[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-17 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-921934672 Thanks @tqchen for the explanation of two viewpoints of how we could see the constraints. I do agree that we should put more constraints progressively to guide the compiler in

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-17 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-921782373 @tqchen , > To say it in another way, we cannot say that "global" definitely mean no stack allocation. The current issue is in the device "CPU" && 'global' certainly

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-16 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-921152468 Thanks @tqchen . In the R4, I was not suggesting to change TOPI but saying we could just do a Pass to change the storage scope. What are your thoughts about making the pass to

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-16 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-921044001 Thanks @tqchen for summarizing the ideas and presenting possible resolutions. The two needs seems very valid. For N0, The operators should really be tagged with 'local'

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-16 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-920880447 I feel that is a workaround for an optimization that certain set of CPUs require it. Moreover, TVMBAWs allows an abstraction to use an Arena that could be used in any memory

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-15 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-920580279 Our proposal is to add a check to that loop whether it has 'local' storage_scope before we place them into the stack as it is the solution that works for the wider definition of the

[GitHub] [tvm] manupa-arm commented on issue #9022: [Bug] BuiltinLower does not use alloca for storage on kDLCPU target devices

2021-09-15 Thread GitBox
manupa-arm commented on issue #9022: URL: https://github.com/apache/tvm/issues/9022#issuecomment-920577170 @tqchen @mbs-octoml , This is not specific to Arm(R) Ethos(TM)-U codegen and its generally applicable for any micro controller where we would want to avoid creating allocation