; Cc: Yang, Rong R <rong.r.y...@intel.com>
> Subject: [Beignet] [PATCH] GBE: remove stacksize 64KB limitation.
>
> From: "Yang, Rong R" <rong.r.y...@intel.com>
>
> If stacksize large 64KB, the formula of calculate the stackptr should
> change, form "
From: "Yang, Rong R"
If stacksize large 64KB, the formula of calculate the stackptr should
change, form "threadId * perThreadSize + laneId*perLaneSize" to
"(threadId * simdWidth + laneId)*perLaneSize", to avoid Dword * Dword.
V2: Only support UD * UW, and UD is IMM must
> > + p->MUL(stackptr, stackptr, GenRegister::immud(perLaneSize)); //
> > + (threadId
> > * simdWidth + laneId)*perLaneSize
> According to Hardware Spec:
> For IVB and HSW, When both src0 and src1 are of type D or UD, only the low
> 16 bits of each element of src1 are used. The accumulator
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Yang Rong
> Sent: Sunday, February 14, 2016 2:42 PM
> To: beignet@lists.freedesktop.org
> Cc: Yang, Rong R <rong.r.y...@intel.com>
> Subject: [Beignet] [PATCH]
If stacksize large 64KB, the formula of calculate the stackptr should
change, form "threadId * perThreadSize + laneId*perLaneSize" to
"(threadId * simdWidth + laneId)*perLaneSize", to avoid Dword * Dword.
Signed-off-by: Yang Rong
---
backend/src/backend/context.cpp