Skip to content

Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

C++

Intel Arc B-series (Battlemage, e.g. Arc Pro B70) runs like any Arc: pictures on by default with the CPU reader, one plain line that the card is new and how to report a problem

Niko12216577b69d5d87
diff --git a/docs/GYRO.md b/docs/GYRO.md
index e5ec06b5a..4a84aabf0 100644
--- a/docs/GYRO.md
+++ b/docs/GYRO.md
@@ -69,7 +69,7 @@ Setup lists them for an Arc like for any card and warns, never refuses, when VRA
card: an Arc A750 (8 GB, i915 driver) in a PC with a Ryzen 5 5600X (6 cores, AVX2) and 62 GB of RAM, on the first-release files of Gyro-S and
Gyro-M (the same size as the pinned ones, a different SHA-256). The pinned Gyro-S passes the same parity run (16 cases, 0 failures) and
answers correctly through the server; it was not compared with CUDA.
-Not tested: a B-series card, two Arc cards (the second-card expert helper refuses a Gyro pack and says so), an Arc on the xe driver. The Battlemage AOT build
+Not tested by us: a B-series card (supported since 2026-10-22; `strata launch` says it is new and how to report a problem), two Arc cards (the second-card expert helper refuses a Gyro pack and says so), an Arc on the xe driver. The Battlemage AOT build
(`bmg-g31`) compiles with the trellis kernels; it was not run.
The trellis decode kernels do the same integer and float operations in the same order as the CUDA ones (int8 codebook, q8_1 activations),
diff --git a/docs/INTEL.md b/docs/INTEL.md
index 1aa957a30..beaa6ad79 100644
--- a/docs/INTEL.md
+++ b/docs/INTEL.md
@@ -797,8 +797,8 @@ the fix).
## Not done
- Images on the llama.cpp config: not wired. On the SYCL port they are wired, see "Pictures on an Arc" below. Open there: a
- Battlemage card (the B-series) keeps pictures opt-in because of the Arc Pro B70 fault, and nobody has run the SYCL picture reader
- on a card of 24 GB or more.
+ nobody has run the SYCL picture reader on a card of 24 GB or more. Pictures are on for a Battlemage card too since 2026-10-22 (see "Default" below),
+ with the CPU reader.
- Speculative decoding on the llama.cpp path (the GGUF carries no draft layer llama.cpp can use). The SYCL
port has it (MTP draft layer, `--mtp`).
@@ -816,12 +816,13 @@ the card from 24 GB. `strata_control/vision_reader.py` is the one place for it.
start, the server falls back to the CPU reader with one log line. On the A750: 0.16 s per picture instead of 1.73 s (after a 30 s
first start that compiles the kernels), and the embeddings match the CPU reader's (cosine 0.9959 to 0.9999 over 8 pictures). It takes
VRAM the experts would use, which is why the rule gives it to 24 GB cards only.
-- **Default.** On for an Alchemist card (A750, A770, ...) with the CPU reader. An Arc of 24 GB or more also keeps the CPU reader by
- default: the SYCL reader has not run on such a card (the B70 is gone), so `--vision gpu` asks for it and
- `vision_reader.INTEL_GPU_READER_BY_DEFAULT` is the switch to flip once one has. A Battlemage card (B580, B70, ...) keeps pictures
- **off** by default: the first picture on the Arc Pro B70 (engine 0.1.45, IQ3_S, 2026-10-19, CPU reader) logged an xe engine reset
- (ccs + bcs) and `Fault response: Unsuccessful -EINVAL`, and the request ended 503. The CPU reader was in use, so the fault is in the engine on xe, not in a GPU reader; it
- has not been looked at since. `--vision yes` turns pictures on there; setup says so.
+- **Default.** On for every Arc card with the CPU reader, a B-series (Battlemage) card included. An Arc of 24 GB or more also keeps the CPU reader by
+ default: the SYCL reader has not run on such a card, so `--vision gpu` asks for it and `vision_reader.INTEL_GPU_READER_BY_DEFAULT` is the switch to
+ flip once one has. **B-series cards are supported and untested by us** (the owner decision of 2026-10-22: users will report failures). One known
+ risk: the first picture on the Arc Pro B70 (engine 0.1.45, IQ3_S, 2026-10-19, CPU reader) logged an xe engine reset (ccs + bcs) and
+ `Fault response: Unsuccessful -EINVAL`, and the request ended 503. The CPU reader was in use, so the fault is in the engine on xe, not in a GPU
+ reader; it has not been looked at since. It is not a refusal: setup and `strata launch` print one plain line that the card is new to Strata and
+ how to report a problem, and `--vision no` (setup) or `--vision off` (`strata launch`) turns pictures off.
Tested on the A750 (IQ3_XXS, 8K context, engine from f-021-int, JIT build, 2026-10-19): a picture with a red circle, a blue square and the
words "STRATA 42", 8 variants, answered correctly by the CPU reader 10 of 10 (a first set with smaller text read "42" as "412"), and
diff --git a/packaging/README.md b/packaging/README.md
index b6e746322..9020c6129 100644
--- a/packaging/README.md
+++ b/packaging/README.md
@@ -22,7 +22,7 @@ A backend name alone is not a target. Each row is an OS x CPU ABI x backend tupl
| `windows-x86_64-hip` | NSIS `.exe` (Tauri) | `tools/hip/build_windows.bat` | **built; never run on a real AMD card.** |
| `linux-x86_64-cuda` | AppImage (Tauri) | `setup.py`; built in a glibc 2.35 chroot | **cuda13 built and tested** (RTX 3060 and RTX A4000: app in the browser, a real model, a picture). cuda12 built. |
| `linux-x86_64-hip` | AppImage (Tauri) | `packaging/build_hip_linux.sh` | **RX 9000 (gfx1200 / gfx1201) built and tested** on a Radeon AI PRO R9700. **RX 7000 (gfx1100 / 1101 / 1102) built; not run on an RX 7000 card.** **RX 6000 (gfx1030 / 1031, 0.2.2, engine package `rx6000` only) built; not tested on a real RX 6000 card.** |
-| `linux-x86_64-sycl` | | `packaging/build_sycl_linux.sh` (one JIT package for Alchemist and Battlemage) | **engine package built; tested on an Arc A750** (Intel Arc is an official platform since 0.1.40.2, docs/INTEL_ARC.md). No B-series card was available for the package. `sycl/setup_intel.py` (Docker) remains the source-build route. |
+| `linux-x86_64-sycl` | | `packaging/build_sycl_linux.sh` (one JIT package for Alchemist and Battlemage) | **engine package built; tested on an Arc A750** (Intel Arc is an official platform since 0.1.40.2, docs/INTEL_ARC.md). Arc B-series (Battlemage) cards are supported and untested by us: no B-series card was available for the package (the B70 tests of 2026-10 were with the source build), and the Terminal says so once and tells the person how to report a problem. `sycl/setup_intel.py` (Docker) remains the source-build route. |
| `windows-x86_64-sycl` | none | none | **gated.** No native Windows SYCL path exists upstream; it needs a separate feasibility gate, not a recipe. |
| `macos-x86_64-*` | none | none | **unsupported.** The macOS engine is Apple-Silicon only (`metal/CMakeLists.txt` requires Apple Silicon for the Metal runtime). |
diff --git a/strata_cli/launch/catalog.py b/strata_cli/launch/catalog.py
index 4d31076c4..f4d00f626 100644
--- a/strata_cli/launch/catalog.py
+++ b/strata_cli/launch/catalog.py
@@ -138,6 +138,17 @@ class Computer:
from strata_control import gpu_fit
return gpu_fit.other_vendor_note(self.gpus, self.engine_vendor)
+ @property
+ def new_card_note(self) -> tuple[str, str] | None:
+ """(what, what to do) when a usable card is an Intel Arc B-series (Battlemage): allowed, but new to Strata and untested by the maintainers."""
+ import re
+ from strata_control import gpu_fit
+ for c in gpu_fit.usable_cards(gpu_fit.for_vendor(self.gpus, self.engine_vendor)[0])[0]:
+ if c.get("vendor") == "intel" and re.search(r"Arc (?:Pro )?B\d", str(c.get("name", ""))):
+ return (f"The {str(c['name']).replace('Intel ', '')} is a B-series Arc card. It is new to Strata and we have not tested it ourselves, so pictures or long answers may fail on it.",
+ "If something fails, start again with --vision off to turn pictures off, and tell us at https://github.com/Niko1221/Strata/issues (say it is a B-series card and attach the engine log).")
+ return None
+
@property
def main_vram_gb(self) -> float:
return max((float(g.get("vram_gb") or 0) for g in self.usable_gpus), default=0.0)
@@ -431,8 +442,9 @@ class Catalog:
return "off"
if self.is_mac():
return VR.default_mode(None, unified=True) # a Mac (unified memory): the Metal reader, nothing to keep free for a card
- vram = self.computer().main_vram_gb
- return VR.default_mode(vram) if vram > 0 else "cpu" # no usable card: nothing to read on, the CPU reader
+ comp = self.computer()
+ vram = comp.main_vram_gb
+ return VR.default_mode(vram, "intel" if comp.engine_vendor == "intel" else "") if vram > 0 else "cpu" # no usable card: nothing to read on, the CPU reader
def is_mac(self) -> bool:
"""An Apple Silicon Mac: the chip's memory is the system's (the Metal engine; strata_control/vision_reader.py)."""
diff --git a/strata_cli/launch/cmd.py b/strata_cli/launch/cmd.py
index 40200055b..a9b546ec7 100644
--- a/strata_cli/launch/cmd.py
+++ b/strata_cli/launch/cmd.py
@@ -364,6 +364,9 @@ def gpu_notices(s: S.LaunchSettings, comp) -> list[S.Notice]:
note = comp.other_vendor_note
if note:
out.append(S.Notice("warn", note, ""))
+ new_card = comp.new_card_note
+ if new_card:
+ out.append(S.Notice("warn", new_card[0], new_card[1]))
return out
diff --git a/strata_control/vision_reader.py b/strata_control/vision_reader.py
index a4a058762..7ae7b7283 100644
--- a/strata_control/vision_reader.py
+++ b/strata_control/vision_reader.py
@@ -37,7 +37,7 @@ GPU_MIN_VRAM_GB = 23.5
#: The one exception to that rule. An Intel Arc of 24 GB or more keeps the CPU reader by default: the SYCL GPU reader works (tested on an A750,
#: forced with --vision gpu: 0.16 s a picture against 1.7 s on the CPU, embeddings cosine >= 0.995 to the CPU reader's) but it needs the oneAPI
-#: runtime on the PC and has not run on a 24 GB+ card (the B70 is gone). --vision gpu asks for it. Set True when a 24 GB+ Arc has run it.
+#: runtime on the PC and has not run on a 24 GB+ card (an Arc Pro B70 is allowed since 2026-10-22; the CPU reader is what it gets). --vision gpu asks for it. Set True when a 24 GB+ Arc has run it.
INTEL_GPU_READER_BY_DEFAULT = False
diff --git a/sycl/setup_intel.py b/sycl/setup_intel.py
index 420cb0f97..617ea2771 100644
--- a/sycl/setup_intel.py
+++ b/sycl/setup_intel.py
@@ -11,7 +11,7 @@ setup.py's, unchanged. What is replaced:
sized from sysfs;
- the engine step: the SYCL build (build-sycl-aot/strata, run in the strata-sycl-dev image by
sycl/serve/strata-sycl.sh; docs/INTEL.md) instead of a CUDA/HIP build;
- - pictures (on by default on an Alchemist card, see pictures_on_by_default): read by the portable CPU reader (`strata-vision`, the program the
+ - pictures (on by default on every Arc card, see pictures_on_by_default): read by the portable CPU reader (`strata-vision`, the program the
AMD packages carry beside their engine, compiled here by setup's own step) on the host; it hands the image embeddings to the engine through
/dev/shm, which strata-sycl.sh shares with the container. The reader choice is strata_control/vision_reader.py's one rule (CPU reader under
24 GB of VRAM); an Arc of 24 GB or more also keeps the CPU reader unless --vision gpu asks for the SYCL one (`strata-vision-sycl`, compiled
@@ -133,16 +133,26 @@ def drop(args, name, value=False):
# Pictures on an Intel Arc: the CPU reader encodes on the host and the embeddings reach the engine through /dev/shm. Tested end to end
# on the Arc A750 (i915, Alchemist; 2026-10-19, docs/INTEL.md): 47 picture requests (shapes, text, two pictures in one request, a picture in the
# history, text after pictures) through the CPU reader and the same through the SYCL GPU reader, in the container and outside it, no GPU fault.
-# The first picture request on the Arc Pro B70 (Battlemage, xe, engine 0.1.45 SYCL, IQ3_S, 2026-10-19) hung the GPU with the CPU reader: an xe
-# engine reset (ccs + bcs) and "Fault response: Unsuccessful -EINVAL", the request ended 503. That fault is in the engine on xe and has not been
-# looked at again (the B70 is gone), so a Battlemage card keeps pictures opt-in (--vision yes) and is told so.
-PICTURES_ON_BY_DEFAULT_IDS = ALCHEMIST_IDS
+# Owner decision 2026-10-22: an Arc B-series (Battlemage, xe) card is allowed to run like any Arc, pictures included. The first picture request on the
+# Arc Pro B70 (engine 0.1.45 SYCL, IQ3_S, 2026-10-19) hung the GPU with the CPU reader: an xe engine reset (ccs + bcs) and "Fault response:
+# Unsuccessful -EINVAL", the request ended 503. That fault has not been looked at again (the B70 is gone), so it is a known risk, not a refusal:
+# pictures are on, the CPU reader is kept, and the person is told once that the card is new to Strata and how to report a problem.
+PICTURES_ON_BY_DEFAULT_IDS = ALCHEMIST_IDS | frozenset(INTEL_ARC) # every Arc card setup knows
def pictures_on_by_default(driver: str, devid: str = "") -> bool:
- """Pictures are on in setup's default for an Alchemist card (i915, or the A-series on xe), which has run them; a Battlemage card
- (xe, the B-series) keeps them opt-in until the first-picture GPU fault of the B70 is understood."""
- return driver == "i915" or devid in PICTURES_ON_BY_DEFAULT_IDS
+ """Pictures are on in setup's default for every Arc card (they were opt-in on a B-series card until 2026-10-22)."""
+ return True
+
+
+def is_b_series(driver: str, devid: str = "") -> bool:
+ """A Battlemage (B-series) card: on xe and not an Alchemist id (an unknown id on xe counts as one: it is newer than the A-series)."""
+ return driver != "i915" and devid not in ALCHEMIST_IDS
+
+
+B_SERIES_LINE = ("This Intel Arc B-series card is new to Strata and we have not tested it ourselves, so pictures and long answers may fail on it. "
+ "If something fails, run again with --vision no to turn pictures off, and tell us at https://github.com/Niko1221/Strata/issues "
+ "(say it is a B-series card and attach the engine log).")
READER_DIR = ROOT / "engine-sycl" # the folder setup treats as the engine's: BUILD.json and, for pictures, the CPU reader strata-vision
@@ -537,10 +547,8 @@ def install(argv) -> None:
sys.argv = [str(ROOT / "setup.py"), *argv]
if "--backend" not in argv:
sys.argv += ["--backend", "hip"]
- if "--vision" not in argv and not pictures_on_by_default(intel[0]["driver"], intel[0].get("devid", "")):
- sys.argv += ["--vision", "none"]
- S.say(" Pictures are off by default on this card (Arc Pro B70 test of 2026-10-19: the first picture reset the GPU, docs/INTEL.md): "
- "--vision yes turns them on")
+ if is_b_series(intel[0]["driver"], intel[0].get("devid", "")):
+ S.say(" " + B_SERIES_LINE)
if "--low-ram" not in argv:
sys.argv += ["--low-ram", "off"]
sys.exit(S.main())
diff --git a/tests/terminal/test_launch.py b/tests/terminal/test_launch.py
index 5a3625a1f..7603323bc 100644
--- a/tests/terminal/test_launch.py
+++ b/tests/terminal/test_launch.py
@@ -353,6 +353,26 @@ class SetupArgs(unittest.TestCase):
off = Runner(S.LaunchSettings(family="qwen", model="IQ3_XXS", context=32768, vision="off"), catalog=catalog(NV))
self.assertEqual(off.setup_args()[off.setup_args().index("--vision") + 1], "no")
+ def test_an_arc_b_series_card_is_allowed_with_pictures_on_the_cpu_reader_and_one_plain_line(self):
+ """Owner decision 2026-10-22: Arc B-series (Battlemage) runs like any Arc. An Arc Pro B70 (32 GB) gets pictures on, with the CPU reader (the GPU
+ reader has not run on a 24 GB+ Arc), and a warn-level note that it is new and how to report a problem. An A-series card gets no such note."""
+ from strata_cli.launch import cmd
+ for name, vram, note in (("Intel Arc Pro B70", 32.0, True), ("Intel Arc B580", 12.0, True), ("Intel Arc A750", 8.0, False)):
+ hw = {"ram_gb": 64.0, "os": "Linux", "cpu": {"avx2": True, "name": "x"},
+ "gpus": [{"index": 0, "name": name, "vendor": "intel", "vram_gb": vram, "usable": True}]}
+ with Home():
+ from strata_cli.launch.catalog import Computer
+ cat = catalog(hw)
+ comp = Computer(ram_gb=64.0, gpus=hw["gpus"], disk_free_gb=500.0, os="Linux", avx2=True, raw=hw, engine_vendor="intel")
+ with mock.patch.object(type(cat), "reader_state", lambda self: "present"), mock.patch.object(type(cat), "computer", lambda self, fresh=False: comp):
+ self.assertEqual(cat.picture_default(True), "cpu", name)
+ self.assertEqual(bool(comp.new_card_note), note, name)
+ if note:
+ text, nxt = comp.new_card_note
+ self.assertIn("new to Strata", text)
+ self.assertIn("--vision off", nxt)
+ self.assertIn("github.com/Niko1221/Strata/issues", nxt)
+
def test_a_mac_defaults_pictures_to_the_metal_reader_whatever_its_memory(self):
for gb in (16.0, 32.0, 128.0):
hw = {"ram_gb": gb, "os": "Darwin", "cpu": {"avx2": None, "name": "Apple M4"},
diff --git a/tools/test_setup_sycl.py b/tools/test_setup_sycl.py
index 339243c85..98dac7aa7 100644
--- a/tools/test_setup_sycl.py
+++ b/tools/test_setup_sycl.py
@@ -264,17 +264,21 @@ class ToSycl(unittest.TestCase):
self.assertEqual(c["vision"], section) # host paths: the reader runs on the host, not in the container
self.assertIn("--vision", c["args"])
- def test_pictures_are_on_by_default_on_an_alchemist_and_opt_in_on_a_battlemage(self):
- """The A750 ran 47 picture requests without a fault; the first picture on the Arc Pro B70 reset the GPU (xe, 2026-10-19), so a
- Battlemage card keeps them opt-in (--vision yes) and setup says so."""
+ def test_pictures_are_on_by_default_on_every_arc_and_a_battlemage_is_told_it_is_new(self):
+ """Owner decision 2026-10-22: an Arc B-series card runs like any Arc. The first picture on the Arc Pro B70 reset the GPU (xe, 2026-10-19) is a
+ known risk, not a refusal: pictures are on (the CPU reader) and setup prints one plain line about it."""
m = self.mod()
- self.assertTrue(m.pictures_on_by_default("i915", "56a1")) # A750
- self.assertTrue(m.pictures_on_by_default("i915", ""))
- self.assertTrue(m.pictures_on_by_default("xe", "56a0")) # A770
- self.assertFalse(m.pictures_on_by_default("xe", "e223")) # B70
- self.assertFalse(m.pictures_on_by_default("xe", "e20b")) # B580
+ for driver, devid in (("i915", "56a1"), ("i915", ""), ("xe", "56a0"), ("xe", "e223"), ("xe", "e20b"), ("xe", "e211"), ("xe", "")):
+ self.assertTrue(m.pictures_on_by_default(driver, devid), (driver, devid))
+ self.assertTrue(m.is_b_series("xe", "e223")) # B70
+ self.assertTrue(m.is_b_series("xe", "e20b")) # B580
+ self.assertFalse(m.is_b_series("i915", "56a1")) # A750
+ self.assertFalse(m.is_b_series("xe", "56a0")) # A770 on xe
+ self.assertIn("new to Strata", m.B_SERIES_LINE)
+ self.assertIn("issues", m.B_SERIES_LINE)
src = (ROOT / "sycl" / "setup_intel.py").read_text(encoding="utf-8")
- self.assertIn('if "--vision" not in argv and not pictures_on_by_default(', src)
+ self.assertIn("B_SERIES_LINE", src)
+ self.assertNotIn('sys.argv += ["--vision", "none"]', src) # no B-series card has pictures switched off for it
self.assertIn("S.hip_vision = intel_vision", src)
self.assertIn("S.build_vision_cpu(", src) # the CPU reader is compiled by setup's own step
self.assertIn("build_vision_sycl(", src) # --vision gpu: the SYCL reader, when the oneAPI compiler is there