[Issue]: [gfx1151] MIOpen CK grouped conv library missing — XTTS inference 3-5x slower than CPU due to GemmFwdRest fallback

[EvilSquirrels]

Problem Description

On gfx1151 (Strix Halo / Radeon 8060S), the MIOpen CK grouped convolution library and kernel database for gfx1151 are missing from the TheRock nightly build, causing all convolution operations to fall back to the slow GemmFwdRest solver with zero workspace allocation. This makes real-world inference workloads (specifically XTTS v2 text-to-speech) 3-5x slower on GPU than on CPU, defeating the purpose of GPU acceleration.

The following warnings appear on every inference call:

MIOpen(HIP): Warning [ParseAndLoadDb] File is unreadable: ".../site-packages/_rocm_sdk_libraries_gfx1151/share/miopen/db/gfx1151_20.HIP.fdb.txt"
MIOpen(HIP): Warning [OpenRuntimeLibraryForDevice] CK grouped conv library not found for device gfx1151: libMIOpenCKGroupedConv_gfx1151.so: cannot open shared object file: No such file or directory
MIOpen(HIP): Warning [IsEnoughWorkspace] [EvaluateInvokers] Solver <GemmFwdRest>, workspace required: 42172416, provided ptr: 0 size: 0

Attempted workaround: Setting HSA_OVERRIDE_GFX_VERSION=11.0.0 (gfx1100 spoof) eliminates the MIOpen warnings but causes hipErrorInvalidImage on torchaudio resampling kernels used by XTTS, making it non-viable.

Note: The segfault present in rocm6.2 builds is resolved in this nightly — basic GPU compute works correctly. The issue is specific to the MIOpen convolution path.

Operating System

Ubuntu 26.04 LTS (Resolute Raccoon), Linux kernel 7.0

CPU

AMD Ryzen AI Max+ 395 (Strix Halo, 16-core Zen 5)

GPU

AMD Radeon 8060S — gfx1151 (integrated, Strix Halo), 128GB unified memory (shared between CPU and GPU), GTT size unlocked via GRUB: amdgpu.gttsize=122880

ROCm Version

TheRock nightly: rocm7.13.0a20260425, PyTorch: 2.11.0+rocm7.13.0a20260425, Python: 3.11.15

ROCm Component

MIOpen

Steps to Reproduce

1. Install TheRock nightly for gfx1151:

pip install --upgrade --index-url https://rocm.nightlies.amd.com/v2/gfx1151/ torch torchaudio torchvision

2. Install XTTS via Coqui TTS:

pip install TTS torchaudio

3. Run inference:

import torch
from TTS.api import TTS

tts = TTS('tts_models/multilingual/multi-dataset/xtts_v2').to('cuda')

import time
start = time.time()
tts.tts_to_file(
    text='Hello, this is a test of GPU text to speech.',
    speaker_wav='any_reference_voice.wav',
    language='en',
    file_path='/tmp/output.wav'
)
print(f'Time: {time.time()-start:.2f}s')

4. Observe MIOpen warnings and ~15-25 second generation time vs ~5-10 seconds on CPU

Simpler reproduction (confirms missing library without TTS dependency):

import torch
x = torch.rand(100, 100).cuda()
y = x @ x
print('Basic compute OK:', y.shape)
# Then run any torchaudio resampling on GPU:
import torchaudio
audio = torch.rand(1, 22050).cuda()
resampled = torchaudio.functional.resample(audio, 22050, 16000)
# This produces hipErrorInvalidImage when HSA_OVERRIDE_GFX_VERSION=11.0.0

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

ROCk module is loaded

HSA System Attributes

Runtime Version: 1.18
Runtime Ext Version: 1.15
System Timestamp Freq.: 1000.000000MHz
Sig. Max Wait Duration: 18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model: LARGE
System Endianness: LITTLE
Mwaitx: DISABLED
XNACK enabled: NO
DMAbuf Support: YES
VMM Support: YES

==========
HSA Agents

---

Agent 1

---

Name: AMD RYZEN AI MAX+ 395 w/ Radeon 8060S
Uuid: CPU-XX
Marketing Name: AMD RYZEN AI MAX+ 395 w/ Radeon 8060S
Vendor Name: CPU
Feature: None specified
Profile: FULL_PROFILE
Float Round Mode: NEAR
Max Queue Number: 0(0x0)
Queue Min Size: 0(0x0)
Queue Max Size: 0(0x0)
Queue Type: MULTI
Node: 0
Device Type: CPU
Cache Info:
L1: 49152(0xc000) KB
Chip ID: 0(0x0)
ASIC Revision: 0(0x0)
Cacheline Size: 64(0x40)
Max Clock Freq. (MHz): 5187
BDFID: 0
Internal Node ID: 0
Compute Unit: 32
SIMDs per CU: 0
Shader Engines: 0
Shader Arrs. per Eng.: 0
WatchPts on Addr. Ranges:1
Memory Properties:
Features: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: FINE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 3
Segment: GLOBAL; FLAGS: KERNARG, FINE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 4
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
ISA Info:

---

Agent 2

---

Name: gfx1151
Uuid: GPU-XX
Marketing Name: AMD Radeon Graphics
Vendor Name: AMD
Feature: KERNEL_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 128(0x80)
Queue Min Size: 64(0x40)
Queue Max Size: 131072(0x20000)
Queue Type: MULTI
Node: 1
Device Type: GPU
Cache Info:
L1: 32(0x20) KB
L2: 2048(0x800) KB
L3: 32768(0x8000) KB
Chip ID: 5510(0x1586)
ASIC Revision: 0(0x0)
Cacheline Size: 128(0x80)
Max Clock Freq. (MHz): 2900
BDFID: 50432
Internal Node ID: 1
Compute Unit: 40
SIMDs per CU: 2
Shader Engines: 2
Shader Arrs. per Eng.: 2
WatchPts on Addr. Ranges:4
Coherent Host Access: FALSE
Memory Properties: APU
Features: KERNEL_DISPATCH
Fast F16 Operation: TRUE
Wavefront Size: 32(0x20)
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Max Waves Per CU: 32(0x20)
Max Work-item Per CU: 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 2147483647(0x7fffffff)
y 65535(0xffff)
z 65535(0xffff)
Max fbarriers/Workgrp: 32
Packet Processor uCode:: 34
SDMA engine uCode:: 18
IOMMU Support:: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 64065360(0x3d18f50) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 64065360(0x3d18f50) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 3
Segment: GROUP
Size: 64(0x40) KB
Allocatable: FALSE
Alloc Granule: 0KB
Alloc Recommended Granule:0KB
Alloc Alignment: 0KB
Accessible by all: FALSE
ISA Info:
ISA 1
Name: amdgcn-amd-amdhsa--gfx1151
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 2147483647(0x7fffffff)
y 65535(0xffff)
z 65535(0xffff)
FBarrier Max Size: 32
ISA 2
Name: amdgcn-amd-amdhsa--gfx11-generic
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 2147483647(0x7fffffff)
y 65535(0xffff)
z 65535(0xffff)
FBarrier Max Size: 32

---

Agent 3

---

Name: aie2p
Uuid: AIE-XX
Marketing Name: RyzenAI-npu5
Vendor Name: AMD
Feature: AGENT_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 1(0x1)
Queue Min Size: 64(0x40)
Queue Max Size: 64(0x40)
Queue Type: SINGLE
Node: 0
Device Type: DSP
Cache Info:
L2: 2048(0x800) KB
L3: 32768(0x8000) KB
Chip ID: 0(0x0)
ASIC Revision: 0(0x0)
Cacheline Size: 0(0x0)
Max Clock Freq. (MHz): 0
BDFID: 0
Internal Node ID: 0
Compute Unit: 0
SIMDs per CU: 0
Shader Engines: 0
Shader Arrs. per Eng.: 0
WatchPts on Addr. Ranges:0
Memory Properties:
Features: AGENT_DISPATCH
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: KERNARG, COARSE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 65536(0x10000) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:0KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 3
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 128130724(0x7a31ea4) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
ISA Info:
*** Done ***

Additional Information

Environment variables currently in use:

export HSA_OVERRIDE_GFX_VERSION=11.5.1
export HSA_ENABLE_SDMA=0
export MIOPEN_USER_DB_PATH="/home/user/.config/miopen"

rocm-smi shows GPU utilization spiking to 100% during inference, confirming the GPU is being used. SCLK reports N/A (expected behavior for Strix Halo integrated GPU with unified memory architecture — not a power state issue).

The rocm-sdk-device-gfx1151 package from the multi-arch index installs successfully but the missing .so and .fdb.txt files are not included. This is the specific gap that needs to be filled for gfx1151 MIOpen performance to be viable.