Displaying 6 results from an estimated 6 matches for "performat".
Did you mean:
performant
2014 Apr 29
4
[LLVMdev] Proposal: add intrinsics for safe division
On 04/29/2014 10:44 AM, Filip Pizlo wrote:
> LD;DR: Your desire to use trapping on x86 only further convinces me
> that Michael's proposed intrinsics are the best way to go.
I'm still not convinced, but am not going to actively oppose it either.
I'm leery of designing a solution with major assumptions we don't have
data to backup.
I worry your assumptions about
2014 May 01
2
[LLVMdev] Proposal: add intrinsics for safe division
...owledged.
>
> Note that I want *both* deopt *and* branching, because in this case, a
> branch is the fastest overall way of detecting when to deopt. In the
> future, I will want to implement the deopt in terms of branching, and
> when we do this, I believe that the most sound and performat approach
> would be using Michael's intrinsics. This is subtle and I'll try to
> explain why it's the case.
>
> The point is that you wouldn't want to do deoptimization by spilling
> state on the main path or by using a patchpoint for the main path of
> the di...
2014 May 02
3
[LLVMdev] Proposal: add intrinsics for safe division
...wledged.
>
> Note that I want *both* deopt *and* branching, because in this case, a
> branch is the fastest overall way of detecting when to deopt. In the
> future, I will want to implement the deopt in terms of branching, and when
> we do this, I believe that the most sound and performat approach would be
> using Michael's intrinsics. This is subtle and I'll try to explain why it's
> the case.
>
> The point is that you wouldn't want to do deoptimization by spilling state
> on the main path or by using a patchpoint for the main path of the divisio...
2014 May 01
6
[LLVMdev] Proposal: add intrinsics for safe division
...both* deopt *and* branching, because in this case,
>>> a branch is the fastest overall way of detecting when to deopt. In
>>> the future, I will want to implement the deopt in terms of
>>> branching, and when we do this, I believe that the most sound and
>>> performat approach would be using Michael's intrinsics. This is
>>> subtle and I'll try to explain why it's the case.
>>>
>>> The point is that you wouldn't want to do deoptimization by spilling
>>> state on the main path or by using a patchpoint for the...
2014 May 02
1
[LLVMdev] Proposal: add intrinsics for safe division
...ant *both* deopt *and* branching, because in this case, a
>> > branch is the fastest overall way of detecting when to deopt. In the
>> > future, I will want to implement the deopt in terms of branching,
>> and when
>> > we do this, I believe that the most sound and performat approach
>> would be
>> > using Michael's intrinsics. This is subtle and I'll try to explain
>> why it's
>> > the case.
>> >
>> > The point is that you wouldn't want to do deoptimization by
>> spilling state
>> > on t...
2014 May 02
5
[LLVMdev] Proposal: add intrinsics for safe division
...cknowledged.
>
> Note that I want *both* deopt *and* branching, because in this case, a
> branch is the fastest overall way of detecting when to deopt. In the
> future, I will want to implement the deopt in terms of branching, and when
> we do this, I believe that the most sound and performat approach would be
> using Michael's intrinsics. This is subtle and I'll try to explain why it's
> the case.
>
> The point is that you wouldn't want to do deoptimization by spilling state
> on the main path or by using a patchpoint for the main path of the division.
&...