RFR(S)JDK-8214074: Ghash optimization using AVX instructions
smita.kamath at intel.com
Mon Nov 19 22:39:47 UTC 2018
The performance gain that I reported was over the earlier implementation of GHASH using clmul instructions (generate_ghash_process_blocks).
From: Florian Weimer [mailto:fweimer at redhat.com]
Sent: Monday, November 19, 2018 1:55 PM
To: Kamath, Smita <smita.kamath at intel.com>
Cc: 'Vladimir Kozlov' <vladimir.kozlov at oracle.com>; Anthony Scarpino <anthony.scarpino at oracle.com>; core-libs-dev at openjdk.java.net; hotspot compiler <hotspot-compiler-dev at openjdk.java.net>
Subject: Re: RFR(S)JDK-8214074: Ghash optimization using AVX instructions
* Smita Kamath:
> I'd like to contribute an optimization for GHASH Algorithm using AVX
> Instructions. I have tested this optimization on SKX x86_64 platform
> and it shows ~20-30% performance improvement for larger message sizes
> (for example 8k).
Performance improvement against what? The pure Java implementation? I find this a bit surprising. I would have expected a much larger performance improvement from the use of CLMUL instructions.
More information about the hotspot-compiler-dev