rasterizer/core/api.cpp

/****************************************************************************
* Copyright (C) 2014-2016 Intel Corporation.   All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
*
* @file api.cpp
*
* @brief API implementation
*
******************************************************************************/

#include <cfloat>
#include <cmath>
#include <cstdio>
#include <new>

#include "core/api.h"
#include "core/backend.h"
#include "core/context.h"
#include "core/depthstencil.h"
#include "core/frontend.h"
#include "core/rasterizer.h"
#include "core/rdtsc_core.h"
#include "core/threads.h"
#include "core/tilemgr.h"
#include "core/clip.h"
#include "core/utils.h"

#include "common/simdintrin.h"
#include "common/os.h"

static const SWR_RECT g_MaxScissorRect = { 0, 0, KNOB_MAX_SCISSOR_X, KNOB_MAX_SCISSOR_Y };

void SetupDefaultState(SWR_CONTEXT *pContext);

static INLINE SWR_CONTEXT* GetContext(HANDLE hContext)
{
    return (SWR_CONTEXT*)hContext;
}

void WakeAllThreads(SWR_CONTEXT *pContext)
{
    pContext->FifosNotEmpty.notify_all();
}

//////////////////////////////////////////////////////////////////////////
/// @brief Create SWR Context.
/// @param pCreateInfo - pointer to creation info.
HANDLE SwrCreateContext(
    SWR_CREATECONTEXT_INFO* pCreateInfo)
{
    RDTSC_RESET();
    RDTSC_INIT(0);

    void* pContextMem = AlignedMalloc(sizeof(SWR_CONTEXT), KNOB_SIMD_WIDTH * 4);
    memset(pContextMem, 0, sizeof(SWR_CONTEXT));
    SWR_CONTEXT *pContext = new (pContextMem) SWR_CONTEXT();

    pContext->privateStateSize = pCreateInfo->privateStateSize;

    pContext->dcRing.Init(KNOB_MAX_DRAWS_IN_FLIGHT);
    pContext->dsRing.Init(KNOB_MAX_DRAWS_IN_FLIGHT);

    pContext->pMacroTileManagerArray = (MacroTileMgr*)AlignedMalloc(sizeof(MacroTileMgr) * KNOB_MAX_DRAWS_IN_FLIGHT, 64);
    pContext->pDispatchQueueArray = (DispatchQueue*)AlignedMalloc(sizeof(DispatchQueue) * KNOB_MAX_DRAWS_IN_FLIGHT, 64);

    for (uint32_t dc = 0; dc < KNOB_MAX_DRAWS_IN_FLIGHT; ++dc)
    {
        pContext->dcRing[dc].pArena = new CachingArena(pContext->cachingArenaAllocator);
        new (&pContext->pMacroTileManagerArray[dc]) MacroTileMgr(*pContext->dcRing[dc].pArena);
        new (&pContext->pDispatchQueueArray[dc]) DispatchQueue();

        pContext->dsRing[dc].pArena = new CachingArena(pContext->cachingArenaAllocator);
    }

    pContext->threadInfo.MAX_WORKER_THREADS        = KNOB_MAX_WORKER_THREADS;
    pContext->threadInfo.MAX_NUMA_NODES            = KNOB_MAX_NUMA_NODES;
    pContext->threadInfo.MAX_CORES_PER_NUMA_NODE   = KNOB_MAX_CORES_PER_NUMA_NODE;
    pContext->threadInfo.MAX_THREADS_PER_CORE      = KNOB_MAX_THREADS_PER_CORE;
    pContext->threadInfo.SINGLE_THREADED           = KNOB_SINGLE_THREADED;

    if (pCreateInfo->pThreadInfo)
    {
        pContext->threadInfo = *pCreateInfo->pThreadInfo;
    }

    memset(&pContext->WaitLock, 0, sizeof(pContext->WaitLock));
    memset(&pContext->FifosNotEmpty, 0, sizeof(pContext->FifosNotEmpty));
    new (&pContext->WaitLock) std::mutex();
    new (&pContext->FifosNotEmpty) std::condition_variable();

    CreateThreadPool(pContext, &pContext->threadPool);

    pContext->ppScratch = new uint8_t*[pContext->NumWorkerThreads];
    pContext->pStats = new SWR_STATS[pContext->NumWorkerThreads];

#if defined(KNOB_ENABLE_AR)
    // Setup ArchRast thread contexts which includes +1 for API thread.
    pContext->pArContext = new HANDLE[pContext->NumWorkerThreads+1];
    pContext->pArContext[pContext->NumWorkerThreads] = ArchRast::CreateThreadContext(ArchRast::AR_THREAD::API);
#endif

    // Allocate scratch space for workers.
    ///@note We could lazily allocate this but its rather small amount of memory.
    for (uint32_t i = 0; i < pContext->NumWorkerThreads; ++i)
    {
#if defined(_WIN32)
        uint32_t numaNode = pContext->threadPool.pThreadData ?
            pContext->threadPool.pThreadData[i].numaId : 0;
        pContext->ppScratch[i] = (uint8_t*)VirtualAllocExNuma(
            GetCurrentProcess(), nullptr, 32 * sizeof(KILOBYTE),
            MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE,
            numaNode);
#else
        pContext->ppScratch[i] = (uint8_t*)AlignedMalloc(32 * sizeof(KILOBYTE), KNOB_SIMD_WIDTH * 4);
#endif

#if defined(KNOB_ENABLE_AR)
        // Initialize worker thread context for ArchRast.
        pContext->pArContext[i] = ArchRast::CreateThreadContext(ArchRast::AR_THREAD::WORKER);
#endif
    }

    // State setup AFTER context is fully initialized
    SetupDefaultState(pContext);

    // initialize hot tile manager
    pContext->pHotTileMgr = new HotTileMgr();

    // initialize function pointer tables
    InitClearTilesTable();

    // initialize callback functions
    pContext->pfnLoadTile = pCreateInfo->pfnLoadTile;
    pContext->pfnStoreTile = pCreateInfo->pfnStoreTile;
    pContext->pfnClearTile = pCreateInfo->pfnClearTile;
    pContext->pfnUpdateSoWriteOffset = pCreateInfo->pfnUpdateSoWriteOffset;
    pContext->pfnUpdateStats = pCreateInfo->pfnUpdateStats;
    pContext->pfnUpdateStatsFE = pCreateInfo->pfnUpdateStatsFE;


    // pass pointer to bucket manager back to caller
#ifdef KNOB_ENABLE_RDTSC
    pCreateInfo->pBucketMgr = &gBucketMgr;
#endif

    pCreateInfo->contextSaveSize = sizeof(API_STATE);

    StartThreadPool(pContext, &pContext->threadPool);

    return (HANDLE)pContext;
}

void CopyState(DRAW_STATE& dst, const DRAW_STATE& src)
{
    memcpy(&dst.state, &src.state, sizeof(API_STATE));
}

template<bool IsDraw>
void QueueWork(SWR_CONTEXT *pContext)
{
    DRAW_CONTEXT* pDC = pContext->pCurDrawContext;
    uint32_t dcIndex = pDC->drawId % KNOB_MAX_DRAWS_IN_FLIGHT;

    if (IsDraw)
    {
        pDC->pTileMgr = &pContext->pMacroTileManagerArray[dcIndex];
        pDC->pTileMgr->initialize();
    }

    // Each worker thread looks at a DC for both FE and BE work at different times and so we
    // multiply threadDone by 2.  When the threadDone counter has reached 0 then all workers
    // have moved past this DC. (i.e. Each worker has checked this DC for both FE and BE work and
    // then moved on if all work is done.)
    pContext->pCurDrawContext->threadsDone = pContext->NumFEThreads + pContext->NumBEThreads;

    if (IsDraw)
    {
        InterlockedIncrement((volatile LONG*)&pContext->drawsOutstandingFE);
    }

    _ReadWriteBarrier();
    {
        std::unique_lock<std::mutex> lock(pContext->WaitLock);
        pContext->dcRing.Enqueue();
    }

    if (pContext->threadInfo.SINGLE_THREADED)
    {
        // flush denormals to 0
        uint32_t mxcsr = _mm_getcsr();
        _mm_setcsr(mxcsr | _MM_FLUSH_ZERO_ON | _MM_DENORMALS_ZERO_ON);

        if (IsDraw)
        {
            uint32_t curDraw[2] = { pContext->pCurDrawContext->drawId, pContext->pCurDrawContext->drawId };
            WorkOnFifoFE(pContext, 0, curDraw[0]);
            WorkOnFifoBE(pContext, 0, curDraw[1], pContext->singleThreadLockedTiles, 0, 0);
        }
        else
        {
            uint32_t curDispatch = pContext->pCurDrawContext->drawId;
            WorkOnCompute(pContext, 0, curDispatch);
        }

        // Dequeue the work here, if not already done, since we're single threaded (i.e. no workers).
        while (CompleteDrawContext(pContext, pContext->pCurDrawContext) > 0) {}

        // restore csr
        _mm_setcsr(mxcsr);
    }
    else
    {
        AR_API_BEGIN(APIDrawWakeAllThreads, pDC->drawId);
        WakeAllThreads(pContext);
        AR_API_END(APIDrawWakeAllThreads, 1);
    }

    // Set current draw context to NULL so that next state call forces a new draw context to be created and populated.
    pContext->pPrevDrawContext = pContext->pCurDrawContext;
    pContext->pCurDrawContext = nullptr;
}

INLINE void QueueDraw(SWR_CONTEXT* pContext)
{
    QueueWork<true>(pContext);
}

INLINE void QueueDispatch(SWR_CONTEXT* pContext)
{
    QueueWork<false>(pContext);
}

DRAW_CONTEXT* GetDrawContext(SWR_CONTEXT *pContext, bool isSplitDraw = false)
{
    AR_API_BEGIN(APIGetDrawContext, 0);
    // If current draw context is null then need to obtain a new draw context to use from ring.
    if (pContext->pCurDrawContext == nullptr)
    {
        // Need to wait for a free entry.
        while (pContext->dcRing.IsFull())
        {
            _mm_pause();
        }

        uint64_t curDraw = pContext->dcRing.GetHead();
        uint32_t dcIndex = curDraw % KNOB_MAX_DRAWS_IN_FLIGHT;

        if ((pContext->frameCount - pContext->lastFrameChecked) > 2 ||
            (curDraw - pContext->lastDrawChecked) > 0x10000)
        {
            // Take this opportunity to clean-up old arena allocations
            pContext->cachingArenaAllocator.FreeOldBlocks();

            pContext->lastFrameChecked = pContext->frameCount;
            pContext->lastDrawChecked = curDraw;
        }

        DRAW_CONTEXT* pCurDrawContext = &pContext->dcRing[dcIndex];
        pContext->pCurDrawContext = pCurDrawContext;

        // Assign next available entry in DS ring to this DC.
        uint32_t dsIndex = pContext->curStateId % KNOB_MAX_DRAWS_IN_FLIGHT;
        pCurDrawContext->pState = &pContext->dsRing[dsIndex];

        // Copy previous state to current state.
        if (pContext->pPrevDrawContext)
        {
            DRAW_CONTEXT* pPrevDrawContext = pContext->pPrevDrawContext;

            // If we're splitting our draw then we can just use the same state from the previous
            // draw. In this case, we won't increment the DS ring index so the next non-split
            // draw can receive the state.
            if (isSplitDraw == false)
            {
                CopyState(*pCurDrawContext->pState, *pPrevDrawContext->pState);

                // Should have been cleaned up previously
                SWR_ASSERT(pCurDrawContext->pState->pArena->IsEmpty() == true);

                pCurDrawContext->pState->pPrivateState = nullptr;

                pContext->curStateId++;  // Progress state ring index forward.
            }
            else
            {
                // If its a split draw then just copy the state pointer over
                // since its the same draw.
                pCurDrawContext->pState = pPrevDrawContext->pState;
                SWR_ASSERT(pPrevDrawContext->cleanupState == false);
            }
        }
        else
        {
            SWR_ASSERT(pCurDrawContext->pState->pArena->IsEmpty() == true);
            pContext->curStateId++;  // Progress state ring index forward.
        }

        SWR_ASSERT(pCurDrawContext->pArena->IsEmpty() == true);

        // Reset dependency
        pCurDrawContext->dependent = false;
        pCurDrawContext->dependentFE = false;

        pCurDrawContext->pContext = pContext;
        pCurDrawContext->isCompute = false; // Dispatch has to set this to true.

        pCurDrawContext->doneFE = false;
        pCurDrawContext->FeLock = 0;
        pCurDrawContext->threadsDone = 0;
        pCurDrawContext->retireCallback.pfnCallbackFunc = nullptr;

        pCurDrawContext->dynState.Reset(pContext->NumWorkerThreads);

        // Assign unique drawId for this DC
        pCurDrawContext->drawId = pContext->dcRing.GetHead();

        pCurDrawContext->cleanupState = true;

    }
    else
    {
        SWR_ASSERT(isSplitDraw == false, "Split draw should only be used when obtaining a new DC");
    }

    AR_API_END(APIGetDrawContext, 0);
    return pContext->pCurDrawContext;
}

API_STATE* GetDrawState(SWR_CONTEXT *pContext)
{
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);
    SWR_ASSERT(pDC->pState != nullptr);

    return &pDC->pState->state;
}

void SwrDestroyContext(HANDLE hContext)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    pDC->FeWork.type = SHUTDOWN;
    pDC->FeWork.pfnWork = ProcessShutdown;

    //enqueue
    QueueDraw(pContext);

    DestroyThreadPool(pContext, &pContext->threadPool);

    // free the fifos
    for (uint32_t i = 0; i < KNOB_MAX_DRAWS_IN_FLIGHT; ++i)
    {
        delete[] pContext->dcRing[i].dynState.pStats;
        delete pContext->dcRing[i].pArena;
        delete pContext->dsRing[i].pArena;
        pContext->pMacroTileManagerArray[i].~MacroTileMgr();
        pContext->pDispatchQueueArray[i].~DispatchQueue();
    }

    AlignedFree(pContext->pDispatchQueueArray);
    AlignedFree(pContext->pMacroTileManagerArray);

    // Free scratch space.
    for (uint32_t i = 0; i < pContext->NumWorkerThreads; ++i)
    {
#if defined(_WIN32)
        VirtualFree(pContext->ppScratch[i], 0, MEM_RELEASE);
#else
        AlignedFree(pContext->ppScratch[i]);
#endif

#if defined(KNOB_ENABLE_AR)
        ArchRast::DestroyThreadContext(pContext->pArContext[i]);
#endif
    }

    delete[] pContext->ppScratch;
    delete[] pContext->pStats;

    delete(pContext->pHotTileMgr);

    pContext->~SWR_CONTEXT();
    AlignedFree(GetContext(hContext));
}

void SWR_API SwrSaveState(
    HANDLE hContext,
    void* pOutputStateBlock,
    size_t memSize)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    auto pSrc = GetDrawState(pContext);
    SWR_ASSERT(pOutputStateBlock && memSize >= sizeof(*pSrc));

    memcpy(pOutputStateBlock, pSrc, sizeof(*pSrc));
}

void SWR_API SwrRestoreState(
    HANDLE hContext,
    const void* pStateBlock,
    size_t memSize)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    auto pDst = GetDrawState(pContext);
    SWR_ASSERT(pStateBlock && memSize >= sizeof(*pDst));

    memcpy(pDst, pStateBlock, sizeof(*pDst));
}

void SetupDefaultState(SWR_CONTEXT *pContext)
{
    API_STATE* pState = GetDrawState(pContext);

    pState->rastState.cullMode = SWR_CULLMODE_NONE;
    pState->rastState.frontWinding = SWR_FRONTWINDING_CCW;

    pState->depthBoundsState.depthBoundsTestEnable = false;
    pState->depthBoundsState.depthBoundsTestMinValue = 0.0f;
    pState->depthBoundsState.depthBoundsTestMaxValue = 1.0f;
}

void SwrSync(HANDLE hContext, PFN_CALLBACK_FUNC pfnFunc, uint64_t userData, uint64_t userData2, uint64_t userData3)
{
    SWR_ASSERT(pfnFunc != nullptr);

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    AR_API_BEGIN(APISync, 0);

    pDC->FeWork.type = SYNC;
    pDC->FeWork.pfnWork = ProcessSync;

    // Setup callback function
    pDC->retireCallback.pfnCallbackFunc = pfnFunc;
    pDC->retireCallback.userData = userData;
    pDC->retireCallback.userData2 = userData2;
    pDC->retireCallback.userData3 = userData3;

    //enqueue
    QueueDraw(pContext);

    AR_API_END(APISync, 1);
}

void SwrWaitForIdle(HANDLE hContext)
{
    SWR_CONTEXT *pContext = GetContext(hContext);

    AR_API_BEGIN(APIWaitForIdle, 0);

    while (!pContext->dcRing.IsEmpty())
    {
        _mm_pause();
    }

    AR_API_END(APIWaitForIdle, 1);
}

void SwrWaitForIdleFE(HANDLE hContext)
{
    SWR_CONTEXT *pContext = GetContext(hContext);

    AR_API_BEGIN(APIWaitForIdle, 0);

    while (pContext->drawsOutstandingFE > 0)
    {
        _mm_pause();
    }

    AR_API_END(APIWaitForIdle, 1);
}

void SwrSetVertexBuffers(
    HANDLE hContext,
    uint32_t numBuffers,
    const SWR_VERTEX_BUFFER_STATE* pVertexBuffers)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    for (uint32_t i = 0; i < numBuffers; ++i)
    {
        const SWR_VERTEX_BUFFER_STATE *pVB = &pVertexBuffers[i];
        pState->vertexBuffers[pVB->index] = *pVB;
    }
}

void SwrSetIndexBuffer(
    HANDLE hContext,
    const SWR_INDEX_BUFFER_STATE* pIndexBuffer)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->indexBuffer = *pIndexBuffer;
}

void SwrSetFetchFunc(
    HANDLE hContext,
    PFN_FETCH_FUNC    pfnFetchFunc)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->pfnFetchFunc = pfnFetchFunc;
}

void SwrSetSoFunc(
    HANDLE hContext,
    PFN_SO_FUNC    pfnSoFunc,
    uint32_t streamIndex)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    SWR_ASSERT(streamIndex < MAX_SO_STREAMS);

    pState->pfnSoFunc[streamIndex] = pfnSoFunc;
}

void SwrSetSoState(
    HANDLE hContext,
    SWR_STREAMOUT_STATE* pSoState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->soState = *pSoState;
}

void SwrSetSoBuffers(
    HANDLE hContext,
    SWR_STREAMOUT_BUFFER* pSoBuffer,
    uint32_t slot)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    SWR_ASSERT((slot < 4), "There are only 4 SO buffer slots [0, 3]\nSlot requested: %d", slot);

    pState->soBuffer[slot] = *pSoBuffer;
}

void SwrSetVertexFunc(
    HANDLE hContext,
    PFN_VERTEX_FUNC pfnVertexFunc)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->pfnVertexFunc = pfnVertexFunc;
}

void SwrSetFrontendState(
    HANDLE hContext,
    SWR_FRONTEND_STATE *pFEState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));
    pState->frontendState = *pFEState;
}

void SwrSetGsState(
    HANDLE hContext,
    SWR_GS_STATE *pGSState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));
    pState->gsState = *pGSState;
}

void SwrSetGsFunc(
    HANDLE hContext,
    PFN_GS_FUNC pfnGsFunc)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));
    pState->pfnGsFunc = pfnGsFunc;
}

void SwrSetCsFunc(
    HANDLE hContext,
    PFN_CS_FUNC pfnCsFunc,
    uint32_t totalThreadsInGroup,
    uint32_t totalSpillFillSize)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));
    pState->pfnCsFunc = pfnCsFunc;
    pState->totalThreadsInGroup = totalThreadsInGroup;
    pState->totalSpillFillSize = totalSpillFillSize;
}

void SwrSetTsState(
    HANDLE hContext,
    SWR_TS_STATE *pState)
{
    API_STATE* pApiState = GetDrawState(GetContext(hContext));
    pApiState->tsState = *pState;
}

void SwrSetHsFunc(
    HANDLE hContext,
    PFN_HS_FUNC pfnFunc)
{
    API_STATE* pApiState = GetDrawState(GetContext(hContext));
    pApiState->pfnHsFunc = pfnFunc;
}

void SwrSetDsFunc(
    HANDLE hContext,
    PFN_DS_FUNC pfnFunc)
{
    API_STATE* pApiState = GetDrawState(GetContext(hContext));
    pApiState->pfnDsFunc = pfnFunc;
}

void SwrSetDepthStencilState(
    HANDLE hContext,
    SWR_DEPTH_STENCIL_STATE *pDSState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->depthStencilState = *pDSState;
}

void SwrSetBackendState(
    HANDLE hContext,
    SWR_BACKEND_STATE *pBEState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->backendState = *pBEState;
}

void SwrSetDepthBoundsState(
    HANDLE hContext,
    SWR_DEPTH_BOUNDS_STATE *pDBState)
{
    API_STATE* pState = GetDrawState(GetContext(hContext));

    pState->depthBoundsState = *pDBState;
}

void SwrSetPixelShaderState(
    HANDLE hContext,
    SWR_PS_STATE *pPSState)
{
    API_STATE *pState = GetDrawState(GetContext(hContext));
    pState->psState = *pPSState;
}

void SwrSetBlendState(
    HANDLE hContext,
    SWR_BLEND_STATE *pBlendState)
{
    API_STATE *pState = GetDrawState(GetContext(hContext));
    memcpy(&pState->blendState, pBlendState, sizeof(SWR_BLEND_STATE));
}

void SwrSetBlendFunc(
    HANDLE hContext,
    uint32_t renderTarget,
    PFN_BLEND_JIT_FUNC pfnBlendFunc)
{
    SWR_ASSERT(renderTarget < SWR_NUM_RENDERTARGETS);
    API_STATE *pState = GetDrawState(GetContext(hContext));
    pState->pfnBlendFunc[renderTarget] = pfnBlendFunc;
}

// update guardband multipliers for the viewport
void updateGuardbands(API_STATE *pState)
{
    uint32_t numGbs = pState->gsState.emitsRenderTargetArrayIndex ? KNOB_NUM_VIEWPORTS_SCISSORS : 1;

    for(uint32_t i = 0; i < numGbs; ++i)
    {
        // guardband center is viewport center
        pState->gbState.left[i] = KNOB_GUARDBAND_WIDTH / pState->vp[i].width;
        pState->gbState.right[i] = KNOB_GUARDBAND_WIDTH / pState->vp[i].width;
        pState->gbState.top[i] = KNOB_GUARDBAND_HEIGHT / pState->vp[i].height;
        pState->gbState.bottom[i] = KNOB_GUARDBAND_HEIGHT / pState->vp[i].height;
    }
}

void SwrSetRastState(
    HANDLE hContext,
    const SWR_RASTSTATE *pRastState)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    API_STATE* pState = GetDrawState(pContext);

    memcpy(&pState->rastState, pRastState, sizeof(SWR_RASTSTATE));
}

void SwrSetViewports(
    HANDLE hContext,
    uint32_t numViewports,
    const SWR_VIEWPORT* pViewports,
    const SWR_VIEWPORT_MATRICES* pMatrices)
{
    SWR_ASSERT(numViewports <= KNOB_NUM_VIEWPORTS_SCISSORS,
        "Invalid number of viewports.");

    SWR_CONTEXT *pContext = GetContext(hContext);
    API_STATE* pState = GetDrawState(pContext);

    memcpy(&pState->vp[0], pViewports, sizeof(SWR_VIEWPORT) * numViewports);
    // @todo Faster to copy portions of the SOA or just copy all of it?
    memcpy(&pState->vpMatrices, pMatrices, sizeof(SWR_VIEWPORT_MATRICES));

    updateGuardbands(pState);
}

void SwrSetScissorRects(
    HANDLE hContext,
    uint32_t numScissors,
    const SWR_RECT* pScissors)
{
    SWR_ASSERT(numScissors <= KNOB_NUM_VIEWPORTS_SCISSORS,
        "Invalid number of scissor rects.");

    API_STATE* pState = GetDrawState(GetContext(hContext));
    memcpy(&pState->scissorRects[0], pScissors, numScissors * sizeof(pScissors[0]));
};

void SetupMacroTileScissors(DRAW_CONTEXT *pDC)
{
    API_STATE *pState = &pDC->pState->state;
    uint32_t numScissors = pState->gsState.emitsViewportArrayIndex ? KNOB_NUM_VIEWPORTS_SCISSORS : 1;
    pState->scissorsTileAligned = true;

    for (uint32_t index = 0; index < numScissors; ++index)
    {
        SWR_RECT &scissorInFixedPoint = pState->scissorsInFixedPoint[index];

        // Set up scissor dimensions based on scissor or viewport
        if (pState->rastState.scissorEnable)
        {
            scissorInFixedPoint = pState->scissorRects[index];
        }
        else
        {
            // the vp width and height must be added to origin un-rounded then the result round to -inf.
            // The cast to int works for rounding assuming all [left, right, top, bottom] are positive.
            scissorInFixedPoint.xmin = (int32_t)pState->vp[index].x;
            scissorInFixedPoint.xmax = (int32_t)(pState->vp[index].x + pState->vp[index].width);
            scissorInFixedPoint.ymin = (int32_t)pState->vp[index].y;
            scissorInFixedPoint.ymax = (int32_t)(pState->vp[index].y + pState->vp[index].height);
        }

        // Clamp to max rect
        scissorInFixedPoint &= g_MaxScissorRect;

        // Test for tile alignment
        bool tileAligned;
        tileAligned  = (scissorInFixedPoint.xmin % KNOB_TILE_X_DIM) == 0;
        tileAligned &= (scissorInFixedPoint.ymin % KNOB_TILE_Y_DIM) == 0;
        tileAligned &= (scissorInFixedPoint.xmax % KNOB_TILE_X_DIM) == 0;
        tileAligned &= (scissorInFixedPoint.ymax % KNOB_TILE_Y_DIM) == 0;

        pState->scissorsTileAligned &= tileAligned;

        // Scale to fixed point
        scissorInFixedPoint.xmin *= FIXED_POINT_SCALE;
        scissorInFixedPoint.xmax *= FIXED_POINT_SCALE;
        scissorInFixedPoint.ymin *= FIXED_POINT_SCALE;
        scissorInFixedPoint.ymax *= FIXED_POINT_SCALE;

        // Make scissor inclusive
        scissorInFixedPoint.xmax -= 1;
        scissorInFixedPoint.ymax -= 1;
    }
}

// templated backend function tables
extern PFN_BACKEND_FUNC gBackendNullPs[SWR_MULTISAMPLE_TYPE_COUNT];
extern PFN_BACKEND_FUNC gBackendSingleSample[SWR_INPUT_COVERAGE_COUNT][2][2];
extern PFN_BACKEND_FUNC gBackendPixelRateTable[SWR_MULTISAMPLE_TYPE_COUNT][SWR_MSAA_SAMPLE_PATTERN_COUNT][SWR_INPUT_COVERAGE_COUNT][2][2][2];
extern PFN_BACKEND_FUNC gBackendSampleRateTable[SWR_MULTISAMPLE_TYPE_COUNT][SWR_INPUT_COVERAGE_COUNT][2][2];
void SetupPipeline(DRAW_CONTEXT *pDC)
{
    SWR_CONTEXT* pContext = pDC->pContext;
    DRAW_STATE* pState = pDC->pState;
    const SWR_RASTSTATE &rastState = pState->state.rastState;
    const SWR_PS_STATE &psState = pState->state.psState;
    BACKEND_FUNCS& backendFuncs = pState->backendFuncs;
    const uint32_t forcedSampleCount = (rastState.forcedSampleCount) ? 1 : 0;

    // setup backend
    if (psState.pfnPixelShader == nullptr)
    {
        backendFuncs.pfnBackend = gBackendNullPs[pState->state.rastState.sampleCount];
    }
    else
    {
        const bool bMultisampleEnable = ((rastState.sampleCount > SWR_MULTISAMPLE_1X) || rastState.forcedSampleCount) ? 1 : 0;
        const uint32_t centroid = ((psState.barycentricsMask & SWR_BARYCENTRIC_CENTROID_MASK) > 0) ? 1 : 0;
        const uint32_t canEarlyZ = (psState.forceEarlyZ || (!psState.writesODepth && !psState.usesSourceDepth && !psState.usesUAV)) ? 1 : 0;

        SWR_BARYCENTRICS_MASK barycentricsMask = (SWR_BARYCENTRICS_MASK)psState.barycentricsMask;

        // select backend function
        switch(psState.shadingRate)
        {
        case SWR_SHADING_RATE_PIXEL:
            if(bMultisampleEnable)
            {
                // always need to generate I & J per sample for Z interpolation
                barycentricsMask = (SWR_BARYCENTRICS_MASK)(barycentricsMask | SWR_BARYCENTRIC_PER_SAMPLE_MASK);
                backendFuncs.pfnBackend = gBackendPixelRateTable[rastState.sampleCount][rastState.samplePattern][psState.inputCoverage][centroid][forcedSampleCount][canEarlyZ];
            }
            else
            {
                // always need to generate I & J per pixel for Z interpolation
                barycentricsMask = (SWR_BARYCENTRICS_MASK)(barycentricsMask | SWR_BARYCENTRIC_PER_PIXEL_MASK);
                backendFuncs.pfnBackend = gBackendSingleSample[psState.inputCoverage][centroid][canEarlyZ];
            }
            break;
        case SWR_SHADING_RATE_SAMPLE:
            SWR_ASSERT(rastState.samplePattern == SWR_MSAA_STANDARD_PATTERN);
            // always need to generate I & J per sample for Z interpolation
            barycentricsMask = (SWR_BARYCENTRICS_MASK)(barycentricsMask | SWR_BARYCENTRIC_PER_SAMPLE_MASK);
            backendFuncs.pfnBackend = gBackendSampleRateTable[rastState.sampleCount][psState.inputCoverage][centroid][canEarlyZ];
            break;
        default:
            SWR_ASSERT(0 && "Invalid shading rate");
            break;
        }
    }

    PFN_PROCESS_PRIMS pfnBinner;
    switch (pState->state.topology)
    {
    case TOP_POINT_LIST:
        pState->pfnProcessPrims = ClipPoints;
        pfnBinner = BinPoints;
        break;
    case TOP_LINE_LIST:
    case TOP_LINE_STRIP:
    case TOP_LINE_LOOP:
    case TOP_LINE_LIST_ADJ:
    case TOP_LISTSTRIP_ADJ:
        pState->pfnProcessPrims = ClipLines;
        pfnBinner = BinLines;
        break;
    default:
        pState->pfnProcessPrims = ClipTriangles;
        pfnBinner = GetBinTrianglesFunc((rastState.conservativeRast > 0));
        break;
    };


    // disable clipper if viewport transform is disabled
    if (pState->state.frontendState.vpTransformDisable)
    {
        pState->pfnProcessPrims = pfnBinner;
    }

    if ((pState->state.psState.pfnPixelShader == nullptr) &&
        (pState->state.depthStencilState.depthTestEnable == FALSE) &&
        (pState->state.depthStencilState.depthWriteEnable == FALSE) &&
        (pState->state.depthStencilState.stencilTestEnable == FALSE) &&
        (pState->state.depthStencilState.stencilWriteEnable == FALSE) &&
        (pState->state.backendState.numAttributes == 0))
    {
        pState->pfnProcessPrims = nullptr;
    }

    if (pState->state.soState.rasterizerDisable == true)
    {
        pState->pfnProcessPrims = nullptr;
    }


    // set up the frontend attribute count
    pState->state.feNumAttributes = 0;
    const SWR_BACKEND_STATE& backendState = pState->state.backendState;
    if (backendState.swizzleEnable)
    {
        // attribute swizzling is enabled, iterate over the map and record the max attribute used
        for (uint32_t i = 0; i < backendState.numAttributes; ++i)
        {
            pState->state.feNumAttributes = std::max(pState->state.feNumAttributes, (uint32_t)backendState.swizzleMap[i].sourceAttrib + 1);
        }
    }
    else
    {
        pState->state.feNumAttributes = pState->state.backendState.numAttributes;
    }

    if (pState->state.soState.soEnable)
    {
        uint32_t streamMasks = 0;
        for (uint32_t i = 0; i < 4; ++i)
        {
            streamMasks |= pState->state.soState.streamMasks[i];
        }

        DWORD maxAttrib;
        if (_BitScanReverse(&maxAttrib, streamMasks))
        {
            pState->state.feNumAttributes = std::max(pState->state.feNumAttributes, (uint32_t)(maxAttrib + 1));
        }
    }

    // complicated logic to test for cases where we don't need backing hottile memory for a draw
    // have to check for the special case where depth/stencil test is enabled but depthwrite is disabled.
    pState->state.depthHottileEnable = ((!(pState->state.depthStencilState.depthTestEnable &&
                                           !pState->state.depthStencilState.depthWriteEnable &&
                                           !pState->state.depthBoundsState.depthBoundsTestEnable &&
                                           pState->state.depthStencilState.depthTestFunc == ZFUNC_ALWAYS)) &&
                                        (pState->state.depthStencilState.depthTestEnable ||
                                         pState->state.depthStencilState.depthWriteEnable ||
                                         pState->state.depthBoundsState.depthBoundsTestEnable)) ? true : false;

    pState->state.stencilHottileEnable = (((!(pState->state.depthStencilState.stencilTestEnable &&
                                             !pState->state.depthStencilState.stencilWriteEnable &&
                                              pState->state.depthStencilState.stencilTestFunc == ZFUNC_ALWAYS)) ||
                                          // for stencil we have to check the double sided state as well
                                          (!(pState->state.depthStencilState.doubleSidedStencilTestEnable &&
                                             !pState->state.depthStencilState.stencilWriteEnable &&
                                              pState->state.depthStencilState.backfaceStencilTestFunc == ZFUNC_ALWAYS))) &&
                                          (pState->state.depthStencilState.stencilTestEnable  ||
                                           pState->state.depthStencilState.stencilWriteEnable)) ? true : false;

    uint32_t numRTs = pState->state.psState.numRenderTargets;
    pState->state.colorHottileEnable = 0;
    if (psState.pfnPixelShader != nullptr)
    {
        for (uint32_t rt = 0; rt < numRTs; ++rt)
        {
            pState->state.colorHottileEnable |=
                (!pState->state.blendState.renderTarget[rt].writeDisableAlpha ||
                 !pState->state.blendState.renderTarget[rt].writeDisableRed ||
                 !pState->state.blendState.renderTarget[rt].writeDisableGreen ||
                 !pState->state.blendState.renderTarget[rt].writeDisableBlue) ? (1 << rt) : 0;
        }
    }

    // Setup depth quantization function
    if (pState->state.depthHottileEnable)
    {
        switch (pState->state.rastState.depthFormat)
        {
        case R32_FLOAT_X8X24_TYPELESS: pState->state.pfnQuantizeDepth = QuantizeDepth < R32_FLOAT_X8X24_TYPELESS > ; break;
        case R32_FLOAT: pState->state.pfnQuantizeDepth = QuantizeDepth < R32_FLOAT > ; break;
        case R24_UNORM_X8_TYPELESS: pState->state.pfnQuantizeDepth = QuantizeDepth < R24_UNORM_X8_TYPELESS > ; break;
        case R16_UNORM: pState->state.pfnQuantizeDepth = QuantizeDepth < R16_UNORM > ; break;
        default: SWR_ASSERT(false, "Unsupported depth format for depth quantiztion.");
            pState->state.pfnQuantizeDepth = QuantizeDepth < R32_FLOAT > ;
        }
    }
    else
    {
        // set up pass-through quantize if depth isn't enabled
        pState->state.pfnQuantizeDepth = QuantizeDepth < R32_FLOAT > ;
    }
}

//////////////////////////////////////////////////////////////////////////
/// @brief InitDraw
/// @param pDC - Draw context to initialize for this draw.
void InitDraw(
    DRAW_CONTEXT *pDC,
    bool isSplitDraw)
{
    // We don't need to re-setup the scissors/pipeline state again for split draw.
    if (isSplitDraw == false)
    {
        SetupMacroTileScissors(pDC);
        SetupPipeline(pDC);
    }


}

//////////////////////////////////////////////////////////////////////////
/// @brief We can split the draw for certain topologies for better performance.
/// @param totalVerts - Total vertices for draw
/// @param topology - Topology used for draw
uint32_t MaxVertsPerDraw(
    DRAW_CONTEXT* pDC,
    uint32_t totalVerts,
    PRIMITIVE_TOPOLOGY topology)
{
    API_STATE& state = pDC->pState->state;

    uint32_t vertsPerDraw = totalVerts;

    if (state.soState.soEnable)
    {
        return totalVerts;
    }

    switch (topology)
    {
    case TOP_POINT_LIST:
    case TOP_TRIANGLE_LIST:
        vertsPerDraw = KNOB_MAX_PRIMS_PER_DRAW;
        break;

    case TOP_PATCHLIST_1:
    case TOP_PATCHLIST_2:
    case TOP_PATCHLIST_3:
    case TOP_PATCHLIST_4:
    case TOP_PATCHLIST_5:
    case TOP_PATCHLIST_6:
    case TOP_PATCHLIST_7:
    case TOP_PATCHLIST_8:
    case TOP_PATCHLIST_9:
    case TOP_PATCHLIST_10:
    case TOP_PATCHLIST_11:
    case TOP_PATCHLIST_12:
    case TOP_PATCHLIST_13:
    case TOP_PATCHLIST_14:
    case TOP_PATCHLIST_15:
    case TOP_PATCHLIST_16:
    case TOP_PATCHLIST_17:
    case TOP_PATCHLIST_18:
    case TOP_PATCHLIST_19:
    case TOP_PATCHLIST_20:
    case TOP_PATCHLIST_21:
    case TOP_PATCHLIST_22:
    case TOP_PATCHLIST_23:
    case TOP_PATCHLIST_24:
    case TOP_PATCHLIST_25:
    case TOP_PATCHLIST_26:
    case TOP_PATCHLIST_27:
    case TOP_PATCHLIST_28:
    case TOP_PATCHLIST_29:
    case TOP_PATCHLIST_30:
    case TOP_PATCHLIST_31:
    case TOP_PATCHLIST_32:
        if (pDC->pState->state.tsState.tsEnable)
        {
            uint32_t vertsPerPrim = topology - TOP_PATCHLIST_BASE;
            vertsPerDraw = vertsPerPrim * KNOB_MAX_TESS_PRIMS_PER_DRAW;
        }
        break;

    // The Primitive Assembly code can only handle 1 RECT at a time.
    case TOP_RECT_LIST:
        vertsPerDraw = 3;
        break;

    default:
        // We are not splitting up draws for other topologies.
        break;
    }

    return vertsPerDraw;
}


//////////////////////////////////////////////////////////////////////////
/// @brief DrawInstanced
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param numVerts - How many vertices to read sequentially from vertex data (per instance).
/// @param startVertex - Specifies start vertex for draw. (vertex data)
/// @param numInstances - How many instances to render.
/// @param startInstance - Which instance to start sequentially fetching from in each buffer (instanced data)
void DrawInstanced(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t numVertices,
    uint32_t startVertex,
    uint32_t numInstances = 1,
    uint32_t startInstance = 0)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    AR_API_BEGIN(APIDraw, pDC->drawId);
    AR_API_EVENT(DrawInstancedEvent(pDC->drawId, topology, numVertices, startVertex, numInstances, startInstance));

    uint32_t maxVertsPerDraw = MaxVertsPerDraw(pDC, numVertices, topology);
    uint32_t primsPerDraw = GetNumPrims(topology, maxVertsPerDraw);
    uint32_t remainingVerts = numVertices;

    API_STATE    *pState = &pDC->pState->state;
    pState->topology = topology;
    pState->forceFront = false;

    // disable culling for points/lines
    uint32_t oldCullMode = pState->rastState.cullMode;
    if (topology == TOP_POINT_LIST)
    {
        pState->rastState.cullMode = SWR_CULLMODE_NONE;
        pState->forceFront = true;
    }
    else if (topology == TOP_RECT_LIST)
    {
        pState->rastState.cullMode = SWR_CULLMODE_NONE;
    }


    int draw = 0;
    while (remainingVerts)
    {
        uint32_t numVertsForDraw = (remainingVerts < maxVertsPerDraw) ?
        remainingVerts : maxVertsPerDraw;

        bool isSplitDraw = (draw > 0) ? true : false;
        DRAW_CONTEXT* pDC = GetDrawContext(pContext, isSplitDraw);
        InitDraw(pDC, isSplitDraw);

        pDC->FeWork.type = DRAW;
        pDC->FeWork.pfnWork = GetProcessDrawFunc(
            false,  // IsIndexed
            false, // bEnableCutIndex
            pState->tsState.tsEnable,
            pState->gsState.gsEnable,
            pState->soState.soEnable,
            pDC->pState->pfnProcessPrims != nullptr);
        pDC->FeWork.desc.draw.numVerts = numVertsForDraw;
        pDC->FeWork.desc.draw.startVertex = startVertex;
        pDC->FeWork.desc.draw.numInstances = numInstances;
        pDC->FeWork.desc.draw.startInstance = startInstance;
        pDC->FeWork.desc.draw.startPrimID = draw * primsPerDraw;
        pDC->FeWork.desc.draw.startVertexID = draw * maxVertsPerDraw;

        pDC->cleanupState = (remainingVerts == numVertsForDraw);

        //enqueue DC
        QueueDraw(pContext);

        remainingVerts -= numVertsForDraw;
        draw++;
    }

    // restore culling state
    pDC = GetDrawContext(pContext);
    pDC->pState->state.rastState.cullMode = oldCullMode;


    AR_API_END(APIDraw, numVertices * numInstances);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrDraw
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param startVertex - Specifies start vertex in vertex buffer for draw.
/// @param primCount - Number of vertices.
void SwrDraw(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t startVertex,
    uint32_t numVertices)
{
    DrawInstanced(hContext, topology, numVertices, startVertex);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrDrawInstanced
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param numVertsPerInstance - How many vertices to read sequentially from vertex data.
/// @param numInstances - How many instances to render.
/// @param startVertex - Specifies start vertex for draw. (vertex data)
/// @param startInstance - Which instance to start sequentially fetching from in each buffer (instanced data)
void SwrDrawInstanced(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t numVertsPerInstance,
    uint32_t numInstances,
    uint32_t startVertex,
    uint32_t startInstance
    )
{
    DrawInstanced(hContext, topology, numVertsPerInstance, startVertex, numInstances, startInstance);
}

//////////////////////////////////////////////////////////////////////////
/// @brief DrawIndexedInstanced
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param numIndices - Number of indices to read sequentially from index buffer.
/// @param indexOffset - Starting index into index buffer.
/// @param baseVertex - Vertex in vertex buffer to consider as index "0". Note value is signed.
/// @param numInstances - Number of instances to render.
/// @param startInstance - Which instance to start sequentially fetching from in each buffer (instanced data)
void DrawIndexedInstance(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t numIndices,
    uint32_t indexOffset,
    int32_t baseVertex,
    uint32_t numInstances = 1,
    uint32_t startInstance = 0)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);
    API_STATE* pState = &pDC->pState->state;

    AR_API_BEGIN(APIDrawIndexed, pDC->drawId);
    AR_API_EVENT(DrawIndexedInstancedEvent(pDC->drawId, topology, numIndices, indexOffset, baseVertex, numInstances, startInstance));

    uint32_t maxIndicesPerDraw = MaxVertsPerDraw(pDC, numIndices, topology);
    uint32_t primsPerDraw = GetNumPrims(topology, maxIndicesPerDraw);
    uint32_t remainingIndices = numIndices;

    uint32_t indexSize = 0;
    switch (pState->indexBuffer.format)
    {
    case R32_UINT: indexSize = sizeof(uint32_t); break;
    case R16_UINT: indexSize = sizeof(uint16_t); break;
    case R8_UINT: indexSize = sizeof(uint8_t); break;
    default:
        SWR_ASSERT(0);
    }

    int draw = 0;
    uint8_t *pIB = (uint8_t*)pState->indexBuffer.pIndices;
    pIB += (uint64_t)indexOffset * (uint64_t)indexSize;

    pState->topology = topology;
    pState->forceFront = false;

    // disable culling for points/lines
    uint32_t oldCullMode = pState->rastState.cullMode;
    if (topology == TOP_POINT_LIST)
    {
        pState->rastState.cullMode = SWR_CULLMODE_NONE;
        pState->forceFront = true;
    }
    else if (topology == TOP_RECT_LIST)
    {
        pState->rastState.cullMode = SWR_CULLMODE_NONE;
    }


    while (remainingIndices)
    {
        uint32_t numIndicesForDraw = (remainingIndices < maxIndicesPerDraw) ?
        remainingIndices : maxIndicesPerDraw;

        // When breaking up draw, we need to obtain new draw context for each iteration.
        bool isSplitDraw = (draw > 0) ? true : false;

        pDC = GetDrawContext(pContext, isSplitDraw);
        InitDraw(pDC, isSplitDraw);

        pDC->FeWork.type = DRAW;
        pDC->FeWork.pfnWork = GetProcessDrawFunc(
            true,   // IsIndexed
            pState->frontendState.bEnableCutIndex,
            pState->tsState.tsEnable,
            pState->gsState.gsEnable,
            pState->soState.soEnable,
            pDC->pState->pfnProcessPrims != nullptr);
        pDC->FeWork.desc.draw.pDC = pDC;
        pDC->FeWork.desc.draw.numIndices = numIndicesForDraw;
        pDC->FeWork.desc.draw.pIB = (int*)pIB;
        pDC->FeWork.desc.draw.type = pDC->pState->state.indexBuffer.format;

        pDC->FeWork.desc.draw.numInstances = numInstances;
        pDC->FeWork.desc.draw.startInstance = startInstance;
        pDC->FeWork.desc.draw.baseVertex = baseVertex;
        pDC->FeWork.desc.draw.startPrimID = draw * primsPerDraw;

        pDC->cleanupState = (remainingIndices == numIndicesForDraw);

        //enqueue DC
        QueueDraw(pContext);

        pIB += maxIndicesPerDraw * indexSize;
        remainingIndices -= numIndicesForDraw;
        draw++;
    }

    // Restore culling state
    pDC = GetDrawContext(pContext);
    pDC->pState->state.rastState.cullMode = oldCullMode;


    AR_API_END(APIDrawIndexed, numIndices * numInstances);
}


//////////////////////////////////////////////////////////////////////////
/// @brief DrawIndexed
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param numIndices - Number of indices to read sequentially from index buffer.
/// @param indexOffset - Starting index into index buffer.
/// @param baseVertex - Vertex in vertex buffer to consider as index "0". Note value is signed.
void SwrDrawIndexed(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t numIndices,
    uint32_t indexOffset,
    int32_t baseVertex
    )
{
    DrawIndexedInstance(hContext, topology, numIndices, indexOffset, baseVertex);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrDrawIndexedInstanced
/// @param hContext - Handle passed back from SwrCreateContext
/// @param topology - Specifies topology for draw.
/// @param numIndices - Number of indices to read sequentially from index buffer.
/// @param numInstances - Number of instances to render.
/// @param indexOffset - Starting index into index buffer.
/// @param baseVertex - Vertex in vertex buffer to consider as index "0". Note value is signed.
/// @param startInstance - Which instance to start sequentially fetching from in each buffer (instanced data)
void SwrDrawIndexedInstanced(
    HANDLE hContext,
    PRIMITIVE_TOPOLOGY topology,
    uint32_t numIndices,
    uint32_t numInstances,
    uint32_t indexOffset,
    int32_t baseVertex,
    uint32_t startInstance)
{
    DrawIndexedInstance(hContext, topology, numIndices, indexOffset, baseVertex, numInstances, startInstance);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrInvalidateTiles
/// @param hContext - Handle passed back from SwrCreateContext
/// @param attachmentMask - The mask specifies which surfaces attached to the hottiles to invalidate.
/// @param invalidateRect - The pixel-coordinate rectangle to invalidate.  This will be expanded to
///                         be hottile size-aligned.
void SWR_API SwrInvalidateTiles(
    HANDLE hContext,
    uint32_t attachmentMask,
    const SWR_RECT& invalidateRect)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    pDC->FeWork.type = DISCARDINVALIDATETILES;
    pDC->FeWork.pfnWork = ProcessDiscardInvalidateTiles;
    pDC->FeWork.desc.discardInvalidateTiles.attachmentMask = attachmentMask;
    pDC->FeWork.desc.discardInvalidateTiles.rect = invalidateRect;
    pDC->FeWork.desc.discardInvalidateTiles.rect &= g_MaxScissorRect;
    pDC->FeWork.desc.discardInvalidateTiles.newTileState = SWR_TILE_INVALID;
    pDC->FeWork.desc.discardInvalidateTiles.createNewTiles = false;
    pDC->FeWork.desc.discardInvalidateTiles.fullTilesOnly = false;

    //enqueue
    QueueDraw(pContext);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrDiscardRect
/// @param hContext - Handle passed back from SwrCreateContext
/// @param attachmentMask - The mask specifies which surfaces attached to the hottiles to discard.
/// @param rect - The pixel-coordinate rectangle to discard.  Only fully-covered hottiles will be
///               discarded.
void SWR_API SwrDiscardRect(
    HANDLE hContext,
    uint32_t attachmentMask,
    const SWR_RECT& rect)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    // Queue a load to the hottile
    pDC->FeWork.type = DISCARDINVALIDATETILES;
    pDC->FeWork.pfnWork = ProcessDiscardInvalidateTiles;
    pDC->FeWork.desc.discardInvalidateTiles.attachmentMask = attachmentMask;
    pDC->FeWork.desc.discardInvalidateTiles.rect = rect;
    pDC->FeWork.desc.discardInvalidateTiles.rect &= g_MaxScissorRect;
    pDC->FeWork.desc.discardInvalidateTiles.newTileState = SWR_TILE_RESOLVED;
    pDC->FeWork.desc.discardInvalidateTiles.createNewTiles = true;
    pDC->FeWork.desc.discardInvalidateTiles.fullTilesOnly = true;

    //enqueue
    QueueDraw(pContext);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrDispatch
/// @param hContext - Handle passed back from SwrCreateContext
/// @param threadGroupCountX - Number of thread groups dispatched in X direction
/// @param threadGroupCountY - Number of thread groups dispatched in Y direction
/// @param threadGroupCountZ - Number of thread groups dispatched in Z direction
void SwrDispatch(
    HANDLE hContext,
    uint32_t threadGroupCountX,
    uint32_t threadGroupCountY,
    uint32_t threadGroupCountZ)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    AR_API_BEGIN(APIDispatch, pDC->drawId);
    AR_API_EVENT(DispatchEvent(pDC->drawId, threadGroupCountX, threadGroupCountY, threadGroupCountZ));
    pDC->isCompute = true;      // This is a compute context.

    COMPUTE_DESC* pTaskData = (COMPUTE_DESC*)pDC->pArena->AllocAligned(sizeof(COMPUTE_DESC), 64);

    pTaskData->threadGroupCountX = threadGroupCountX;
    pTaskData->threadGroupCountY = threadGroupCountY;
    pTaskData->threadGroupCountZ = threadGroupCountZ;

    uint32_t totalThreadGroups = threadGroupCountX * threadGroupCountY * threadGroupCountZ;
    uint32_t dcIndex = pDC->drawId % KNOB_MAX_DRAWS_IN_FLIGHT;
    pDC->pDispatch = &pContext->pDispatchQueueArray[dcIndex];
    pDC->pDispatch->initialize(totalThreadGroups, pTaskData, &ProcessComputeBE);

    QueueDispatch(pContext);
    AR_API_END(APIDispatch, threadGroupCountX * threadGroupCountY * threadGroupCountZ);
}

// Deswizzles, converts and stores current contents of the hot tiles to surface
// described by pState
void SWR_API SwrStoreTiles(
    HANDLE hContext,
    uint32_t attachmentMask,
    SWR_TILE_STATE postStoreTileState,
    const SWR_RECT& storeRect)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    AR_API_BEGIN(APIStoreTiles, pDC->drawId);

    pDC->FeWork.type = STORETILES;
    pDC->FeWork.pfnWork = ProcessStoreTiles;
    pDC->FeWork.desc.storeTiles.attachmentMask = attachmentMask;
    pDC->FeWork.desc.storeTiles.postStoreTileState = postStoreTileState;
    pDC->FeWork.desc.storeTiles.rect = storeRect;
    pDC->FeWork.desc.storeTiles.rect &= g_MaxScissorRect;

    //enqueue
    QueueDraw(pContext);

    AR_API_END(APIStoreTiles, 1);
}

//////////////////////////////////////////////////////////////////////////
/// @brief SwrClearRenderTarget - Clear attached render targets / depth / stencil
/// @param hContext - Handle passed back from SwrCreateContext
/// @param attachmentMask - combination of SWR_ATTACHMENT_*_BIT attachments to clear
/// @param renderTargetArrayIndex - the RT array index to clear
/// @param clearColor - color use for clearing render targets
/// @param z - depth value use for clearing depth buffer
/// @param stencil - stencil value used for clearing stencil buffer
/// @param clearRect - The pixel-coordinate rectangle to clear in all cleared buffers
void SWR_API SwrClearRenderTarget(
    HANDLE hContext,
    uint32_t attachmentMask,
    uint32_t renderTargetArrayIndex,
    const float clearColor[4],
    float z,
    uint8_t stencil,
    const SWR_RECT& clearRect)
{
    if (KNOB_TOSS_DRAW)
    {
        return;
    }

    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    AR_API_BEGIN(APIClearRenderTarget, pDC->drawId);

    pDC->FeWork.type = CLEAR;
    pDC->FeWork.pfnWork = ProcessClear;
    pDC->FeWork.desc.clear.rect = clearRect;
    pDC->FeWork.desc.clear.rect &= g_MaxScissorRect;
    pDC->FeWork.desc.clear.attachmentMask = attachmentMask;
    pDC->FeWork.desc.clear.renderTargetArrayIndex = renderTargetArrayIndex;
    pDC->FeWork.desc.clear.clearDepth = z;
    pDC->FeWork.desc.clear.clearRTColor[0] = clearColor[0];
    pDC->FeWork.desc.clear.clearRTColor[1] = clearColor[1];
    pDC->FeWork.desc.clear.clearRTColor[2] = clearColor[2];
    pDC->FeWork.desc.clear.clearRTColor[3] = clearColor[3];
    pDC->FeWork.desc.clear.clearStencil = stencil;

    // enqueue draw
    QueueDraw(pContext);

    AR_API_END(APIClearRenderTarget, 1);
}

//////////////////////////////////////////////////////////////////////////
/// @brief Returns a pointer to the private context state for the current
///        draw operation. This is used for external componets such as the
///        sampler.
///        SWR is responsible for the allocation of the private context state.
/// @param hContext - Handle passed back from SwrCreateContext
VOID* SwrGetPrivateContextState(
    HANDLE hContext)
{
    SWR_CONTEXT* pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);
    DRAW_STATE* pState = pDC->pState;

    if (pState->pPrivateState == nullptr)
    {
        pState->pPrivateState = pState->pArena->AllocAligned(pContext->privateStateSize, KNOB_SIMD_WIDTH*sizeof(float));
    }

    return pState->pPrivateState;
}

//////////////////////////////////////////////////////////////////////////
/// @brief Clients can use this to allocate memory for draw/dispatch
///        operations. The memory will automatically be freed once operation
///        has completed. Client can use this to allocate binding tables,
///        etc. needed for shader execution.
/// @param hContext - Handle passed back from SwrCreateContext
/// @param size - Size of allocation
/// @param align - Alignment needed for allocation.
VOID* SwrAllocDrawContextMemory(
    HANDLE hContext,
    uint32_t size,
    uint32_t align)
{
    SWR_CONTEXT* pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    return pDC->pState->pArena->AllocAligned(size, align);
}

//////////////////////////////////////////////////////////////////////////
/// @brief Enables stats counting
/// @param hContext - Handle passed back from SwrCreateContext
/// @param enable - If true then counts are incremented.
void SwrEnableStatsFE(
    HANDLE hContext,
    bool enable)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    pDC->pState->state.enableStatsFE = enable;
}

//////////////////////////////////////////////////////////////////////////
/// @brief Enables stats counting
/// @param hContext - Handle passed back from SwrCreateContext
/// @param enable - If true then counts are incremented.
void SwrEnableStatsBE(
    HANDLE hContext,
    bool enable)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    pDC->pState->state.enableStatsBE = enable;
}

//////////////////////////////////////////////////////////////////////////
/// @brief Mark end of frame - used for performance profiling
/// @param hContext - Handle passed back from SwrCreateContext
void SWR_API SwrEndFrame(
    HANDLE hContext)
{
    SWR_CONTEXT *pContext = GetContext(hContext);
    DRAW_CONTEXT* pDC = GetDrawContext(pContext);

    RDTSC_ENDFRAME();
    AR_API_EVENT(FrameEndEvent(pContext->frameCount, pDC->drawId));

    pContext->frameCount++;
}