[AMD Official Use Only - General]
Hi Jerin,
> -Original Message-
> From: Jerin Jacob
> Sent: Tuesday, October 17, 2023 8:53 AM
> To: Tummala, Sivaprasad
> Cc: harry.van.haa...@intel.com; anatoly.bura...@intel.com; dev@dpdk.org;
> Yigit,
> Ferruh ; david.h...@intel.com
> Subject: Re: [P
On Wed, Oct 18, 2023 at 12:38 PM Tummala, Sivaprasad
wrote:
>
> [AMD Official Use Only - General]
>
> Hi Jerin,
>
> > -Original Message-
> > From: Jerin Jacob
> > Sent: Tuesday, October 17, 2023 8:53 AM
> > To: Tummala, Sivaprasad
> > Cc: harry.van.haa...@intel.com; anatoly.bura...@intel
On Tue, Oct 17, 2023 at 4:48 PM Ankur Dwivedi wrote:
>
> While adding flow in npc_flow_list, the flow can be added before the
> current flow iterator. The function returns after adding this flow.
> This prevents flow to be added in age flow list correctly. This patch moves
> the addition of age fl
> -Original Message-
> From: Robin Jarry
> Sent: Tuesday, October 17, 2023 5:47 PM
> To: Sunil Kumar Kori
> Cc: dev@dpdk.org; Thomas Monjalon ; Jerin Jacob
>
> Subject: [EXT] Re: [PATCH v2 1/1] usertools/rss: add CNXK RSS key
>
> External Email
>
> -
> -Original Message-
> From: Amit Prakash Shukla
> Sent: Wednesday, October 18, 2023 12:24 AM
> To: Vamsi Krishna Attunuru
> Cc: dev@dpdk.org; Jerin Jacob Kollanukkaran ;
> fengcheng...@huawei.com; kevin.la...@intel.com;
> bruce.richard...@intel.com; conor.wa...@intel.com; g.si...@nxp.
From: Sunil Kumar Kori
This patch adds RSS key for CNXK platforms. CNXK platform uses
48 bytes long key for hash calculations.
For the same patch also updates help mesaages to provide range
information for supporting NICs/platforms.
Also CNXK uses reta size as 64 so to get correct offset to ret
This patch series try to enhance the bond framework to support the
offload feature better:
* Add new API to make the member port can access some information of the
bond port which belongs.
* Add new API to get the result of whether bond port is created by the
member port.
* Add two command line
From: Long Wu
Bonding PMD does not let member ports know the bonding port's
information, like how many member ports the bonding port has,
what mode the bonding port is in and so on.
Add the notification interface for bonding port to let member
port know it is added to a bonding port and what the
From: Long Wu
After bonding port notification, member port hardware may create the
bonding port. We want to get the result of creatition, so we add this
API to do the getting action.
Signed-off-by: Long Wu
Reviewed-by: James Hershaw
Reviewed-by: Chaoyong He
---
drivers/net/bonding/rte_eth_bo
From: Long Wu
Include the following new arguments for bonding ports:
- "notify_member" to enable/disable member notification.
- "dedicated_queue" to enable/disable dedicated queue.
Add these two arguments in initial argument.
Signed-off-by: Long Wu
Reviewed-by: James Hershaw
Reviewed-by: Chao
From: Long Wu
Several ports may share the same PCI address, like nfp representor.
So we cannot add this type of ports to bonding port by "--vdev"
argument in dpdk-testpmd. But the port's data name is unique between
them, we include an option to add such ports to the bonding port.
After adding th
From: Long Wu
Add API to support checking if the port id is a bonding
port id.
Signed-off-by: Long Wu
Reviewed-by: James Hershaw
Reviewed-by: Chaoyong He
---
drivers/net/bonding/rte_eth_bond.h | 13 +
drivers/net/bonding/rte_eth_bond_api.c | 7 +++
drivers/net/bonding/ve
From: Long Wu
Add some commands to support bonding port notification in
dpdk-testpmd.
1. We can enable the notification by command:
"set bonding notify_member (port_id) (enable|disable)"
2. If member port hardware try to create the bonding port after
notification we can get the status by comman
This series adds new fastpath routines for cn10k & cn9k endpoint
devices and supports 32B Tx desciptor format which improves the
performance.
V5 changes:
- Series rebased
v4 changes:
- Use rte_atomic_xxx instead of __atomic_xxx built-ins
v2 & v3 changes:
- Fixed CI
Shijith Thotton (1):
net/oc
From: Shijith Thotton
Update input queue setup to consider descriptor size in driver conf.
The default instruction size for otx2 and cnxk devices has been updated
to 32 bytes.
Signed-off-by: Shijith Thotton
---
drivers/net/octeon_ep/cnxk_ep_vf.c| 10 +-
drivers/net/octeon_ep/otx2_e
Patch improves Rx routine and pkt count update routines,
packet count update routines need to drain inflight ISM
memory updates while decrementing the packet count register.
Signed-off-by: Vamsi Attunuru
---
drivers/net/octeon_ep/otx_ep_rxtx.c | 164
1 file changed,
Adds new fastpath routines for cn10k & cn9k endpoint
devices and assigns the fastpath routines based on
the offload flags.
Patch also adds misc changes to improve performance
and code-readability.
Signed-off-by: Vamsi Attunuru
---
drivers/net/octeon_ep/cnxk_ep_rx.c| 310
Calling an experimental symbol from an inline helper triggers a warning
when such code is not compiled with experimental API.
This can be seen when rte_bitops.h gets (indirectly) included in OVS
builds.
On the other hand, rte_clz32, rte_clz64, rte_ctz32, rte_ctz64,
rte_popcount32, rte_popcount64 a
18/10/2023 09:26, Sunil Kumar Kori:
> From: Robin Jarry
> > From: Sunil Kumar Kori
> > >
> > > This patch adds RSS key for CNXK platforms. CNXK platform uses
> > > 48 bytes long key for hash calculations.
> > >
> > > For the same patch also updates help mesaages to provide range
> > > information
On Wed, Oct 18, 2023 at 1:17 PM Vamsi Krishna Attunuru
wrote:
>
>
>
> > -Original Message-
> > From: Amit Prakash Shukla
> > Sent: Wednesday, October 18, 2023 12:24 AM
> > To: Vamsi Krishna Attunuru
> > Cc: dev@dpdk.org; Jerin Jacob Kollanukkaran ;
> > fengcheng...@huawei.com; kevin.la..
driver_sdk_headers is used to expose headers that may be used by
external drivers.
Don't export ml/cnxk internal headers.
Fixes: fe83ffd9ec2e ("ml/cnxk: add skeleton")
Signed-off-by: David Marchand
---
drivers/ml/cnxk/meson.build | 7 ---
1 file changed, 7 deletions(-)
diff --git a/drivers
Thomas Monjalon, Oct 18, 2023 at 11:14:
18/10/2023 09:26, Sunil Kumar Kori:
> From: Robin Jarry
> > From: Sunil Kumar Kori
> > >
> > > This patch adds RSS key for CNXK platforms. CNXK platform uses
> > > 48 bytes long key for hash calculations.
> > >
> > > For the same patch also updates help m
As an external header, rte_thash_x86_gfni.h should be self sufficient.
Fixes: 3d4e27fd7ff0 ("use abstracted bit count functions")
Signed-off-by: David Marchand
---
lib/hash/rte_thash_x86_gfni.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/hash/rte_thash_x86_gfni.h b/lib/hash/rte_thas
> -Original Message-
> From: David Marchand
> Sent: 18 October 2023 14:46
> To: dev@dpdk.org
> Cc: Jerin Jacob Kollanukkaran ; tho...@monjalon.net;
> Srikanth Yalavarthi ; Prince Takkar
>
> Subject: [EXT] [PATCH] ml/cnxk: don't export internal headers
>
> External Email
>
>
On Wed, Oct 18, 2023 at 10:48:14AM +0530, Jerin Jacob wrote:
> On Tue, Oct 17, 2023 at 10:21 PM Jerin Jacob wrote:
> >
> > On Tue, Oct 17, 2023 at 9:45 PM Bruce Richardson
> > wrote:
> > >
> > > On Tue, Oct 17, 2023 at 09:34:04PM +0530, Jerin Jacob wrote:
> > > > On Tue, Oct 17, 2023 at 9:32 PM B
> -Original Message-
> From: Robin Jarry
> Sent: Wednesday, October 18, 2023 2:48 PM
> To: Thomas Monjalon ; Sunil Kumar Kori
>
> Cc: dev@dpdk.org; Jerin Jacob
> Subject: Re: [EXT] Re: [PATCH v2 1/1] usertools/rss: add CNXK RSS key
>
> Thomas Monjalon, Oct 18, 2023 at 11:14:
> > 18/10/2
On Mon, Oct 9, 2023 at 5:06 AM Ma, WenwuX wrote:
> > From a pci bus API pov, nothing prevents a driver from mixing memory
> > mapped with vfio and ioport resources (iow, calls to
> > rte_pci_map_resource() and rte_pci_ioport_map()).
> > IOW, it may not be the case with the net/virtio driver but, i
On Wed, Oct 18, 2023 at 12:05 PM wrote:
>
> From: Rakesh Kudurumalla
>
> Adds an use case l3fwd. It contains a dedicated l3fwd.cli file
> mentioning commands to configure the required resources.
>
> Once application successfully parses the l3fwd.cli then a graph is
> created having below nodes:
>
Series-acked-by: Chengwen Feng
On 2023/10/18 9:39, Jie Hai wrote:
> This patch adds telemetry commands for TM capabilities and make some
> bufix for hns3 driver.
>
> Jie Hai (4):
> net/hns3: fix a typo
> ethdev: add telemetry command for TM capabilities
> ethdev: add telemetry command for
On Tue, Oct 10, 2023 at 7:55 PM Hernan Vargas wrote:
>
> Implement de-ratematch pre-processing for ACC100 SW corner cases.
> Some specific 5GUL FEC corner cases may cause unintended back pressure
> and in some cases a potential stability issue on the ACC100.
> The PMD can detect such code block co
On Wed, May 31, 2023 at 11:52 AM David Marchand
wrote:
>
> (I reformatted the mail a bit)
>
> On Wed, May 31, 2023 at 10:51 AM Elena Agostini wrote:
> > > On Wed, May 31, 2023 at 10:44 AM Elena Agostini eagost...@nvidia.com
> > > wrote:
> > > > > On Tue, May 30, 2023 at 1:48 PM eagost...@nvidia.
On Tue, Oct 17, 2023 at 2:29 PM Aaron Conole wrote:
> David Marchand writes:
> > Add missing package so manuals are generated as part of the docs check.
> >
> > Signed-off-by: David Marchand
> Reviewed-by: Aaron Conole
Applied, thanks.
--
David Marchand
> As per IPSEC ESP RFC 4303, for both tunnel mode or transport mode,
> next proto 50, so we cannot identify a packet is for tunnel mode or
> transport mode by just packet parsing.
> Am I missing something ?
You are absolutely correct, the only way to tell the difference is
to parse the next_proto f
This series adds new fastpath routines for cn10k & cn9k endpoint
devices and supports 32B Tx descriptor format which improves the
performance.
V6 changes:
- Use __atomic_xxx built-ins to fix CI build
V5 changes:
- Series rebased
v4 changes:
- Use rte_atomic_xxx instead of __atomic_xxx built-ins
From: Shijith Thotton
Update input queue setup to consider descriptor size in driver conf.
The default instruction size for otx2 and cnxk devices has been updated
to 32 bytes.
Signed-off-by: Shijith Thotton
---
drivers/net/octeon_ep/cnxk_ep_vf.c| 10 +-
drivers/net/octeon_ep/otx2_e
Patch improves Rx routine and pkt count update routines,
packet count update routines need to drain inflight ISM
memory updates while decrementing the packet count register.
Signed-off-by: Vamsi Attunuru
---
drivers/net/octeon_ep/otx_ep_rxtx.c | 162
1 file changed,
Adds new fastpath routines for cn10k & cn9k endpoint
devices and assigns the fastpath routines based on
the offload flags.
Patch also adds misc changes to improve performance
and code-readability.
Signed-off-by: Vamsi Attunuru
---
drivers/net/octeon_ep/cnxk_ep_rx.c| 309
Hello Bruce,
On Tue, Oct 17, 2023 at 7:08 PM Bruce Richardson
wrote:
> > > sure I like it that much as a feature :-) I rather like having unique
> > > prefixes for each command. I wasn't actually aware of the testpmd "help
> > > " command at all. I will have to look into it.
> >
> > Let me propo
hns3 patch can be stripped from this series.
ltgm,
Acked-by: Huisong Li
在 2023/10/18 9:39, Jie Hai 写道:
This patch fixes a typo.
Fixes: c09c7847d892 ("net/hns3: support traffic management")
Cc: sta...@dpdk.org
Signed-off-by: Jie Hai
---
drivers/net/hns3/hns3_tm.c | 4 ++--
1 file changed,
On Wed, Oct 18, 2023 at 01:21:40PM +0200, David Marchand wrote:
> Hello Bruce,
>
>
> On Tue, Oct 17, 2023 at 7:08 PM Bruce Richardson
> wrote:
> > > > sure I like it that much as a feature :-) I rather like having unique
> > > > prefixes for each command. I wasn't actually aware of the testpmd "
On Wed, Oct 18, 2023 at 10:48:14AM +0530, Jerin Jacob wrote:
> On Tue, Oct 17, 2023 at 10:21 PM Jerin Jacob wrote:
> >
> > On Tue, Oct 17, 2023 at 9:45 PM Bruce Richardson
> > wrote:
> > >
> > > On Tue, Oct 17, 2023 at 09:34:04PM +0530, Jerin Jacob wrote:
> > > > On Tue, Oct 17, 2023 at 9:32 PM B
The eventdevs based on vdevs, rather than on e.g. HW PCI devices, were,
as a rule, not setting the ".dev" pointer in the eventdev structure.
This caused issues as a NULL pointer was returned in calls to info_get,
triggering crashes if the pointer is passed unchecked to e.g.
rte_dev_name() to print
When calling rte_eventdev_info_get() the driver name string field should
be populated.
Fixes: 929da5e6 ("event/skeleton: add skeleton eventdev driver")
Cc: sta...@dpdk.org
Signed-off-by: Bruce Richardson
---
drivers/event/skeleton/skeleton_eventdev.c | 2 ++
1 file changed, 2 insertions(+)
On 10/18/2023 3:35 PM, David Marchand wrote:
On Mon, Oct 9, 2023 at 5:06 AM Ma, WenwuX wrote:
From a pci bus API pov, nothing prevents a driver from mixing memory
mapped with vfio and ioport resources (iow, calls to
rte_pci_map_resource() and rte_pci_ioport_map()).
IOW, it may not be the ca
On Wed, Oct 18, 2023 at 2:26 PM Bruce Richardson
wrote:
>
> When calling rte_eventdev_info_get() the driver name string field should
> be populated.
>
> Fixes: 929da5e6 ("event/skeleton: add skeleton eventdev driver")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Bruce Richardson
event/dpaa2 seem
On Wed, Oct 18, 2023 at 02:44:11PM +0200, David Marchand wrote:
> On Wed, Oct 18, 2023 at 2:26 PM Bruce Richardson
> wrote:
> >
> > When calling rte_eventdev_info_get() the driver name string field should
> > be populated.
> >
> > Fixes: 929da5e6 ("event/skeleton: add skeleton eventdev driver"
Some cryptodev driver specific tests were in the "driver-tests" suite,
while similar tests for other drivers were not.
Signed-off-by: David Marchand
---
app/test/test_cryptodev.c | 20 ++--
app/test/test_cryptodev_asym.c | 9 +++--
2 files changed, 13 insertions(+), 16
On Wed, Oct 18, 2023 at 2:26 PM Bruce Richardson
wrote:
>
> The eventdevs based on vdevs, rather than on e.g. HW PCI devices, were,
> as a rule, not setting the ".dev" pointer in the eventdev structure.
> This caused issues as a NULL pointer was returned in calls to info_get,
> triggering crashes
On Wed, Oct 18, 2023 at 02:55:42PM +0200, David Marchand wrote:
> Some cryptodev driver specific tests were in the "driver-tests" suite,
> while similar tests for other drivers were not.
>
> Signed-off-by: David Marchand
> ---
Spotted that inconsistency myself but never got around to fixing it.
The eventdevs based on vdevs, rather than on e.g. HW PCI devices, were,
as a rule, not setting the ".dev" pointer in the eventdev structure.
This caused issues as a NULL pointer was returned in calls to info_get,
triggering crashes if the pointer is passed unchecked to e.g.
rte_dev_name() to print
> -Original Message-
> From: Feifei Wang
> Sent: Tuesday, September 12, 2023 7:39 AM
> Cc: dev@dpdk.org; n...@arm.com; Feifei Wang
> Subject: [PATCH v3 0/3] fix test-pipeline issues
>
> For test-pipeline application, there are some problems with the normal
> operation of the program a
Rather than relying on the individual drivers to always populated the
driver name field in the info structure - something missed by some
drivers, we can do so in the eventdev rte_event_dev_info_get() function.
This fixes issues
Fixes: 929da5e6 ("event/skeleton: add skeleton eventdev driver")
F
On Wed, Oct 18, 2023 at 01:52:08PM +0100, Bruce Richardson wrote:
> On Wed, Oct 18, 2023 at 02:44:11PM +0200, David Marchand wrote:
> > On Wed, Oct 18, 2023 at 2:26 PM Bruce Richardson
> > wrote:
> > >
> > > When calling rte_eventdev_info_get() the driver name string field should
> > > be populate
On Wed, Oct 18, 2023 at 2:39 PM Gupta, Nipun wrote:
> On 10/18/2023 3:35 PM, David Marchand wrote:
> > On Mon, Oct 9, 2023 at 5:06 AM Ma, WenwuX wrote:
> >>> From a pci bus API pov, nothing prevents a driver from mixing memory
> >>> mapped with vfio and ioport resources (iow, calls to
> >>> rte_
This patch series is an implementation of revised ml/cnxk driver
to support models compiled with TVM compiler framework. TVM models
use a hybrid mode for execution, with regions of the model executing
on the ML accelerator and the rest executing on CPU cores.
This series of commits reorganizes the
Dropped support for device argument "poll_mem" for cnxk
ML driver. Support to use registers for polling is removed
and DDR addresses would be used for polling.
Signed-off-by: Srikanth Yalavarthi
---
Depends-on: patch-132887 ("ml/cnxk: don't export internal headers")
doc/guides/mldevs/cnxk.rst
Introduce generic cnxk device structure. This structure is
a top level device structure for the driver, which would
encapsulate the target / platform specific device structure.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 316 ++--
drivers/ml/cnxk/cn1
Introduced generic xstats structures and renamed cn10k
xstats enumerations with cnxk prefix.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 86 +---
drivers/ml/cnxk/cn10k_ml_model.h | 6 +-
drivers/ml/cnxk/cn10k_ml_ops.c | 169 ++---
Renamed cn10k ML ops structure with cnxk prefix.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 2 +-
drivers/ml/cnxk/cn10k_ml_ops.c | 73 +-
drivers/ml/cnxk/cn10k_ml_ops.h | 34 +++-
drivers/ml/cnxk/cnxk_ml_ops.c | 36 ++
Introduce generic cnxk model and layer structure. These
structures would enable supporting models with multiple
layers. A model is a collection of multiple independent
layers with flow dependencies between the layers.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 9
Implement CNXK wrapper functions for dev_info_get,
dev_configure, dev_close, dev_start and dev_stop. The
wrapper functions allocate / release common resources
for the ML driver and invoke device specific functions.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 230 ++---
Added generic cnxk request structure. Moved common fields
from cn10k structures to cnxk structure. Moved job related
structures and enumerations to ops headers.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 72 +++
drivers/ml/cnxk/cn10k_ml_dev.h | 269 +
Added cnxk wrapper function to handle ML device queue-pairs.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 135 +
drivers/ml/cnxk/cn10k_ml_ops.h | 7 +-
drivers/ml/cnxk/cnxk_ml_ops.c | 153 -
drivers/ml/cnxk
Implemented cnxk wrapper functions to load and unload
ML models. Wrapper functions would invoke the cn10k
model load and unload functions.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_model.c | 244 -
drivers/ml/cnxk/cn10k_ml_model.h | 26 ++-
drivers/
Implemented cnxk wrapper functions to start and stop
ML models. Wrapper functions would invoke the cn10k
model start and stop functions.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ocm.c | 28 ++--
drivers/ml/cnxk/cn10k_ml_ocm.h | 12 +-
drivers/ml/cnxk/cn10k_ml_ops.c | 282
Added cnxk wrapper function to update model params and
fetch model info.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 38 ++-
drivers/ml/cnxk/cn10k_ml_ops.h | 5 ++--
drivers/ml/cnxk/cnxk_ml_ops.c | 48 --
3 fil
Added cnxk wrapper functions to quantize input data and
dequantize output data.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 164 -
drivers/ml/cnxk/cn10k_ml_ops.h | 7 --
drivers/ml/cnxk/cnxk_ml_io.c | 95 +++
driver
Added cnxk wrapper for device dump and selftest debug
functions.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_model.c | 118 +
drivers/ml/cnxk/cn10k_ml_model.h | 1 +
drivers/ml/cnxk/cn10k_ml_ocm.c | 8 +-
drivers/ml/cnxk/cn10k_ml_ocm.h | 2 +-
dr
Added cnxk wrapper function to handle ML device stats
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 32 --
drivers/ml/cnxk/cn10k_ml_ops.h | 2 --
drivers/ml/cnxk/cnxk_ml_ops.c | 36 --
3 files changed, 34 ins
Added cnxk wrapper function to handle ML device and model
extended stats. Handling resources for the xstats is done
in the cnxk layer. Introduced internal xstats group.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 4 -
drivers/ml/cnxk/cn10k_ml_ops.c | 531 +++
Implemented cnxk layer fast-path functions and added support
for model specific fast-path functions. CNXK layer functions
would invoke model specific fast-path functions.
Added support for model specific poll handling functions and
updated internal inference sync function. Drop use of rte_ml_op
as
Move error type structures to cnxk layer. cn10k layer to
handle fw and hw error sub-types only.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 41 ++-
drivers/ml/cnxk/cn10k_ml_ops.c | 93 +-
drivers/ml/cnxk/cnxk_ml_dev.c | 8
Introduced model type, sub-type and layer type. Added
internal structures for TVM model objects.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ocm.c | 3 ++
drivers/ml/cnxk/cn10k_ml_ops.c | 6 ++-
drivers/ml/cnxk/cnxk_ml_model.h | 66 +++-
dri
Added support to configure and close TVMDP library based
on ML device configuration options.
Updated meson build to enable Jansson, TVM runtime, TVMDP
library as build dependencies.
Signed-off-by: Srikanth Yalavarthi
---
doc/guides/mldevs/cnxk.rst | 78
dr
Enable support to parse model buffer to identify the
model type and model sub-type. Enabled basic checks
for Glow model type buffer.
Signed-off-by: Srikanth Yalavarthi
Signed-off-by: Anup Prabhu
---
drivers/ml/cnxk/cnxk_ml_model.c | 49
drivers/ml/cnxk/cnxk_ml_mode
Added support to parse TVM model objects from the model
archive buffer. Added support to check for all expected
objects and copy TVM model objects to internal buffers.
Signed-off-by: Srikanth Yalavarthi
Signed-off-by: Anup Prabhu
---
drivers/ml/cnxk/cnxk_ml_ops.c| 5 ++-
drivers/ml/cnxk/mv
Added support to fetch TVM model layer information and
update internal structures based on the layer information
Set callback functions for layer load and unload and
enable model loading using TVMDP library. Added support
to fetch full metadata after model load.
Signed-off-by: Srikanth Yalavarthi
Enable unloading model using external tvmdp library. Updated
layer unload callback to support multiple layers.
Signed-off-by: Srikanth Yalavarthi
Signed-off-by: Anup Prabhu
---
drivers/ml/cnxk/cn10k_ml_ops.c | 8 +---
drivers/ml/cnxk/cnxk_ml_ops.c| 7 +--
drivers/ml/cnxk/mvtvm_m
Enabled updating internal IO info structures for TVM model.
Compute static fields related to the model I/O.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cnxk_ml_ops.c| 4 ++
drivers/ml/cnxk/mvtvm_ml_model.c | 111 +++
drivers/ml/cnxk/mvtvm_ml_model.h |
From: Anup Prabhu
Enabled check for OCM size requirement for multi-layer
TVM model. Compute OCM scratch and WB requirement for
all layers during the load stage.
Signed-off-by: Anup Prabhu
---
drivers/ml/cnxk/cnxk_ml_ops.c | 60 +++
1 file changed, 60 insertions(
Added support to start and stop TVM models. TVM model
start would invoke layer start for all Glow layers part
of the model. TVM model stop would invoke layer stop
for all Glow layers part of the model.
Signed-off-by: Srikanth Yalavarthi
Signed-off-by: Anup Prabhu
---
drivers/ml/cnxk/cn10k_ml_op
Enabled support to print TVM model layer info.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cnxk_ml_model.c | 7 +++-
drivers/ml/cnxk/mvtvm_ml_model.c | 59
drivers/ml/cnxk/mvtvm_ml_model.h | 2 ++
drivers/ml/cnxk/mvtvm_ml_stubs.c | 8 +
drivers
From: Prince Takkar
Added support to update internal model info structure
for TVM models.
Signed-off-by: Prince Takkar
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/mvtvm_ml_model.c | 65
drivers/ml/cnxk/mvtvm_ml_model.h | 2 +
drivers/ml/cnxk/mvtvm_
Added model xstats entries to compute runtime latency.
Allocated internal resources for TVM model xstats.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 9 +++
drivers/ml/cnxk/cn10k_ml_ops.h | 2 +
drivers/ml/cnxk/cnxk_ml_ops.c| 131 ++
Implemented generic ML malloc and free callbacks
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 30 ++
drivers/ml/cnxk/cn10k_ml_ops.h | 3 +++
drivers/ml/cnxk/mvtvm_ml_ops.c | 2 ++
3 files changed, 35 insertions(+)
diff --git a/drivers/ml/
From: Prince Takkar
Added support for quantize and dequantize callback
functions for TVM models.
Signed-off-by: Prince Takkar
---
drivers/ml/cnxk/mvtvm_ml_ops.c | 129 +
drivers/ml/cnxk/mvtvm_ml_ops.h | 4 +
2 files changed, 133 insertions(+)
diff --git a/dri
Implemented callback functions for IO allocation and free
for Glow layers.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 87 ++
drivers/ml/cnxk/cn10k_ml_ops.h | 3 ++
drivers/ml/cnxk/mvtvm_ml_ops.c | 2 +
3 files changed, 92 insertions(
From: Anup Prabhu
Enable fast-path ops support for TVM models. Models would
use TVMDP library function calls to execute inference
operations for Hybrid and LLVM model sub-types.
For TVM MRVL model subtypes that have a single MRVL layer,
the inference requests are directly enqueued to hardware
by
Enable support to create a mvtvm virtual device on system's
without a PCI based ML HW accelerator.
Signed-off-by: Srikanth Yalavarthi
---
doc/guides/mldevs/cnxk.rst | 49 +++-
drivers/ml/cnxk/cn10k_ml_dev.c | 8 ++
drivers/ml/cnxk/cn10k_ml_dev.h | 3 +
drivers/ml/cnxk/cnxk_ml_
On Wed, Oct 18, 2023 at 7:24 PM Srikanth Yalavarthi
wrote:
>
> This patch series is an implementation of revised ml/cnxk driver
> to support models compiled with TVM compiler framework. TVM models
> use a hybrid mode for execution, with regions of the model executing
> on the ML accelerator and th
On Wed, Oct 18, 2023 at 3:03 PM Srikanth Yalavarthi
wrote:
>
> > -Original Message-
> > From: David Marchand
> > Sent: 18 October 2023 14:46
> > To: dev@dpdk.org
> > Cc: Jerin Jacob Kollanukkaran ; tho...@monjalon.net;
> > Srikanth Yalavarthi ; Prince Takkar
> >
> > Subject: [EXT] [PATCH
Add a credits file of past contributors to DPDK.
There are obviously more names that should be added but
lets start this with Venky.
Signed-off-by: Stephen Hemminger
Acked-by: Jerin Jacob
---
v2 - reword opening, fix spelling
CREDITS | 9 +
1 file changed, 9 insertions(+)
create mode
On Tue, Aug 08, 2023 at 05:09:50PM -0700, Stephen Hemminger wrote:
> These API's have all ben around for several releases.
>
> Signed-off-by: Stephen Hemminger
> ---
Acked-by: Bruce Richardson
On Tue, Aug 08, 2023 at 05:09:58PM -0700, Stephen Hemminger wrote:
> This API was added in 2020 and should no longer be experimental.
>
> Signed-off-by: Stephen Hemminger
> ---
Acked-by: Bruce Richardson
On Tue, Aug 08, 2023 at 05:10:02PM -0700, Stephen Hemminger wrote:
> These were added in 20.11 time now to remove experimental flag.
>
> Signed-off-by: Stephen Hemminger
> ---
Acked-by: Bruce Richardson
On Tue, Aug 08, 2023 at 05:10:03PM -0700, Stephen Hemminger wrote:
> The API's for rte_meter_trtcm were never properly flagged
> as experimental; missing __rte_experimental but there was
> an experimental comment in the docbook comment.
> Remove the comment.
>
> Signed-off-by: Stephen Hemminger
>
On Tue, Aug 08, 2023 at 05:10:05PM -0700, Stephen Hemminger wrote:
> The function rte_kvargs_get_with_value was added in 21.11
> so experimental flag can be removed.
>
> Signed-off-by: Stephen Hemminger
> ---
Acked-by: Bruce Richardson
On Wed, Oct 18, 2023 at 7:52 PM Srikanth Yalavarthi
wrote:
>
> Added support to configure and close TVMDP library based
> on ML device configuration options.
>
> Updated meson build to enable Jansson, TVM runtime, TVMDP
> library as build dependencies.
>
> Signed-off-by: Srikanth Yalavarthi
> ---
Hello Stephen,
On Wed, Aug 9, 2023 at 2:10 AM Stephen Hemminger
wrote:
>
> Since 23.11 is an LTS release it is time to remove the experimental
> bandaid off many API's. There are about 850 API's marked with experimental
> on current main branch. This addresses the easy to remove ones and
> gets i
On Wed, Aug 9, 2023 at 6:44 PM Stephen Hemminger
wrote:
>
> About 80 function in EAL were marked experimental
> and should have been made stable by now.
>
> Stephen Hemminger (15):
> eal: make bitops a stable API
> eal: mark rte_dev API's as stable
> eal: make rte_class API's stable
> eal:
1 - 100 of 151 matches
Mail list logo
