GPU Side Channel Attacks are Practical
 

This paper explains the problem:


[url]http://www.cs.ucr.edu/~zhiyunq/pub/ccs18_gpu_side_channel.pdf[/url]

Fixed, maybe.

[URL="https://nvidia.custhelp.com/app/answers/detail/a_id/4738/~/security-notice%3A-nvidia-response-to-%E2%80%9Crendered-insecure%3A-gpu-side-channel"]https://nvidia.custhelp.com/app/answers/detail/a_id/4738/~/security-notice%3A-nvidia-response-to-%E2%80%9Crendered-insecure%3A-gpu-side-channel[/URL]

[QUOTE=nomead;514929]Fixed, maybe.

[URL]https://nvidia.custhelp.com/app/answers/detail/a_id/4738/~/security-notice%3A-nvidia-response-to-%E2%80%9Crendered-insecure%3A-gpu-side-channel[/URL][/QUOTE]


very cool, however this addresses only NVIDIA.

[QUOTE=SELROC;514932]very cool, however this addresses only NVIDIA.[/QUOTE]


Proof of concept was on NVIDIA; using CUDA and OpenGl.
[url]https://www.tomshardware.co.uk/researchers-publish-side-channel-attacks-nvidia-gpu,news-59413.html[/url]
"The paper focused on Nvidia graphics cards, but the researchers also informed AMD and Intel of the vulnerabilities and will continue to investigate them."
...
"After informing AMD and Intel about the attacks, the researchers moved to investigate the possibility of these attacks working on Android smartphones."


Patches for eight vulnerabilities released: [url]https://www.tomshardware.co.uk/nvidia-graphics-card-security-vulnerabilities-drivers,news-60056.html[/url]

[QUOTE=kriesel;514944]Proof of concept was on NVIDIA; using CUDA and OpenGl.
[URL]https://www.tomshardware.co.uk/researchers-publish-side-channel-attacks-nvidia-gpu,news-59413.html[/URL]
"The paper focused on Nvidia graphics cards, but the researchers also informed AMD and Intel of the vulnerabilities and will continue to investigate them."
...
"After informing AMD and Intel about the attacks, the researchers moved to investigate the possibility of these attacks working on Android smartphones."


Patches for eight vulnerabilities released: [URL]https://www.tomshardware.co.uk/nvidia-graphics-card-security-vulnerabilities-drivers,news-60056.html[/URL][/QUOTE]




As I said: it only addresses NVIDIA.

Another more generic paper
 

[QUOTE=SELROC;514917]This paper explains the problem:


[URL]http://www.cs.ucr.edu/~zhiyunq/pub/ccs18_gpu_side_channel.pdf[/URL][/QUOTE]




* A Novel Side-Channel Timing Attack on GPUs *



[URL="http://delivery.acm.org/10.1145/3070000/3060462/p167-jiang.pdf?ip=84.253.145.166&id=3060462&acc=CHORUS&key=4D4702B0C3E38B35%2E4D4702B0C3E38B35%2E4D4702B0C3E38B35%2E6D218144511F3437&__acm__=1556431073_2c6e20a52e4c16f774bd59c9fc09fe60"]http://delivery.acm.org/10.1145/3070000/3060462/p167-jiang.pdf?ip=84.253.145.166&id=3060462&acc=CHORUS&key=4D4702B0C3E38B35%2E4D4702B0C3E38B35%2E4D4702B0C3E38B35%2E6D218144511F3437&__acm__=1556431073_2c6e20a52e4c16f774bd59c9fc09fe60[/URL]








More, but OpenCL related: Scalable and efficient implementation of correlation power analysis using graphics processing units (GPUs)


[url]https://www.researchgate.net/publication/266659555_Scalable_and_efficient_implementation_of_correlation_power_analysis_using_graphics_processing_units_GPUs[/url]