thr3ads.net - llvm dev - [LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!! [Jul 2014]

If this information is useful, please help other people find it:
Share via:

Wan, Xiaofei

2014-Jul-10 07:53 UTC

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

You may need implement these two functions in your runtime environment
'__gnu_f2h_ieee' and '__gnu_h2f_ieee'

From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On
Behalf Of yalong at multicorewareinc.com
Sent: Friday, July 11, 2014 6:49 AM
To: Andrea Di Biagio; Matt Arsenault
Cc: llvmdev
Subject: Re: [LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select:
0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

Hi Andrea
    Thank you your replying.
    I do like your letter. Add following to line to MipsISelLowering.cpp. As
your words,   @llvm.convert.to.fp16  can compile successfully. However, the
runtime is not right.

+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);
+  setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);

Robin

________________________________
yalong at multicorewareinc.com<mailto:yalong at multicorewareinc.com>

From: Andrea Di Biagio<mailto:andrea.dibiagio at gmail.com>
Date: 2014-07-09 14:20
To: Matt Arsenault<mailto:Matthew.Arsenault at amd.com>
CC: yalong at multicorewareinc.com<mailto:yalong at multicorewareinc.com>;
Kevin Qin<mailto:kevinqindev at gmail.com>; llvmdev<mailto:llvmdev at
cs.uiuc.edu>
Subject: Re: [LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select:
0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!
Not sure if this can help, but
if you really really want to have minimal half float support on Mips,
then one thing you could try to do is to hack MipsISelLowering.cpp
adding rules to expand float-half conversion SDNodes into library
calls.

+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);
+  setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);

(The MVT::i32 on the second rule is required because type i16 is
promoted to i32).

If you then convert every occurrence of 'fptrunc' from float to half
with calls to @llvm.convert.to.fp16, then you should be able to
compile (hopefully) with no errors.
That means, in your original example you would convert the following
IR statement:
  %Vt3_1 = fptrunc float %Vt_2 to half
into
  %Vt3_1 = call i16 @llvm.convert.to.fp16(float %Vt_2)

The downside is that you will have to add definitions for
'__gnu_f2h_ieee' and '__gnu_h2f_ieee' in the compiler runtime.
That is
because the backend will expand all the float-half conversions into
library calls...
This workaround should work assuming that a) you can hack the backend,
and b) it is acceptable (i.e. a reasonable solution in your case) to
have a library call for every float-half conversion in your code.

On Wed, Jul 9, 2014 at 8:51 PM, Matt Arsenault
<Matthew.Arsenault at amd.com<mailto:Matthew.Arsenault at amd.com>>
wrote:> I think that support for the half type is only implemented for ARM. Last I
> tried to use it, I found that none of it works even on x86, and the current
> handling of the half conversion SDNodes seem to rely on ARM specific
> assumptions
Just for the record,
since revision 212293 (committed only five days ago), the x86 backend
supports float half conversions.
On x86, if the target has F16C, there are ISel patterns to map
float-half conversions to specific instructions. If there is no F16C
support, then the backend expands float-half conversions into runtime
library calls.

Cheers,
Andrea
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140710/4ae33c05/attachment.html>

yalong at multicorewareinc.com

2014-Jul-10 22:48 UTC

head link

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

Hi Andrea    Thank you your replying.    I do like your letter. Add following to
line to MipsISelLowering.cpp. As your words,   @llvm.convert.to.fp16  can
compile successfully. However, the runtime is not right.
+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);+ 
setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);
Robin

yalong at multicorewareinc.com
 From: Andrea Di BiagioDate: 2014-07-09 14:20To: Matt ArsenaultCC: yalong at
multicorewareinc.com; Kevin Qin; llvmdevSubject: Re: [LLVMdev] Help!!!!Help!!!!
" LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750
[ID=16] " problem!!!!!!!!!!!!!!!!!!Not sure if this can help, but
if you really really want to have minimal half float support on Mips,
then one thing you could try to do is to hack MipsISelLowering.cpp
adding rules to expand float-half conversion SDNodes into library
calls.

+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);
+  setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);

(The MVT::i32 on the second rule is required because type i16 is
promoted to i32).

If you then convert every occurrence of 'fptrunc' from float to half
with calls to @llvm.convert.to.fp16, then you should be able to
compile (hopefully) with no errors.
That means, in your original example you would convert the following
IR statement:
  %Vt3_1 = fptrunc float %Vt_2 to half
into
  %Vt3_1 = call i16 @llvm.convert.to.fp16(float %Vt_2)

The downside is that you will have to add definitions for
'__gnu_f2h_ieee' and '__gnu_h2f_ieee' in the compiler runtime.
That is
because the backend will expand all the float-half conversions into
library calls...
This workaround should work assuming that a) you can hack the backend,
and b) it is acceptable (i.e. a reasonable solution in your case) to
have a library call for every float-half conversion in your code.

On Wed, Jul 9, 2014 at 8:51 PM, Matt Arsenault
<Matthew.Arsenault at amd.com> wrote:> I think that support for the half type is only implemented for ARM. Last I
> tried to use it, I found that none of it works even on x86, and the current
> handling of the half conversion SDNodes seem to rely on ARM specific
> assumptions 
Just for the record,
since revision 212293 (committed only five days ago), the x86 backend
supports float half conversions.
On x86, if the target has F16C, there are ISel patterns to map
float-half conversions to specific instructions. If there is no F16C
support, then the backend expands float-half conversions into runtime
library calls.

Cheers,
Andrea

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140710/c0df4b12/attachment.html>

yalong at multicorewareinc.com

2014-Jul-10 23:06 UTC

head link

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

Hi Wan,Xiaofei,    Thank you your replying.    I see the letter of Andrea, This
only resolve translate bug, but no runtime. I will do some research
about '__gnu_f2h_ieee' and '__gnu_h2f_ieee'. Then implement
them.Cheer,Robin


yalong at multicorewareinc.com
 From: Wan, XiaofeiDate: 2014-07-10 00:53To: yalong at multicorewareinc.com;
Andrea Di Biagio; Matt ArsenaultCC: llvmdevSubject: RE: [LLVMdev]
Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16
0x9fc0750 [ID=16] "	problem!!!!!!!!!!!!!!!!!!







You may need implement these two functions in your runtime environment
'__gnu_f2h_ieee' and '__gnu_h2f_ieee'
 


From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu]
On Behalf Of yalong at multicorewareinc.com

Sent: Friday, July 11, 2014 6:49 AM

To: Andrea Di Biagio; Matt Arsenault

Cc: llvmdev

Subject: Re: [LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select:
0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!


 

Hi Andrea


    Thank you your replying.


    I do like your letter. Add following to line to MipsISelLowering.cpp. As
your words,   @llvm.convert.to.fp16  can compile
 successfully. However, the runtime is not right.


 



+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);


+  setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);


 


Robin



 







yalong at multicorewareinc.com





 




From: Andrea
 Di Biagio


Date: 2014-07-09 14:20


To: Matt
 Arsenault


CC: yalong at multicorewareinc.com;
Kevin Qin; 
llvmdev


Subject: Re: [LLVMdev] Help!!!!Help!!!! " LLVM
 ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] "
problem!!!!!!!!!!!!!!!!!!





Not sure if this can help, but


if you really really want to have minimal half float support on Mips,


then one thing you could try to do is to hack MipsISelLowering.cpp


adding rules to expand float-half conversion SDNodes into library


calls.


 


+  setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);


+  setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);


 


(The MVT::i32 on the second rule is required because type i16 is


promoted to i32).


 


If you then convert every occurrence of 'fptrunc' from float to half


with calls to @llvm.convert.to.fp16, then you should be able to


compile (hopefully) with no errors.


That means, in your original example you would convert the following


IR statement:


  %Vt3_1 = fptrunc float %Vt_2 to half


into


  %Vt3_1 = call i16 @llvm.convert.to.fp16(float %Vt_2)


 


The downside is that you will have to add definitions for


'__gnu_f2h_ieee' and '__gnu_h2f_ieee' in the compiler runtime.
That is


because the backend will expand all the float-half conversions into


library calls...


This workaround should work assuming that a) you can hack the backend,


and b) it is acceptable (i.e. a reasonable solution in your case) to


have a library call for every float-half conversion in your code.


 


On Wed, Jul 9, 2014 at 8:51 PM, Matt Arsenault


<Matthew.Arsenault at amd.com> wrote:

> I think that support for the half type is only implemented for ARM. Last I
> tried to use it, I found that none of it works even on x86, and the current
> handling of the half conversion SDNodes seem to rely on ARM specific
> assumptions

 


Just for the record,


since revision 212293 (committed only five days ago), the x86 backend


supports float half conversions.


On x86, if the target has F16C, there are ISel patterns to map


float-half conversions to specific instructions. If there is no F16C


support, then the backend expands float-half conversions into runtime


library calls.


 


Cheers,


Andrea







-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140710/2d307544/attachment.html>

llvm dev - Jul 2014 - [LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!

[LLVMdev] Help!!!!Help!!!! " LLVM ERROR: Cannot select: 0x9fc9680: i32 = fp32_to_fp16 0x9fc0750 [ID=16] " problem!!!!!!!!!!!!!!!!!!