AccurateArithmetic.jl Versions Save

AccurateArithmetic v0.3.8

Diff since v0.3.7

Merged pull requests:

• Typo (#30) (@PallHaraldsson)
• CompatHelper: bump compat for "VectorizationBase" to "0.21" (#31) (@github-actions[bot])

AccurateArithmetic v0.3.7

Diff since v0.3.6

Closed issues:

• Cannot install any version past 0.3.0 (#25)

Merged pull requests:

• CompatHelper: bump compat for "VectorizationBase" to "0.20" (#28) (@github-actions[bot])

AccurateArithmetic v0.3.6

Diff since v0.3.5

This release is the first to be compatible with Julia v1.6

Changes:

• Switch to GitHub Actions for CI (#22)
• Drop dependency to SIMDPirates and add support for VectorizationBase >v0.15 and Julia 1.6 (#23, #24, #25)
• Drop unnecessary dependency to JSON

Closed issues:

• Transition off of SIMDPirates (#23)

Merged pull requests:

• MassInstallAction: Install the CI workflow on this repository (#22) (@timholy)
• Drop SIMDPirates and update to new VectorizationBase (#24) (@chriselrod)

AccurateArithmetic v0.3.5

Diff since v0.3.4

• Added sum_mixed & dot_mixed: mixed-precision implementations of summation and dot product for 32-bit numbers (#20)
• Fixed four_sum (#18, #21)

Merged pull requests:

• Mixed precision implementations for operations on single-precision inputs (#20) (@ffevotte)
• correct four_sum (#21) (@JeffreySarnoff)

AccurateArithmetic v0.3.4

Diff since v0.3.3

This release introduces a @fusible counterpart to SIMDops.@explicit, which allows using SIMD instructions from SIMDPirates.jl in all cases, whether exact instructions are wanted, or fused operations are allowed to happen. EFTs now use this when possible. The API for @explicit changed: it now affects the following expressions (instead of the whole scope inside which it is placed).

This release also adds support for cache prefetching, which significantly improves performance for large vectors. This was actually the explanation for the performance discrepancies observed for large vectors and mentioned in #7. Default values for the cache prefetching mechanism are probably not optimal for all architectures. Users interested in the last 10% performance outside the cache are invited to customize this parameter.

Closed issues:

• Naive dot product performance (#7)

Merged pull requests:

• CompatHelper: bump compat for "SIMDPirates" to "0.8" (#17) (@github-actions[bot])
• @explicit/@fusible + cache prefetching (#19) (@ffevotte)

AccurateArithmetic v0.3.3

Diff since v0.3.2

This version adjusts the SIMDPirates compatibility bounds. Hopefully it fixes issue gh-11 once and for all.

Closed issues:

• Changes in SIMDPirates.jl broke this package (#11)

Merged pull requests:

• Update compat bounds (#13) (@ffevotte)
• CompatHelper: bump compat for "SIMDPirates" to "0.6" (#14) (@github-actions[bot])
• CompatHelper: bump compat for "SIMDPirates" to "0.7" (#15) (@github-actions[bot])
• Increase SIMDPirates lower bound to 0.2 or above (#16) (@chriselrod)

AccurateArithmetic v0.3.2

Diff since v0.3.1

This version fixes the broken compatibility bounds affecting v0.3.1 (which is currently uninstallable, see gh-11).

Closed issues:

• Requesting a new release (#9)

Merged pull requests:

• Install TagBot as a GitHub Action (#10) (@JuliaTagBot)
• Update compat bounds (#12) (@ffevotte)

v0.3.1 (2020-01-18)

Diff since v0.3.0

Closed issues:

• EFT overflows (#8)
• delete Package.toml (#6)

Merged pull requests:

• Avoid allocations by inlining functions accepting accumulators (#5) (chriselrod)