MaidSafe Dev Update - February 1, 2018

Still wants you to log in as a user :confused:

When you go to https://maidsafe.atlassian.net/browse/MAID-2475?jql=labels%20%3D%20dc-1 itā€™s asking you to log in?

If so, itā€™s not asking me to and has me wondering why. Itā€™s also showing the dc-1 items.

1 Like

works here [20]ā€¦

1 Like

Works now (20 characters)

1 Like

Thanks Viv! Thats good to know. I started looking at the JIRA project and noticed that 1000+ tasks have been already completed by the teamā€¦ That is huge effort! I see only 32 tasks that are open right now. Wondering if the team uses JIRA only for implementation tasks? I would expected much bigger backlog for a project at the stage MAIDSAFE is right now.

Thanks again!

1 Like

Iā€™ve made some improvements at FOSDEM:

30 Likes

ā€¦and until I read the Guardian article last weekI was blissfully ignorant of MaidSafe.

Iā€™m now trying to catch up!

12 Likes

Iā€™d say you have quite a lot of reading to do, some of which will send tingles down your spine! :slight_smile:

Starting with the new primer would be a good first stop if you havenā€™t read it already! https://maidsafe.net/docs/Safe%20Network%20Primer.pdf

7 Likes

Thanks for the link. I thought I was being quite thorough, but that PDF had evaded me.

7 Likes

Thereā€™s a web version too if the PDF takes too long to download.
https://primer.safenetwork.org/

8 Likes

Isnā€™t XSalsa20 many times slower than AES in the usual case when the latter is supported by hardware? Is XSalsa20 the choice because it is faster without acceleration? Iā€™m not sure if that is the case, just asking.

1 Like

It can be when you are talking about AES-NG side of things, where hardware acceleration can make a difference. AES-SIV is possibly a better choice (I am plying about with it) for more safety (nonce reuse etc. or at least IV handling). An issue we have with HA is the platform support really, that makes me wary, plus recently where things have been accelerated in hardware we have now spectre to consider as well. It is a minefield, so security first is probably a good thing.

In terms of NACL though there are choices in crypto and djb is really trying to ensure as much safety as possible first, but he seems very concerned with benchmarks as well. it is a pretty deep area but the xsalsa poly combination does appear to be pretty quick, I do not have the benchmarks to hand, but you can check them out.

For speed the balance to watch for here is bottleknecks, cpu bound, bandwidth bound etc. It is a constant consideration, the enc though here is client side so wont affect vaults as much, but always worth watching.

10 Likes

Thanks for shedding some light on the decision! The Salsa20 design is very simple, so maybe it will get hardware acceleration later. It seems it could be GPU accelerated, though that may come with security implications.

EDIT 2: ASIC/FGPA: http://www.engr.mun.ca/~howard/PAPERS/ccece07_yan.pdf

EDIT: GPU: https://www.researchgate.net/publication/266002588_New_Speed_Records_for_Salsa20_Stream_Cipher_Using_an_Autotuning_Framework_on_GPUs

Since the introduction of the CUDA programming model, GPUs are considered a viable platform for accelerating non-graphical applications. Many cryptographic algorithms have been reported to achieve remarkable performance speedups, especially block ciphers. For stream ciphers, however, the lack of reported GPU acceleration endeavors is due to their inherent iterative structures that prohibit parallelization. In this paper, we propose an efficient implementation methodology for data-parallel cryptographic functions in a batch processing fashion on modern GPUs in general and optimizations for Salsa20 in particular. We present an autotuning framework to reach the most optimized set of device and application parameters for Salsa20 kernel variants with throughput maximization as a figure of merit. The peak performance achieved by our implementation for Salsa20/12 is 2.7 GBps and 43.44 GBps with and without memory transfers respectively on NVIDIA GeForce GTX 590. These figures beat the fastest reported GPU implementation of any stream cipher in the eSTREAM portfolio including Salsa20/12, as well as the block cipher AES optimized by hand-tuning, and thus, to the best of our knowledge set a new speed record.

3 Likes

I only just got to thisā€¦
The banner reminding folk to check that their IP is updated is BRILLIANTā€¦
why did it take so long :wink:


or has it been around and I am late to the party :thinking:

11 Likes

Yeah, I also like these type of messages, that was recently introduced, thatā€™s the first version of Peruse to have such a message as part of the efforts to enhance the browser UX/UI.

14 Likes

great job, thanks :hugs:

3 Likes

Same here. Great reading.

1 Like