similar to: REG_SEQUENCE use question

Dan Gohman wrote: > There certainly are wishlist items for TableGen and TableGen-based > instruction descriptions, though I don't know of an official list. > Offhand, > a few things that come to mind are the ability to handle nodes with > multiple results, Is there an official workaround, BTW? - Volodya

On Jul 28, 2010, at 12:25 PM, Carlos Sánchez de La Lama wrote: > Which after register coalescing gets transformed into: > > 36 %reg16404:1<def> = LDWr %reg16384, 0; mem:LD4[<unknown>] > 76 %reg16394<def> = LDWr %reg16386<kill>, 0; mem:LD4[<unknown>] > 124 %reg16404<def> = INSERT_SUBREG %reg16404, %reg16394<kill>, 2 > 132

Hi all, We are working on a backend for a machine that has 4-wide vector register & ops, *but* not vector loads. All the vector register elements are directly accesible, so VI1 reg (Vector Integer 1) has I4, I5, I6 and I7 as its (integer) subregisters. Subregisters of same reg *never* overlap. Therefore, vector loads are lowered to scalar loads followed by a chain of INSERT_VECTOR_ELTs. Then

On Sep 9, 2010, at 12:59 PM, Borja Ferrer wrote: > Hello, i've noticed a new possible missed optimization while testing more trivial code. > This time it's not a with a xor but with a multiplication instruction and the example is little bit more involved. > > C code: > > typedef short t; > t foo(t a, t b) > { > t a4 = a*b; > return a4; > } >

Hello, i've noticed a new possible missed optimization while testing more trivial code. This time it's not a with a xor but with a multiplication instruction and the example is little bit more involved. C code: typedef short t; t foo(t a, t b) { t a4 = a*b; return a4; } argument "a" is passed in R15:R14, argument "b" in R13:R12, the return value is stored in

On May 31, 2013, at 4:07 PM, Joe Matarazzo <joe.matarazzo at gmail.com> wrote: > The register coalescer treats virtual super register classes -- a sequential register range composed of multiple hardware registers -- as a register with sub registers. When making coalescing decisions it thinks that the virtual super reg interferes with sub reg instances, even though in reality they

The register coalescer treats virtual super register classes -- a sequential register range composed of multiple hardware registers -- as a register with sub registers. When making coalescing decisions it thinks that the virtual super reg interferes with sub reg instances, even though in reality they shouldn't conflict. That is, they are individual registers and would be better compared as

On May 31, 2013, at 4:59 PM, Joe Matarazzo <joe.matarazzo at gmail.com> wrote: > I think the last time I pulled from trunk was probably end of last year. Some time ago. Does your reply intimate it's fixed on trunk? Yes, it’s been fixed recently. /jakob

I think the last time I pulled from trunk was probably end of last year. Some time ago. Does your reply intimate it's fixed on trunk? That would be great. (I don't sync too often to avoid churn with my TD.) Joe On Fri, May 31, 2013 at 4:21 PM, Jakob Stoklund Olesen <stoklund at 2pi.dk>wrote: > > On May 31, 2013, at 4:07 PM, Joe Matarazzo <joe.matarazzo at gmail.com>

> I've disassembled the failing MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4 and compared it to > the one from the LLVM 3.6.0 test runs. There's nothing obvious. We've removed some useless > 'addiu $sp,$sp,0', eliminated two (seemingly redundant) sign extends, and the addresses of > functions+data has changed slightly. I've investigated further and I'm

Was it the subreg lane masks / mapping that was added to address the missed coalescing? This solution is nice, but I don't think it'll work for me. I have 8-element vector registers that can be grouped into virtual super regs for bulk save/restore, and as soon as I have more than 4 in a tuple, the unsigned int used to hold the lane masks overflows and switches over to the "bit 31 set

This is a bit premature to be considered a code review, but given how unfamiliar I am with SelectionDAG (& that I'm seeing somewhat more 'interesting' results compared to my change to FastISel) I wanted to get a bit of feedback to see if I was on the right track or had missed any obvious cases. I've attached my patch in progress (including a modification to the existing test

I try to define a register class def GPR64 : RegisterClass<"mytarget", [i64], 64, [T0, T1.....] to simulate even/odd pair of GPR32 register. Actually, I just use GPR64 as a temporary register. My CPU just support i32 Integer type directly. I use FDR to save f64. def FDR : RegisterClass<"mytarget", [f64], 64,[FD0, FD1, ....] When I move f64 to even/odd pair register, I

Dear Evan Chang: I register incorrect Register class for MVT::f64. I have fixed it. Thanks your advice. "-view-legalize-dags" is very good option. But I don't know why my LLC do not know " -view-legalize-type-dags" option. By the way, I use llvm 2.5 merged from llvm2.4. Best Regards, Ren Kun --- 09年3月31日，周二, Evan Cheng <echeng at apple.com> 写道： 发件人: Evan Cheng

> On Nov 11, 2017, at 11:03 AM, Hal Finkel via llvm-dev <llvm-dev at lists.llvm.org> wrote: > > > On 11/11/2017 12:44 PM, Amara Emerson wrote: >> >>> On Nov 10, 2017, at 10:04 PM, Aditya Nandakumar <proaditya at gmail.com <mailto:proaditya at gmail.com>> wrote: >>>> >>>> The current DAGCombine, being constructed on top of

>>Note that the isCommutable flag is only really useful for two-address instructions. If the two inputs are not constrained, nothing is really won by swapping them. Ahh i see, good to know that. >> Does the -view-*-dags output look correct? They do look correct, there are three Xmul_lohi blocks, one returns the low part copied into R14 and the rest of combinations get added and merged

On 11/11/2017 12:44 PM, Amara Emerson wrote: > >> On Nov 10, 2017, at 10:04 PM, Aditya Nandakumar <proaditya at gmail.com >> <mailto:proaditya at gmail.com>> wrote: >>> >>> The current DAGCombine, being constructed on top of SDAG, has a kind >>> of built-in CSE and automatic DCE. How will things change, if >>> they'll change, in

Preamble -------- While working on an IR-level optimisation completely unrelated to register allocation I happened to trigger some really strange register allocator behaviour causing a large regression in bzip2 in spec2006. I've been trying to fix that regression before getting the optimisation patch committed, because I don't want to regress spec2006, but I'm basically fumbling in

From: Akira Hatanaka [mailto:ahatanak at gmail.com] Sent: Wednesday, September 05, 2012 12:44 PM To: Villmow, Micah Cc: reed kotler; llvmdev at cs.uiuc.edu Subject: Re: [LLVMdev] 64 bit special purpose registers Micah, Do you mean we should make GPR64 available to register allocator by calling addRegisterClass? addRegisterClass(MVT::i64, &GPR64RegClass) If we add register class GPR64, type

If no i64 reg classes are registered, then type-legalization will expand a 32b x 32b = 64b multiply node into a 32-bit mult node with two i32 results (for example, SMUL_LOHI). The problem is that there isn't an easy way to have RA assign two consecutive hi/lo registers to the two i32 registers, once the 64-bit result is split into two 32-bit results. Is there a constraint I can use (something