cfallin opened issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
alexcrichton labeled issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
alexcrichton labeled issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
alexcrichton labeled issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
alexcrichton unlabeled issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
cfallin closed issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
cfallin reopened issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
cfallin commented on issue #3810:
Reopened; GH mis-parsed "once we fix #3810" as "fix #3810" in #4231's commit.
alexcrichton closed issue #3810:
In #3809, a new extension to our fuzzing found that some of our Wasm-SIMD support requires nominally optional ISA features, at least on x86-64.
I think we should try to have fallbacks such that we can support all relevant features on any machine that supports the base ISA. On x86-64, this means just SSE2. Or if this is very difficult for some reason, we should document why and then possibly consider raising our minimum required ISA version.
alexcrichton commented on issue #3810:
I believe that this is now done, so closing.
Last updated: Nov 22 2024 at 17:03 UTC