On Thu, 24 Jul 2025 10:29:15 GMT, Galder Zamarreño wrote:
> I've added support to vectorize `MoveD2L`, `MoveL2D`, `MoveF2I` and `MoveI2F`
> nodes. The implementation follows a similar pattern to what is done with
> conversion (`Conv*`) nodes. The tests in `TestCompatibleUseDefTypeSize` have
>
On Mon, 14 Jul 2025 21:43:58 GMT, Ian Graves wrote:
>> Adding SUADD an associative operation in the Vector API. Saturated addition
>> on fixed-width unsigned integers is provably associative.
>
> Ian Graves has updated the pull request incrementally with one additional
> commit since the last r
On Thu, 3 Apr 2025 12:54:01 GMT, Per Minborg wrote:
>> src/java.base/share/classes/java/lang/StableValue.java line 402:
>>
>>> 400: * race. Since stable functions and stable collections are built on
>>> top of
>>> 401: * {@linkplain StableValue#orElseSet(Supplier) orElseSet()} they too
>>> a
On Wed, 2 Apr 2025 13:22:44 GMT, Per Minborg wrote:
>> Implement JEP 502.
>>
>> The PR passes tier1-tier3 tests.
>
> Per Minborg has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Add info that Map#values and Map#entrySet are stable
src/java
On Mon, 31 Mar 2025 14:22:16 GMT, Maurizio Cimadamore
wrote:
>> Per Minborg has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> Finish and clean up benchmarks
>
> src/java.base/share/classes/jdk/internal/lang/stable/StableValueImpl.java
>
On Fri, 21 Mar 2025 20:25:51 GMT, Andrew Dinn wrote:
>>> In all these cases there is a fast path: e.g. when we know we have already
>>> warned for enable native access, or for Unsafe. In the SV API, the fast
>>> path is when we know that the SV is set already. In my experience, the
>>> volatil
On Fri, 21 Mar 2025 15:50:05 GMT, Maurizio Cimadamore
wrote:
> In all these cases there is a fast path: e.g. when we know we have already
> warned for enable native access, or for Unsafe. In the SV API, the fast path
> is when we know that the SV is set already. In my experience, the volatile
On Fri, 6 Dec 2024 16:30:47 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch improves the performance of a typical `Arena::allocate` in several
> ways:
>
> - Delay the creation of the NativeMemorySegmentImpl. This avoids the merge of
> the instance with the one obtained
On Thu, 13 Mar 2025 16:21:07 GMT, Maurizio Cimadamore
wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> expand wildcard imports
>
> Marked as reviewed by mcimadamore (Reviewer).
@mc
On Wed, 12 Mar 2025 08:09:04 GMT, Per Minborg wrote:
>> If you have an `@Stable Object[]`, then the elements are also considered
>> `@Stable`. Then you can do something like:
>>
>> ReentrantLock[] locks;
>>
>> T get(int idx) {
>> Object x = backing[idx];
>> if (x == nul
On Tue, 11 Mar 2025 11:19:13 GMT, Per Minborg wrote:
>> src/java.base/share/classes/java/util/ImmutableCollections.java line 777:
>>
>>> 775: private final IntFunction mapper;
>>> 776: @Stable
>>> 777: private final StableValueImpl[] backing;
>>
>> You can use a backing
On Mon, 10 Mar 2025 18:11:23 GMT, Per Minborg wrote:
> Implement JEP 502.
>
> The PR passes tier1-tier3 tests.
src/java.base/share/classes/java/util/ImmutableCollections.java line 777:
> 775: private final IntFunction mapper;
> 776: @Stable
> 777: private final StableVa
On Wed, 12 Mar 2025 11:15:29 GMT, Maurizio Cimadamore
wrote:
>> Quan Anh Mai has updated the pull request with a new target base due to a
>> merge or a rebase. The incremental webrev excludes the unrelated changes
>> brought in by the merge/rebase. The pull request contai
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request incrementally with one additional
commit sin
On Fri, 7 Mar 2025 17:38:13 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch improves the performance of a typical `Arena::allocate` in
>> several ways:
>>
>> - Delay the creation of the NativeMemorySegmentImpl. This avoids the merge
>> of the insta
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request with a new target base due to a merge
or a rebase. The i
On Wed, 5 Feb 2025 10:18:01 GMT, Maurizio Cimadamore
wrote:
>> Only the return value of `CALLOC` is converted to `MemorySegment` in an
>> equivalent way, I believe passing a `MemorySegment` to a downcall involves
>> acquiring the corresponding segment? As a result, `FREE` is made to accept
>>
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request with a new target base due to a merge
or a rebase. The i
On Fri, 24 Jan 2025 11:57:06 GMT, Maurizio Cimadamore
wrote:
>> The main goal here is that passing/receiving raw addresses as longs is
>> cheaper than as `MemorySegment`s.
>
> But the address is immediately converted to memory segment using
> MemorySegment::ofAddress, which is what the linker
On Mon, 16 Dec 2024 11:01:11 GMT, Maurizio Cimadamore
wrote:
>> Quan Anh Mai has updated the pull request incrementally with two additional
>> commits since the last revision:
>>
>> - wrong init
>> - move segment instance creation to SegmentFactories
>
&g
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request with a new target base due to a merge
or a rebase. The i
On Wed, 18 Dec 2024 16:22:51 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch adds the missing null checks to the shuffle arguments of the
> various `rearrangeTemplate` method. For all other API points, we null check
> arguments by calling `check` on them. As `rearrange` implement
On Wed, 18 Dec 2024 17:51:08 GMT, Paul Sandoz wrote:
>> Looks good, thanks for fixing.
>>
>> This is really a bug rather than an enhancement. In the package JavaDoc we
>> state:
>>> * For every class in this package,
>>> * unless specified otherwise, any method arguments of reference
>>> * type
On Wed, 18 Dec 2024 17:44:02 GMT, Paul Sandoz wrote:
>> Hi,
>>
>> This patch adds the missing null checks to the shuffle arguments of the
>> various `rearrangeTemplate` method. For all other API points, we null check
>> arguments by calling `check` on them. As `rearrange` implementations cast
Hi,
This patch adds the missing null checks to the shuffle arguments of the various
`rearrangeTemplate` method. For all other API points, we null check arguments
by calling `check` on them. As `rearrange` implementations cast the argument to
the concrete type, there is no need for species check
On Tue, 17 Sep 2024 16:13:55 GMT, Quan Anh Mai wrote:
> Hi,
>
> This is just a redo of https://github.com/openjdk/jdk/pull/13093. mostly just
> the revert of the backout.
>
> Regarding the related issues:
>
> - [JDK-8306008](https://bugs.openjdk.org/browse/JDK-8306
On Wed, 11 Dec 2024 18:50:18 GMT, Maurizio Cimadamore
wrote:
>> I want to put it here so that even if `SegmentFactories::allocateNative` is
>> not inlined, this small method will be more likely to be inlined and the
>> segment can be non-escape, then. I have added a comment regarding class
>>
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request incrementally with two additional
commits since the last r
ke it more straightforward.
> - Make `VM.pageAlignDirectMemory` a `Boolean` instead of a `boolean` so that
> `false` value can be constant folded.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has updated the pull request incrementally with one additional
c
On Mon, 9 Dec 2024 12:48:07 GMT, Maurizio Cimadamore
wrote:
>> src/java.base/share/classes/jdk/internal/foreign/SegmentFactories.java line
>> 213:
>>
>>> 211:
>>> 212: @DontInline
>>> 213: private static long allocateNativeOveraligned(long byteSize, long
>>> byteAlignment, MemorySess
On Fri, 6 Dec 2024 18:29:46 GMT, Jorn Vernee wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> address review
>
> src/java.base/share/classes/jdk/internal/foreign/Ar
On Fri, 6 Dec 2024 16:30:47 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch improves the performance of a typical `Arena::allocate` in several
> ways:
>
> - Delay the creation of the NativeMemorySegmentImpl. This avoids the merge of
> the instance with the one obtained
On Tue, 10 Dec 2024 16:10:09 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This is just a redo of https://github.com/openjdk/jdk/pull/13093. mostly
>> just the revert of the backout.
>>
>> Regarding the related issues:
>>
>> - [JDK-8306008](https://b
On Tue, 10 Dec 2024 09:02:40 GMT, Jatin Bhateja wrote:
>> This is called in an `assert` so I think throwing `AssertionError` seems
>> more reasonable, the original implementation also throws `AssertionError`
>
> I see, you want to throw an error here and not just use an assert statement
> that
On Tue, 10 Dec 2024 08:34:30 GMT, Jatin Bhateja wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> Change wording on VectorLoadShuffleNode
>>
>> Co-authored-by: Jatin Bhateja
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Mon, 9 Dec 2024 13:40:16 GMT, Jatin Bhateja wrote:
>> @merykitty do you want me to initiate tier 1 to 3 tests?
>
>> @PaulSandoz @sviswa7 Thanks for your advice, I have made the PR ready for
>> review
>>
>> @iwanowww Could you take another look at this, please?
>>
>> @jatin-bhateja Could you
On Tue, 10 Dec 2024 07:01:27 GMT, Jatin Bhateja wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> adverb order
>
> src/jdk.incubator.vector/share/classes/jdk/incubator/vector/B
On Tue, 10 Dec 2024 07:44:57 GMT, Jatin Bhateja wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> Change wording on VectorLoadShuffleNode
>>
>> Co-authored-by: Jatin
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Tue, 10 Dec 2024 07:54:55 GMT, Jatin Bhateja wrote:
>> src/jdk.incubator.vector/share/classes/jdk/incubator/vector/Byte128Vector.java
>> line 822:
>>
>>> 820: static final Class ETYPE = byte.class; // used by the JVM
>>> 821:
>>> 822: Byte128Shuffle(byte[] indices) {
>>
>>
On Mon, 9 Dec 2024 14:39:25 GMT, Emanuel Peter wrote:
>> @eme64 I think this PR is orthogonal to the concern you are having. Either
>> we need to refactor the expansion to lowering, then modify this PR to match
>> the semantics, or we integrate this PR first, then do the lowering refactor
>> o
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Mon, 9 Dec 2024 13:59:29 GMT, Emanuel Peter wrote:
>> @eme64 Yes I have thought about that. My idea is that once phase lowering is
>> ready we will move the expansion there (#21599) . This removes the need to
>> have a standalone method that checks if `LoadShuffleNode` is needed. The
>> cur
On Mon, 9 Dec 2024 13:47:10 GMT, Emanuel Peter wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> address reviews
>
> Ok, looks better.
>
> I've thought about this a little.
On Mon, 9 Dec 2024 08:10:20 GMT, Emanuel Peter wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> add comment, extract cast into local variable
>
> src/hotspot/share/opto/vector
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Fri, 6 Dec 2024 17:22:33 GMT, Paul Sandoz wrote:
>> @sviswa7 @PaulSandoz @eme64 @jatin-bhateja Thanks for taking a look, I have
>> merged the PR with a more recent master and resolved the sematic difference
>> with newly added intrinsics, too.
>
> @merykitty do you want me to initiate tier 1
On Fri, 6 Dec 2024 16:20:55 GMT, Paul Sandoz wrote:
>> Ah i see now, you want to ensure an invocation to the final/concrete method.
>> (The IDE's highlighting of the redundant cast is misleading)
>
> The common way we tend to do this in other areas is assign to a local
> variable with the sharp
On Fri, 6 Dec 2024 16:16:04 GMT, Paul Sandoz wrote:
>> This piece of code follows the pattern:
>>
>> if (uncommonCondition) {
>> return uncommonPath();
>> }
>> // Continue the common path
>>
>> So I think it is better to keep it as it is.
>
> Ok, but since it deviates from t
On Fri, 6 Dec 2024 16:30:47 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch improves the performance of a typical `Arena::allocate` in several
> ways:
>
> - Delay the creation of the NativeMemorySegmentImpl. This avoids the merge of
> the instance with the one obtained
Hi,
This patch improves the performance of a typical `Arena::allocate` in several
ways:
- Delay the creation of the NativeMemorySegmentImpl. This avoids the merge of
the instance with the one obtained from the call in the uncommon path,
increasing the chance the object being scalar replaced.
-
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Thu, 5 Dec 2024 20:48:07 GMT, Paul Sandoz wrote:
>> `shuffleFromOp` is a slow path op so I don't think it is. Additionally, our
>> vector multiplication is against a scalar, too. So we can optimize it if
>> `step` is a constant.
>
> I incorrectly read `!=` as `==` :-) as that is the more com
On Wed, 27 Nov 2024 16:12:04 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch resolves some issues I found that the tests wrongly assume
> sufficient alignment from the segment returned by `Arena::allocate`.
>
> Please take a look and leave your reviews, thanks a lot.
This pu
On Wed, 27 Nov 2024 18:56:02 GMT, Maurizio Cimadamore
wrote:
>> Hi,
>>
>> This patch resolves some issues I found that the tests wrongly assume
>> sufficient alignment from the segment returned by `Arena::allocate`.
>>
>> Please take a look and leave your reviews, thanks a lot.
>
> Marked as
On Wed, 27 Nov 2024 16:41:39 GMT, Jorn Vernee wrote:
>> Hi,
>>
>> This patch resolves some issues I found that the tests wrongly assume
>> sufficient alignment from the segment returned by `Arena::allocate`.
>>
>> Please take a look and leave your reviews, thanks a lot.
>
> test/jdk/java/forei
Hi,
This patch resolves some issues I found that the tests wrongly assume
sufficient alignment from the segment returned by `Arena::allocate`.
Please take a look and leave your reviews, thanks a lot.
-
Commit messages:
- Some java/foreign tests wrongly assume aligned memory
Chang
On Sat, 23 Nov 2024 16:29:10 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch sharpens the return types of various internal methods. This helps
> the compiler to have better information when the corresponding methods are
> not inlined.
>
> Please take a look and leave your
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Tue, 29 Oct 2024 20:27:20 GMT, Paul Sandoz wrote:
>> Quan Anh Mai has updated the pull request with a new target base due to a
>> merge or a rebase. The pull request now contains one commit:
>>
>> [vectorapi] Refactor VectorShuffle implementation
>
> src/jdk.
On Tue, 15 Oct 2024 16:25:04 GMT, Emanuel Peter wrote:
>> Quan Anh Mai has updated the pull request with a new target base due to a
>> merge or a rebase. The pull request now contains one commit:
>>
>> [vectorapi] Refactor VectorShuffle implementation
&g
On Fri, 8 Nov 2024 23:17:35 GMT, Sandhya Viswanathan
wrote:
>> I have adapted the patch in accordance with
>> https://github.com/openjdk/jdk/pull/20634, I moved the index wrapping into
>> C2 instead of making it a separate step as I think it seems clearer. Also, I
>> think in the future we ca
qu 0x10(%r10),%xmm2
> movabs $0x7515a0d08,%r10; {oop([I{0x0007515a0d08})}
> vmovdqu 0x10(%r10),%xmm1
> movabs $0x75158afb8,%r10; {oop([I{0x00075158afb8})}
> vmovdqu 0x10(%r10),%xmm0
> vpand -0xddc12(%rip),%xmm0,%xmm0# S
On Sun, 24 Nov 2024 03:26:18 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch sharpens the return types of various internal methods. This helps
>> the compiler to have better information when the corresponding methods are
>> not inlined.
>>
>>
On Mon, 25 Nov 2024 12:26:37 GMT, Jorn Vernee wrote:
>> src/java.base/share/classes/jdk/internal/foreign/AbstractMemorySegmentImpl.java
>> line 144:
>>
>>> 142: }
>>> 143:
>>> 144: private NativeMemorySegmentImpl reinterpretInternal(Class
>>> callerClass, long newSize, MemorySessionIm
On Mon, 25 Nov 2024 07:26:56 GMT, Per Minborg wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> fix failing tests
>
> src/java.base/share/classes/jdk/internal/foreign/NativeMem
> Hi,
>
> This patch sharpens the return types of various internal methods. This helps
> the compiler to have better information when the corresponding methods are
> not inlined.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has upd
On Sat, 23 Nov 2024 17:27:43 GMT, Chen Liang wrote:
>> Quan Anh Mai has refreshed the contents of this pull request, and previous
>> commits have been removed. The incremental views will show differences
>> compared to the previous content of the PR. The pull request co
> Hi,
>
> This patch sharpens the return types of various internal methods. This helps
> the compiler to have better information when the corresponding methods are
> not inlined.
>
> Please take a look and leave your reviews, thanks a lot.
Quan Anh Mai has refreshed the
Hi,
This patch sharpens the return types of various internal methods. This helps
the compiler to have better information when the corresponding methods are not
inlined.
Please take a look and leave your reviews, thanks a lot.
-
Commit messages:
- sharpen the return types of vario
On Fri, 8 Nov 2024 11:56:08 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch fixes the deadlock in `TestMemorySession#testAcquireCloseRace`.
> The lock-step spin lock is implemented as with `lock` being an
> `AtomicInteger`:
>
> // Keep the 2 threads operating on the s
On Fri, 8 Nov 2024 16:59:48 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch fixes the deadlock in `TestMemorySession#testAcquireCloseRace`.
>> The lock-step spin lock is implemented as with `lock` being an
>> `AtomicInteger`:
>>
>> // Keep
; proceed because lock can never reach 2
>
>
> The solution is to not rely on the exact value of the lock but instead
> whether the lock has passed the expected value.
>
> Testing: I have run this test several hundreds times and got no failure while
> without this patch I e
On Fri, 8 Nov 2024 11:56:08 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch fixes the deadlock in `TestMemorySession#testAcquireCloseRace`.
> The lock-step spin lock is implemented as with `lock` being an
> `AtomicInteger`:
>
> // Keep the 2 threads operating on the s
On Fri, 8 Nov 2024 12:22:54 GMT, Maurizio Cimadamore
wrote:
>> Hi,
>>
>> This patch fixes the deadlock in `TestMemorySession#testAcquireCloseRace`.
>> The lock-step spin lock is implemented as with `lock` being an
>> `AtomicInteger`:
>>
>> // Keep the 2 threads operating on the same scop
Hi,
This patch fixes the deadlock in `TestMemorySession#testAcquireCloseRace`. The
lock-step spin lock is implemented as with `lock` being an `AtomicInteger`:
// Keep the 2 threads operating on the same scope
int k = lock.getAndAdd(1) + 1;
while (k != i * 2) {
Thread.onSpinWa
On Thu, 7 Nov 2024 15:12:01 GMT, Maurizio Cimadamore
wrote:
> This small PR fixes a build failure due to usage of sun.misc.Unsafe in a new
> benchmark.
>
> FFM benchmarks appear to be broken -- because of JDK-8332744 -- but we will
> fix that separately.
@mcimadamore Sorry for the failure an
On Thu, 31 Oct 2024 15:52:04 GMT, Quan Anh Mai wrote:
> Hi,
>
> This patch makes `MemorySessionImpl.state` a `Stable` field so that liveness
> check of non-closeable scopes such as the global scope can be elided.
>
> Currently, the `state` field is overloaded with 2 respon
On Fri, 18 Oct 2024 05:37:16 GMT, Quan Anh Mai wrote:
>> src/hotspot/share/opto/vectornode.cpp line 2122:
>>
>>> 2120: // MulL (URShift SRC1 , 32) (URShift SRC2, 32)
>>> 2121: // MulL (URShift SRC1 , 32) ( And SRC2, 0x)
>>> 2122: //
On Fri, 18 Oct 2024 05:35:28 GMT, Vladimir Ivanov wrote:
>> This patch optimizes LongVector multiplication by inferring VPMUL[U]DQ
>> instruction for following IR pallets.
>>
>>
>>MulVL ( AndV SRC1, 0x) ( AndV SRC2, 0x)
>>MulVL (URShiftVL SRC1 , 32)
On Sun, 29 Sep 2024 04:21:19 GMT, Jatin Bhateja wrote:
> This patch optimizes LongVector multiplication by inferring VPMUL[U]DQ
> instruction for following IR pallets.
>
>
>MulVL ( AndV SRC1, 0x) ( AndV SRC2, 0x)
>MulVL (URShiftVL SRC1 , 32) (URShif
On Fri, 18 Oct 2024 05:42:21 GMT, Jatin Bhateja wrote:
>>> The issue is that a node is not immutable.
>>
>> I don't see any issues with mutability here.
>> `MulVLNode::_mult_lower_double_word` is constant, so you have to allocate
>> new node if you want to change its value. (And that's exactly
On Fri, 18 Oct 2024 05:16:04 GMT, Vladimir Ivanov wrote:
>>> I see this is as a twostep optimization, in the first step we do analysis
>>> and annotate additional information on existing IR, which is later used by
>>> instruction selector. I plan to subsume first stage with enhanced dataflow
>
On Fri, 18 Oct 2024 04:16:15 GMT, Jatin Bhateja wrote:
> I see this is as a twostep optimization, in the first step we do analysis and
> annotate additional information on existing IR, which is later used by
> instruction selector. I plan to subsume first stage with enhanced dataflow
> analysi
On Tue, 15 Oct 2024 17:00:26 GMT, Jasmine Karthikeyan
wrote:
>> This patch optimizes LongVector multiplication by inferring VPMUL[U]DQ
>> instruction for following IR pallets.
>>
>>
>>MulVL ( AndV SRC1, 0x) ( AndV SRC2, 0x)
>>MulVL (URShiftVL SRC1
On Mon, 14 Oct 2024 12:12:58 GMT, Jatin Bhateja wrote:
>>> I am having a similar idea that is to group those transformations together
>>> into a `Phase` called `PhaseLowering`
>>
>> I think such a phase could be quite useful in general. Recently I was trying
>> to implement the BMI1 instructio
On Thu, 31 Oct 2024 23:08:40 GMT, Maurizio Cimadamore
wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> fix benchmarks
>
> Actually, @JornVernee pointed out (great observation!)
On Tue, 5 Nov 2024 14:07:12 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch makes `MemorySessionImpl.state` a `Stable` field so that liveness
>> check of non-closeable scopes such as the global scope can be elided.
>>
>> Currently, the `state` field is ov
On Tue, 5 Nov 2024 15:31:11 GMT, Maurizio Cimadamore
wrote:
>> Okay, fair enough.
>
> Using two blocking queues might also be possible:
> * at the start of each iteration a thread does a `get` on its own queue
> * at the end of each iteration, a thread does a put on the other thread's
> queue
>
On Tue, 5 Nov 2024 15:29:22 GMT, Per Minborg wrote:
>> Quan Anh Mai has updated the pull request incrementally with two additional
>> commits since the last revision:
>>
>> - copyright year
>> - address reviews
>
> In `SharedSession`, would it not be nec
On Tue, 5 Nov 2024 14:43:29 GMT, Jorn Vernee wrote:
>> Quan Anh Mai has updated the pull request incrementally with two additional
>> commits since the last revision:
>>
>> - copyright year
>> - address reviews
>
> test/jdk/java/foreign/TestMe
On Fri, 1 Nov 2024 02:44:15 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch makes `MemorySessionImpl.state` a `Stable` field so that liveness
>> check of non-closeable scopes such as the global scope can be elided.
>>
>> Currently, the `state` field is ov
his is the results without this patch:
>
> Benchmark Mode Cnt Score Error Units
> MemorySegmentGetUnsafe.panama avgt 30 0.420 ± 0.019 ns/op
> MemorySegmentGetUnsafe.unsafe avgt 30 0.329 ± 0.003 ns/op
>
> Please kindly review, thanks v
On Mon, 4 Nov 2024 19:08:49 GMT, Jorn Vernee wrote:
>> `checkValidStateRaw` synchronizes with `justClose` using handshakes so an
>> opaque or higher load is only needed in `sharedSessionAlreadyClosed`. A
>> `getOpaque` would probably be adequate. But I believe there is no formal
>> restriction
On Mon, 4 Nov 2024 11:32:12 GMT, Maurizio Cimadamore
wrote:
>> I'm dubious about this. Typically these modes work in a symmetric way - if
>> you `setOpaque` you need to `getOpaque`. From
>> [this](https://gee.cs.oswego.edu/dl/html/j9mm.html):
>>
>>> Opaque mode, obtained using VarHandle getOp
On Fri, 1 Nov 2024 08:55:09 GMT, Per Minborg wrote:
>> Quan Anh Mai has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> wait for the close operation to complete on acquire failures
>
> src/java.base/shar
his is the results without this patch:
>
> Benchmark Mode Cnt Score Error Units
> MemorySegmentGetUnsafe.panama avgt 30 0.420 ± 0.019 ns/op
> MemorySegmentGetUnsafe.unsafe avgt 30 0.329 ± 0.003 ns/op
>
> Please kindly review, thanks v
On Thu, 31 Oct 2024 18:09:55 GMT, Quan Anh Mai wrote:
>> Hi,
>>
>> This patch makes `MemorySessionImpl.state` a `Stable` field so that liveness
>> check of non-closeable scopes such as the global scope can be elided.
>>
>> Currently, the `state` field is ov
1 - 100 of 255 matches
Mail list logo
