Vulkan GPUs: real device names and selectable ordinals (rebase of #7356 onto #7476) (#7498)

* Vulkan GPUs: real device names and selectable ordinals

Rebases the durable half of #7356 onto the inference_gpu transport #7476
landed on main. Those two PRs solve an overlapping problem and disagree on
the data model, so merging #7356 as-is would ship two parallel Vulkan
device concepts with different index semantics. This keeps main's transport
and adds what #7356 had that #7476 does not.

- _vulkan_probe.py emits a 5th column, ggml's device description, sanitized
  for the tab protocol and UTF-8 safe. Reader tolerates 4- or 5-column
  output so an older probe still parses.
- llama_cpp gains _run_vulkan_probe (shared parse) and
  vulkan_device_inventory (names + is_igpu + real totals).
- get_vulkan_inference_gpu_info reports the real name and an explicit
  is_igpu instead of "Vulkan<i>" and a total == 0 guess.
- index_kind becomes "vulkan", not "relative", and gpu_ids picks are
  supported on Vulkan builds once the probe enumerated ordinals. The XPU ban
  no longer applies to them: a Vulkan pick is a ggml ordinal, not a torch-xpu
  index, so it works on an Intel host too.
- Frontend picker reads the Vulkan inventory as the pickable set.

Memory deliberately still comes from _get_gpu_memory, not the inventory.
That path applies _apply_igpu_host_reserve_mib and zeroes a shared total;
budgeting an APU off its raw shared total would hand out the whole machine's
RAM with no OS headroom. Identity is joined onto it by ordinal, so a probe
failure degrades to Vulkan<i> names with the memory readings intact.

Dropped from #7356 as superseded: validate_vulkan_gpu_ids (main's
resolve_requested_gpu_ids already rejects duplicates and
_resolve_gguf_gpu_ids_for_request already probes for existence), the
gguf_devices transport, and the iGPU budget fallback in 71619891e, which
main's aggregateGpuMemoryTotalGb handles better by counting a shared pool
once.

Also keeps #7356's removal of the late diffusion raise, so the graceful
gpu_ids drop stays reachable for a GGUF only classified as diffusion after
download. #7415's real guard, _reject_vulkan_diffusion_gpu_ids_before_
teardown, is untouched.

Verified on Windows + Strix Halo: backend Vulkan/GPU-selection suites at the
same 4 pre-existing failures as main, tests/studio 1671 passed with no new
failures, frontend typecheck clean. Hardware confirmation of the underlying
behavior is on #7356 from @Bebiv24 (RX 9070 XT + RX 480).

Co-authored-by: LeoBorcherding <borchborchmail@gmail.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: LeoBorcherding <borchborchmail@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

studio/backend/core/inference/_vulkan_probe.py

@@ -6,12 +6,14 @@
 Run in a short-lived subprocess (``python _vulkan_probe.py <bindir>``) so the
 Vulkan instance never lives in the long-running backend process. Loads the
 bundled ggml Vulkan backend from ``<bindir>`` and prints one
-``<idx>\\t<free_bytes>\\t<is_igpu>\\t<total_bytes>`` line per device to stdout.
-Indices are ggml's own Vulkan device ordinals, which need not match nvidia-smi
-order. ``is_igpu`` (from ggml's device type) is ``1`` for an integrated GPU
-sharing system RAM. ``total_bytes`` is the device-local heap; the reader uses
-it to reserve absolute headroom on a discrete card (parity with the CUDA/ROCm
-fit) and ignores it for an iGPU, whose "VRAM" is shared system RAM.
+``<idx>\\t<free_bytes>\\t<is_igpu>\\t<total_bytes>\\t<name>`` line per device to
+stdout. Indices are ggml's own Vulkan device ordinals, which need not match
+nvidia-smi order. ``is_igpu`` (from ggml's device type) is ``1`` for an
+integrated GPU sharing system RAM. ``total_bytes`` is the device-local heap;
+the reader uses it to reserve absolute headroom on a discrete card (parity
+with the CUDA/ROCm fit) and ignores it for an iGPU, whose "VRAM" is shared
+system RAM. ``name`` is ggml's device description (the marketing name, e.g.
+"AMD Radeon RX 9070 XT"); empty when the registry lookup fails.
 
 Uses only the standard library so it stays runnable as a bare script.
 """
@@ -24,15 +26,30 @@ import sys
 _GGML_BACKEND_DEVICE_TYPE_IGPU = 2
 
 
-def _igpu_flags(base, lib, count: int) -> list[bool]:
-    """Per-device integrated-GPU flags via ggml's backend registry.
+def _igpu_flags_and_names(base, lib, count: int) -> tuple[list[bool], list[str]]:
+    """Per-device integrated-GPU flags and descriptions via ggml's backend registry.
 
     The Vulkan reg enumerates devices in the same order as
     ``ggml_backend_vk_get_device_memory`` (each context uses ``ctx->device =
-    i``), so reg index == device ordinal. Returns all-False on any failure so
-    the reader never over-caps a discrete card.
+    i``), so reg index == device ordinal. Returns all-False / empty-name on any
+    failure so the reader never over-caps a discrete card and the memory
+    readings still get through.
     """
     flags = [False] * count
+    names = [""] * count
+
+    # The name lookup is bound OUTSIDE the type-detection try: a ggml-base
+    # without ggml_backend_dev_description (older/custom build) must degrade to
+    # unnamed devices, not abort before the iGPU flags are read (which would
+    # count an iGPU's shared RAM as VRAM).
+    describe = None
+    try:
+        base.ggml_backend_dev_description.restype = ctypes.c_char_p
+        base.ggml_backend_dev_description.argtypes = [ctypes.c_void_p]
+        describe = base.ggml_backend_dev_description
+    except Exception:
+        pass
+
     try:
         lib.ggml_backend_vk_reg.restype = ctypes.c_void_p
         lib.ggml_backend_vk_reg.argtypes = []
@@ -45,17 +62,31 @@ def _igpu_flags(base, lib, count: int) -> list[bool]:
 
         reg = lib.ggml_backend_vk_reg()
         if not reg:
-            return flags
+            return flags, names
         dev_count = base.ggml_backend_reg_dev_count(reg)
         for i in range(min(count, dev_count)):
             dev = base.ggml_backend_reg_dev_get(reg, i)
             if dev:
                 flags[i] = base.ggml_backend_dev_type(dev) == _GGML_BACKEND_DEVICE_TYPE_IGPU
+                if describe is not None:
+                    try:
+                        desc = describe(dev)
+                        if desc:
+                            # Tabs/newlines would corrupt the line protocol;
+                            # spaces are safe.
+                            names[i] = (
+                                desc.decode("utf-8", errors = "replace")
+                                .replace("\t", " ")
+                                .replace("\n", " ")
+                                .strip()
+                            )
+                    except Exception:
+                        pass
     except Exception:
-        # Best-effort: any failure degrades to "discrete" so the memory
-        # readings still get through instead of crashing the probe.
+        # Best-effort: any failure degrades to "discrete"/"unnamed" so the
+        # memory readings still get through instead of crashing the probe.
         pass
-    return flags
+    return flags, names
 
 
 def main() -> int:
@@ -63,6 +94,14 @@ def main() -> int:
         return 0
     bindir = sys.argv[1]
 
+    # Device names can be non-ASCII (localized drivers); the platform-default
+    # stdout encoding (e.g. cp1252) would raise on them and lose the whole
+    # inventory. The reader decodes UTF-8 with the same error mode.
+    try:
+        sys.stdout.reconfigure(encoding = "utf-8", errors = "replace")
+    except Exception:
+        pass
+
     # Hold add_dll_directory's handle for the rest of main() (the documented
     # idiom) so bindir stays on the search path while the sibling ggml DLLs
     # resolve below.
@@ -96,12 +135,12 @@ def main() -> int:
     ]
 
     count = lib.ggml_backend_vk_get_device_count()
-    igpu = _igpu_flags(base, lib, count)
+    igpu, names = _igpu_flags_and_names(base, lib, count)
     rows = []
     for i in range(count):
         free, total = ctypes.c_size_t(0), ctypes.c_size_t(0)
         lib.ggml_backend_vk_get_device_memory(i, ctypes.byref(free), ctypes.byref(total))
-        rows.append("%d\t%d\t%d\t%d" % (i, free.value, int(igpu[i]), total.value))
+        rows.append("%d\t%d\t%d\t%d\t%s" % (i, free.value, int(igpu[i]), total.value, names[i]))
     sys.stdout.write("\n".join(rows))
     return 0
 

studio/backend/core/inference/llama_cpp.py

@@ -3501,18 +3501,17 @@ class LlamaCppBackend:
             return []
 
     @staticmethod
-    def _get_gpu_free_memory_vulkan(binary: Optional[str] = None) -> list[tuple[int, int, int]]:
-        """Query free (and total) VRAM per device via the bundled ggml Vulkan backend.
+    def _run_vulkan_probe(binary: Optional[str] = None) -> list[dict]:
+        """Run ``_vulkan_probe.py`` and parse its per-device lines.
 
-        Loads ``libggml-vulkan`` in a short-lived subprocess (no Vulkan instance
-        in this process) and returns (device_index, free_mib, total_mib) sorted
-        by index. The index is ggml's compact Vulkan ordinal -- the one the
-        registry names ``Vulkan<index>`` and load_model pins with ``--device``,
-        NOT the raw ``GGML_VK_VISIBLE_DEVICES`` space. A user-set
-        ``GGML_VK_VISIBLE_DEVICES`` is honored by ggml (passed through), so the
-        list already reflects it. iGPUs leave a host-RAM margin (see
-        ``_apply_igpu_host_reserve_mib``) and report total 0; discrete cards pass
-        their real total through. [] when no Vulkan build or device is reachable.
+        Returns raw (uncapped) rows sorted by index:
+        ``{"index", "free_mib", "total_mib", "is_igpu", "name"}``. The index is
+        ggml's compact Vulkan ordinal -- the one the registry names
+        ``Vulkan<index>`` and load_model pins with ``--device``, NOT the raw
+        ``GGML_VK_VISIBLE_DEVICES`` space. A user-set ``GGML_VK_VISIBLE_DEVICES``
+        is honored by ggml (passed through), so the list already reflects it.
+        ``name`` is ggml's device description; "" from an older 4-column probe.
+        [] when no Vulkan build or device is reachable.
         """
         binary = binary or LlamaCppBackend._find_llama_server_binary()
         if not binary:
@@ -3537,10 +3536,13 @@ class LlamaCppBackend:
             )
         probe_script = Path(__file__).with_name("_vulkan_probe.py")
         try:
+            # UTF-8 to match the probe's stdout reconfigure: device names can be
+            # non-ASCII, and the platform-default decode (cp1252) could throw.
             result = subprocess.run(
                 [sys.executable, str(probe_script), str(binary_dir)],
                 capture_output = True,
-                text = True,
+                encoding = "utf-8",
+                errors = "replace",
                 timeout = 15,
                 env = env,
                 **_windows_hidden_subprocess_kwargs(),
@@ -3554,21 +3556,56 @@ class LlamaCppBackend:
             logger.debug(f"vulkan GPU probe failed: {e}")
             return []
 
-        gpus: list[tuple[int, int, int]] = []
+        rows: list[dict] = []
         for line in result.stdout.strip().splitlines():
             parts = line.split("\t")
-            if len(parts) != 4:
+            # 4 columns from an older probe (no name); 5 with the name column.
+            if len(parts) not in (4, 5):
                 continue
             try:
-                idx = int(parts[0])
-                free_mib = int(parts[1]) // (1024 * 1024)
-                is_igpu = parts[2] == "1"
-                # iGPU "total" is shared RAM, not a VRAM budget -> keep 0 so the
-                # fit stays on free*frac (the host reserve below is its
-                # headroom); a discrete card passes its real total through.
-                total_mib = 0 if is_igpu else int(parts[3]) // (1024 * 1024)
+                rows.append(
+                    {
+                        "index": int(parts[0]),
+                        "free_mib": int(parts[1]) // (1024 * 1024),
+                        "is_igpu": parts[2] == "1",
+                        "total_mib": int(parts[3]) // (1024 * 1024),
+                        "name": parts[4].strip() if len(parts) == 5 else "",
+                    }
+                )
             except ValueError:
                 continue
+        rows.sort(key = lambda r: r["index"])
+        return rows
+
+    @staticmethod
+    def vulkan_device_inventory(binary: Optional[str] = None) -> list[dict]:
+        """UI-facing Vulkan device list: the devices llama-server will actually
+        use, with real totals (an iGPU keeps its shared-RAM total here -- the
+        caller labels it, unlike the fit which zeroes it). Same rows as
+        ``_run_vulkan_probe``; names fall back to ``Vulkan<i>``.
+        """
+        rows = LlamaCppBackend._run_vulkan_probe(binary)
+        for row in rows:
+            if not row["name"]:
+                row["name"] = f"Vulkan{row['index']}"
+        return rows
+
+    @staticmethod
+    def _get_gpu_free_memory_vulkan(binary: Optional[str] = None) -> list[tuple[int, int, int]]:
+        """Query free (and total) VRAM per device via the bundled ggml Vulkan backend.
+
+        Fit-oriented view of ``_run_vulkan_probe``: returns (device_index,
+        free_mib, total_mib) sorted by index. iGPUs leave a host-RAM margin (see
+        ``_apply_igpu_host_reserve_mib``) and report total 0; discrete cards pass
+        their real total through. [] when no Vulkan build or device is reachable.
+        """
+        gpus: list[tuple[int, int, int]] = []
+        for row in LlamaCppBackend._run_vulkan_probe(binary):
+            idx, free_mib, is_igpu = row["index"], row["free_mib"], row["is_igpu"]
+            # iGPU "total" is shared RAM, not a VRAM budget -> keep 0 so the
+            # fit stays on free*frac (the host reserve below is its
+            # headroom); a discrete card passes its real total through.
+            total_mib = 0 if is_igpu else row["total_mib"]
             capped = _apply_igpu_host_reserve_mib(free_mib, is_igpu)
             if capped < free_mib:
                 logger.info(
@@ -3577,7 +3614,6 @@ class LlamaCppBackend:
                     f"({free_mib}->{capped}MiB usable)"
                 )
             gpus.append((idx, capped, total_mib))
-        gpus.sort(key = lambda g: g[0])
         if gpus:
             logger.info(
                 "Vulkan GPU memory detected: "
@@ -6635,12 +6671,23 @@ class LlamaCppBackend:
             # Block-diffusion GGUFs (DiffusionGemma) cannot run on llama-server;
             # serve them with the diffusion runner (same OpenAI-compat interface).
             if self._is_diffusion:
-                # Final defense: route and pre-teardown preflights reject before Phase 1.
-                if is_vulkan_backend and gpu_ids:
-                    raise ValueError(_VULKAN_DIFFUSION_GPU_IDS_ERROR)
                 # Not a tensor/layer GGUF: clear any preserved-fallback flag from a
                 # prior load (this path skips the command builder that clears it).
                 self._layer_preserves_tensor_intent = False
+                # On a Vulkan build gpu_ids are ggml Vulkan ordinals, but the diffusion
+                # runner selects its device by CUDA physical index (_diffusion_gpu_arg
+                # forwards gpu_ids[0] as a CUDA/DG_GPU token) with no mapping to them.
+                # The route rejects a CONFIRMED-diffusion pick up front; an uncached GGUF
+                # only classified as diffusion post-download still reaches here with a
+                # pin, so drop it and serve on the default device (like an unpinned load).
+                if gpu_ids and is_vulkan_backend:
+                    logger.warning(
+                        "Ignoring gpu_ids %s for diffusion GGUF on a Vulkan build: "
+                        "the diffusion runner cannot map ggml Vulkan ordinals; "
+                        "serving on the default device.",
+                        gpu_ids,
+                    )
+                    gpu_ids = None
                 with self._lock:
                     if self._cancel_event.is_set():
                         logger.info("Load cancelled before diffusion server start")

studio/backend/main.py

@@ -1249,16 +1249,18 @@ def _get_cached_system_gpu_info(logger) -> tuple[dict[str, Any], dict[str, Any]]
             )
             enriched_devices.append(enriched_dev)
 
-        # Whether GGUF loads accept an explicit gpu_ids pick: /load and
-        # /validate 400 picks on XPU hosts (no visibility mask speaks torch-xpu
-        # ordinals) and on Vulkan-only builds (--device pins ggml's own
-        # ordinals), so the picker must not offer them.
+        # Whether GGUF loads accept an explicit gpu_ids pick. /load and /validate
+        # 400 picks on XPU hosts, where no visibility mask speaks torch-xpu
+        # ordinals. A Vulkan build IS pinnable: its picks are ggml ordinals, the
+        # same space `--device Vulkan<i>` uses, so check it first and let it
+        # through even on an XPU host (the XPU ban is about torch ordinals).
+        is_vulkan_build = False
         try:
             from core.inference.llama_cpp import LlamaCppBackend
             from utils.hardware import DeviceType, get_device
-            gpu_ids_supported = (
-                get_device() != DeviceType.XPU and not LlamaCppBackend._is_vulkan_backend()
-            )
+
+            is_vulkan_build = LlamaCppBackend._is_vulkan_backend()
+            gpu_ids_supported = is_vulkan_build or get_device() != DeviceType.XPU
         except Exception as e:
             logger.debug(f"Could not resolve gpu_ids support: {e}")
             gpu_ids_supported = True
@@ -1284,7 +1286,9 @@ def _get_cached_system_gpu_info(logger) -> tuple[dict[str, Any], dict[str, Any]]
             inference_gpu_info = (
                 {
                     **vulkan_info,
-                    "gguf_gpu_ids_supported": False,
+                    # Pinnable only once the probe actually enumerated devices:
+                    # without ordinals the frontend has nothing valid to offer.
+                    "gguf_gpu_ids_supported": bool(vulkan_info.get("devices")),
                 }
                 if vulkan_info is not None
                 else gpu_info

studio/backend/tests/test_gpu_selection.py

@@ -427,14 +427,16 @@ class TestVisibleGpuUtilization(_GpuCacheResetMixin, unittest.TestCase):
 
         self.assertTrue(result["available"])
         self.assertEqual(result["backend"], "vulkan")
-        self.assertEqual(result["index_kind"], "relative")
+        # ggml Vulkan ordinals are the space `--device Vulkan<i>` pins, so they
+        # are selectable, unlike a torch-xpu relative ordinal.
+        self.assertEqual(result["index_kind"], "vulkan")
         self.assertEqual(result["parent_visible_gpu_ids"], [])
         self.assertEqual(
             result["devices"],
             [
                 {
                     "index": 0,
-                    "index_kind": "relative",
+                    "index_kind": "vulkan",
                     "visible_ordinal": 0,
                     "name": "Vulkan0",
                     "memory_total_gb": 8.0,

studio/backend/tests/test_system_vulkan_gpu_info.py

@@ -56,7 +56,9 @@ def test_system_gpu_info_preserves_vulkan_visibility_metrics(monkeypatch):
     assert gpu["available"] is False
     assert gpu["backend"] == "cpu"
     assert gpu["index_kind"] == "relative"
-    assert gpu["gguf_gpu_ids_supported"] is False
+    # A Vulkan llama.cpp build accepts gpu_ids even when torch training is
+    # CPU-only: the pick is a ggml ordinal, not a torch device index.
+    assert gpu["gguf_gpu_ids_supported"] is True
     assert gpu["devices"] == []
     assert inference_gpu["backend"] == "vulkan"
     assert inference_gpu["devices"] == [vulkan_device]
@@ -124,7 +126,8 @@ def test_system_gpu_info_keeps_forced_vulkan_separate_from_training_metrics(monk
     assert gpu["devices"][0]["vram_used_gb"] == 6.0
     assert inference_gpu["backend"] == "vulkan"
     assert inference_gpu["devices"][0]["vram_used_gb"] == 1.0
-    assert inference_gpu["gguf_gpu_ids_supported"] is False
+    # Probed devices exist, so the ordinals are known and picks are offered.
+    assert inference_gpu["gguf_gpu_ids_supported"] is True
 
 
 def test_system_gpu_info_does_not_merge_metrics_across_backend_index_spaces(monkeypatch):
@@ -169,3 +172,85 @@ def test_system_gpu_info_does_not_merge_metrics_across_backend_index_spaces(monk
 
     assert gpu["devices"] == [vulkan_device]
     assert inference_gpu == gpu
+
+
+def test_vulkan_inference_gpu_uses_real_device_names_and_igpu_flag(monkeypatch):
+    """The picker and the GPU labels need ggml's real device description, not a
+    Vulkan<i> placeholder, and an explicit iGPU flag rather than inferring one
+    from a zero total. Memory still comes from _get_gpu_memory so the iGPU host
+    reserve is applied; budgeting off the raw shared total would hand out the
+    whole machine's RAM with no OS headroom.
+    """
+    from core.inference.llama_cpp import LlamaCppBackend
+    from utils.hardware.hardware import get_vulkan_inference_gpu_info
+
+    monkeypatch.setattr(
+        LlamaCppBackend, "_is_vulkan_backend", staticmethod(lambda binary = None: True)
+    )
+    # Fit view: discrete card keeps its total, iGPU reports 0 with capped free.
+    monkeypatch.setattr(
+        LlamaCppBackend,
+        "_get_gpu_memory",
+        staticmethod(lambda binary = None: [(0, 15 * 1024, 16 * 1024), (1, 12 * 1024, 0)]),
+    )
+    monkeypatch.setattr(
+        LlamaCppBackend,
+        "vulkan_device_inventory",
+        staticmethod(
+            lambda binary = None: [
+                {
+                    "index": 0,
+                    "name": "AMD Radeon RX 9070 XT",
+                    "free_mib": 15 * 1024,
+                    "total_mib": 16 * 1024,
+                    "is_igpu": False,
+                },
+                {
+                    "index": 1,
+                    "name": "AMD Radeon(TM) 8060S Graphics",
+                    "free_mib": 89 * 1024,
+                    "total_mib": 91 * 1024,
+                    "is_igpu": True,
+                },
+            ]
+        ),
+    )
+
+    info = get_vulkan_inference_gpu_info()
+    assert info is not None and info["index_kind"] == "vulkan"
+    dgpu, igpu = info["devices"]
+
+    assert dgpu["name"] == "AMD Radeon RX 9070 XT"
+    assert dgpu["index_kind"] == "vulkan"
+    assert dgpu["shared_memory"] is False
+    assert dgpu["memory_total_gb"] == 16.0
+
+    assert igpu["name"] == "AMD Radeon(TM) 8060S Graphics"
+    assert igpu["shared_memory"] is True
+    # The capped free budget from _get_gpu_memory, NOT the 91 GiB raw total.
+    assert igpu["memory_total_gb"] == 12.0
+
+
+def test_vulkan_inference_gpu_falls_back_to_ordinal_names(monkeypatch):
+    """A probe that cannot resolve descriptions must not lose the device list:
+    names degrade to Vulkan<i> and the memory readings still get through."""
+    from core.inference.llama_cpp import LlamaCppBackend
+    from utils.hardware.hardware import get_vulkan_inference_gpu_info
+
+    monkeypatch.setattr(
+        LlamaCppBackend, "_is_vulkan_backend", staticmethod(lambda binary = None: True)
+    )
+    monkeypatch.setattr(
+        LlamaCppBackend,
+        "_get_gpu_memory",
+        staticmethod(lambda binary = None: [(0, 15 * 1024, 16 * 1024)]),
+    )
+    monkeypatch.setattr(
+        LlamaCppBackend,
+        "vulkan_device_inventory",
+        staticmethod(lambda binary = None: (_ for _ in ()).throw(RuntimeError("probe failed"))),
+    )
+
+    info = get_vulkan_inference_gpu_info()
+    assert info["devices"][0]["name"] == "Vulkan0"
+    assert info["devices"][0]["memory_total_gb"] == 16.0

studio/backend/utils/hardware/hardware.py

@@ -1706,7 +1706,7 @@ def get_visible_gpu_utilization() -> Dict[str, Any]:
         "backend": _backend_label(device),
         "parent_visible_gpu_ids": [],
         "devices": [],
-        "index_kind": "relative",
+        "index_kind": "vulkan",
     }
 
 
@@ -2600,22 +2600,35 @@ def get_vulkan_inference_gpu_info() -> Optional[Dict[str, Any]]:
         "backend_cuda_visible_devices": None,
         "parent_visible_gpu_ids": [],
         "devices": [],
-        "index_kind": "relative",
+        "index_kind": "vulkan",
     }
+    # Identity (real device description, explicit iGPU flag) comes from the
+    # inventory; the memory numbers stay on _get_gpu_memory, which applies the
+    # iGPU host reserve and zeroes a shared total. Budgeting an APU off the raw
+    # shared total instead would hand out the whole machine's RAM with no OS
+    # headroom. Join by ordinal; a probe failure just leaves names unresolved.
+    identity: Dict[int, Dict[str, Any]] = {}
+    try:
+        identity = {row["index"]: row for row in LlamaCppBackend.vulkan_device_inventory()}
+    except Exception as e:
+        logger.debug("Vulkan device inventory failed, falling back to ordinals: %s", e)
+
     try:
         for ordinal, free_mib, total_mib in LlamaCppBackend._get_gpu_memory():
-            # Integrated Vulkan GPUs report total=0 because their memory is
-            # shared. Publish the capped free value as their usable inference
-            # budget and mark it so clients do not add system RAM again.
-            shared_memory = total_mib == 0
+            info = identity.get(ordinal, {})
+            # _get_gpu_memory reports total 0 for a shared pool; prefer the
+            # explicit flag when the inventory resolved this ordinal.
+            shared_memory = bool(info["is_igpu"]) if "is_igpu" in info else total_mib == 0
             budget_mib = total_mib or free_mib
             used_mib = max(0, total_mib - free_mib) if total_mib else None
             result["devices"].append(
                 {
                     "index": ordinal,
-                    "index_kind": "relative",
+                    # ggml Vulkan ordinals are the space `--device Vulkan<i>` pins,
+                    # so unlike a torch-xpu relative ordinal these are selectable.
+                    "index_kind": "vulkan",
                     "visible_ordinal": ordinal,
-                    "name": f"Vulkan{ordinal}",
+                    "name": info.get("name") or f"Vulkan{ordinal}",
                     "memory_total_gb": round(budget_mib / 1024, 2),
                     "vram_used_gb": round(used_mib / 1024, 2) if used_mib is not None else None,
                     "vram_free_gb": round(free_mib / 1024, 2),
@@ -2727,7 +2740,7 @@ def get_backend_visible_gpu_info() -> Dict[str, Any]:
         "backend_cuda_visible_devices": os.environ.get("CUDA_VISIBLE_DEVICES"),
         "parent_visible_gpu_ids": [],
         "devices": [],
-        "index_kind": "relative",
+        "index_kind": "vulkan",
     }
 
 

studio/frontend/src/hooks/use-gpu-info.ts

@@ -100,13 +100,29 @@ function toGpuInfo(
 }
 
 function toGpuDevices(data: SystemInfoResponse | null): SystemGpuDevice[] {
-  // Unpinnable configurations must hide every pick surface: XPU indices are
-  // torch-xpu ordinals no applicator speaks, and Vulkan-only builds pin ggml's
-  // own ordinals -- /load and /validate 400 picks on both, so the backend
-  // reports gpu.gguf_gpu_ids_supported and every gate keyed on physicalIndex
-  // (picker, persisted-pick reconcile) follows it. The device flavor lives on
-  // the TOP-LEVEL device_backend field; absent support info defaults to
-  // pinnable (older backend).
+  // GGUF loads run through llama-server, so on a Vulkan build the pickable set
+  // is the inference inventory, not the torch view: it can see cards torch
+  // cannot, and its indices are the ggml ordinals `--device Vulkan<i>` pins.
+  // The XPU ban does not apply there, it is about torch-xpu ordinals that no
+  // applicator speaks; a Vulkan pick does not use them.
+  const inference = data?.inference_gpu;
+  if (inference?.backend === "vulkan" && (inference.devices ?? []).length) {
+    const picksAccepted = inference.gguf_gpu_ids_supported !== false;
+    return (inference.devices ?? [])
+      .filter((d) => typeof d.index === "number")
+      .map((d) => ({
+        index: d.index as number,
+        name: d.name ?? `GPU ${d.index}`,
+        memoryTotalGb: d.memory_total_gb ?? 0,
+        memoryFreeGb: d.vram_free_gb ?? 0,
+        physicalIndex: picksAccepted && d.index_kind === "vulkan",
+      }));
+  }
+  // Otherwise the torch view is the pickable set. Unpinnable configurations
+  // must hide every pick surface: XPU indices are torch-xpu ordinals no
+  // applicator speaks, so /load and /validate 400 them, and the backend reports
+  // gpu.gguf_gpu_ids_supported. Absent support info defaults to pinnable
+  // (older backend).
   const pinnableBackend =
     data?.device_backend !== "xpu" &&
     data?.gpu?.gguf_gpu_ids_supported !== false;

tests/studio/test_model_picker_contracts.py

@@ -578,3 +578,24 @@ def test_legacy_migration_is_idempotent_and_non_destructive():
     # Layer 3: non-overwriting merge skips an existing (or default) key, so even a
     # forced re-run cannot duplicate or clobber a user's config.
     assert "if (isDefaultConfig(migrated) || Object.hasOwn(map, key)) {" in src
+
+
+def test_vulkan_inference_devices_are_the_pickable_set():
+    """GGUF loads run through llama-server, so on a Vulkan build the picker must
+    offer the inference inventory (ggml ordinals, the space `--device Vulkan<i>`
+    pins) rather than the torch view, which can miss cards llama-server drives.
+    The XPU ban must not apply there: it is about torch-xpu ordinals no
+    applicator speaks, and a Vulkan pick does not use them.
+    """
+    src = " ".join(_read("hooks/use-gpu-info.ts").split())
+    # The Vulkan inventory is consulted first, and only when it has devices.
+    assert (
+        "const inference = data?.inference_gpu; "
+        'if (inference?.backend === "vulkan" && (inference.devices ?? []).length) {' in src
+    )
+    # Pinnable on the ggml ordinal space, gated on the backend's own support flag.
+    assert "const picksAccepted = inference.gguf_gpu_ids_supported !== false;" in src
+    assert 'physicalIndex: picksAccepted && d.index_kind === "vulkan",' in src
+    # The torch fallback keeps its physical-only gate and the XPU ban.
+    assert 'data?.device_backend !== "xpu" &&' in src
+    assert 'physicalIndex: pinnableBackend && d.index_kind === "physical",' in src