Comments (10)
Hi, that use_external_launcher flag has not been used since we published LumiBench. It was only used back when the simulator was first published and only supported the Intel driver. Could you run this in debug mode and show us where the simulator crashes when running GPGPU workloads?
from vulkan-sim.
Hi,
I source'ed environment setting with debug and run with LUD application in Rodinia (also used in lumibench paper). Here are the last a few line of output:
GPGPU-Sim PTX: finished parsing EMBEDDED .ptx file lud-rodinia-3.4.sm_62.ptx
GPGPU-Sim PTX: loading globals with explicit initializers...
GPGPU-Sim PTX: finished loading globals (0 bytes total).
GPGPU-Sim PTX: loading constants with explicit initializers... done.
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.1.sm_60.ptx
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.3.sm_60.ptx
GPGPU-Sim PTX: Kernel '_Z12lud_internalPfii' : regs=23, lmem=0, smem=2048, cmem=336
GPGPU-Sim PTX: Kernel '_Z13lud_perimeterPfii' : regs=32, lmem=0, smem=3072, cmem=336
GPGPU-Sim PTX: Kernel '_Z12lud_diagonalPfii' : regs=32, lmem=0, smem=1024, cmem=336
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.2.sm_62.ptx
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.4.sm_62.ptx
GPGPU-Sim PTX: Kernel '_Z12lud_internalPfii' : regs=23, lmem=0, smem=2048, cmem=336
GPGPU-Sim PTX: Kernel '_Z13lud_perimeterPfii' : regs=32, lmem=0, smem=3072, cmem=336
GPGPU-Sim PTX: Kernel '_Z12lud_diagonalPfii' : regs=32, lmem=0, smem=1024, cmem=336
GPGPU-Sim PTX: __cudaRegisterFunction _Z13lud_perimeterPfii : hostFun 0x0x563b94ccf580, fat_cubin_handle = 2
GPGPU-Sim PTX: __cudaRegisterFunction _Z12lud_diagonalPfii : hostFun 0x0x563b94ccf740, fat_cubin_handle = 2
WG size of kernel = 16 X 16
Generate input matrix internally, size =256
Creating matrix internally size=256
Could not find 0xc0000000 in Vulkan address map
gpgpusim: WARNING: Memory backing buffer not found for address 0xc0000000. This data write may be invalid
Segmentation fault (core dumped)```
from vulkan-sim.
I think as long as the program doesn't use any of the ray tracing functions we added, you can set the use_external_launcher flag to true and it should restore the original gpgpusim behavior.
vulkan-sim/src/cuda-sim/vulkan_ray_tracing.cc
Line 120 in dc742fa
from vulkan-sim.
Hi thanks for replying fast,
I also tried to change this flag. but it still doesn't work. log is attched.
GPGPU-Sim PTX: finished parsing EMBEDDED .ptx file lud-rodinia-3.4.sm_62.ptx
GPGPU-Sim PTX: loading globals with explicit initializers...
GPGPU-Sim PTX: finished loading globals (0 bytes total).
GPGPU-Sim PTX: loading constants with explicit initializers... done.
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.1.sm_60.ptx
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.3.sm_60.ptx
GPGPU-Sim PTX: Kernel '_Z12lud_internalPfii' : regs=23, lmem=0, smem=2048, cmem=336
GPGPU-Sim PTX: Kernel '_Z13lud_perimeterPfii' : regs=32, lmem=0, smem=3072, cmem=336
GPGPU-Sim PTX: Kernel '_Z12lud_diagonalPfii' : regs=32, lmem=0, smem=1024, cmem=336
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.2.sm_62.ptx
GPGPU-Sim PTX: Loading PTXInfo from lud-rodinia-3.4.sm_62.ptx
GPGPU-Sim PTX: Kernel '_Z12lud_internalPfii' : regs=23, lmem=0, smem=2048, cmem=336
GPGPU-Sim PTX: Kernel '_Z13lud_perimeterPfii' : regs=32, lmem=0, smem=3072, cmem=336
GPGPU-Sim PTX: Kernel '_Z12lud_diagonalPfii' : regs=32, lmem=0, smem=1024, cmem=336
GPGPU-Sim PTX: __cudaRegisterFunction _Z13lud_perimeterPfii : hostFun 0x0x55a1783b7580, fat_cubin_handle = 2
GPGPU-Sim PTX: __cudaRegisterFunction _Z12lud_diagonalPfii : hostFun 0x0x55a1783b7740, fat_cubin_handle = 2
WG size of kernel = 16 X 16
Generate input matrix internally, size =256
Creating matrix internally size=256
Segmentation fault (core dumped)```
from vulkan-sim.
Could you also attach the gdb backtrace? Thanks!
from vulkan-sim.
Attached, Thank you so much!!
(gdb) bt
#0 0x00007f7d1c49d9fb in std::__cxx11::list<stream_operation, std::allocator<stream_operation> >::begin (this=0x8)
at /usr/include/c++/9/bits/stl_list.h:942
#1 0x00007f7d1c49d875 in std::__cxx11::list<stream_operation, std::allocator<stream_operation> >::front (this=0x8)
at /usr/include/c++/9/bits/stl_list.h:1111
#2 0x00007f7d1c49d6f6 in CUstream_st::front (this=0x0) at stream_manager.h:235
#3 0x00007f7d1c49c911 in stream_manager::register_finished_kernel (this=0x55e6c3f28210, grid_uid=3287445632) at stream_manager.cc:293
#4 0x00007f7d1c49c8c6 in stream_manager::register_finished_kernel (this=0x55e6c3f28210) at stream_manager.cc:286
#5 0x00007f7d1c48dddb in gpgpu_sim_thread_concurrent (ctx_ptr=0x55e6c3f0e190) at gpgpusim_entrypoint.cc:155
#6 0x00007f7d1b6e4609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#7 0x00007f7d1ba39353 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
from vulkan-sim.
Hi, thanks for sharing all the details! There were a few changes that we made to run Rodinia workloads which weren't included with the Vulkan-Sim release. I will sort out those changes and get back to you in a couple of days :)
from vulkan-sim.
Hi @lucylufei ,
Thank you so much!
Have a good weekend!
from vulkan-sim.
Hi @Yfeng-44, I've pushed the changes to a separate cuda branch. The use_external_launcher
variable has been moved into the gpgpusim.config
file so please add -gpgpu_rt_external_launch 1
to the configs when running CUDA applications. You can change this back to -gpgpu_rt_external_launch 0
for Vulkan applications. You may also need to modify the -gpgpu_max_simulated_rt_kernels
parameter to simulate multiple kernel launches on CUDA applications.
Please let me know if you still encounter any issues.
from vulkan-sim.
Hi @lucylufei ,
Thanks for the efforts! I will try the change.
Will close this issue for now, and will let you know if there's any issues.
Thanks!
from vulkan-sim.
Related Issues (20)
- RTV6: Driver does not support the 0x46a6 PCI ID HOT 3
- Run./ray tracer-scene 6-height 320-width 448 "Trace dumped" does not appear. HOT 1
- Generating frames using Vulkan Sim HOT 1
- Segmentation fault while running vulkan-sim 2.0.0 HOT 10
- Wrong Descriptor Set index HOT 3
- Scene 12 (PartyTug) generates an empty image HOT 4
- Running Lumibench benchmarks for --width and --height greater than 128 HOT 1
- Scene numbers mismatch in Lumibench HOT 2
- Error while running Lumibench HOT 7
- Bug in NIR to PTX translation? HOT 1
- Running provided traces on cluster CPUs without a graphics card HOT 2
- Running multiple jobs in parallel? HOT 2
- Number of pipeline stage mismatch for other GPU configurations
- Incorrect selection of closest hit primitive
- Do RT cores use general purpose ldst unit? HOT 2
- Program Aborts or Gets killed with Modified Parameters HOT 5
- Question BVH properties HOT 2
- The number of pixel data is inconsistent with the expecting amount
- Whether to support coherent engine? HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from vulkan-sim.