github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9999] [ENHANCE] Add Ubuntu Support for AMD Ryzen AI 9 HX 370 w/ Radeon 890M (gfx1150) #6554

New Issue
Closed
opened 2026-04-12 18:10:35 -05:00 by GiteaMirror · 13 comments
GiteaMirror commented 2026-04-12 18:10:35 -05:00
Owner
Copy Link

Originally created by @liuchuan01 on GitHub (Mar 26, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9999

Description

I'm testing a new AMD Ryzen AI 9 HX 370 APU with integrated Radeon 890M graphics (LLVM target gfx1150) on Ubuntu 24.04. This GPU supports dynamic memory allocation (1-16GB), making it promising for local LLM inference. However, Ollama currently lacks native support for this architecture.

Environment

OS: Ubuntu 24.04 LTS
ROCm: 6.10.5
Ollama: 0.6.2 (official .deb)
GPU: Radeon 890M (gfx1150, 16GB dynamic VRAM)
Memory: 64GB DDR5

Verify GPU detection via rocminfo:

ROCk module version 6.10.5 is loaded
=====================
HSA System Attributes
=====================
Runtime Version:         1.14
Runtime Ext Version:     1.6
System Timestamp Freq.:  1000.000000MHz
Sig. Max Wait Duration:  18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model:           LARGE
System Endianness:       LITTLE
Mwaitx:                  DISABLED
DMAbuf Support:          YES

==========
HSA Agents
==========
*******
Agent 1
*******
  Name:                    AMD Ryzen AI 9 HX 370 w/ Radeon 890M
  Uuid:                    CPU-XX
  Marketing Name:          AMD Ryzen AI 9 HX 370 w/ Radeon 890M
  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):   4367
  BDFID:                   0
  Internal Node ID:        0
  Compute Unit:            24
  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:                    48928408(0x2ea9698) 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:                    48928408(0x2ea9698) 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:                    48928408(0x2ea9698) 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:                    48928408(0x2ea9698) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:4KB
      Alloc Alignment:         4KB
      Accessible by all:       TRUE
  ISA Info:
*******
Agent 2
*******
  Name:                    gfx1150
  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
  Chip ID:                 5390(0x150e)
  ASIC Revision:           4(0x4)
  Cacheline Size:          64(0x40)
  Max Clock Freq. (MHz):   2900
  BDFID:                   25856
  Internal Node ID:        1
  Compute Unit:            16
  SIMDs per CU:            2
  Shader Engines:          1
  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                        4294967295(0xffffffff)
    y                        4294967295(0xffffffff)
    z                        4294967295(0xffffffff)
  Max fbarriers/Workgrp:   32
  Packet Processor uCode:: 25
  SDMA engine uCode::      11
  IOMMU Support::          None
  Pool Info:
    Pool 1
      Segment:                 GLOBAL; FLAGS: COARSE GRAINED
      Size:                    24464204(0x1754b4c) 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:                    24464204(0x1754b4c) 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--gfx1150
      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                        4294967295(0xffffffff)
        y                        4294967295(0xffffffff)
        z                        4294967295(0xffffffff)
      FBarrier Max Size:       32
*** Done ***

Expected Behavior

The Radeon 890M iGPU (APU) is detected as gfx1150 via rocminfo,If I start it directly，this happen,then ollama run with CPU

(supported types:[gfx1010 gfx1012 gfx1030 gfx1100 gfx1101 gfx1102 gfx1151 gfx1200 gfx1201 gfx900 gfx906 gfx908 gfx90a gfx942])”

By forcing HSA_OVERRIDE_GFX_VERSION=11.5.1 enables GPU detection but results in 6.42 tokens/s inference speed - 2.5x slower than CPU-only mode's 15.94 tokens/s.

Actual Results

GPU(version 1151):

total duration:       1m31.251975429s
load duration:        13.21737ms
prompt eval count:    1799 token(s)
prompt eval duration: 10.001967ms
prompt eval rate:     179864.62 tokens/s
eval count:           586 token(s)
eval duration:        1m31.228035288s
eval rate:            6.42 tokens/s

CPU:
total duration:       56.576003535s
load duration:        13.20631ms
prompt eval count:    736 token(s)
prompt eval duration: 11.71417167s
prompt eval rate:     62.83 tokens/s
eval count:           715 token(s)
eval duration:        44.847692753s
eval rate:            15.94 tokens/s

Request

Add official gfx1150 support to Ollama's ROCm backend

Originally created by @liuchuan01 on GitHub (Mar 26, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9999 ## Description I'm testing a new AMD Ryzen AI 9 HX 370 APU with integrated Radeon 890M graphics (LLVM target gfx1150) on Ubuntu 24.04. This GPU supports dynamic memory allocation (1-16GB), making it promising for local LLM inference. However, Ollama currently lacks native support for this architecture. ## Environment OS: Ubuntu 24.04 LTS ROCm: 6.10.5 Ollama: 0.6.2 (official .deb) GPU: Radeon 890M (gfx1150, 16GB dynamic VRAM) Memory: 64GB DDR5 Verify GPU detection via rocminfo: ```bash ROCk module version 6.10.5 is loaded ===================== HSA System Attributes ===================== Runtime Version: 1.14 Runtime Ext Version: 1.6 System Timestamp Freq.: 1000.000000MHz Sig. Max Wait Duration: 18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count) Machine Model: LARGE System Endianness: LITTLE Mwaitx: DISABLED DMAbuf Support: YES ========== HSA Agents ========== ******* Agent 1 ******* Name: AMD Ryzen AI 9 HX 370 w/ Radeon 890M Uuid: CPU-XX Marketing Name: AMD Ryzen AI 9 HX 370 w/ Radeon 890M 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): 4367 BDFID: 0 Internal Node ID: 0 Compute Unit: 24 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: 48928408(0x2ea9698) 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: 48928408(0x2ea9698) 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: 48928408(0x2ea9698) 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: 48928408(0x2ea9698) KB Allocatable: TRUE Alloc Granule: 4KB Alloc Recommended Granule:4KB Alloc Alignment: 4KB Accessible by all: TRUE ISA Info: ******* Agent 2 ******* Name: gfx1150 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 Chip ID: 5390(0x150e) ASIC Revision: 4(0x4) Cacheline Size: 64(0x40) Max Clock Freq. (MHz): 2900 BDFID: 25856 Internal Node ID: 1 Compute Unit: 16 SIMDs per CU: 2 Shader Engines: 1 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 4294967295(0xffffffff) y 4294967295(0xffffffff) z 4294967295(0xffffffff) Max fbarriers/Workgrp: 32 Packet Processor uCode:: 25 SDMA engine uCode:: 11 IOMMU Support:: None Pool Info: Pool 1 Segment: GLOBAL; FLAGS: COARSE GRAINED Size: 24464204(0x1754b4c) 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: 24464204(0x1754b4c) 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--gfx1150 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 4294967295(0xffffffff) y 4294967295(0xffffffff) z 4294967295(0xffffffff) FBarrier Max Size: 32 *** Done *** ``` ## Expected Behavior The Radeon 890M iGPU (APU) is detected as gfx1150 via rocminfo,If I start it directly，this happen,then ollama run with CPU ```bash (supported types:[gfx1010 gfx1012 gfx1030 gfx1100 gfx1101 gfx1102 gfx1151 gfx1200 gfx1201 gfx900 gfx906 gfx908 gfx90a gfx942])” ``` By forcing HSA_OVERRIDE_GFX_VERSION=11.5.1 enables GPU detection but results in 6.42 tokens/s inference speed - 2.5x slower than CPU-only mode's 15.94 tokens/s. Actual Results ```bash GPU(version 1151): total duration: 1m31.251975429s load duration: 13.21737ms prompt eval count: 1799 token(s) prompt eval duration: 10.001967ms prompt eval rate: 179864.62 tokens/s eval count: 586 token(s) eval duration: 1m31.228035288s eval rate: 6.42 tokens/s CPU: total duration: 56.576003535s load duration: 13.20631ms prompt eval count: 736 token(s) prompt eval duration: 11.71417167s prompt eval rate: 62.83 tokens/s eval count: 715 token(s) eval duration: 44.847692753s eval rate: 15.94 tokens/s ``` ## Request Add official gfx1150 support to Ollama's ROCm backend
GiteaMirror added the feature requestamd labels 2026-04-12 18:10:35 -05:00
GiteaMirror commented 2026-04-12 18:10:39 -05:00
Author
Owner
Copy Link

@orrinwitt commented on GitHub (Mar 28, 2025):

Adding my request for this as well. Framework's new laptops with these Ryzen AI chips start shipping next month, and I will definitely be wanting to use Ollama with it. (I ordered the one with this same processor and maxxed out 96gb ram. Can't wait! 😁)

<!-- gh-comment-id:2761438254 --> @orrinwitt commented on GitHub (Mar 28, 2025): Adding my request for this as well. Framework's new laptops with these Ryzen AI chips start shipping next month, and I will definitely be wanting to use Ollama with it. (I ordered the one with this same processor and maxxed out 96gb ram. Can't wait! 😁)
GiteaMirror commented 2026-04-12 18:10:39 -05:00
Author
Owner
Copy Link

@abraxaswd commented on GitHub (Mar 31, 2025):

Actually this is what I am looking for too. Too bad AMD does not deliver drivers for their Chipset, which are compatible with Ollama / Docker and ROCm.
Performance is bad when using HSA_OVERRIDE_GFX_VERSION.

<!-- gh-comment-id:2765737153 --> @abraxaswd commented on GitHub (Mar 31, 2025): Actually this is what I am looking for too. Too bad AMD does not deliver drivers for their Chipset, which are compatible with Ollama / Docker and ROCm. Performance is bad when using HSA_OVERRIDE_GFX_VERSION.
GiteaMirror commented 2026-04-12 18:10:40 -05:00
Author
Owner
Copy Link

@DocMAX commented on GitHub (Apr 6, 2025):

Any updates? So still no full GPU/NPU acceleration in ollama?

<!-- gh-comment-id:2781309299 --> @DocMAX commented on GitHub (Apr 6, 2025): Any updates? So still no full GPU/NPU acceleration in ollama?
GiteaMirror commented 2026-04-12 18:10:41 -05:00
Author
Owner
Copy Link

@ghost commented on GitHub (Apr 11, 2025):

There is a recent fork of ollama that supposedly adds this functionality with not too many lines of code.

https://github.com/rjmalagon/ollama-linux-amd-apu/tree/main

<!-- gh-comment-id:2796471372 --> @ghost commented on GitHub (Apr 11, 2025): There is a recent fork of ollama that supposedly adds this functionality with not too many lines of code. https://github.com/rjmalagon/ollama-linux-amd-apu/tree/main
GiteaMirror commented 2026-04-12 18:10:42 -05:00
Author
Owner
Copy Link

@taikai-zz commented on GitHub (May 15, 2025):

https://github.com/ollama/ollama/pull/5426
1df7ac813e (diff-7ee7b1925642f957e6b8274a2abdd37661c060e369669cbe8e6dad6cddf01bc2)
Can solve the problem of 890M GPU

<!-- gh-comment-id:2881941251 --> @taikai-zz commented on GitHub (May 15, 2025): https://github.com/ollama/ollama/pull/5426 https://github.com/ollama/ollama/pull/5426/files/1df7ac813eb12e12885c78585a614425f0a7bcc8#diff-7ee7b1925642f957e6b8274a2abdd37661c060e369669cbe8e6dad6cddf01bc2 Can solve the problem of 890M GPU
GiteaMirror commented 2026-04-12 18:10:43 -05:00
Author
Owner
Copy Link

@ian-cowley commented on GitHub (Jun 23, 2025):

I ran the windows version... this was in the logs

time=2025-06-23T20:21:05.923+01:00 level=WARN source=amd_windows.go:138 msg="amdgpu is not supported (supported types:[gfx1030 gfx1100 gfx1101 gfx1102 gfx1151 gfx906])" gpu_type=gfx1150 gpu=0 library=C:\Users\Ian\AppData\Local\Programs\Ollama\lib\ollama\rocm

CPU: AMD Ryzen AI 9 HX 370 w/ Radeon 890M

The Meaning of the Log Message
Let's break down the crucial line:
msg="amdgpu is not supported (supported types:[... gfx1151 ...])" gpu_type=gfx1150

gpu_type=gfx1150: The Windows driver is identifying your specific GPU architecture as gfx1150.
supported types:[... gfx1151 ...]: The Ollama for Windows software has a hardcoded list of GPU architectures it supports. Crucially, your gfx1150 is not on that list, but a very similar architecture, gfx1151, is.
This is the entire problem. The native Windows version of Ollama knows about one version of the new AMD APUs, but not the specific one in your machine.

What are gfx1150 and gfx1151?
My research confirms that both gfx1150 and gfx1151 are part of AMD's latest "Strix Point" family of processors, which includes the Radeon 800M series graphics.

gfx1151: This corresponds to the Radeon 890M found in processors like the Ryzen AI 9 HX 370. This GPU is officially on Ollama's supported list.
gfx1150: This corresponds to a different, slightly more powerful configuration in the same family, sometimes referred to as "Strix Halo."
Your laptop, an EVO-X1, appears to have the gfx1150 variant. Because your hardware is so new, the Ollama developers have added support for one model from the family but not the other one yet.

PLEASE add support.. :)

<!-- gh-comment-id:2997724685 --> @ian-cowley commented on GitHub (Jun 23, 2025): I ran the windows version... this was in the logs time=2025-06-23T20:21:05.923+01:00 level=WARN source=amd_windows.go:138 msg="amdgpu is not supported (supported types:[gfx1030 gfx1100 gfx1101 gfx1102 gfx1151 gfx906])" gpu_type=gfx1150 gpu=0 library=C:\Users\Ian\AppData\Local\Programs\Ollama\lib\ollama\rocm CPU: AMD Ryzen AI 9 HX 370 w/ Radeon 890M The Meaning of the Log Message Let's break down the crucial line: msg="amdgpu is not supported (supported types:[... gfx1151 ...])" gpu_type=gfx1150 gpu_type=gfx1150: The Windows driver is identifying your specific GPU architecture as gfx1150. supported types:[... gfx1151 ...]: The Ollama for Windows software has a hardcoded list of GPU architectures it supports. Crucially, your gfx1150 is not on that list, but a very similar architecture, gfx1151, is. This is the entire problem. The native Windows version of Ollama knows about one version of the new AMD APUs, but not the specific one in your machine. What are gfx1150 and gfx1151? My research confirms that both gfx1150 and gfx1151 are part of AMD's latest "Strix Point" family of processors, which includes the Radeon 800M series graphics. gfx1151: This corresponds to the Radeon 890M found in processors like the Ryzen AI 9 HX 370. This GPU is officially on Ollama's supported list. gfx1150: This corresponds to a different, slightly more powerful configuration in the same family, sometimes referred to as "Strix Halo." Your laptop, an EVO-X1, appears to have the gfx1150 variant. Because your hardware is so new, the Ollama developers have added support for one model from the family but not the other one yet. PLEASE add support.. :)
GiteaMirror commented 2026-04-12 18:10:44 -05:00
Author
Owner
Copy Link

@jmorille commented on GitHub (Jun 25, 2025):

+1
PLEASE add support for the Ryzen AI 9 HX 370
I am in the same case, with the gfx1150

<!-- gh-comment-id:3006116198 --> @jmorille commented on GitHub (Jun 25, 2025): +1 PLEASE add support for the Ryzen AI 9 HX 370 I am in the same case, with the gfx1150
GiteaMirror commented 2026-04-12 18:10:46 -05:00
Author
Owner
Copy Link

@6d72646f65 commented on GitHub (Jul 13, 2025):

+1
same case, Ryzen AI 9 HX 370 with the gfx1150

<!-- gh-comment-id:3067277266 --> @6d72646f65 commented on GitHub (Jul 13, 2025): +1 same case, Ryzen AI 9 HX 370 with the gfx1150
GiteaMirror commented 2026-04-12 18:10:48 -05:00
Author
Owner
Copy Link

@meetmangukiya commented on GitHub (Jul 15, 2025):

i was able to use the iGPU, but you have to set the version to gfx1151, that was the closest version supported and available for me. check the comment here: https://github.com/ollama/ollama/issues/5186#issuecomment-2994386318

<!-- gh-comment-id:3074481250 --> @meetmangukiya commented on GitHub (Jul 15, 2025): i was able to use the iGPU, but you have to set the version to gfx1151, that was the closest version supported and available for me. check the comment here: https://github.com/ollama/ollama/issues/5186#issuecomment-2994386318
GiteaMirror commented 2026-04-12 18:10:48 -05:00
Author
Owner
Copy Link

@cherio commented on GitHub (Aug 7, 2025):

Using gfx1151 on my Ryzen AI 9 HX 370 box didn't work. iGPU runs unutilized.
Please add support :)

<!-- gh-comment-id:3162044461 --> @cherio commented on GitHub (Aug 7, 2025): Using gfx1151 on my Ryzen AI 9 HX 370 box didn't work. iGPU runs unutilized. Please add support :)
GiteaMirror commented 2026-04-12 18:10:49 -05:00
Author
Owner
Copy Link

@shymega commented on GitHub (Aug 8, 2025):

I think we've established the issue - please to new commenters, can we not post with a '+1 me too', please? A lot of us are subscribed to this issue interested in development updates, or workarounds, and it's getting a bit noisy.

<!-- gh-comment-id:3168429854 --> @shymega commented on GitHub (Aug 8, 2025): I think we've established the issue - please to new commenters, can we not post with a '+1 me too', please? A lot of us are subscribed to this issue interested in development updates, or workarounds, and it's getting a bit noisy.
GiteaMirror commented 2026-04-12 18:10:50 -05:00
Author
Owner
Copy Link

@RobertoMaurizzi commented on GitHub (Oct 5, 2025):

AMD now released official support for gfx1150... in ROCm6.4.4, latest ollama is STILL not including it in the list of supported architectures but it DOES work if you set HSA_OVERRIDE_GFX_VERSION=11.5.1 when running ollama.

For people that can't get it to run: dynamic VRAM allocation detection was broken in amdgpu before kernel 6.16.9, the driver would report only the minimum memory allocation set by the BIOS or the kernel so that might be your problem.
Either set a fixed VRAM size in BIOS or upgrade the kernel (the first being much simpler than the second): see https://github.com/ROCm/ROCm/issues/5444

<!-- gh-comment-id:3368853790 --> @RobertoMaurizzi commented on GitHub (Oct 5, 2025): AMD now released official support for gfx1150... in ROCm6.4.4, latest ollama is STILL not including it in the list of supported architectures but it DOES work if you set `HSA_OVERRIDE_GFX_VERSION=11.5.1` when running ollama. For people that can't get it to run: dynamic VRAM allocation detection was broken in `amdgpu` before kernel 6.16.9, the driver would report only the minimum memory allocation set by the BIOS or the kernel so that might be your problem. Either set a fixed VRAM size in BIOS or upgrade the kernel (the first being much simpler than the second): see https://github.com/ROCm/ROCm/issues/5444
GiteaMirror commented 2026-04-12 18:10:51 -05:00
Author
Owner
Copy Link

@fsyllkn commented on GitHub (Nov 3, 2025):

2025.10.30，AMD add gfx1150 & gfx1151 support

<!-- gh-comment-id:3480607117 --> @fsyllkn commented on GitHub (Nov 3, 2025): 2025.10.30，AMD add gfx1150 & gfx1151 support
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label amd feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6554
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?