sakjain92 / fractional-gpus Goto Github PK
View Code? Open in Web Editor NEWSplits single Nvidia GPU into multiple partitions with complete compute and memory isolation (wrt to performace) between the partitions
Splits single Nvidia GPU into multiple partitions with complete compute and memory isolation (wrt to performace) between the partitions
Build env: ubuntu16(4.15.0-142-generic), cuda9.1, modified nvidia driver, cmake 3.22, gcc/g++ 5 or 6
I used scripts/evlaution.sh to run as Reverse Engineering mode. Got the following prompt:
FGPU:Couldn't open shmem
fgpu_init() failed
Init failed
Error: Reverse engineering code failed
After reading doc/FAQ.md , I tried to build with
"FGPU_COMP_COLORING_ENABLE=ON" "FGPU_MEM_COLORING_ENABLED=ON" "FGPU_TEST_MEM_COLORING_ENABLED=ON"
or
"FGPU_COMP_COLORING_ENABLE=ON" "FGPU_MEM_COLORING_ENABLED=OFF" "FGPU_TEST_MEM_COLORING_ENABLED=ON"
But neither works correctly.
Can you tell me how I should solve this problem in order to make the project work properly?
Dear author(s),
I run this project to reverse-engineer the DRAM Hash functions of GTX 1060 3GB. The result is only one hash function:
Hash Function for DRAM Banks:
Bit(10) ^ Bit(13) ^ Bit(18) ^ Bit(20) ^ Bit(21) ^ Bit(23) ^ Bit(24) ^ Bit(26) ^ Bit(28) ^ Bit(30) ^ Bit(31)
I wonder how to get other bits of bank index through this code.
Thanks in advance.
Hi,
I try to porting your modification for the uvm driver for the 390 version nvidia driver to the 418.87.00, but I encounter fgpu_server reporting it cannot get color info error.
I add some printk inside the nvidia-uvm driver, and find get color ioctl never issued at all. With adding more printk, I find the ioctl seems being hijacked by MPS server, not really issued by the fgpu_server itself. And when fgpu_server try to issue the UVM_GET_DEVICE_COLOR_INFO, MPS hijack it, and issue UVM_UNMAP_EXTERNAL_ALLOCATION and UVM_FREE.
For MPS is closed source, I don't know why it happen. I wonder whether you encounter similar issue when doing this work over 390 version driver.
If I apply your repo, can I divide the GPU like this and check it on nvidia-smi?
like this: https://docs.run.ai/Researcher/Walkthroughs/walkthrough-fractions/
The Paper says " we use the Nvidia MPS and bypass this limitation without the need to merge applications", does it mean that FGPU only support Nvidia Volta architecture?
Dear Dr. Jain,
I trying to print out GPU physical address for reverse engineering. Could you elaborate what modification on driver you did for print out GPU physical address? Like what files you changed, and what functions you added in the driver?
Thanks for your help.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.