Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

OpenCl: optimize cn-v8 div #2086

Merged
merged 1 commit into from
Nov 21, 2018
Merged

OpenCl: optimize cn-v8 div #2086

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions xmrstak/backend/amd/amd_gpu/opencl/cryptonight.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -401,7 +401,7 @@ __kernel void JOIN(cn0,ALGO)(__global ulong *input, __global uint4 *Scratchpad,
AES2[i] = rotate(tmp, 16U);
AES3[i] = rotate(tmp, 24U);
}

__local ulong State_buf[8 * 25];

barrier(CLK_LOCAL_MEM_FENCE);
Expand Down Expand Up @@ -474,12 +474,12 @@ __kernel void JOIN(cn0,ALGO)(__global ulong *input, __global uint4 *Scratchpad,
}

mem_fence(CLK_LOCAL_MEM_FENCE);

// cryptonight_heavy || cryptonight_haven || cryptonight_bittube2
#if (ALGO == 4 || ALGO == 9 || ALGO == 10)
__local uint4 xin[8][8];
{


/* Also left over threads perform this loop.
* The left over thread results will be ignored
Expand Down Expand Up @@ -530,7 +530,7 @@ __kernel void JOIN(cn0,ALGO)(__global ulong *input, __global uint4 *Scratchpad,

)==="
R"===(

// cryptonight_monero_v8 && NVIDIA
#if(ALGO==11 && defined(__NV_CL_C_VERSION))
# define SCRATCHPAD_CHUNK(N) (*(__local uint4*)((__local uchar*)(scratchpad_line) + (idxS ^ (N << 4))))
Expand Down Expand Up @@ -630,7 +630,7 @@ __kernel void JOIN(cn1,ALGO) (__global uint4 *Scratchpad, __global ulong *states
tweak1_2 ^= as_uint2(states[24]);
#endif
}

mem_fence(CLK_LOCAL_MEM_FENCE);

#if(COMP_MODE==1)
Expand Down Expand Up @@ -755,7 +755,7 @@ __kernel void JOIN(cn1,ALGO) (__global uint4 *Scratchpad, __global ulong *states
#endif

((uint4 *)a)[0] ^= tmp;

// cryptonight_monero_v8
#if (ALGO == 11)
# if defined(__NV_CL_C_VERSION)
Expand Down Expand Up @@ -815,7 +815,7 @@ __kernel void JOIN(cn2,ALGO) (__global uint4 *Scratchpad, __global ulong *states
__local uint4 xin1[8][8];
__local uint4 xin2[8][8];
#endif

#if(COMP_MODE==1)
// do not use early return here
if(gIdx < Threads)
Expand Down
74 changes: 24 additions & 50 deletions xmrstak/backend/amd/amd_gpu/opencl/fast_int_math_v2.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,61 +68,35 @@ inline uint get_reciprocal(const __local uchar *RCP, uint a)

inline uint2 fast_div_v2(const __local uint *RCP, ulong a, uint b)
{
const uint r = get_reciprocal((const __local uchar *)RCP, b);
const ulong k = mul_hi(as_uint2(a).s0, r) + ((ulong)(r) * as_uint2(a).s1) + a;

ulong q;
((uint*)&q)[0] = as_uint2(k).s1;

#if defined(cl_amd_device_attribute_query) && (OPENCL_DRIVER_MAJOR == 14)
/* The AMD driver 14.XX is not able to compile `(k < a)`
* https://github.com/fireice-uk/xmr-stak/issues/1922
* This is a workaround for the broken compiler.
*/
ulong whyAMDwhy;
((uint*)&whyAMDwhy)[0] = as_uint2(k).s0;
((uint*)&whyAMDwhy)[1] = as_uint2(k).s1;
((uint*)&q)[1] = (whyAMDwhy < a) ? 1U : 0U;
#else
((uint*)&q)[1] = (k < a) ? 1U : 0U;
#endif

const long tmp = a - q * b;
const bool overshoot = (tmp < 0);
const bool undershoot = (tmp >= b);

return (uint2)(
as_uint2(q).s0 + (undershoot ? 1U : 0U) - (overshoot ? 1U : 0U),
as_uint2(tmp).s0 + (overshoot ? b : 0U) - (undershoot ? b : 0U)
);
const uint r = get_reciprocal((const __local uchar *)RCP, b);

const ulong k = mul_hi(as_uint2(a).s0, r) + ((ulong)(r) * as_uint2(a).s1) + a;
const uint q = as_uint2(k).s1;
long tmp = a - ((ulong)(q) * b);
((int*)&tmp)[1] -= (as_uint2(k).s1 < as_uint2(a).s1) ? b : 0;
const int overshoot = ((int*)&tmp)[1] >> 31;
const int undershoot = as_int2(as_uint(b - 1) - tmp).s1 >> 31;
return (uint2)(q + overshoot - undershoot, as_uint2(tmp).s0 + (as_uint(overshoot) & b) - (as_uint(undershoot) & b));
}

inline uint fast_sqrt_v2(const ulong n1)
{
float x = as_float((as_uint2(n1).s1 >> 9) + ((64U + 127U) << 23));

float x1 = native_rsqrt(x);
x = native_sqrt(x);

// The following line does x1 *= 4294967296.0f;
x1 = as_float(as_uint(x1) + (32U << 23));

const uint x0 = as_uint(x) - (158U << 23);
const long delta0 = n1 - (((long)(x0) * x0) << 18);
const float delta = convert_float_rte(as_int2(delta0).s1) * x1;

uint result = (x0 << 10) + convert_int_rte(delta);
const uint s = result >> 1;
const uint b = result & 1;

const ulong x2 = (ulong)(s) * (s + b) + ((ulong)(result) << 32) - n1;
if ((long)(x2 + b) > 0) --result;
if ((long)(x2 + 0x100000000UL + s) < 0) ++result;

return result;
float x = as_float((as_uint2(n1).s1 >> 9) + ((64U + 127U) << 23));
float x1 = native_rsqrt(x);
x = native_sqrt(x);
// The following line does x1 *= 4294967296.0f;
x1 = as_float(as_uint(x1) + (32U << 23));
const uint x0 = as_uint(x) - (158U << 23);
const long delta0 = n1 - (as_ulong((uint2)(mul24(x0, x0), mul_hi(x0, x0))) << 18);
const float delta = convert_float_rte(as_int2(delta0).s1) * x1;
uint result = (x0 << 10) + convert_int_rte(delta);
const uint s = result >> 1;
const uint b = result & 1;
const ulong x2 = (ulong)(s) * (s + b) + ((ulong)(result) << 32) - n1;
if ((long)(x2 + as_int(b - 1)) >= 0) --result;
if ((long)(x2 + 0x100000000UL + s) < 0) ++result;
return result;
}

#endif

)==="