fixedW[3]);
+	RND(E,F,G,H,A,B,C,D, fixedW[4]);
+	RND(D,E,F,G,H,A,B,C, fixedW[5]);
+	RND(C,D,E,F,G,H,A,B, fixedW[6]);
+	RND(B,C,D,E,F,G,H,A, fixedW[7]);
+	RND(A,B,C,D,E,F,G,H, fixedW[8]);
+	RND(H,A,B,C,D,E,F,G, fixedW[9]);
+	RND(G,H,A,B,C,D,E,F, fixedW[10]);
+	RND(F,G,H,A,B,C,D,E, fixedW[11]);
+	RND(E,F,G,H,A,B,C,D, fixedW[12]);
+	RND(D,E,F,G,H,A,B,C, fixedW[13]);
+	RND(C,D,E,F,G,H,A,B, fixedW[14]);
+	RND(B,C,D,E,F,G,H,A, fixedW[15]);
+	RND(A,B,C,D,E,F,G,H, fixedW[16]);
+	RND(H,A,B,C,D,E,F,G, fixedW[17]);
+	RND(G,H,A,B,C,D,E,F, fixedW[18]);
+	RND(F,G,H,A,B,C,D,E, fixedW[19]);
+	RND(E,F,G,H,A,B,C,D, fixedW[20]);
+	RND(D,E,F,G,H,A,B,C, fixedW[21]);
+	RND(C,D,E,F,G,H,A,B, fixedW[22]);
+	RND(B,C,D,E,F,G,H,A, fixedW[23]);
+	RND(A,B,C,D,E,F,G,H, fixedW[24]);
+	RND(H,A,B,C,D,E,F,G, fixedW[25]);
+	RND(G,H,A,B,C,D,E,F, fixedW[26]);
+	RND(F,G,H,A,B,C,D,E, fixedW[27]);
+	RND(E,F,G,H,A,B,C,D, fixedW[28]);
+	RND(D,E,F,G,H,A,B,C, fixedW[29]);
+	RND(C,D,E,F,G,H,A,B, fixedW[30]);
+	RND(B,C,D,E,F,G,H,A, fixedW[31]);
+	RND(A,B,C,D,E,F,G,H, fixedW[32]);
+	RND(H,A,B,C,D,E,F,G, fixedW[33]);
+	RND(G,H,A,B,C,D,E,F, fixedW[34]);
+	RND(F,G,H,A,B,C,D,E, fixedW[35]);
+	RND(E,F,G,H,A,B,C,D, fixedW[36]);
+	RND(D,E,F,G,H,A,B,C, fixedW[37]);
+	RND(C,D,E,F,G,H,A,B, fixedW[38]);
+	RND(B,C,D,E,F,G,H,A, fixedW[39]);
+	RND(A,B,C,D,E,F,G,H, fixedW[40]);
+	RND(H,A,B,C,D,E,F,G, fixedW[41]);
+	RND(G,H,A,B,C,D,E,F, fixedW[42]);
+	RND(F,G,H,A,B,C,D,E, fixedW[43]);
+	RND(E,F,G,H,A,B,C,D, fixedW[44]);
+	RND(D,E,F,G,H,A,B,C, fixedW[45]);
+	RND(C,D,E,F,G,H,A,B, fixedW[46]);
+	RND(B,C,D,E,F,G,H,A, fixedW[47]);
+	RND(A,B,C,D,E,F,G,H, fixedW[48]);
+	RND(H,A,B,C,D,E,F,G, fixedW[49]);
+	RND(G,H,A,B,C,D,E,F, fixedW[50]);
+	RND(F,G,H,A,B,C,D,E, fixedW[51]);
+	RND(E,F,G,H,A,B,C,D, fixedW[52]);
+	RND(D,E,F,G,H,A,B,C, fixedW[53]);
+	RND(C,D,E,F,G,H,A,B, fixedW[54]);
+	RND(B,C,D,E,F,G,H,A, fixedW[55]);
+	RND(A,B,C,D,E,F,G,H, fixedW[56]);
+	RND(H,A,B,C,D,E,F,G, fixedW[57]);
+	RND(G,H,A,B,C,D,E,F, fixedW[58]);
+	RND(F,G,H,A,B,C,D,E, fixedW[59]);
+	RND(E,F,G,H,A,B,C,D, fixedW[60]);
+	RND(D,E,F,G,H,A,B,C, fixedW[61]);
+	RND(C,D,E,F,G,H,A,B, fixedW[62]);
+	RND(B,C,D,E,F,G,H,A, fixedW[63]);
+	
+#undef A
+#undef B
+#undef C
+#undef D
+#undef E
+#undef F
+#undef G
+#undef H
+	*state0 += S0;
+	*state1 += S1;
+}
+
+void shittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[1].x,B[2].y,B[3].z,B[0].w);
+	tmp[1] = (uint4)(B[2].x,B[3].y,B[0].z,B[1].w);
+	tmp[2] = (uint4)(B[3].x,B[0].y,B[1].z,B[2].w);
+	tmp[3] = (uint4)(B[0].x,B[1].y,B[2].z,B[3].w);
+	
+
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[5].x,B[6].y,B[7].z,B[4].w);
+	tmp[1] = (uint4)(B[6].x,B[7].y,B[4].z,B[5].w);
+	tmp[2] = (uint4)(B[7].x,B[4].y,B[5].z,B[6].w);
+	tmp[3] = (uint4)(B[4].x,B[5].y,B[6].z,B[7].w);
+	
+
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void unshittify(uint4 B[8])
+{
+	uint4 tmp[4];
+	tmp[0] = (uint4)(B[3].x,B[2].y,B[1].z,B[0].w);
+	tmp[1] = (uint4)(B[0].x,B[3].y,B[2].z,B[1].w);
+	tmp[2] = (uint4)(B[1].x,B[0].y,B[3].z,B[2].w);
+	tmp[3] = (uint4)(B[2].x,B[1].y,B[0].z,B[3].w);
+	
+
+	for(uint i=0; i<4; ++i)
+		B[i] = EndianSwap(tmp[i]);
+
+	tmp[0] = (uint4)(B[7].x,B[6].y,B[5].z,B[4].w);
+	tmp[1] = (uint4)(B[4].x,B[7].y,B[6].z,B[5].w);
+	tmp[2] = (uint4)(B[5].x,B[4].y,B[7].z,B[6].w);
+	tmp[3] = (uint4)(B[6].x,B[5].y,B[4].z,B[7].w);
+	
+
+	for(uint i=0; i<4; ++i)
+		B[i+4] = EndianSwap(tmp[i]);
+}
+
+void salsa(uint4 B[8])
+{
+	uint4 w[4];
+
+
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i]^=B[i+4]);
+
+
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+
+	for(uint i=0; i<4; ++i)
+		w[i] = (B[i+4]^=(B[i]+=w[i]));
+
+
+	for(uint i=0; i<4; ++i)
+	{
+		w[0] ^= rotl(w[3]     +w[2]     , 7U);
+		w[1] ^= rotl(w[0]     +w[3]     , 9U);
+		w[2] ^= rotl(w[1]     +w[0]     ,13U);
+		w[3] ^= rotl(w[2]     +w[1]     ,18U);
+		w[2] ^= rotl(w[3].wxyz+w[0].zwxy, 7U);
+		w[1] ^= rotl(w[2].wxyz+w[3].zwxy, 9U);
+		w[0] ^= rotl(w[1].wxyz+w[2].zwxy,13U);
+		w[3] ^= rotl(w[0].wxyz+w[1].zwxy,18U);
+	}
+
+
+	for(uint i=0; i<4; ++i)
+		B[i+4] += w[i];
+}
+
+#define Coord(x,y,z) x+y*(x ## SIZE)+z*(y ## SIZE)*(x ## SIZE)
+#define CO Coord(z,x,y)
+
+void scrypt_core(uint4 X[8], __global uint4*restrict lookup)
+{
+	shittify(X);
+	const uint zSIZE = 8;
+	const uint ySIZE = (1024/LOOKUP_GAP+(1024%LOOKUP_GAP>0));
+	const uint xSIZE = CONCURRENT_THREADS;
+	uint x = get_global_id(0)%xSIZE;
+
+	for(uint y=0; y<1024/LOOKUP_GAP; ++y)
+	{
+
+		for(uint z=0; z<zSIZE; ++z)
+			lookup[CO] = X[z];
+		for(uint i=0; i<LOOKUP_GAP; ++i) 
+			salsa(X);
+	}
+	for (uint i=0; i<1024; ++i) 
+	{
+		uint j = X[7].x & K[85];
+		uint y = (j/LOOKUP_GAP);
+
+		if (j&1)
+		{
+			uint4 V[8];
+			for(uint z=0; z<zSIZE; ++z)
+				V[z] = lookup[CO];
+
+			salsa(V);
+
+			for(uint z=0; z<zSIZE; ++z)
+				X[z] ^= V[z];
+		} else {
+			for(uint z=0; z<zSIZE; ++z)
+				X[z] ^= lookup[CO];
+		}
+
+		salsa(X);
+	}
+	unshittify(X);
+}
+
+#define FOUND (0xFF)
+#define SETFOUND(Xnonce) output[output[FOUND]++] = Xnonce
+
+__attribute__((reqd_work_group_size(WORKSIZE, 1, 1)))
+__kernel void search(__global const uint4 * restrict input,
+volatile __global uint*restrict output, __global uint4*restrict padcache,
+const uint4 midstate0, const uint4 midstate16, const uint target)
+{
+	uint gid = get_global_id(0);
+	uint4 X[8];
+	uint4 tstate0, tstate1, ostate0, ostate1, tmp0, tmp1;
+	uint4 data = (uint4)(input[4].x,input[4].y,input[4].z,gid);
+	uint4 pad0 = midstate0, pad1 = midstate16;
+
+	SHA256(&pad0,&pad1, data, (uint4)(K[84],0,0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[86]));
+	SHA256_fresh(&ostate0,&ostate1, pad0^ K[82], pad1^ K[82], K[82], K[82]);
+	SHA256_fresh(&tstate0,&tstate1, pad0^ K[83], pad1^ K[83], K[83], K[83]);
+
+	tmp0 = tstate0;
+	tmp1 = tstate1;
+	SHA256(&tstate0, &tstate1, input[0],input[1],input[2],input[3]);
+
+
+	for (uint i=0; i<4; i++) 
+	{
+		pad0 = tstate0;
+		pad1 = tstate1;
+		X[i<<1 ] = ostate0;
+		X[(i<<1)+1] = ostate1;
+
+		SHA256(&pad0,&pad1, data, (uint4)(i+1,K[84],0,0), (uint4)(0,0,0,0), (uint4)(0,0,0, K[87]));
+		SHA256(X+(i<<1),X+(i<<1)+1, pad0, pad1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+	}
+	scrypt_core(X,padcache);
+	SHA256(&tmp0,&tmp1, X[0], X[1], X[2], X[3]);
+	SHA256(&tmp0,&tmp1, X[4], X[5], X[6], X[7]);
+	SHA256_fixed(&tmp0,&tmp1);
+	SHA256(&ostate0,&ostate1, tmp0, tmp1, (uint4)(K[84], 0U, 0U, 0U), (uint4)(0U, 0U, 0U, K[88]));
+
+	bool result = (EndianSwap(ostate1.w) <= target);
+	if (result)
+		SETFOUND(gid);
+}
--- a/miner.h
+++ b/miner.h
@ -372,7 +372,10 @@ enum dev_enable {
				@@ -372,7 +372,10 @@ enum dev_enable {

 enum cl_kernels {
 	KL_NONE,
-	KL_SCRYPT,
+	KL_ALEXKARNEW,
+	KL_ALEXKAROLD,
+	KL_CKOLIVAS,
+	KL_ZUIKKIS,
 };

 enum dev_reason {
--- a/ocl.c
+++ b/ocl.c
@ -400,8 +400,8 @@ _clState *initCl(unsigned int gpu, char *name, size_t nameSize)
				@@ -400,8 +400,8 @@ _clState *initCl(unsigned int gpu, char *name, size_t nameSize)
 	char numbuf[16];

 	if (cgpu->kernel == KL_NONE) {
-		applog(LOG_INFO, "Selecting scrypt kernel");
-		clState->chosen_kernel = KL_SCRYPT;
+		applog(LOG_INFO, "Selecting kernel ckolivas");
+		clState->chosen_kernel = KL_CKOLIVAS;
 		cgpu->kernel = clState->chosen_kernel;
 	} else {
 		clState->chosen_kernel = cgpu->kernel;
@ -415,12 +415,32 @@ _clState *initCl(unsigned int gpu, char *name, size_t nameSize)
				@@ -415,12 +415,32 @@ _clState *initCl(unsigned int gpu, char *name, size_t nameSize)
 	else if (preferred_vwidth > 2)
 		preferred_vwidth = 2;

+	/* All available kernels only support vector 1 */
+	cgpu->vwidth = 1;
+
 	switch (clState->chosen_kernel) {
-		case KL_SCRYPT:
-			strcpy(filename, SCRYPT_KERNNAME".cl");
-			strcpy(binaryfilename, SCRYPT_KERNNAME);
-			/* Scrypt only supports vector 1 */
-			cgpu->vwidth = 1;
+		case KL_ALEXKARNEW:
+			applog(LOG_WARNING, "Kernel alexkarnew is experimental.");
+			strcpy(filename, ALEXKARNEW_KERNNAME".cl");
+			strcpy(binaryfilename, ALEXKARNEW_KERNNAME);
+			break;
+		case KL_ALEXKAROLD:
+			applog(LOG_WARNING, "Kernel alexkarold is experimental.");
+			strcpy(filename, ALEXKAROLD_KERNNAME".cl");
+			strcpy(binaryfilename, ALEXKAROLD_KERNNAME);
+			break;
+		case KL_CKOLIVAS:
+			strcpy(filename, CKOLIVAS_KERNNAME".cl");
+			strcpy(binaryfilename, CKOLIVAS_KERNNAME);
+			break;
+		case KL_ZUIKKIS:
+			applog(LOG_WARNING, "Kernel zuikkis is experimental.");
+			strcpy(filename, ZUIKKIS_KERNNAME".cl");
+			strcpy(binaryfilename, ZUIKKIS_KERNNAME);
+			/* Kernel only supports lookup-gap 2 */
+			cgpu->lookup_gap = 2;
+			/* Kernel only supports worksize 256 */
+			cgpu->work_size = 256;
 			break;
 		case KL_NONE: /* Shouldn't happen */
 			break;
--- a/sgminer.c
+++ b/sgminer.c
@ -4205,8 +4205,17 @@ void write_config(FILE *fcfg)
				@@ -4205,8 +4205,17 @@ void write_config(FILE *fcfg)
 			switch (gpus[i].kernel) {
 				case KL_NONE: // Shouldn't happen
 					break;
-				case KL_SCRYPT:
-					fprintf(fcfg, "scrypt");
+				case KL_ALEXKARNEW:
+					fprintf(fcfg, ALEXKARNEW_KERNNAME);
+					break;
+				case KL_ALEXKAROLD:
+					fprintf(fcfg, ALEXKAROLD_KERNNAME);
+					break;
+				case KL_CKOLIVAS:
+					fprintf(fcfg, CKOLIVAS_KERNNAME);
+					break;
+				case KL_ZUIKKIS:
+					fprintf(fcfg, ZUIKKIS_KERNNAME);
 					break;
 			}
 		}