[turbofan] Implemented the optional Float32RoundTruncate operator.

[turbofan] Implemented the optional Float32RoundTruncate operator.

The Float32RoundTruncate operator rounds float32 numbers towards zero.
The operator is currently implemented on x64, ia32, arm, and arm64.

Additionally I added support for the float32 vrintz, vrintn, and vrinta
instructions to the arm simulator.

R=titzer@chromium.org

Committed: https://crrev.com/19741ac977ea0fae9a01fb78356cb84d546629b8
Cr-Commit-Position: refs/heads/master@{#32301}

[ahaas, 2015-11-25 09:55:15]

ahaas@chromium.org changed reviewers:
+ v8-arm-ports@googlegroups.com

[jbramley, 2015-11-25 10:54:48]

jacob.bramley@arm.com changed reviewers:
+ jacob.bramley@arm.com

[jbramley, 2015-11-25 10:54:49]

The ARM and ARM64 parts look good.

[jbramley, 2015-11-25 11:05:53]

On 2015/11/25 10:54:49, jbramley wrote:
> The ARM and ARM64 parts look good.

Actually, there are vrintz double tests in test/cctest/test-assembler-arm.cc.
Could you add float tests too please?

[ahaas, 2015-11-25 16:46:49]

On 2015/11/25 at 11:05:53, jacob.bramley wrote:
> On 2015/11/25 10:54:49, jbramley wrote:
> > The ARM and ARM64 parts look good.
> 
> Actually, there are vrintz double tests in test/cctest/test-assembler-arm.cc.
Could you add float tests too please?

I added tests for all float vrint instructions now, and all vrint instructions
are supported in the disassembler. Additionally I change the rounding functions
in the tests and in the arm simulator to the std ones (e.g. std::floor).

[ahaas, 2015-11-25 17:00:20]

On 2015/11/25 at 16:46:49, ahaas wrote:
> On 2015/11/25 at 11:05:53, jacob.bramley wrote:
> > On 2015/11/25 10:54:49, jbramley wrote:
> > > The ARM and ARM64 parts look good.
> > 
> > Actually, there are vrintz double tests in
test/cctest/test-assembler-arm.cc. Could you add float tests too please?
> 
> I added tests for all float vrint instructions now, and all vrint instructions
are supported in the disassembler. Additionally I change the rounding functions
in the tests and in the arm simulator to the std ones (e.g. std::floor).

Apparently std::nearbyint does not exist on all platforms. For consistency I use
the C rounding functions now everywhere.

[jbramley, 2015-11-25 17:04:10]

On 2015/11/25 17:00:20, ahaas wrote:
> On 2015/11/25 at 16:46:49, ahaas wrote:
> > On 2015/11/25 at 11:05:53, jacob.bramley wrote:
> > > On 2015/11/25 10:54:49, jbramley wrote:
> > > > The ARM and ARM64 parts look good.
> > > 
> > > Actually, there are vrintz double tests in
> test/cctest/test-assembler-arm.cc. Could you add float tests too please?
> > 
> > I added tests for all float vrint instructions now, and all vrint
instructions
> are supported in the disassembler. Additionally I change the rounding
functions
> in the tests and in the arm simulator to the std ones (e.g. std::floor).
> 
> Apparently std::nearbyint does not exist on all platforms. For consistency I
use
> the C rounding functions now everywhere.

Perhaps dcarney could comment on this, since he changed them to 'std::' in
https://codereview.chromium.org/207543003.

[titzer, 2015-11-25 17:16:21]

On 2015/11/25 17:04:10, jbramley wrote:
> On 2015/11/25 17:00:20, ahaas wrote:
> > On 2015/11/25 at 16:46:49, ahaas wrote:
> > > On 2015/11/25 at 11:05:53, jacob.bramley wrote:
> > > > On 2015/11/25 10:54:49, jbramley wrote:
> > > > > The ARM and ARM64 parts look good.
> > > > 
> > > > Actually, there are vrintz double tests in
> > test/cctest/test-assembler-arm.cc. Could you add float tests too please?
> > > 
> > > I added tests for all float vrint instructions now, and all vrint
> instructions
> > are supported in the disassembler. Additionally I change the rounding
> functions
> > in the tests and in the arm simulator to the std ones (e.g. std::floor).
> > 
> > Apparently std::nearbyint does not exist on all platforms. For consistency I
> use
> > the C rounding functions now everywhere.
> 
> Perhaps dcarney could comment on this, since he changed them to 'std::' in
> https://codereview.chromium.org/207543003.

Unfortunately he left Google earlier this year.

[jbramley, 2015-11-25 17:43:44]

Ok, the ARM(64) bits now look good as far as I'm concerned.

- It'd be nice to have some guidance about std::<op> versus <op>f. They were
changed at some point and there must have been some justification for that.
(Note that the std:: versions are templated, and I think we might rely on that
in a couple of places in ARM64.)

- This patch also now adds vrinta and vrintn, so the commit message should
reflect that.

[ahaas, 2015-11-25 17:52:03]

Description was changed from

==========
[turbofan] Implemented the optional Float32RoundTruncate operator.

The Float32RoundTruncate operator rounds float32 numbers towards zero.
The operator is currently implemented on x64, ia32, arm, and arm64.

I had to add the vrintz instruction to the arm simulator.

R=titzer@chromium.org
==========

to

==========
[turbofan] Implemented the optional Float32RoundTruncate operator.

The Float32RoundTruncate operator rounds float32 numbers towards zero.
The operator is currently implemented on x64, ia32, arm, and arm64.

Additionally I added support for the float32 vrintz, vrintn, and vrinta 
instructions to the arm simulator.

R=titzer@chromium.org
==========

[ahaas, 2015-11-25 17:56:55]

On 2015/11/25 at 17:43:44, jacob.bramley wrote:
> Ok, the ARM(64) bits now look good as far as I'm concerned.
> 
> - It'd be nice to have some guidance about std::<op> versus <op>f. They were
changed at some point and there must have been some justification for that.
(Note that the std:: versions are templated, and I think we might rely on that
in a couple of places in ARM64.)

I think that <op>f is the C version of these functions, whereas std::<op> is the
C++ version. That's why in my first change I wanted to use the std::<op>
version. However, not all std::<op>s existed on all platforms, so to make it
consistent I changed everything to <op>f.

> 
> - This patch also now adds vrinta and vrintn, so the commit message should
reflect that.

done.

[titzer, 2015-11-25 18:28:10]

On 2015/11/25 17:56:55, ahaas wrote:
> On 2015/11/25 at 17:43:44, jacob.bramley wrote:
> > Ok, the ARM(64) bits now look good as far as I'm concerned.
> > 
> > - It'd be nice to have some guidance about std::<op> versus <op>f. They were
> changed at some point and there must have been some justification for that.
> (Note that the std:: versions are templated, and I think we might rely on that
> in a couple of places in ARM64.)
> 
> I think that <op>f is the C version of these functions, whereas std::<op> is
the
> C++ version. That's why in my first change I wanted to use the std::<op>
> version. However, not all std::<op>s existed on all platforms, so to make it
> consistent I changed everything to <op>f.
> 
> > 
> > - This patch also now adds vrinta and vrintn, so the commit message should
> reflect that.
> 
> done.

lgtm

[ahaas, 2015-11-25 18:30:42]

The CQ bit was checked by ahaas@chromium.org

[commit-bot: I haz the power, 2015-11-25 18:31:00]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1468303005/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1468303005/40001

[commit-bot: I haz the power, 2015-11-25 19:28:54]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-11-25 19:29:08]

Description was changed from

==========
[turbofan] Implemented the optional Float32RoundTruncate operator.

The Float32RoundTruncate operator rounds float32 numbers towards zero.
The operator is currently implemented on x64, ia32, arm, and arm64.

Additionally I added support for the float32 vrintz, vrintn, and vrinta 
instructions to the arm simulator.

R=titzer@chromium.org
==========

to

==========
[turbofan] Implemented the optional Float32RoundTruncate operator.

The Float32RoundTruncate operator rounds float32 numbers towards zero.
The operator is currently implemented on x64, ia32, arm, and arm64.

Additionally I added support for the float32 vrintz, vrintn, and vrinta
instructions to the arm simulator.

R=titzer@chromium.org

Committed: https://crrev.com/19741ac977ea0fae9a01fb78356cb84d546629b8
Cr-Commit-Position: refs/heads/master@{#32301}
==========

[commit-bot: I haz the power, 2015-11-25 19:29:09]

Patchset 3 (id:??) landed as
https://crrev.com/19741ac977ea0fae9a01fb78356cb84d546629b8
Cr-Commit-Position: refs/heads/master@{#32301}