My parents are off to NY this week.. booyah, houseparty! Anyways they took off this morning and woke me up at 4am... no fun! Having oodles of time to waste...
Hi Tom, We had been telling you lately that GnuPG's performances were 4 times worse than OpenSSL, and that LTM seems to be not too far from OpenSSL. Actually,...
... <axelle_apvrille@y...> wrote: [snip timings] ... but ... Part of the problem as far as LTM is, is that GNUPG and OpenSSL are hardly generic toolkits. They...
... Yes, I know... on one hand this is bad because the code gets awful to read + less portable... but on the other hand their performance is better... ... ...
... Perhaps. But I never said LTM is as best as it could be [for ISO C]. Though you have to admit if it comes that close and the others are in hardcoded asm...
Sorry, I posted this answer yesterday night but I forgot I hadn't subscribed with that email... ... ___________________________________________________________...
Hi, I have worked on doing a few pre-computations for modular exponentiation: * the window size & its content, in bn_mp_exptmod_fast.c * dr in bn_mp_exptmod.c ...
Another interesting result: LTM and GnuPG WITHOUT assembly code are close in performance. Tests have been done on sample modular exponentiation. Tests #13 and...
... i=15 is probably because GnuPG will use CRT. Did you code your RSA via LTM to use CRT or just one straight exptmod? The rest of the nums seem about right....
... No. I wouldn't think so, because this is a signature verification, i.e a public key operation. To my understanding, you only use CRT on private key ...
... You lie. Liar! kiddin. hmm I dunno why GPG is faster if software only is used. Chances are there are still specific tweaks in the algos. One outstanding...
... GPG is using Karatsuba, not Montgomery. But I'm surprised, I would have thought Montgomery would be faster. Also, on my machine, it looks like GPG...
I've fixed several things in both LTM/LTC. In LTC I fixed a bug in rsa_exptmod where if you threw a random "key type" at it it would always just use the "d"...
Hi Tom Before you release next LTM you might want to fix these two small warnings that bcb 5.51 gave me: Warning W8004...
henrik_goldman@...
Nov 7, 2003 1:53 pm
22
... 64: 'str' is assigned a value that is never used in function mp_toradix Fixed this one. It was the *str++ = '\0'; NULL [or NIL as you may prefer :-)] byte...
Greg being the ultra-nice person he is has been plugging the LTM book around. Addison-Wesley turned the book down [too good for them I guess... ;-)] and...
I made a Cygwin makefile that will make a generic [no cygwin required to use the DLL] DLL from LibTomMath. In theory you could make an import lib and then use...
... required ... fine. ... which ... Hehehe, yeah LTC as a DLL is 222KB... it works great and MSVC can see the symbols in the dll [just don't know how to make...
Hmm addition... oh hi people. Just writing to say that I have added additional functions to ... oh ... anyways I updated the prime gen code in both LTC/LTM to...
Hi, I have been using (a derivative of) rsa_make_key to create an RSA keypair. When I attempted to use the key with another piece of software I discovered that...
Trevor Davel
TrevorD@...
Dec 3, 2003 12:36 am
30
Regarding the primes of being one bit longer the point was when you say "I have a k-bit prime" it should be a prime larger than 2^k not 2^ {k-1} as in the case...
Ok, I went a bit mad over the last couple of weeks. If you ever watch a cartoon called Invader Zim beware! So I have exams this week [one today in 24 mins]....
I sped up da hash code by making it able to compress data from source buffers. So if there are no bytes in the state and you pass it a buffer >= the block...
You mean that if I assemble a packet, encrypt it, call SHA-1 on it, and append the hash, that the packet buffer will have been altered? I don't consider this...
Greg Rose
ggr@...
Dec 7, 2003 4:55 pm
35
... and ... don't ... certainly ... Yup. Let me explain using SHA-1 as an example. If I give sha1_process() K bytes it will [using the old code] add those...
For the worried....tv_gen emits the same hash_tv.txt file with the new SHA-1 code as before [it tests messages from 0 to 128 bytes long, double the block...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
