* [PATCH 0/1] Expose AVX_VNNI instruction to guset @ 2021-03-01 5:15 Yang Zhong 2021-03-01 5:15 ` [PATCH 1/1] i386/cpu: " Yang Zhong 0 siblings, 1 reply; 3+ messages in thread From: Yang Zhong @ 2021-03-01 5:15 UTC (permalink / raw) To: qemu-devel; +Cc: yang.zhong, pbonzini This patch will expose AVX_VNNI features to the guest. The related kvm/kernel patches series have been queued as below link: https://lore.kernel.org/kvm/eee07399-df81-83ed-d410-18b42d51e26c@redhat.com/ Yang Zhong (1): i386/cpu: Expose AVX_VNNI instruction to guset target/i386/cpu.c | 4 ++-- target/i386/cpu.h | 2 ++ 2 files changed, 4 insertions(+), 2 deletions(-) -- 2.29.2.334.gfaefdd61ec ^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH 1/1] i386/cpu: Expose AVX_VNNI instruction to guset 2021-03-01 5:15 [PATCH 0/1] Expose AVX_VNNI instruction to guset Yang Zhong @ 2021-03-01 5:15 ` Yang Zhong 2021-03-09 5:16 ` Yang Zhong 0 siblings, 1 reply; 3+ messages in thread From: Yang Zhong @ 2021-03-01 5:15 UTC (permalink / raw) To: qemu-devel; +Cc: yang.zhong, pbonzini Expose AVX (VEX-encoded) versions of the Vector Neural Network Instructions to guest. The bit definition: CPUID.(EAX=7,ECX=1):EAX[bit 4] AVX_VNNI The following instructions are available when this feature is present in the guest. 1. VPDPBUS: Multiply and Add Unsigned and Signed Bytes 2. VPDPBUSDS: Multiply and Add Unsigned and Signed Bytes with Saturation 3. VPDPWSSD: Multiply and Add Signed Word Integers 4. VPDPWSSDS: Multiply and Add Signed Integers with Saturation The release document ref below link: https://software.intel.com/content/www/us/en/develop/download/\ intel-architecture-instruction-set-extensions-programming-reference.html Signed-off-by: Yang Zhong <yang.zhong@intel.com> --- target/i386/cpu.c | 4 ++-- target/i386/cpu.h | 2 ++ 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/target/i386/cpu.c b/target/i386/cpu.c index 6a53446e6a..55545d8692 100644 --- a/target/i386/cpu.c +++ b/target/i386/cpu.c @@ -996,7 +996,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = { .type = CPUID_FEATURE_WORD, .feat_names = { NULL, NULL, NULL, NULL, - NULL, "avx512-bf16", NULL, NULL, + "avx-vnni", "avx512-bf16", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, @@ -3273,7 +3273,7 @@ static X86CPUDefinition builtin_x86_defs[] = { MSR_ARCH_CAP_SKIP_L1DFL_VMENTRY | MSR_ARCH_CAP_MDS_NO | MSR_ARCH_CAP_PSCHANGE_MC_NO | MSR_ARCH_CAP_TAA_NO, .features[FEAT_7_1_EAX] = - CPUID_7_1_EAX_AVX512_BF16, + CPUID_7_1_EAX_AVX_VNNI | CPUID_7_1_EAX_AVX512_BF16, /* * Missing: XSAVES (not supported by some Linux versions, * including v4.1 to v4.12). diff --git a/target/i386/cpu.h b/target/i386/cpu.h index 8be39cfb62..ef5f145626 100644 --- a/target/i386/cpu.h +++ b/target/i386/cpu.h @@ -803,6 +803,8 @@ typedef uint64_t FeatureWordArray[FEATURE_WORDS]; /* Speculative Store Bypass Disable */ #define CPUID_7_0_EDX_SPEC_CTRL_SSBD (1U << 31) +/* AVX VNNI Instruction */ +#define CPUID_7_1_EAX_AVX_VNNI (1U << 4) /* AVX512 BFloat16 Instruction */ #define CPUID_7_1_EAX_AVX512_BF16 (1U << 5) -- 2.29.2.334.gfaefdd61ec ^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH 1/1] i386/cpu: Expose AVX_VNNI instruction to guset 2021-03-01 5:15 ` [PATCH 1/1] i386/cpu: " Yang Zhong @ 2021-03-09 5:16 ` Yang Zhong 0 siblings, 0 replies; 3+ messages in thread From: Yang Zhong @ 2021-03-09 5:16 UTC (permalink / raw) To: pbonzini; +Cc: yang.zhong, qemu-devel Paolo, please help review this patch when you are not busy. The related kernel patch has been merged by you last time, thanks! Regards, Yang On Mon, Mar 01, 2021 at 01:15:52PM +0800, Yang Zhong wrote: > Expose AVX (VEX-encoded) versions of the Vector Neural Network > Instructions to guest. > > The bit definition: > CPUID.(EAX=7,ECX=1):EAX[bit 4] AVX_VNNI > > The following instructions are available when this feature is > present in the guest. > 1. VPDPBUS: Multiply and Add Unsigned and Signed Bytes > 2. VPDPBUSDS: Multiply and Add Unsigned and Signed Bytes with Saturation > 3. VPDPWSSD: Multiply and Add Signed Word Integers > 4. VPDPWSSDS: Multiply and Add Signed Integers with Saturation > > The release document ref below link: > https://software.intel.com/content/www/us/en/develop/download/\ > intel-architecture-instruction-set-extensions-programming-reference.html > > Signed-off-by: Yang Zhong <yang.zhong@intel.com> > --- > target/i386/cpu.c | 4 ++-- > target/i386/cpu.h | 2 ++ > 2 files changed, 4 insertions(+), 2 deletions(-) > > diff --git a/target/i386/cpu.c b/target/i386/cpu.c > index 6a53446e6a..55545d8692 100644 > --- a/target/i386/cpu.c > +++ b/target/i386/cpu.c > @@ -996,7 +996,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = { > .type = CPUID_FEATURE_WORD, > .feat_names = { > NULL, NULL, NULL, NULL, > - NULL, "avx512-bf16", NULL, NULL, > + "avx-vnni", "avx512-bf16", NULL, NULL, > NULL, NULL, NULL, NULL, > NULL, NULL, NULL, NULL, > NULL, NULL, NULL, NULL, > @@ -3273,7 +3273,7 @@ static X86CPUDefinition builtin_x86_defs[] = { > MSR_ARCH_CAP_SKIP_L1DFL_VMENTRY | MSR_ARCH_CAP_MDS_NO | > MSR_ARCH_CAP_PSCHANGE_MC_NO | MSR_ARCH_CAP_TAA_NO, > .features[FEAT_7_1_EAX] = > - CPUID_7_1_EAX_AVX512_BF16, > + CPUID_7_1_EAX_AVX_VNNI | CPUID_7_1_EAX_AVX512_BF16, > /* > * Missing: XSAVES (not supported by some Linux versions, > * including v4.1 to v4.12). > diff --git a/target/i386/cpu.h b/target/i386/cpu.h > index 8be39cfb62..ef5f145626 100644 > --- a/target/i386/cpu.h > +++ b/target/i386/cpu.h > @@ -803,6 +803,8 @@ typedef uint64_t FeatureWordArray[FEATURE_WORDS]; > /* Speculative Store Bypass Disable */ > #define CPUID_7_0_EDX_SPEC_CTRL_SSBD (1U << 31) > > +/* AVX VNNI Instruction */ > +#define CPUID_7_1_EAX_AVX_VNNI (1U << 4) > /* AVX512 BFloat16 Instruction */ > #define CPUID_7_1_EAX_AVX512_BF16 (1U << 5) > > -- > 2.29.2.334.gfaefdd61ec ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2021-03-09 5:31 UTC | newest] Thread overview: 3+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2021-03-01 5:15 [PATCH 0/1] Expose AVX_VNNI instruction to guset Yang Zhong 2021-03-01 5:15 ` [PATCH 1/1] i386/cpu: " Yang Zhong 2021-03-09 5:16 ` Yang Zhong
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).