The verdict splits, and not where anyone expected. Measured on one machine, one
build, one day, physical console, uncapped, 4x MSAA on both arms, validation off:
Stationary LIGHT scene (6,675 entities, 780-870 FPS)
CPU p50 GL 1.127 ms -> VK 1.294 ms MISS +14.8%
CPU p99 GL 1.407 ms -> VK 1.531 ms MISS +8.8%
GPU p50 GL 0.651 ms -> VK 0.160 ms PASS -75.4%
Alloc/frm GL 77,664 B -> VK 11,440 B PASS -85.3%
Working GL 943.7 MiB -> VK 877.1 MiB PASS -7.0%
Stationary DENSE scene (21,024 entities, identical on both arms)
CPU p50 GL 5.934 ms -> VK 5.775 ms PASS -2.7%
CPU p99 GL 8.867 ms -> VK 7.354 ms PASS -17.1%
GPU p50 GL 1.673 ms -> VK 0.909 ms PASS -45.7%
Alloc/frm GL 82,016 B -> VK 15,752 B PASS -80.8%
Process CPU GL 1.246 cores -> VK 1.016 PASS -18.5% (Windows, not ours)
Canonical nine-stop route, identical world at all nine stops
Frames GL 30,378 -> VK 38,683 PASS +27.3%
CPU p50 GL 11.718 ms -> VK 9.166 ms PASS -21.8%
GPU p99 GL 2.325 ms -> VK 1.193 ms PASS -48.7%
Vulkan loses two rows in exactly one configuration: a stationary field at a frame
rate no player will ever see. The reason is measured rather than argued. A
temporary probe on BOTH arms, now stripped, attributes 0.148 ms/frame to required
Vulkan WSI and synchronisation calls - vkQueuePresentKHR 0.070, vkQueueSubmit2
0.027, the timeline wait 0.026, vkAcquireNextImageKHR 0.025 - against roughly
0.014 ms for GL's whole SwapBuffers. That cost is FIXED per frame, so it is 12%
of a 1.13 ms frame, 2.5% of a 5.9 ms one and under 1% of a dense-town frame,
while the GPU and allocation savings scale with the work. The sign of the CPU
comparison flips as soon as the frame contains a town.
The campaign's named cost centre is closed rather than carried a fourth time.
Bindings 4, 6, 7 and 8 costing a descriptor write per draw - forward-carried
since V6i-3 as the thing to fix if CPU were short - measures 0.031 ms for ALL
~216 draws of the frame, about 140 ns each and 2.4% of it. No Vulkan code was
changed to chase the miss: every lever the V8 row named was already taken
(coherent rings, one submit per frame), irrelevant to p50 (pipeline pre-warm),
measured and small (descriptors), or would have traded real memory for nothing
(a fourth swapchain image, when acquire is call cost and not waiting).
The methodological finding is worth reading before the numbers. The R6 soak is
NOT the vehicle the founding numbers came from - the G5 production profile states
its own conditions and they exclude the probe, the artifact owner and the
screenshot oracle - and it is biased AGAINST Vulkan, because
VulkanGraphicsContext arms retainBackbufferCapture exactly when
ACDREAM_AUTOMATION_ARTIFACT_DIR is set, making every Vulkan frame copy the whole
swapchain image while GL reads on demand. On one binary in one hour the soak
reports CPU p50 7.3 ms and 2,531 KiB/frame where the ordinary profile reports
1.13 ms and 77 KiB. The route table above is therefore conservative in Vulkan's
favour: it wins on the vehicle that charges it extra.
Gates: Release build green; App tests 4,152 / 3 skipped, the pre-slice baseline,
no #250-family failure; strict GL offline pixel gate against
|
||
|---|---|---|
| .github | ||
| .vscode | ||
| docs | ||
| memory | ||
| references | ||
| src | ||
| tests | ||
| tools | ||
| .gitattributes | ||
| .gitignore | ||
| .gitmodules | ||
| AcDream.slnx | ||
| AGENTS.md | ||
| analyze_flap_live.py | ||
| CLAUDE.md | ||
| find_burst.py | ||
| launch-a6-issue98-capture.ps1 | ||
| launch-a6-issue98-cottage-gfxobj-dump.ps1 | ||
| launch-a6-issue98-polydump.ps1 | ||
| launch-flap-capture.ps1 | ||
| launch-flap-churn.ps1 | ||
| launch-flap-verify.ps1 | ||
| NOTICE.md | ||
| README.md | ||
acdream
A modern open-source C# / .NET 10 Asheron's Call client.
acdream ports the observable behaviour of the September 2013 retail client to Silk.NET and a modern, plugin-friendly architecture. The code is modern; the behaviour is retail.
Status: playable pre-alpha. M3, “Cast a spell,” landed on 2026-07-21 and
M4, “Live in the world,” is active. The graphical client supports the connected
combat, magic, movement, portal, inventory, loot, and retained-UI loops used by
the current test characters. The presentation-independent GameRuntime and
the Linux/Windows multi-session headless host are complete. Native Linux
graphics are intentionally parked at the L1 capability checkpoint; Windows is
the currently validated graphical platform.
The documentation map is the entry point for current milestones, roadmap state, architecture, issues, retail divergences, research, and durable project memory.
Technology
- Runtime: C# and .NET 10
- Graphics: Silk.NET, OpenGL 4.3 core, bindless textures, shader draw parameters, SSBOs, and multi-draw indirect
- Audio: OpenAL through Silk.NET
- Content: retail DAT files plus a machine-local, memory-mapped
acdream.pakproduced byAcDream.Bake - Networking: custom UDP, ISAAC cipher, and game-message layers compatible with ACEmulator
- UI: retained retail gameplay UI plus opt-in ImGui developer tools
- Automation: the same presentation-independent
GameRuntimeis hosted by both the graphical client andAcDream.Headless
The modern renderer is mandatory. There is no legacy renderer fallback. Startup reports an actionable error if the required OpenGL capabilities are missing.
What works
- ACE login, character selection, world entry, chat, client commands, reconnect, and graceful logout.
- Outdoor, building, cellar, and dungeon streaming with prepared terrain, scenery, buildings, EnvCells, collision, portal visibility, sky, fog, lighting, audio, and day/night presentation.
- Local and observed movement, animation, jumping, selection, radar, combat stances, melee, bows, crossbows, spell projectiles, death, corpses, chests, and looting.
- Inventory bags, stable server ordering, stack splitting, ground drops, paperdoll equipment, weapon switching, quick bars, item use, cooldowns, and giving items to NPCs.
- Retail-style retained UI for vitals, chat, toolbar, inventory, character, attributes, skills, spellbook, components, effects, combat/spell/jump bars, radar/compass, dialogs, external containers, and assessment.
- Complete end-of-retail spell catalog, learned and favorite spells, component preflight, connected casts, enchantments, DAT-driven projectiles and effects, recall, portal-space travel, Hidden/UnHide, and remote materialization.
- One presentation-independent runtime owner for session, entities, objects, inventory, character state, selection, interactions, combat, magic, movement, physics, projectiles, world environment, and portal transit.
- A no-window Windows/Linux host with deterministic bot commands/events, shared immutable content, multi-session scheduling, isolation, reconnect, resource telemetry, and tested 1/5/10/30-session ownership.
- Plugin loading, shared command/input abstractions, retained markup panels,
and permanent ImGui developer tools behind
ACDREAM_DEVTOOLS=1.
Current boundaries
- The active M4 prelude is world interaction completion. Slices 1–3, including assessment and its final formula/icon/layout correction, are user-accepted. Equipped-child picking and vendor browse/buy/sell are the next uncompleted slices.
- Issue
#225retains the lifestone/particle shared-alpha visual comparison. Its connected lifetime and performance routes already pass. - Narrow carried behaviour debt includes issue
#153(an unstreamed far-teleport edge), issue#116(slide feel), issue#235(30 Hz capped/RDP jump presentation), and the live temporary-stopgap rows in the retail divergence register. - Native Linux graphics are deferred. L0 portability and L1 backend/capability
reporting are implemented; WSLg reaches the GPU through Mesa D3D12 but does
not expose mandatory
GL_ARB_bindless_texture. Resume with a supported physical Linux AMD/NVIDIA driver before beginning later Slice L work. - Advanced vendor/trade/crafting/social surfaces and larger M4 quest, character-creation, and emote bodies remain roadmap work.
Prerequisites
- .NET 10 SDK
- Your own retail Asheron's Call DAT directory containing:
client_portal.datclient_cell_1.datclient_highres.datclient_local_English.dat
- A machine-local
acdream.pakbuilt from those DATs - A running ACE server for connected play; the examples use
127.0.0.1:9000 - For the graphical client, a driver exposing the mandatory modern OpenGL capabilities
The project does not distribute Microsoft/Turbine DAT files or derived prepared packages.
Build and test
dotnet restore AcDream.slnx
dotnet build AcDream.slnx -c Release
dotnet test AcDream.slnx -c Release --no-build
The current baseline is a successful Release build with 8,826 passing tests
and 5 intentional skips. The build currently reports 17 test-project
warnings tracked by #228;
production compilation has zero errors.
Prepare content
Production rendering and collision use the validated prepared package rather than decoding world meshes on the frame path:
dotnet run --project src\AcDream.Bake\AcDream.Bake.csproj -c Release -- `
--dat-dir "C:\Games\Asheron's Call" `
--out "C:\Games\Asheron's Call\acdream.pak"
A complete package is approximately 30 GB. It is machine-local and must not be
committed. ACDREAM_PAK_PATH overrides the default
<DAT directory>\acdream.pak.
Run the graphical client
$env:ACDREAM_DAT_DIR = "C:\Games\Asheron's Call"
$env:ACDREAM_PAK_PATH = "C:\Games\Asheron's Call\acdream.pak"
$env:ACDREAM_LIVE = "1"
$env:ACDREAM_TEST_HOST = "127.0.0.1"
$env:ACDREAM_TEST_PORT = "9000"
$env:ACDREAM_TEST_USER = "testaccount"
$env:ACDREAM_TEST_PASS = "testpassword"
$env:ACDREAM_RETAIL_UI = "1"
dotnet run --project src\AcDream.App\AcDream.App.csproj -c Release
The DAT directory can instead be supplied as the first positional argument:
dotnet run --project src\AcDream.App\AcDream.App.csproj -c Release -- `
"C:\Games\Asheron's Call"
Run a headless session
AcDream.Headless loads no App, UI, OpenGL, native-window, or audio assembly.
Create a version-1 configuration such as bot.json:
{
"version": 1,
"process": {
"content": {
"datDirectory": "/opt/ac",
"preparedAssetPath": "/opt/ac/acdream.pak"
}
},
"sessions": [
{
"id": "bot-1",
"endpoint": { "host": "127.0.0.1", "port": 9000 },
"account": "testaccount",
"character": { "index": 0 },
"policy": { "id": "idle" },
"credential": {
"provider": "environment",
"reference": "ACDREAM_BOT_PASSWORD"
}
}
]
}
Then validate and run it:
export ACDREAM_BOT_PASSWORD='testpassword'
dotnet run --project src/AcDream.Headless/AcDream.Headless.csproj -c Release -- \
validate --config bot.json
dotnet run --project src/AcDream.Headless/AcDream.Headless.csproj -c Release -- \
run --config bot.json
For a single local session, run also accepts
--user <account> --password <password>. Add uniquely identified session
entries and credential references for a multi-session process. Available
built-in policies are idle, lifecycle-smoke, observer-movement, and
portal-route-smoke.
Useful startup options
| Variable | Effect |
|---|---|
ACDREAM_DAT_DIR |
Retail DAT directory |
ACDREAM_PAK_PATH |
Prepared package path; defaults to <DAT dir>/acdream.pak |
ACDREAM_LIVE=1 |
Enable connected mode |
ACDREAM_TEST_HOST / ACDREAM_TEST_PORT |
ACE endpoint |
ACDREAM_TEST_USER / ACDREAM_TEST_PASS |
Graphical-client credentials |
ACDREAM_RETAIL_UI=1 |
Enable the retained retail gameplay UI |
ACDREAM_DEVTOOLS=1 |
Enable ImGui developer tools |
ACDREAM_NO_AUDIO=1 |
Suppress OpenAL initialization |
ACDREAM_UNCAPPED_RENDER=1 |
Disable normal frame pacing for diagnostics |
ACDREAM_DISPLAY_PROTOCOL=auto|x11|wayland |
Select the Linux GLFW backend |
ACDREAM_DAY_GROUP=N |
Force a day-group index for weather/lighting comparisons |
ACDREAM_STREAM_RADIUS=N |
Legacy override over configured streaming radii |
ACDREAM_DUMP_SKY=1 |
Dump sky interpolation and draw diagnostics |
ACDREAM_DUMP_MOTION=1 |
Dump inbound movement and motion-cycle decisions |
Additional diagnostic and budget controls are documented beside their typed owners and in the linked research plans; they are not stable user settings.
Repository layout
src/
AcDream.Runtime/ presentation-independent GameRuntime
AcDream.App/ graphical host, retained UI, renderer, audio
AcDream.Headless/ Windows/Linux no-window multi-session host
AcDream.Core/ retail gameplay, movement, physics, world logic
AcDream.Core.Net/ UDP, ISAAC, protocol and message routing
AcDream.Content/ GL-free DAT and prepared-package content
AcDream.Bake/ offline acdream.pak builder
AcDream.Cli/ offline DAT inspector
AcDream.UI.Abstractions/ shared UI/input models and contracts
AcDream.UI.ImGui/ developer-tool presentation
AcDream.Plugin.Abstractions/ BCL-only plugin contracts
AcDream.Plugins.Smoke/ example plugin
tests/
AcDream.*.Tests/ layer-matched xUnit projects
docs/
README.md documentation authority and current map
architecture/ ownership, structure, divergence, WB inventory
plans/ milestone, roadmap, and execution plans
research/ retail pseudocode, traces, fixtures, evidence
audit/ completion and conformance audits
memory/ durable engineering references
references/ gitignored external reference repositories
Development workflow
All AC-specific behaviour starts from the named retail oracle in
docs/research/named-retail/:
- Search the named retail pseudo-C and headers by
class::method. - Use the older Ghidra chunks only when the named oracle is insufficient.
- Cross-reference ACE and the relevant client/viewer implementation.
- Record readable pseudocode and exact constants/order.
- Port the retail mechanism into the correct modern owner.
- Add conformance, lifecycle, and failure-boundary tests.
- Run the automated gate and the appropriate connected or visual gate.
- Update architecture, roadmap, divergences, and durable memory with the same change.
Guessing at AC-specific algorithms is forbidden. See AGENTS.md, CLAUDE.md, and the architecture guide for the full rules.
Reference projects
- ACE / ACEmulator: authoritative server and protocol behaviour
- ACViewer: character appearance and DAT presentation cross-check
- WorldBuilder: extracted Silk.NET DAT/rendering foundation
- Chorizite.ACProtocol: clean-room protocol reference
- holtburger: broad non-retail client behaviour reference
- AC2D: terrain and movement-packet cross-checks
The retail binary/decomp remains the behavioural oracle when references disagree.
Licence and game assets
The acdream source has not yet been assigned a top-level licence and is not ready for public redistribution. External reference code retains its own licence.
Asheron's Call DAT files, art, names, and other game assets remain the property of Microsoft/Turbine. This repository does not distribute them; users must supply their own retail installation.