RE: [PATCH] config/arm: add Ampere Altra/AltraMax/AmpereOne platform

[Yutang Jiang <yutang.jiang@amperecomputing.com>]

> -----Original Message-----
> From: Ruifeng Wang <Ruifeng.Wang@arm.com>
> Sent: Tuesday, June 13, 2023 4:18 PM
> To: Yutang Jiang <yutang.jiang@amperecomputing.com>
> Cc: Open Source Submission <patches@amperecomputing.com>;
> dev@dpdk.org; Yutang Jiang OS <jiangyutang@os.amperecomputing.com>;
> Juraj Linkeš <juraj.linkes@pantheon.tech>; nd <nd@arm.com>
> Subject: RE: [PATCH] config/arm: add Ampere Altra/AltraMax/AmpereOne
> platform
> 
> [EXTERNAL EMAIL NOTICE: This email originated from an external sender.
> Please be mindful of safe email handling and proprietary information
> protection practices.]
> 
> 
> > -----Original Message-----
> > From: Yutang Jiang <yutang.jiang@amperecomputing.com>
> > Sent: Monday, June 12, 2023 10:48 PM
> > To: Ruifeng Wang <Ruifeng.Wang@arm.com>
> > Cc: Open Source Submission <patches@amperecomputing.com>;
> > dev@dpdk.org; Yutang Jiang OS <jiangyutang@os.amperecomputing.com>
> > Subject: RE: [PATCH] config/arm: add Ampere Altra/AltraMax/AmpereOne
> > platform
> >
> > Hi Ruifeng,
> >
> >     Please help me review this patch. Thank you very much.
> >
> > Best Regards,
> > Yutang Jiang
> >
> > -----Original Message-----
> > From: Yutang Jiang OS <jiangyutang@os.amperecomputing.com>
> > Sent: Sunday, June 11, 2023 1:03 PM
> > To: dev@dpdk.org
> > Cc: Open Source Submission <patches@amperecomputing.com>; Yutang
> Jiang
> > <yutang.jiang@amperecomputing.com>; Yutang Jiang OS
> > <jiangyutang@os.amperecomputing.com>
> > Subject: [PATCH] config/arm: add Ampere Altra/AltraMax/AmpereOne
> > platform
> >
> > This patch add Ampere series platform support for DPDK:
> > 1. Merging the eMAG and AmpereOne to the one struct
> implementer_ampere.
> > 2. The microarchitecture of Altra/AltraMax is N1, which
> implementer/part_number
> >    is defined in arm, so the definition of
> RTE_MAX_LCORE/RTE_MAX_NUMA_NODES
> >    refers to the quadrant sub-numa definition in AltraMax 2P system.
> > 3. Added basic definition of AmpereOne.
> >
> > Signed-off-by: Yutang Jiang <yutang.jiang@amperecomputing.com>
> > Signed-off-by: Yutang Jiang <jiangyutang@os.amperecomputing.com>
> > ---
> >  config/arm/arm64_altra_linux_gcc     | 16 +++++++++
> >  config/arm/arm64_ampereone_linux_gcc | 16 +++++++++
> >  config/arm/meson.build               | 52 ++++++++++++++++++++++++----
> >  3 files changed, 78 insertions(+), 6 deletions(-)  create mode 100644
> > config/arm/arm64_altra_linux_gcc  create mode 100644
> > config/arm/arm64_ampereone_linux_gcc
> >
> > diff --git a/config/arm/arm64_altra_linux_gcc
> > b/config/arm/arm64_altra_linux_gcc
> > new file mode 100644
> > index 0000000000..ce0667ebe2
> > --- /dev/null
> > +++ b/config/arm/arm64_altra_linux_gcc
> > @@ -0,0 +1,16 @@
> > +[binaries]
> > +c = ['ccache', 'aarch64-linux-gnu-gcc'] cpp = ['ccache',
> > +'aarch64-linux-gnu-g++'] ar = 'aarch64-linux-gnu-gcc-ar'
> > +strip = 'aarch64-linux-gnu-strip'
> > +pkgconfig = 'aarch64-linux-gnu-pkg-config'
> > +pcap-config = ''
> > +
> > +[host_machine]
> > +system = 'linux'
> > +cpu_family = 'aarch64'
> > +cpu = 'armv8.2-a'
> > +endian = 'little'
> > +
> > +[properties]
> > +platform = 'altra'
> > diff --git a/config/arm/arm64_ampereone_linux_gcc
> > b/config/arm/arm64_ampereone_linux_gcc
> > new file mode 100644
> > index 0000000000..8964432a74
> > --- /dev/null
> > +++ b/config/arm/arm64_ampereone_linux_gcc
> > @@ -0,0 +1,16 @@
> > +[binaries]
> > +c = ['ccache', 'aarch64-linux-gnu-gcc'] cpp = ['ccache',
> > +'aarch64-linux-gnu-g++'] ar = 'aarch64-linux-gnu-gcc-ar'
> > +strip = 'aarch64-linux-gnu-strip'
> > +pkgconfig = 'aarch64-linux-gnu-pkg-config'
> > +pcap-config = ''
> > +
> > +[host_machine]
> > +system = 'linux'
> > +cpu_family = 'aarch64'
> > +cpu = 'armv8.6-a'
> > +endian = 'little'
> > +
> > +[properties]
> > +platform = 'ampereone'
> > diff --git a/config/arm/meson.build b/config/arm/meson.build index
> > 43f6a551a2..57424b88dc
> > 100644
> > --- a/config/arm/meson.build
> > +++ b/config/arm/meson.build
> > @@ -71,8 +71,8 @@ part_number_config_arm = {
> >              ['RTE_MACHINE', '"neoverse-n1"'],
> >              ['RTE_ARM_FEATURE_ATOMICS', true],
> >              ['RTE_MAX_MEM_MB', 1048576],
> > -            ['RTE_MAX_LCORE', 160],
> > -            ['RTE_MAX_NUMA_NODES', 2]
> > +            ['RTE_MAX_LCORE', 256],
> > +            ['RTE_MAX_NUMA_NODES', 8]
> >          ]
> >      },
> >      '0xd40': {
> > @@ -166,16 +166,29 @@ implementer_cavium = {  implementer_ampere =
> {
> >      'description': 'Ampere Computing',
> >      'flags': [
> > -        ['RTE_MACHINE', '"emag"'],
> >          ['RTE_CACHE_LINE_SIZE', 64],
> > -        ['RTE_MAX_LCORE', 32],
> > -        ['RTE_MAX_NUMA_NODES', 1]
> > +        ['RTE_MAX_LCORE', 1024],
> > +        ['RTE_MAX_NUMA_NODES', 8]
> >      ],
> >      'part_number_config': {
> >          '0x0': {
> >              'march': 'armv8-a',
> 
> eMag is defined as armv8.2-a in cross file.
> Do you want to update march to armv8.2-a?
[Yutang Jiang] The march definition in eMAG cross file and here are both 'armv8-a', I have not made any changes. The definition in the newly added Altra cross file is armv8.2-a
> 
> >              'march_features': ['crc', 'crypto'],
> > -            'compiler_options':  ['-mtune=emag']
> > +            'compiler_options':  ['-mtune=emag'],
> > +            'flags': [
> > +                ['RTE_MACHINE', '"eMAG"'],
> > +                ['RTE_MAX_LCORE', 32],
> > +                ['RTE_MAX_NUMA_NODES', 1]
> > +            ]
> > +        },
> > +        '0xac3': {
> > +            'march': 'armv8.6-a',
> > +            'march_features': ['crc', 'crypto'],
> > +            'compiler_options':  ['-mcpu=ampere1'],
> > +            'flags': [
> > +                ['RTE_MACHINE', '"AmpereOne"'],
> > +                ['RTE_MAX_LCORE', 320]
> 
> Add MAX_NUMA_NODES here.
[Yutang Jiang] Yes, I will add MAX_NUMA_NODES definitions in the next version patch.
> 
> > +            ]
> >          }
> >      }
> >  }
> > @@ -266,6 +279,7 @@ implementers = {
> >      '0x43': implementer_cavium,
> >      '0x48': implementer_hisilicon,
> >      '0x50': implementer_ampere,
> > +    '0xc0': implementer_ampere,
> >      '0x51': implementer_qualcomm,
> >      '0x70': implementer_phytium,
> >  }
> > @@ -347,6 +361,28 @@ soc_emag = {
> >      'part_number': '0x0'
> >  }
> >
> > +soc_altra = {
> > +    'description': 'Ampere Altra/AltraMax',
> > +    'implementer': '0x41',
> > +    'part_number': '0xd0c',
> > +    'flags': [
> > +        ['RTE_MAX_LCORE', 256],
> > +        ['RTE_MAX_NUMA_NODES', 8]
> 
> These two flags are redundant. MAX_LCORE & MAX_NUMA_NODES are
> already in corresponding part_number_config.
[Yutang Jiang] Yes, I will remove these two duplicate definitions in the next version patch.
> 
> > +    ],
> > +    'numa': true
> > +}
> > +
> > +soc_ampereone = {
> > +    'description': 'Ampere AmpereOne',
> > +    'implementer': '0xc0',
> > +    'part_number': '0xac3',
> > +    'flags': [
> > +        ['RTE_MAX_LCORE', 320],
> > +        ['RTE_MAX_NUMA_NODES', 8]
> 
> Same here.
[Yutang Jiang] Yes, I will remove these two duplicate definitions in the next version patch.
> 
> > +    ],
> > +    'numa': true
> > +}
> > +
> >  soc_ft2000plus = {
> >      'description': 'Phytium FT-2000+',
> >      'implementer': '0x70',
> > @@ -469,6 +505,8 @@ cn9k:            Marvell OCTEON 9
> >  cn10k:           Marvell OCTEON 10
> >  dpaa:            NXP DPAA
> >  emag:            Ampere eMAG
> > +altra:           Ampere Altra/AltraMax
> > +ampereone:       Ampere AmpereOne
> >  ft2000plus:      Phytium FT-2000+
> >  tys2500:         Phytium TengYun S2500
> >  graviton2:       AWS Graviton2
> > @@ -496,6 +534,8 @@ socs = {
> >      'cn10k' : soc_cn10k,
> >      'dpaa': soc_dpaa,
> >      'emag': soc_emag,
> > +    'altra': soc_altra,
> > +    'ampereone': soc_ampereone,
> >      'ft2000plus': soc_ft2000plus,
> >      'tys2500': soc_tys2500,
> >      'graviton2': soc_graviton2,
> > --
> > 2.40.0