Fix a couple of bugs in AMD split buffer mode

- FASTEXIT only works well on AMD, disable it for non-AMD GPUs.
- Seems the number DAG entries is not always even!
- Reduce GPU memory footprint where possible.

Author: jean-m-cyr

libethash-cl/CLMiner.cpp

@@ -22,7 +22,7 @@ namespace eth
 // WARNING: Do not change the value of the following constant
 // unless you are prepared to make the neccessary adjustments
 // to the assembly code for the binary kernels.
-const size_t c_maxSearchResults = 15;
+const size_t c_maxSearchResults = 4;
 
 struct CLChannel : public LogChannel
 {
@@ -625,6 +625,11 @@ bool CLMiner::initDevice()
         cllog << "Unrecognized Platform";
         return false;
     }
+    if (!m_settings.noExit && (m_hwmoninfo.deviceType != HwMonitorInfoType::AMD))
+    {
+        m_settings.noExit = true;
+        cllog << "no exit option enabled for non AMD opencl device";
+    }
 
     if (m_deviceDescriptor.clPlatformVersionMajor == 1 &&
         (m_deviceDescriptor.clPlatformVersionMinor == 0 ||
@@ -839,8 +844,20 @@ bool CLMiner::initEpoch_internal()
                   << dev::getFormattedMemory(
                          (double)(m_deviceDescriptor.totalMemory - RequiredMemory));
             m_dag.clear();
-            m_dag.push_back(cl::Buffer(m_context[0], CL_MEM_READ_ONLY, m_epochContext.dagSize / 2));
-            m_dag.push_back(cl::Buffer(m_context[0], CL_MEM_READ_ONLY, m_epochContext.dagSize / 2));
+            if (m_epochContext.dagNumItems & 1)
+            {
+                m_dag.push_back(
+                    cl::Buffer(m_context[0], CL_MEM_READ_ONLY, m_epochContext.dagSize / 2 + 64));
+                m_dag.push_back(
+                    cl::Buffer(m_context[0], CL_MEM_READ_ONLY, m_epochContext.dagSize / 2 - 64));
+            }
+            else
+            {
+                m_dag.push_back(
+                    cl::Buffer(m_context[0], CL_MEM_READ_ONLY, (m_epochContext.dagSize) / 2));
+                m_dag.push_back(
+                    cl::Buffer(m_context[0], CL_MEM_READ_ONLY, (m_epochContext.dagSize) / 2));
+            }
             cllog << "Loading kernels";
 
             // If we have a binary kernel to use, let's try it

libethash-cl/kernels/bin/ethash_baffin_lws128.bin

@@ -31,7 +31,7 @@
         .arg target, "ulong", ulong
     .text
 
-        MAX_SEARCH_RESULTS = 15
+        MAX_SEARCH_RESULTS = 4
         result_count_offset = (MAX_SEARCH_RESULTS * 64)
 
         mix0 = %v24

libethash-cl/kernels/bin/ethash_baffin_lws128_exit.bin

@@ -31,7 +31,7 @@
         .arg target, "ulong", ulong
     .text
 
-        MAX_SEARCH_RESULTS = 15
+        MAX_SEARCH_RESULTS = 4
         result_count_offset = (MAX_SEARCH_RESULTS * 64)
 
         mix0 = %v24