Commit Graph

3916 Commits

Author SHA1 Message Date
whitequark 18c394976e runtime: disable the Nagle algorithm entirely.
See also commit feed91d; that commit fixed the test_rpc_timing test,
but caused frequent hangs elsewhere, which were also caused by buggy
Nagle implementation. Just disable this entirely, as with our
explicit buffering it provides no benefit anyway.
2016-11-13 00:33:24 +00:00
whitequark feed91d8b2 runtime: buffer RPC send packets.
This brings mean RPC time from ~45ms to ~2ms.

The cause of the slowness without buffering is, primarily, that lwip
is severely pessimized by small writes, whether with Nagle on or off.
(In fact, disabling Nagle makes it function *better* on many small
writes, which begs the question of what's the point of having Nagle
there in the first place.) In practical terms, the slowness appears
only when writing a 4-byte buffer (the synchronization segment);
writing buffers of other sizes does not trigger the problem.

This all is extremely confusing and the fix is partly palliative,
but since it seems to work reliably and we're migrating off lwip
I think it is unwise to spend any more time debugging this.
2016-11-12 23:06:33 +00:00
whitequark 3ce1826891 runtime: don't print debug messages to the UART.
It takes ~4ms to print an empty log line because of how slow
the UART is. This makes the log timestamps useless for debugging
performance problems.

After this commit, it takes ~75us to print an empty log line instead,
which pessimizes test_rpc_timing by less than 2ms with tracing
enabled.
2016-11-12 20:26:32 +00:00
whitequark acc5e53b32 runtime: print microsecond timestamps in debug messages. 2016-11-12 20:26:32 +00:00
whitequark dca3fb5c96 artiq_devtool: abort if build failed. 2016-11-12 20:26:32 +00:00
whitequark 7c2b1155ef conda: bump llvmlite-artiq dep. 2016-11-12 13:58:00 +00:00
whitequark 5eb940deb7 conda: bump llvmlite-artiq dep. 2016-11-12 04:09:34 +00:00
whitequark d3ee858d16 llvm_ir_generator: use !{→unconditionally.}invariant.load metadata.
This helps LICM, among other things.
2016-11-12 04:08:58 +00:00
whitequark 8b6418c604 artiq_devtool: more robust port forwarding. 2016-11-11 15:22:07 +00:00
Sebastien Bourdeauducq fbc2420443 setup: remove paramiko dependency (optional and developer-only) 2016-11-11 11:05:37 +08:00
whitequark 3b6cbb1f06 artiq_devtool: implement. 2016-11-10 20:25:15 +00:00
whitequark 0e76cbc414 artiq_compile: actually disable attribute writeback.
I wrote both halves of this condition but forgot to hook
them together.

Fixes #586.
2016-11-10 01:04:36 +00:00
whitequark 67e743d74a conda: use pythonparser 1.0. 2016-11-09 15:19:27 +00:00
whitequark 71f1d38860 conda: tighten pythonparser dependency (fixes #600). 2016-11-09 15:10:53 +00:00
whitequark eee8d05398 doc: clarify kernel_invariant doc (fixes #609). 2016-11-09 14:59:57 +00:00
David Nadlinger bfbdba9205 compiler: Emit all-kernel_invariant objects as LLVM constants
This enables constant propagation optimisations, as verified by
the included test case. This is only a first stop-gap measure, though;
we should support optimisation based on kernel invariants on a more
fine-grained level.
2016-11-09 14:50:48 +00:00
whitequark 124b257e05 conda: update for LLVM 3.9. 2016-11-08 23:24:03 +00:00
Sebastien Bourdeauducq 99ad9b5917 add has_dds, use config flags 2016-11-08 23:33:03 +08:00
whitequark ec8fe6f8bd Revert "Revert "Revert "Revert "Update for LLVM 3.9.""""
This reverts commit 7b81ed1d18.
2016-11-08 14:22:47 +00:00
whitequark 7b81ed1d18 Revert "Revert "Revert "Update for LLVM 3.9."""
This reverts commit 0d7688017b.
2016-11-08 12:58:20 +00:00
whitequark a8fd697d41 runtime: unbreak 453e8b7.
Running
  rustc --cfg 'foo="1"'
does not result in a statement of the form
  do_thing()
to be compilex in.
2016-11-08 12:57:06 +00:00
whitequark e0297039c9 artiq_run: fix bitrot in .ll/.bc runners. 2016-11-08 12:08:20 +00:00
whitequark 0d7688017b Revert "Revert "Update for LLVM 3.9.""
This reverts commit 5f5975844a.
2016-11-08 11:59:16 +00:00
whitequark 798a5f70df Revert "runtime: remove some redundant libm functions copied inline."
This reverts commit fee75bd50f.

LLVM does not have lround as a libcall and this inhibits LICM.
2016-11-08 11:54:17 +00:00
David Nadlinger 6e77f65d50 compiler: Clarify recv_rpc value names and documentation [nfc]
Previously, the phi emitted for the pointer parameter to recv_rpc was –
rather confusingly – called "size", and the pseudo-code in the comment
had bit-rotted.

Signed-off-by: David Nadlinger <code@klickverbot.at>
2016-11-07 14:38:21 +00:00
Sebastien Bourdeauducq 453e8b7eb3 runtime: support configurations without moninj, log or dds 2016-11-06 23:53:10 +08:00
Robert Jördens d158c69be0 phaser: fix frequency comment 2016-11-05 16:54:23 +01:00
David Nadlinger 7dcc987dd7 compiler: Fix break/continue targets in loop else blocks
Previously, they would still target the respective labels in the
just-exited loop.

Signed-off-by: David Nadlinger <code@klickverbot.at>
2016-11-05 02:35:01 +00:00
whitequark f102f2d4e6 manual: fix rustc invocation. 2016-11-04 09:11:45 +00:00
Sebastien Bourdeauducq 8ec73cb9ec dashboard: pack moninj widgets (#603) 2016-11-04 16:20:46 +08:00
Sebastien Bourdeauducq 9d58b4516c manual: Vivado is now default 2016-11-04 11:54:04 +08:00
whitequark 208f3f8568 manual: fix Rust installation instructions. 2016-11-04 03:35:59 +00:00
Sebastien Bourdeauducq 77bc247744 manual: add missing cd 2016-11-04 00:04:26 +08:00
Sebastien Bourdeauducq a252b88a36 manual: fix binutils patch download URL 2016-11-03 23:56:13 +08:00
whitequark b30734a105 runtime: fix a race condition with async RPCs.
session.rs has code like:

    while !rpc_queue::empty() {
        try!(process_kern_queued_rpc(stream, &mut session))
    }

    // A

    if mailbox::receive() != 0 {
        try!(process_kern_message(waiter, Some(stream), &mut session));
    }

If both an async and a mailbox RPC (async or large sync) are posted
at point A then they will be processed out of order.
This commit fixes the issue by flushing the async RPC queue before
posting any RPC to the mailbox.
2016-11-01 13:22:22 +00:00
whitequark 6fcd57a41a runtime: fix remaining async RPC bugs. 2016-11-01 10:33:57 +00:00
whitequark 2095d01b84 runtime: dirty hacks to remove allocations in ksupport. 2016-11-01 08:55:49 +00:00
whitequark c1e6d4b67c runtime: fix multiple async RPC bugs. 2016-11-01 06:51:44 +00:00
whitequark 636d4efe81 gateware: rewrite mailbox to use bits_for. 2016-11-01 06:28:43 +00:00
Sebastien Bourdeauducq 43cd970100 make set_dataset and mutate_dataset async RPCs 2016-11-01 11:11:41 +08:00
whitequark 18ae8d54a3 gateware: fix mailbox. 2016-11-01 02:33:00 +00:00
whitequark 898a716b91 runtime: work around mor1kx ignoring low bits of reset address.
Fixes #599.
2016-10-31 18:13:15 +00:00
whitequark 617e345d16 gateware: fix kernel CPU exec address. 2016-10-31 15:16:35 +00:00
Sebastien Bourdeauducq 2392113bb6 kc705: use misoc clock for false path 2016-10-30 11:16:04 +08:00
whitequark cd68577dbc compiler: add support for async RPCs. 2016-10-30 00:57:31 +00:00
whitequark 2ac85cd40f runtime: implement prototype background RPCs. 2016-10-29 21:34:25 +00:00
Sebastien Bourdeauducq c656a53532 kc705: clean up clock constraints 2016-10-29 21:28:01 +08:00
Sebastien Bourdeauducq ed4d57c638 use new Migen signal attribute API 2016-10-29 21:19:58 +08:00
Robert Jördens b14fcd41e4 ksupport: add ad9154* 2016-10-28 02:39:24 +02:00
Robert Jördens 2a1e529dcf phaser: DDS config dummies 2016-10-28 01:58:08 +02:00