As part of recent decisions regarding namespace
naming conventions, all namespaces will be changed
to snake case.
::Stats became ::statistics.
"statistics" was chosen over "stats" to avoid generating
conflicts with the already existing variables (there are
way too many "stats" in the codebase), which would make
this patch even more disturbing for the users.
Change-Id: If877b12d7dac356f86e3b3d941bf7558a4fd8719
Signed-off-by: Daniel R. Carvalho <odanrc@yahoo.com.br>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/45421
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
As part of recent decisions regarding namespace
naming conventions, all namespaces will be changed
to snake case.
sim_clock::Int became sim_clock::as_int.
"as_int" was chosen because "int" is a reserved
keyword, and this namespace acts as a selector of
how to read the internal variables.
Another possibility to resolve this would be to
remove the namespaces "Float" and "Int" and use
unions instead.
Change-Id: I65f47608d2212424bed1731c7f53d242d5a7d89a
Signed-off-by: Daniel R. Carvalho <odanrc@yahoo.com.br>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/45436
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Hoa Nguyen <hoanguyen@ucdavis.edu>
Maintainer: Gabe Black <gabe.black@gmail.com>
GPU MTYPE is currently set using a global config passed to the
PACoalescer. This patch enables MTYPE to be set by the shader on a
per-request bases. In real hardware, the MTYPE is extracted from a
GPUVM PTE during address translation. However, our current simulator
only models x86 page tables which do not have the appropriate bits for
GPU MTYPES. Rather than hacking non-x86 bits into our x86 page table
models, this patch instead keeps an interval tree of all pages that
request custom MTYPES in the driver itself. This is currently
only used to map host pages to the GPU as uncacheable, but is easily
extensible to other MTYPES.
Change-Id: I7daab0ffae42084b9131a67c85cd0aa4bbbfc8d6
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/42216
Maintainer: Matthew Poremba <matthew.poremba@amd.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
New topology ripped from Fiji to support dGPU. A dGPU flag is added to
the config which is propogated to the driver. The emulated driver is
now able to properly deal with dGPU ioctls and mmaps. For now, dGPU
physical memory is allocated from the host, but this is easy to change
once we get a GPU memory controller up and running.
Change-Id: I594418482b12ec8fb2e4018d8d0371d56f4f51c8
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/42214
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Use SConsopts files local to individual domains to pull
non-foundational build code out of SConstruct. This greatly simplifies
SConstruct, and also makes it easier to find build configuration having
to do with particular pieces of gem5.
This change also converts some python level variables, all_protocols,
protocol_dirs, and slicc_includes, into the environment where the timing
of their initialization is more flexible.
Change-Id: Ie61ceb75ae9e5557cc400603c972a9582e99c1ea
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/40872
Maintainer: Gabe Black <gabe.black@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Gabe Black <gabe.black@gmail.com>
This commit removes functions that indexed into the
vectors that held the operands. Instead, for-each loops
are used, iterating through one of 6 vectors
(src, dst, srcScalar, srcVec, dstScalar, dstVec)
that all hold various (potentially overlapping)
combinations of the operands.
Change-Id: Ia3a857c8f6675be86c51ba2f77e3d85bfea9ffdb
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/42212
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
This change removes the GPUDynInstPtr argument from
getRegisterIndex(). The dynamic inst was only needed
to get access to its parent WF's state so it could
determine the number of scalar registers the wave was
allocated. However, we can simply pass the number of
scalar registers directly. This cuts down on shared
pointer usage.
Change-Id: I29ab8d9a3de1f8b82b820ef421fc653284567c65
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/42210
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
The only thing left in isa_traits.hh are two constants, one for the
number of bytes in a page, and one for how far to shift an address to
get the page number. To make it clear that this is the only thing
isa_traits.hh should be used for from this point forward (until it is
entirely eliminated), this change renames it to the much less generic
page_size.hh.
Also, because isa_traits.hh used to have *much* more stuff in it, it was
included in a lot of places it didn't need to be. This change also
clears out all these legacy includes while updating the actually needed
ones to the new name.
Change-Id: I939b01b117c53d620b6b0a98982f6f21dc2ada72
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/40179
Reviewed-by: Gabe Black <gabe.black@gmail.com>
Maintainer: Gabe Black <gabe.black@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
dGPUs in all versions of ROCm and APUs starting with ROCM 2.2 can
under-allocate scratch resources. This patch adds support for
the CP to trigger a recoverable error so that the host can attempt to
re-allocate scratch to satisfy the currently stalled kernel.
Note that this patch does not include a mechanism to handle dynamic
scratch allocation for queues with in-flight kernels, as these queues
would first need to be drained and descheduled, which would require some
additional effort in the hsaPP and HW queue scheduler. If the CP
encounters this scenerio it will assert. I suspect this is not a
particularly common occurence in most of our applications so it is left
as a TODO.
This patch also fixes a few memory leaks and updates the old DMA callback
object interface to use a much cleaner c++11 lambda interface.
Change-Id: Ica8a5fc88888283415507544d6cc49fa748fe84d
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/42201
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
The systemc dir was not included in this fix.
First it was identified that there were only occurrences
at 0, 1, and 2 levels of indentation (and 2 of 2 spaces,
1 of 3 spaces and 2 of 12 spaces), using:
grep -nrE --exclude-dir=systemc \
"^ *enum [A-Za-z].* {$" src/
Then the following commands were run to replace:
<indent level>enum X ... {
by:
<indent level>enum X ...
<indent level>{
Level 0:
grep -nrl --exclude-dir=systemc \
"^enum [A-Za-z].* {$" src/ | \
xargs sed -Ei \
's/^enum ([A-Za-z].*) \{$/enum \1\n\{/g'
Level 1:
grep -nrl --exclude-dir=systemc \
"^ enum [A-Za-z].* {$" src/ | \
xargs sed -Ei \
's/^ enum ([A-Za-z].*) \{$/ enum \1\n \{/g'
and so on.
Change-Id: Ib186cf379049098ceaec20dfe4d1edcedd5f940d
Signed-off-by: Daniel R. Carvalho <odanrc@yahoo.com.br>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/43326
Reviewed-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Gabe Black <gabe.black@gmail.com>
Maintainer: Giacomo Travaglini <giacomo.travaglini@arm.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Due the compute unit pipeline being executed in reverse order, there
exists a scenario where a compute unit will execute an extra
instruction when it's supposed to be stopped at a barrier. It occurs
as follows:
* The ScheduleStage sets a barrier instruction ready to execute.
* The ScoreboardCheckStage adds another instruction to the readyList.
This is where the barrier is checked, but because the barrier isn't
executing yet, the instruction can be passed along to ScheduleStage
* The barrier executes, and stalls
* The ScheduleStage sees that there's a new instruction and schedules
it to be executed.
* Only now will the ScoreboardCheckStage realize a barrier is active
and stall accordingly
* The subsequent instruction executes
This patch sets the wavefront status to be S_BARRIER in ScheduleStage
instead of in the barrier instruction execution in order to have
ScoreboardCheckStage realize that we're going to execute a barrier,
preventing it from marking another instruciton as ready.
Change-Id: Ib683e2c68f361d7ee60a3beaf53b4b6c888c9f8d
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/41573
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Alexandru Duțu <alexandru.dutu@amd.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
We have a fail_if in attachDriver to prevent driver from being
overwritten. However, the fail_if only checks for if the driver
is not nullptr.
Previously, in some cases driver was set to garbage, which made
the fail_if trip the first time we were assigning the driver.
This patch explicitly sets driver to nullptr in the constructor, thus
ensuring that it will be nullptr the first time we call attachDriver
Change-Id: I325f6033e785025a912e3af3888c66cee0332f40
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/41973
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
The patch is using the newly defined PARAMS macro to replace
custom params() getters in derived class.
The patch is also removing redundant _params:
Instead of creating yet another _params field, SimObject descendants
should use params() to expose the real type of SimObject::_params they
already have.
Change-Id: I43394cebb9661fe747bdbb332236f0f0181b3dba
Signed-off-by: Alexander Klimov <Alexander.Klimov@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/39900
Reviewed-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Maintainer: Giacomo Travaglini <giacomo.travaglini@arm.com>
Tested-by: kokoro <noreply+kokoro@google.com>
The mismatches were from places where Params structs had been declared
as classes instead of structs, and ruby's MachineID struct.
A comment describing why the warning had been disabled said that it was
because of libstdc++ version 4.8. As far as I can tell, that version is
old enough to be outside the window we support, and so that should no
longer be a problem. It looks like the oldest version of gcc we
support, 5.0, corresponds with approximately libstdc++ version 6.0.21.
https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html#abi.versioning
Change-Id: I75ad92f3723a1883bd47e3919c5572a353344047
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/40953
Reviewed-by: Gabe Black <gabe.black@gmail.com>
Maintainer: Gabe Black <gabe.black@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Event creation and management support from emulated drivers is required
to support interruptible signals in HSA and this support was not
available. This changeset adds the event creation and management support
in the emulated driver. With this patch, each interruptible signal
created by the HSA runtime is associated with a signal event. The HSA
runtime can then put a thread waiting on a signal condition to sleep
asking the driver to monitor the event associated with that signal. If
the signal is modified by the GPU, the dispatcher notifies the driver
about signal value change. If the modifier is a CPU thread, the thread
will have to make HSA API calls to modify the signal and these API calls
will notify the driver about signal value change. Once the driver is
notified about a change in the signal value, the driver checks to see if
any thread is sleeping on that signal and wake up the sleeping thread
associated with that event. The driver has also implemented the time_out
wakeup that can wake up the thread after a certain time period has
expired. This is also true for barrier packets.
Each signal has an event address in a kernel managed and allocated
event page that can be used as a mailbox pointer to notify an event.
However, this feature used by non-CPU agents to communicate with the
driver is not implemented by this changeset because the non-CPU HSA
agents in our model can directly communicate with driver in our
implementation. Having said that, adding that feature should be trivial
because the event address and event pages are correctly setup by this
changeset and just adding the event page's virtual address to our PIO
doorbell interface in the page tables and registering that pio address
to the driver should be sufficient. Managing mailbox pointer for an
event is based on event ID and using this event ID as an index into
event page, this changeset already provides a unique mailbox pointer for
each event.
Change-Id: Ic62794076ddd47526b1f952fdb4c1bad632bdd2e
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/38335
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
This commit makes it so LGKM count is decremented in a single place
(after completeAcc), which fixes a couple of potential bugs
1. Data is only written by completeAcc, not after initiateAcc. LGKM
count is supposed to be decremented after data is written.
2. LGKM count is now properly decremented for atomics without return
Change-Id: Ic791af3b42e04f7baaa0ce50cb2a2c6286c54f5a
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/39396
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Matthew Poremba <matthew.poremba@amd.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
A prior commit (f6ec145fc0) fixed early LGKM decrementing for flat loads
and stores, but failed to address flat atomics.
Per the GCN3 ISA, LGKM count is decremented on flat atomics with return
when the data has been returned. This patch checks if the flat
instruction is an atomic with return, and decrements LGKM count if so.
Change-Id: I5c0c2c205a8b21327d4c42ba71c59842c15bd63b
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/39155
Reviewed-by: Matthew Poremba <matthew.poremba@amd.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matthew Poremba <matthew.poremba@amd.com>
Tested-by: kokoro <noreply+kokoro@google.com>
SimplePoolManager doesn't allow mapping of two WGs
simultaneously on the same Compute Unit (provided
the previous WG has been mapped to all the SIMDs)
even if there is sufficient VRF and SRF space
available.
DynPoolManager takes care of that by dynamically
allocating and deallocating register file space
to wavefronts
Change-Id: I2255c68d4b421615d7b231edc05d3ebb27cbd66c
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/32034
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Reviewed-by: Alexandru Duțu <alexandru.dutu@amd.com>
FLAT instructions used to decrement lgkm count on execute, while the
GCN3 ISA specifies that lgkm count should be decremented on data being
returned or data being written.
This patch changes it so that lgkm is decremented after initiateAcc (for
stores) and after completeAcc (for loads) to better reflect the ISA
definition.
This fixes a bug where waitcnts would be satisfied even though the
memory access wasn't completed, which lead to instructions using the
wrong data.
Change-Id: I596cb031af9cda8d47a1b5e146e4a4ffd793d36c
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/38696
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Matthew Poremba <matthew.poremba@amd.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
exp_cnt (expInstsIssued in the code) is used in the waitcnt instruction
to track that data has been read out of VGPRs in previous global
memory instructions, making it safe to overwrite the VGPRs used in said
global memory instructions.
Previously, exp_cnt wasn't being tracked at all, which lead to the
waitcnt finishing immediately, leading to the memory instruction's VPGRs
getting overwritten by subsequent instructions, causing errors.
This patch makes it so waitcnts waiting on exp_cnt will wait for MUBUF
buffer store instructions to read their VGPRs before completing
Change-Id: Idd2b59511bc086cf316217da27b7a228272b0b0f
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/37555
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Alexandru Duțu <alexandru.dutu@amd.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
HSA packet processor will now accept and process agent packets.
Type field in packet is command type.
For now:
AgentCmd::Nop = 0
AgentCmd::Steal = 1
Steal command steals the completion signal for a running kernel.
This enables a benchmark to use hsa primitives to send an agent
packet to steal the signal, then wait on that signal.
Minimal working example to be added in gem5-resources.
Change-Id: I37f8a4b7ea1780b471559aecbf4af1050353b0b1
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/37015
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Matthew Poremba <matthew.poremba@amd.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
