gpu-parity: pin check requires a clean packages/cultmath and non-empty SHAs; README matches the run
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
This commit is contained in:
parent
d1ae7ae1e3
commit
2d30db03e7
2 changed files with 6 additions and 4 deletions
|
|
@ -32,7 +32,9 @@ internal static class Program
|
|||
string head = Git(cultlib, "rev-parse HEAD");
|
||||
string pin = Git(cultlib, $"rev-parse {PinTag}^{{commit}}");
|
||||
Console.WriteLine($"CultLib {cultlib}\n HEAD {head}\n {PinTag} {pin}");
|
||||
if (head != pin) { Console.Error.WriteLine("PIN FAILED: CultLib HEAD is not the pinned tag."); return 3; }
|
||||
string dirty = Git(cultlib, "status --porcelain -- packages/cultmath");
|
||||
if (head.Length == 0 || head != pin || dirty.Length != 0)
|
||||
{ Console.Error.WriteLine("PIN FAILED: CultLib HEAD is not the pinned tag, or packages/cultmath is modified."); return 3; }
|
||||
|
||||
var pts = BuildPoints(out int tieCount);
|
||||
Console.WriteLine($"points: {pts.Length / 4} ({tieCount} tie-constructed, {pts.Length / 4 - tieCount} seeded random)");
|
||||
|
|
|
|||
|
|
@ -18,9 +18,9 @@ dotnet run -c Release --project tools/gpu-parity/GpuParity.csproj
|
|||
```
|
||||
|
||||
It compiles the kernel twice (default flags, where FMA contraction is allowed, and `/Gis` IEEE strict),
|
||||
dispatches once per flag (about 130k threads), and prints max abs / max relative error per function,
|
||||
dispatches once per flag (about 164k threads), and prints max abs / max relative error per function,
|
||||
split into tie-constructed points and a seeded random sweep. Exit code 0 means within `Tolerance`
|
||||
(in `Program.cs`) and no branch mismatches.
|
||||
(in `Program.cs`). Branch mismatches are reported, not failed.
|
||||
|
||||
Points: dyadic coordinates on the simplex tie planes (`x0.x == x0.y`, `y == z`, `x == z`, offsets
|
||||
k in -2..2) with 0, 1 and 2 ulp neighbours, plus 3 x 40000 seeded uniform points at scales 4, 64, 1024.
|
||||
|
|
@ -55,7 +55,7 @@ the first steps differently), not with the noise: 16x more input magnitude gave
|
|||
the first two tiers. Single GPU and driver; a different driver may need the bounds re-measured.
|
||||
|
||||
Tie handling: under both flag sets the GPU took a different simplex ordering branch than C# at
|
||||
190 of 35814 tie points (float3) and 39 (float2) with |p| <= 8, 2 and 4 in the middle tier, 0 above
|
||||
190 of 35814 tie points (float3) and 39 (float2) with |p| <= 8, 2 and 4 in the middle tier, 0 above among tie points (the random sweep above 128 had 1 float3 mismatch)
|
||||
(the same counts under `/Gis`, so it is not FMA contraction; the probe is a separate expression tree,
|
||||
so it is evidence of a GPU/CPU rounding difference in x0, not of the noise body's own path). Every one of
|
||||
those points is inside the value and gradient tolerance: the noise is continuous across the tie, so a
|
||||
|
|
|
|||
Loading…
Reference in a new issue