Adding initial SVE support to TVM #8655
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Prototype containing initial VLA and predication implementation
MeeraN7
requested review from
ZihengJiang,
areusch,
comaniac,
jcf94,
jroesch,
junrushao,
kazum,
kparzysz-quic,
liangfu,
masahi,
merrymercy,
tmoreau89,
tqchen,
vinx13,
yzhliu and
zhiics
as code owners
Author
Contributor
|
@tqchen @jcf94 @junrushao1994 you may also be interested |
huajsj
reviewed
Aug 7, 2021
Comment on lines
+96
to
+102
| # ctx = remote.context(target) | ||
| # # launch the kernel. | ||
| # n = nn | ||
| # a = tvm.nd.array(np.random.uniform(size=(n + base, stride)).astype(A.dtype), ctx) | ||
| # c = tvm.nd.array(np.zeros((n, stride), dtype=C.dtype), ctx) | ||
| # f(a, c) | ||
| # tvm.testing.assert_allclose(c.asnumpy(), a.asnumpy()[base:] + 1) |
Comment on lines
+71
to
+76
| # ctx = remote.context(target) | ||
| # # launch the kernel. | ||
| # a = tvm.nd.empty((n,), A.dtype, ctx).copyfrom(np.random.uniform(size=(n, lanes))) | ||
| # c = tvm.nd.empty((n,), C.dtype, ctx) | ||
| # f(a, c) | ||
| # tvm.testing.assert_allclose(c.asnumpy(), a.asnumpy() + 1) |
Comment on lines
+39
to
+45
| # ctx = remote.context(target) | ||
| # # launch the kernel. | ||
| # n = nn | ||
| # a = tvm.nd.array(np.random.uniform(size=(n + base)).astype(A.dtype), ctx) | ||
| # c = tvm.nd.array(np.zeros(n, dtype=C.dtype), ctx) | ||
| # f(a, c) | ||
| # tvm.testing.assert_allclose(c.asnumpy(), a.asnumpy()[::-1][:n]) |
| } | ||
|
|
||
| // All pattern | ||
| int all_pattern = 31; |
Comment on lines
+373
to
+375
| // auto const str_to_parse = os.str(); | ||
| // auto pos = str_to_parse.find("x"); | ||
| // auto stem= str_to_parse.substr(0, pos); |
Comment on lines
+81
to
+91
| // DataType(int code, int bits) { | ||
| // data_.code = static_cast<uint8_t>(code); | ||
| // data_.bits = static_cast<uint8_t>(bits); | ||
| // is_scalable_ = true; | ||
| // std::cout<<bits<<std::endl; | ||
| // data_.lanes = uint16_t(128) / static_cast<uint16_t>(8); // minimal lanes | ||
| // | ||
| //// if (code == kBFloat) { | ||
| //// ICHECK_EQ(bits, 16); | ||
| //// } | ||
| // } |
| bool is_scalable() const { return is_scalable_; } | ||
|
|
||
| DataType with_scalable_lanes() const { | ||
| int min_num_lanes = 128 / bits(); |
Contributor
There was a problem hiding this comment.
macro? and does here only support vector size 128 or should be configurable between 128 - 2048?
| bool is_vector_bool() const { return is_vector() && bits() == 1; } | ||
| /*! \return whether type is a Void type. */ | ||
| bool is_void() const { return code() == DataType::kHandle && bits() == 0 && lanes() == 0; } | ||
| bool is_scalable() const { return is_scalable_; } |
| bool operator==(const DataType& other) const { | ||
| return data_.code == other.data_.code && data_.bits == other.data_.bits && | ||
| data_.lanes == other.data_.lanes; | ||
| data_.lanes == other.data_.lanes; // && is_scalable_ == other.is_scalable_; |
Contributor
There was a problem hiding this comment.
seem like no diff with original code, is the comment for "is_scalable_ == other.is_scalable" intend logic?
| bool is_void() const { return code() == DataType::kHandle && bits() == 0 && lanes() == 0; } | ||
| bool is_scalable() const { return is_scalable_; } | ||
|
|
||
| DataType with_scalable_lanes() const { |
Contributor
There was a problem hiding this comment.
how this function cooperate with is_scalable_? what happen if is_scalable_ is false but this function get called?
Member
|
This PR appears to be out of date, please feel free to reopen it if this is not the case. As part of the new year we are attempting to triage the project's open pull requests to ensure that code which Thanks again for your contribution, and feel free to reach out to discuss these changes. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Prototype for the addition of Arm Architecture's Scalable Vector Extension (SVE) in TVM containing initial VLA and predication implementation, based on earlier work by Giuseppe Rossini.
The RFC can be found here