Optimization for unused symbols in NR UE
After processing TX frequency domain data skip IFFT call for symbols
that aren't used, the time domain buffer is memset instead.
Biggest benefit is when UE does not transmit PUSCH, or PUSCH is using <
14 symbols
feat(oran_fhlib_5g): O-RAN 7.2 FrontHaul Interface on Arm
This MR is enabling the 7.2 FHI on Arm systems.
Even though the performance are yet very bad (UE connection is very
difficult, 100Mbps DL, almost no UL), still I would like to merge this
now since the matter of making the FHI work is rather orthogonal to the
performance issues.
So merging this feature would provide a reliable settlement on top of
which we could then progressively enhance the porting of OAI to Arm.
Especially that would make our work easier by having less features to
rebase regularly on develop.
The main features this MR is made of are:
1. Make the xran patches dual
We provide patches for both E and F releases of xran that handles x86
targets as well as Arm targets. Both patches have been tested on both
types of machines (armix, broadbill). Thanks to @brobert and @changy
for their work on xran.
2. Perform 9-bits BFP compression with ArmRAL
Since the BFP compression implementation provided with xran does rely
heavily on AVX512, we provide an alternative implementation that is
relying on Arm's RAN Acceleration Library (ArmRAL).
As of writing this commit description, only 9-bits compression is
supported. Other IQ width will be added.
This upcoming change does not prevent review to be performed since the
BFP compression functions of ArmRAL are identical except for the IQ
width. Adding further IQ width is therefore very straightforward.
3. Fixes in oran_fhlib_5g to support Arm
The library oran_fhlib_5g needed few changes to successfully compile and
run on Arm systems. Thanks to @brobert for his work on oran_fhlib_5g.
Remove include of SIMDE Neon header sse_intrinsic.h to compile on ARM. Otherwise the error below is issued.
```
/usr/include/simde/arm/neon/get_lane.h:279:5: error: incompatible type for argument 1 of ‘vget_lane_f16’
279 | SIMDE_CONSTIFY_8_(vget_lane_f16, r, (HEDLEY_UNREACHABLE(), SIMDE_FLOAT16_VALUE(0.0)), lane, v);
removed the inclusion of neon.h
```
remove ul_ch_estimates_time to save memory
remove ul_ch_estimates_time that saves 50MB memory to access in RAM, and
make better quality scope data passing (no race)
Add support for ARM build pipeline
Modify the existing python code to be able to build images where the
image tag can be prepended with a prefix, here arm_. This is preparatory
work to create the ARM build image pipeline, and reuse the existing
internal registry on porcepix to have x86 and ARM images coexist.
Fix various bugs in the build system to allow to build on a system with
as many cores as gracehopper.
Also, fix two programs (usim, nvram) to work correctly under ARM.
Harmonize and update Frequency Range computation to the current values
specified by the standard
3GPP TS 38.101-1 Version 19.0.0 Table 5.1-1: Definition of frequency
ranges
- FR1 from 410 MHz to 7125 MHz
- FR2 from 24.25 GHz to 71 GHz
Various libraries in this commit use the log headers (that is the
direct dependency). The log headers in turn include the T headers
(transitive dependency). Building on gracehopper with its many cores
triggers the build of these libraries before T headers have been
generated.
To resolve this and force the generation of T headers before building
the libraries, correctly express their dependency on the log headers,
which in turn ensures that T headers have been generated.
Depending on log_headers is preferrable because it is really the headers
that we depend on. Using instead e.g., utils, would trigger build of
source files, that we don't need for these libraries. This commit is
part of a larger cleanup that I have that is not merged yet. Follow-up
work will generalize this in the repo, and ensure that every(!) library
builds.
We have to add CONFIG_LIB, as the log module has a direct dependency
onto the configuration module (it uses it). I am not sure why this was
not relevant before.
fix(nrLDPC_decoder): fix numIter increment
The logs of ./nr_ulsim -n100 -m19 -s12 -S12 -z2 -L4 were showing that sometimes
the decoding of one segment was failing while the CRC check succeeded. This was
happening because the iteration counter numIter was not properly managed so that
when the last iteration was successful to decode the segment still numIter at
the exit was equal to numMaxIter plus 1 so that decoding was considered to have
failed. In other words the last iteration if it occured was always wasted.
This commit offers to solve this problem by simply moving the iteration counter
increment at the end of the iteration loop while remaining careful at fixing the
tests relying on this counter.
Post-Scriptum:
After a first review, we decided to make the counter more standard from a C code
perspective. The counter now counts from 0 to numMaxIter minus 1. If it reaches
numMaxIter, it means that decoding was unsuccessful.
The idea is to make LDPC iteration accounting more standard from a C coding perspective by making two changes:
1. Now the iteration counter starts at 0 for the first iteration and goes up to max_ldpc_iterations - 1 for the last iteration.
If it is equal to max_ldpc_iterations, it means that decoding failed.
2. A chain of useless returns in the UE DLSCH decoding call chain was removed.
The logs of `./nr_ulsim -n100 -m19 -s12 -S12 -z2 -L4` were showing that sometimes the decoding of one segment was failing while the CRC check succeeded.
This was happening because the iteration counter `numIter` was not properly managed so that when the last iteration was successful to decode the segment
still `numIter` at the exit was equal to `numMaxIter` plus 1 so that decoding was considered to have failed.
In other words the last iteration if it occured was always wasted.
This commit offers to solve this problem by simply moving the iteration counter increment at the end of the iteration loop
while remaining careful at fixing the tests relying on this counter.
Use memcpy in nr_prach.c to fill txdata buffer.
Remove global tables related to PRACH generation.
Add new function to compute modulo multiplicative inverse and use it
directly to get inverse instead of LUT.
Removed nr_ru table and compute the values in-place.
In UE, added new flag root_seq_computed in nrUE_config to only compute
the sequence once after new nrUE_config is received.
Long PRACH extends beyond a slot. Changes necessary to mark future slots
that may contain prach in ru, phy and mac functions.
In RU, the PRACH IDFT is done only after receiving last slot and marks
for L1 to start processing. The L1 processes at the last prach slot and
reports to MAC with the first slot index.
In MAC scheduler, the VRB for future slots are also marked as used.
PRACH scheduler is called in advance one frame minus num prach slots.
fix(nrLDPC_coding): timers in DL encoding, fix(ldpctest): Confusion in sizes
Main improvements:
1. fix(ldpctest): Confusion in sizes
IMPORTANT: This fix makes that ldpctest now works with BG2 and K' != K!
(Worth double checking with more cases than I did check with though)
The variable block_length had an ambiguous role and was used as both K and
Kprime. This was leading to arrays with wrong sizes and functions with wrong
arguments. Now there is not anymore block_length but K and Kprime.
Kprime can be set with the command line argument -l.
There was also some problems with array allocation for segments with length
not divisible by 8. A proper sizing with ((size_in_bits + 7) & ~7) / 8 was
used.
A proper masking of the last byte containing payload bits was also added to
keep filler bits to 0 after random initialization.
2. fix(nrLDPC_coding): timers in DL encoding
While timers were available for rate matching and interleaving in the new
interface, these timers were not merged to rate matching and interleaving
timers in nr_dlsch_coding.
francescomani