The "stdint.h" header is outside _RTE_TELEMETRY_H_ macro, which cause
this header is uncontrollable. So this patch moves this header to inside
_RTE_TELEMETRY_H_.
Fixes: 99a2dd955fba ("lib: remove librte_ prefix from directory names")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
---
lib/telemet
Some lib telemetry interfaces add the 'u32' and 'u64' data by the
rte_tel_data_add_dict/array_int API. This may cause data conversion
error or data truncation.
The 'u32' data can not be assigned to signed 32-bit integer. However,
assigning to u64 is very wasteful, after all, the buffer capacity of
The 'u32' and 'u64' data can not assigned to 'int' type variable. The 'u32'
data need to use the 'u32' telemetry API to add, and the 'u64' data need to
use the 'u64' telemetry API to add.
Fixes: 2f5c4025abb3 ("mempool: add telemetry endpoint")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
---
l
Currently, 32-bit integer value only is signed in telemetry. The u32 data
can not be assigned to signed 32-bit integer. However, assigning to u64 is
very wasteful, after all, the buffer capacity of each transfer is limited.
So it is necessary for 'u32' data to add usigned 32-bit integer and a
serie
The 'dev_flags', 'rx_offloads', 'tx_offloads' and 'rss_hf' are suitable
for hexadecimal display.
Like:
-->old display by input /ethdev/info,0
"dev_flags": 3,
"rx_offloads": 524288,
"tx_offloads": 65536,
"ethdev_rss_hf": 9100
--> now display
"dev_flags": "0x3",
The 'u32' and 'u64' data can not assigned to 'int' type variable. The 'u32'
data need to use the 'u32' telemetry API to add, and the 'u64' data need to
use the 'u64' telemetry API to add.
Fixes: 58b43c1ddfd1 ("ethdev: add telemetry endpoint for device info")
Cc: sta...@dpdk.org
Signed-off-by: Hui
The 'u32' data can not assigned to 'int' type variable. The 'u32' data need
to use the 'u32' telemetry API to add.
Fixes: d3d98f5ce9d0 ("cryptodev: support telemetry")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
---
lib/cryptodev/rte_cryptodev.c | 2 +-
1 file changed, 1 insertion(+), 1 delet
The 'u32' and 'u64' data can not assigned to 'int' type variable. The 'u32'
data need to use the 'u32' telemetry API to add, and the 'u64' data need to
use the 'u64' telemetry API to add.
Fixes: e6732d0d6e26 ("mem: add telemetry infos")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
---
lib/eal/
Add test cases for u32 telemetry data API.
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
---
app/test/test_telemetry_data.c | 86 +-
app/test/test_telemetry_json.c | 23 -
2 files changed, 105 insertions(+), 4 deletions(-)
diff --git a/app/test/test_tele
> -Original Message-
> From: You, KaisenX
> Sent: Thursday, November 17, 2022 2:57 PM
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Yang, Qiming ; Zhou, YidingX
> ; You, KaisenX ; Wu, Jingjing
> ; Xing, Beilei ; Zhang, Qi Z
>
> Subject: [PATCH] net/iavf:fix slow memory allocation
>
> In s
https://bugs.dpdk.org/show_bug.cgi?id=1064
Bruce Richardson (bruce.richard...@intel.com) changed:
What|Removed |Added
Status|UNCONFIRMED |RESOLVED
Hi Akhil,
On 12/5/22 08:59, Akhil Goyal wrote:
Baseband patches are now split from dpdk-next-crypto,
All baseband patches need to be submitted to dpdk-next-baseband
which will be maintained by Maxime.
Cc: maxime.coque...@redhat.com
Signed-off-by: Akhil Goyal
---
MAINTAINERS | 12 ---
+To: Bruce and Stephen might also have opinions on this.
> From: Huisong Li [mailto:lihuis...@huawei.com]
> Sent: Thursday, 8 December 2022 09.06
>
> The 'dev_flags', 'rx_offloads', 'tx_offloads' and 'rss_hf' are suitable
> for hexadecimal display.
>
> Like:
> -->old display by input /ethdev/in
The offset of format_select_dw_8_6_ext was incorrect.
Update the reserved offsets to correct value.
Fixes: 365cdf5f8ce7 ("net/mlx5/hws: add command layer")
Signed-off-by: Alex Vesker
Reviewed-by: Erez Shitrit
---
drivers/common/mlx5/mlx5_prm.h | 7 ---
1 file changed, 4 insertions(+), 3 del
On 06/12/2022 15:02, Maxime Coquelin wrote:
This patch changes VHOST_USER_SET_VRING_ERR "not
implemented" log level from INFO to DEBUG.
Indeed, it is not required to support this message type,
and since version 7.1, QEMU sends this message for every
queue. It could confuse the end-user, thinking
With recent versions of rdma-core, it becomes important to install,
so some non-standard compiler tricks are avoided.
While at it, disabling unneeded pyVerbs and man pages.
Signed-off-by: Thomas Monjalon
---
doc/guides/nics/mlx4.rst | 3 ++-
doc/guides/platform/mlx5.rst | 3 ++-
2 files cha
29/11/2022 22:33, Tyler Retzlaff:
> On Tue, Nov 29, 2022 at 02:59:00PM +0100, David Marchand wrote:
> > When passing a patch downloaded as mbox, the mail headers were being
> > scanned which resulted in mailing list and other unrelated mail
> > addresses being checked against the contributors list.
On Thu, Dec 08, 2022 at 11:55:16AM +0100, Morten Brørup wrote:
> +To: Bruce and Stephen might also have opinions on this.
>
> > From: Huisong Li [mailto:lihuis...@huawei.com]
> > Sent: Thursday, 8 December 2022 09.06
> >
> > The 'dev_flags', 'rx_offloads', 'tx_offloads' and 'rss_hf' are suitable
08/12/2022 10:46, Maxime Coquelin:
> On 12/5/22 08:59, Akhil Goyal wrote:
> > Baseband patches are now split from dpdk-next-crypto,
> > All baseband patches need to be submitted to dpdk-next-baseband
> > which will be maintained by Maxime.
> >
> > Cc: maxime.coque...@redhat.com
> >
> > Signed-off
06/12/2022 11:27, David Marchand:
> On Tue, Dec 6, 2022 at 11:17 AM Thomas Monjalon wrote:
> >
> > The command "meson build" causes a deprecation warning with meson 0.64:
> >
> > WARNING: Running the setup command as `meson [options]` instead of
> > `meson setup [options]` is ambiguous and
On Thu, Dec 08, 2022 at 12:16:46PM +0100, Thomas Monjalon wrote:
> With recent versions of rdma-core, it becomes important to install,
> so some non-standard compiler tricks are avoided.
>
I think this paragraph could do with a bit more explanation in it, e.g. to
clarify that this is (presumably)
It has been observed that mbufs of some received VLAN packets had the
VLAN tag correctly set in vlan_tci, but ol_flags were missing the
VLAN-indicating flags.
_mm256_shuffle_epi8 operates as two independent 128-bit operations,
not as a single 256-bit operation. To have the RTE_MBUF_F_RX_VLAN* flag
Hi Abhishek,
Please see comments inline.
BRs,
Xiao
> -Original Message-
> From: Maheshwari, Abhishek
> Sent: Tuesday, December 6, 2022 8:55 PM
> To: Wang, Xiao W
> Cc: dev@dpdk.org; sta...@dpdk.org; Xia, Chenbo ;
> Mandal, Purna Chandra ; Maheshwari,
> Abhishek
> Subject: [PATCH v2] v
I'm sorry Patrick that it gives you more work.
Your proposals below don't look possible because a tag is something fixed
forever.
We had an ABI issue in the initial tag so we cannot use the tag v22.11 as
planned.
I don't see how we can better plan except having more tests on release
candidates.
Release status meeting minutes 2022-12-08
=
Agenda:
* Release Dates
* Subtrees
* Roadmaps
* LTS
* Defects
* Opens
Participants:
* ARM
* Debian/Microsoft
* Intel
* Marvell
* Nvidia
* Red Hat
* AMD
Release Dates
-
The following are the proposed
https://bugs.dpdk.org/show_bug.cgi?id=1144
Bug ID: 1144
Summary: virtio may lost pkts when pmd rxq less than
virtio-max_queue_pairs
Product: DPDK
Version: 22.11
Hardware: All
OS: All
Status: UNCO
On 12/7/2022 5:48 PM, Junxiao Shi wrote:
> Hi Ferruh
>
>> On 12/7/2022 2:41 PM, Junxiao Shi wrote:
>> > This allows a DPDK application running with root privilege to create a
>> > memif socket listener with non-root owner uid and gid, which can be
>> > connected from client applications running wi
On 11/17/2022 6:57 AM, Kaisen You wrote:
> In some cases, the DPDK does not allocate hugepage heap memory to
> some sockets due to the user setting parameters
> (e.g. -l 40-79, SOCKET 0 has no memory).
> When the interrupt thread runs on the corresponding core of this
> socket, each allocation/rele
Thomas - thanks for the response. We will proceed with making the necessary
changes for using v22.11.1.
On Thu, Dec 8, 2022 at 8:22 AM Thomas Monjalon wrote:
> I'm sorry Patrick that it gives you more work.
> Your proposals below don't look possible because a tag is something fixed
> forever.
>
On 12/7/2022 3:53 PM, Junxiao Shi wrote:
> This allows a DPDK application running with root privilege to create a
> memif socket listener with non-root owner uid and gid, which can be
> connected from client applications running without root privilege.
>
> Signed-off-by: Junxiao Shi
(moving from
On 12/7/2022 3:53 PM, Junxiao Shi wrote:
> This allows a DPDK application running with root privilege to create a
> memif socket listener with non-root owner uid and gid, which can be
> connected from client applications running without root privilege.
>
> Signed-off-by: Junxiao Shi
(moving from
On Wed, Dec 07, 2022 at 08:07:16PM -0800, Stephen Hemminger wrote:
> On Wed, 7 Dec 2022 14:33:31 -0800
> Tyler Retzlaff wrote:
>
> > On Wed, Dec 07, 2022 at 01:03:41PM -0800, Stephen Hemminger wrote:
> > > On Wed, 7 Dec 2022 11:00:13 -0800
> > > Tyler Retzlaff wrote:
> > >
> > > > +static ch
On 12/8/2022 3:31 AM, Chaoyong He wrote:
> From: Peng Zhang
>
> In the description of nfp document, DPDK PMD supports that the PF
> looks for a firmware file by the PCI name.
>
> After this commit, it will looks for the PCI name format such as
> pci-:04:00.0.nffw rather than pci-04:00.0.nffw
> By default 'test-meson-builds.sh' script clones the repository which the
> script is in, and selects a configured branch ('DPDK_ABI_REF_VERSION')
> as a reference for ABI check.
>
> This patch enables selecting different repository to close for reference
> using 'DPDK_ABI_REF_SRC' environment va
Hi Morten,
we have not looked at that.
Jochen
From: Morten Brørup
Date: Wednesday, December 7, 2022 at 12:28 AM
To: Jochen Behrens
Cc: dev@dpdk.org
Subject: VMware vmxnet3 PMD roadmap question
!! External Email
VMware vmxnet3 PMD maintainer,
Are there any plans for VMware t
Implemented framework for mldev test application. New test cases
can be added using the framework. Support is also enabled to add
options specific to the test cases. User can launch the tests by
specifying the name of test as part of launch arguments.
Code to parse command line arguments is import
Added test case to validate device handling operations. Device ops
test is a collection of multiple sub-tests. Enabled sub-test to
validate device reconfiguration. Set device_ops as the default test.
Signed-off-by: Srikanth Yalavarthi
---
app/test-mldev/meson.build | 1 +
app/test-mldev/
Added common functions used by all tests. Common code
includes functions to check capabilities, options, and
handle ML devices.
Signed-off-by: Srikanth Yalavarthi
---
app/test-mldev/meson.build | 1 +
app/test-mldev/test_common.c | 139 +++
app/test-mldev/test
Added test case to validate model operations. Model ops test
is a collection of sub-tests. Each sub-test invokes the model
operations in a specific order.
Sub-test A: (load -> start -> stop -> unload) x n
Sub-test B: load x n -> start x n -> stop x n -> unload x n
Sub-test C: load x n + (start +
Added an ordered test case to execute inferences with single
or multiple models. In this test case inference requests for
a model are enqueued after completion of all requests for
the previous model. Test supports inference repetitions.
Operations sequence when testing with N models and R reps,
(
Added test case to interleave inference requests from multiple
models. Interleaving would load and start all models and launch
inference requests for the models using available queue-pairs
Operations sequence when testing with N models and R reps,
(load + start) x N -> (enqueue + dequeue) x N x R
Added 'burst_size' support for inference tests. Burst size
controls the number of inference requests handled during
the burst enqueue and dequeue operations of the test case.
Signed-off-by: Srikanth Yalavarthi
---
app/test-mldev/ml_options.c| 26 ++--
app/test-mldev/ml_options.h
Enabled support to validate inference output with reference
output provided by the user. Validation would be successful
only when the inference outputs are within the 'tolerance'
specified through command line option "--tolerance".
Signed-off-by: Srikanth Yalavarthi
---
app/test-mldev/meson.buil
Added support to create multiple queue-pairs per device to
enqueue and dequeue inference requests. Number of queue pairs
to be created can be specified through "--queue_pairs" option.
Support is also enabled to control the number of descriptors
per each queue pair through "--queue_size" option. Inf
Enable reporting driver xstats and inference end-to-end
latency and throughput in mldev inference tests. Reporting
of stats can be enabled using "--stats" option.
Signed-off-by: Srikanth Yalavarthi
---
v3:
* Fixed issue in reporting end-to-end inference stats
v2:
* Fixed typos and formatting iss
Enabled support to execute multiple batches of inferences
per each enqueue request. Input and reference for the test
should be appropriately provided for multi-batch run. Number
of batches can be specified through "--batches" option.
Signed-off-by: Srikanth Yalavarthi
---
app/test-mldev/ml_optio
Machine Learning common code
This patch series implements the common ML code that can be used by
ML drivers. Common code include functions to convert ML IO type to
string, IO format type to string, function get size of ML IO type,
and functions for converting data type
Added initial files for common ML driver code. Implemented ML
type to size conversion, type to string and format to string
conversion utility functions.
Signed-off-by: Srikanth Yalavarthi
---
Depends-on: series-26046 ("app/mldev: implement test framework for mldev")
MAINTAINERS
Type conversion routines transform data from higher to lower
precision data types or vice-versa. These conversion functions
can be used by the ML driver implementations for quantization
and de-quantization.
Added driver routines for type conversion. These driver routines
invoke the architecture sp
Added generic implementations to support conversion of data types.
Support is enabled to handle int8, uint8, int16, uint16, float16,
float32 and bfloat16 types.
Signed-off-by: Srikanth Yalavarthi
---
drivers/common/ml/meson.build| 2 +
drivers/common/ml/ml_utils.c | 86 +---
d
Added ARM NEON intrinsic based implementations to support conversion
of data types. Support is enabled to handle int8, uint8, int16, uint16,
float16, float32 and bfloat16 types.
Signed-off-by: Srikanth Yalavarthi
---
drivers/common/ml/meson.build | 5 +
drivers/common/ml/ml_utils.c |
08/12/2022 19:14, Akhil Goyal:
> > By default 'test-meson-builds.sh' script clones the repository which the
> > script is in, and selects a configured branch ('DPDK_ABI_REF_VERSION')
> > as a reference for ABI check.
> >
> > This patch enables selecting different repository to close for reference
Added ML cnxk headers for register, structure definitions and
ROC layer. Implemented ROC functions, registered logtype for
ML module with the name pmd.ml.cnxk and defined ML hardware ID.
Signed-off-by: Srikanth Yalavarthi
---
Depends-on: series-26047 ("implementation of ML common code")
MAINTAI
Marvell ML CNXK Driver
--
This patch series implements Machine Learning (ML) driver for Marvell
Octeon 10 (cnxk) platform. ML inferencing is supported on cnxk platform
through an integrated ML inferencing processor. The current driver
supports programming the ML hardware engine
ML inference engine on cn10k platform is a PCI based device. Added
driver support to probe and remove the device for cn10k poll mode
driver. The device is named by the PMD as "ml_cn10k".
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 114 +
Added initial source files and build files for ML cnxk driver.
Signed-off-by: Srikanth Yalavarthi
---
Depends-on: patch-120600 ("common/cnxk: add ML headers and ROC code for cnxk")
MAINTAINERS| 2 ++
drivers/meson.build| 1 +
drivers/ml/cnxk/cn10k_ml_dev.c | 8
Added support to get the cn10k ML device information. This is a
driver implementation for the RTE function rte_ml_dev_info_get.
ML device on cn10k supports one queue-pair in lock-free mode and
does not support segmented input output data.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn
Enabled device initialization and firmware load on simulator
platform. Firmware load stage on simulator would involve
launching a firmware handshake request only.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 119 +
1 file changed, 107 in
Implemented driver functions to configure and close ML devices.
Added skeleton code and support to reconfigure ML device. PCI
device remove is enabled in device close.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 2 ++
drivers/ml/cnxk/cn10k_ml_dev.h | 21
Enabled parsing ML firmware path for cn10k. Default path is set
as "/lib/firmware/mlip-fw.bin", when args are not provided. Added
internal structures for ML firmware.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 71 ++
drivers/ml/cnxk/cn
Added cnxk driver implementations to load and unload ML models.
Enabled support in configure stage to allocate model handles
array. Assign model ID and allocate resources per each model
during load stage and release resources during model unload.
Added internal structures to handle ML models.
Sign
Added internal structures to handle derived address fields
and enabled support to compute DMA addresses for model start.
Enabled updating internal model fields.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_model.c | 88
drivers/ml/cnxk/cn10k_ml
Enabled support to load ML firmware on cn10ka ROC model. Reset
MLIP device during dev_close driver operation. Device can't be
reconfigured after a call to close. Job execution is disabled
after firmware load, execution is enabled in device start state.
Added internal request structure.
Signed-off-
Implemented model start driver function. A model start job
is checked for completion in synchronous mode. Tilemask and
OCM slot is calculated before starting the model. Model start
is enqueued through scratch registers. OCM pages are reserved
after model start completion.
Signed-off-by: Srikanth
Implemented ML driver functions to start and stop ML device.
Start / Stop would enable or disable ML device to accept
inference requests.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 42 +++---
1 file changed, 39 insertions(+), 3 deletions(-
Implemented model stop driver function. A model stop job is
enqueued through scratch registers and is checked for
completion through polling in a synchronous mode. OCM pages
are released after model stop completion.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 115
Enabled support to create and destroy device queue-pairs. Updated
configure stage to create array to store queue-pair handles. Added
internal structure for queue-pair, queue and ML inference requests.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 207 +++
Added driver functions to get model information. Added
internal functions to set and get model info.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_model.c | 54
drivers/ml/cnxk/cn10k_ml_model.h | 9 ++
drivers/ml/cnxk/cn10k_ml_ops.c | 37
Added model metadata structure and enabled metadata check
during model load. Remap cnxk IO types with RTE IO types.
Store and update model metadata in model structure.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_model.c | 196 +++
drivers/ml/cnxk/cn10k_ml_mode
Added cnxk driver functions to update model params or weights
and bias after a models is loaded. Updating model params would
not require reloading the model.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 31 +++
1 file changed, 31 insertions(
Added internal structures to handle tile and OCM information and
OCM to model memory mapping. Initialize the fields to platform
specific defaults and compute the OCM / tile requirements for model.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 5 ++
drivers/ml/cnxk/cn
Added JD structures for load, unload and run jobs. Initialize
job command and allocate memory for request structures for slow
path jobs.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 99
drivers/ml/cnxk/cn10k_ml_model.h | 4 ++
driver
Added driver functions to get input and output buffer sizes
for a given batch size. This function would compute the buffer
size based on specific requirements of the device.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 52 ++
1 file chan
Added support to compute OCM tilemask and page start for a
model. The computed tilemask and page start are used during
model start to copy model weights and bias to OCM. OCM slot
for a model is allocated from the tiles with maximum amount
of free memory.
Signed-off-by: Srikanth Yalavarthi
---
dr
Added support to reserve and free OCM pages for a model. OCM
pages are reserved upon completion of model start and are
released after model stop.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ocm.c | 131 -
drivers/ml/cnxk/cn10k_ml_ocm.h | 3 +
Added support to dump device debug information. Debug info on
cn10k device includes model state info, OCM usage info, firmware
debug and exception buffer.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ocm.c | 51 +
drivers/ml/cnxk/cn10k_ml_ocm.h | 1 +
drivers/ml/cnx
Added support for device selftest. Device selftest includes
checking the status of firmware.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 57 ++
1 file changed, 57 insertions(+)
diff --git a/drivers/ml/cnxk/cn10k_ml_ops.c b/drivers/ml/c
Enabled driver support to dequeue inference requests from
internal queue. Dequeue checks for request completion by
polling the status field of the job request.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 61 ++
drivers/ml/cnxk/cn10k_ml_
Enabled support to get and reset ML device stats. Device stats
include number of requests enqueued/dequeued and error count.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 55 --
drivers/ml/cnxk/cn10k_ml_ops.h | 3 ++
2 files changed, 55
Implemented driver functions to quantize / dequantize input
and output data. Support is enabled for multiple batches.
Quantization / dequantization use the type conversion functions
defined in ML common code.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 150 +++
Enabled support to retrieve xstats in either cycles or ns.
Access to sclk is enabled only if an RVU device is probed
during initialization. Driver would return the xstats in
nanoseconds only when an RVU device is probed, else would
fallback to cycles.
Signed-off-by: Srikanth Yalavarthi
---
drive
Enabled driver support to enqueue a burst of inference requests
to ML device. Enqueue uses internal ML request structure to queue
the inferences and job completion through polling.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 96 ++
driv
Enabled support to enable and report DPE warnings from ML
firmware. Configure firmware load flags based on the device
arguments.
Default values:
enable_dpe_errors = 1
report_dpe_errors = 0
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 94 +++
Added internal function to execute ML inference requests
in synchronous mode. Sync mode inference execution is used
to launch inference requests without using a queue-pair.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 53 ++
drivers/ml/c
Added device argument "ocm_alloc_mode" to select OCM allocation
method during model start. Two modes are supported by the driver.
Added implementation for ocm_alloc_mode lowest as default.
ocm_alloc_mode:
lowest: Allocate from first available free slot / lowest
tile ID in OCM (default)
Enabled support for error handling. Added error types and subtypes
supported by ML firmware. Enabled support to get device specific
error code and message for a completed ML request.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 4 +-
drivers/ml/cnxk/cn10k_ml_dev.h |
Added support to handle ML device extended stats. Support
is enabled to get xstats names and stats values and reset
xstats. Supported xstats include avg, min and max hardware
and firmware latency.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.h | 3 +
drivers/ml/cnxk/cn
Added device argument 'cache_model_data' to enable model data
caching. An inference request would be executed with dummy data
in synchronous mode during model start stage. This run would
cache the model weights and bias in the memory and result in
improved inference throughput.
cache_model_data =
Added device argument "hw_queue_lock" to select the JCMDQ enqueue
ROC function to be used in fast path.
hw_queue_lock:
0: Disable, use lock free version of JCMDQ enqueue ROC function for
job queuing. To avoid race condition in request queuing to
hardware, disabling hw_queue_lock
Added device argument "poll_mem" to select the memory
region to be used for polling in fast-path requests.
Implemented support to use scratch registers for polling.
Available pool of scratch registers one-to-one mapped with
the internal request queue.
poll_mem:
ddr: Use DDR memory location f
Added user guide for Marvell cnxk ML driver for Marvell Octeon
cnxk Soc family. Added details about device initialization,
debug options and runtime device args supported by the driver.
Signed-off-by: Srikanth Yalavarthi
---
MAINTAINERS | 1 +
doc/guides/index.rst| 1
Marvell ML CNXK Driver
--
This patch series implements Machine Learning (ML) driver for Marvell
Octeon 10 (cnxk) platform. ML inferencing is supported on cnxk platform
through an integrated ML inferencing processor. The current driver
supports programming the ML hardware engine
Added initial source files and build files for ML cnxk driver.
Signed-off-by: Srikanth Yalavarthi
---
Depends-on: patch-120600 ("common/cnxk: add ML headers and ROC code for cnxk")
MAINTAINERS| 2 ++
drivers/meson.build| 1 +
drivers/ml/cnxk/cn10k_ml_dev.c | 8
Added support to get the cn10k ML device information. This is a
driver implementation for the RTE function rte_ml_dev_info_get.
ML device on cn10k supports one queue-pair in lock-free mode and
does not support segmented input output data.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn
ML inference engine on cn10k platform is a PCI based device. Added
driver support to probe and remove the device for cn10k poll mode
driver. The device is named by the PMD as "ml_cn10k".
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 114 +
Implemented driver functions to configure and close ML devices.
Added skeleton code and support to reconfigure ML device. PCI
device remove is enabled in device close.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 2 ++
drivers/ml/cnxk/cn10k_ml_dev.h | 21
Enabled support to load ML firmware on cn10ka ROC model. Reset
MLIP device during dev_close driver operation. Device can't be
reconfigured after a call to close. Job execution is disabled
after firmware load, execution is enabled in device start state.
Added internal request structure.
Signed-off-
Enabled device initialization and firmware load on simulator
platform. Firmware load stage on simulator would involve
launching a firmware handshake request only.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 119 +
1 file changed, 107 in
Enabled parsing ML firmware path for cn10k. Default path is set
as "/lib/firmware/mlip-fw.bin", when args are not provided. Added
internal structures for ML firmware.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_dev.c | 71 ++
drivers/ml/cnxk/cn
Implemented ML driver functions to start and stop ML device.
Start / Stop would enable or disable ML device to accept
inference requests.
Signed-off-by: Srikanth Yalavarthi
---
drivers/ml/cnxk/cn10k_ml_ops.c | 42 +++---
1 file changed, 39 insertions(+), 3 deletions(-
1 - 100 of 147 matches
Mail list logo
