DFAPacketizer

Project Infrastructure LLVM Dev List Archives
1

``

``

``

``
---
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation

Hi,

I am having problems writing the ProcessorItineraries list. As instructions on my VLIW target have varying size I want to model both cpu units and bundle bits as FUs. The following does not work, to my surprise:

InstrItinData<ALU, [InstrStage<1, [BITS1,BITS2, BITS3, BITS4], 0>,

InstrStage<1, [ALU1, ALU2]>]>

I want to express that there are two ALU’s, and four bundle slots, and in this case while allocating this type of instructions several times, I would like to be allowed only two such instructions per cycle.

ResourceTracker->reserveResources();

ResourceTracker->reserveResources();

ResourceTracker->reserveResources();

ResourceTracker->reserveResources();

ResourceTracker->reserveResources();

DFAPacketizer.cpp:84: void llvm::DFAPacketizer::reserveResources(const llvm::MCInstrDesc*): Assertion `CachedTable.count(StateTrans) != 0’ failed.

It seems to not want to worry about the alu’s, but it is fine as long as there are any FU of the first row (BITS) available.

What is missing?

Thanks,

Jonas Paulsson

2

Hi,

I looked a bit through the mail archives, and found this question answered in Oct 2011 (see below). It is interesting to find this in the ARM backend, considering your answer. Can you give more information about for example is this a temporary deficiency in the DFAPacketizer? What is the IIC_iMOVi itinerary doing below?

Thanks,

Jonas

Thu Oct 6 15:11:25 CDT 2011:

Hello Hal.

> Is there a way to express a multiple pipeline itinerary using the

> current scheme

Yes, surely

> (maybe some trick with setting NextCycles = 0)?

Yep!

> Specifically, consider a case where a floating-point load simultaneously

> uses units from a floating-point pipeline and a load/store pipeline.

Look into ARM itineraries, they contain a decent amount of such examples.

[Head feb 2013:]

InstrItinData<IIC_iMOVi , [InstrStage<1, [SW_DIS0, SW_DIS1, SW_DIS2], 0>,

InstrStage<1, [SW_ALU0, SW_ALU1]>],

[1]>,

3

Hi Jonas,

> It is interesting to find this in the ARM backend, considering your answer.

The ARM backend doesn't use the DFA packetizer. It's only used by Hexagon. At this point, there is no plan to address thisin the DFA packetizer since none of the supported targets needthe functionality.

Thanks
-Anshu

4

Hi Anshu,

Would there be any interest in extending this algorithm to handling more extensive models, such as VLIW scheduling based on FU’s and bundle space… ie handle multiple stages ?

I might do it and commit, if there is acceptance and guidance…

Jonas

5

Would there be any interest in extending this algorithm to handling

more extensive models, such

as VLIW scheduling based on FU's and bundle space... ie handle multiple

stages ?

If there is a supported backend that needs this then yes, we should extend the DFA packetizer. But I am not in favor of extending the DFA packetizer if there are no clients of the added functionality in llvm sources.

-Anshu