E6010

Error · Topology/Hardware

E6010 — The working set placed in a memory space (the sum of its tiles) exceeds capacity. Downgraded to W1028 when the space is declared overcommit.

What the compiler reports

E6010 … (2 tiles) sums to 134217728 bytes

The fragments the test suite holds the compiler to. An ellipsis marks text the test does not constrain; the emitted message also carries a source location.

A program that triggers it

//
// Part of the Vx Project, under the Apache License v2.0 with LLVM Exceptions.
// See LICENSE for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//
// A reshape view adds nothing to the working set, but a second tensor placed in the same space
// still does: two 64 MiB tiles do not fit in 100 MiB.
//

Memory CPU_DRAM {}
Memory GPU_HBM {
  within: Memory::CPU_DRAM, capacity: 100 MiB, bandwidth: 300 GB/s, managed: cached, scope: device
}

fn fill<const R : i32, const C : i32>(t : &mut Tensor<f32, [R, C], Memory::GPU_HBM>) -> void {
  spawn on(Topology::GPU) {
    for r in 0..R {
      for c in 0..C {
        t[r][c] = 1.0;
      }
    }
  }
  return;
}

fn main() -> i32 {
  let h = Tensor<f32, [4096, 4096]>::new();
  let mut t = transfer(h, Memory::GPU_HBM);
  let mut v = t.reshape([16384, 1024]);
  let g = Tensor<f32, [4096, 4096]>::new();
  let mut u = transfer(g, Memory::GPU_HBM);
  fill<16384, 1024>(&mut v);
  fill<4096, 4096>(&mut u);
  return 0;
}

From tests/frontend/fail/second_tile_next_to_a_reshape_view.vx, which asserts this diagnostic on every commit.

Related