Topic: Sandy Bridge CPU's with AES Encryption in Hardware (Read 4396 times)
Have you been hittin' up silkroad ?
using GPU for same purposes, for example, promise 5x or 14x speedup improvement[in Nvidia and AMD case, resepectively], compared to percentages of boost, when its runned by unverified/undisclosed microcode wrapper.
laughable silicon amount waste on such "acceleration"
my conclusion: stay with software-based AES/SHA.
my conclusion: stay with software-based AES/SHA.
uh... really?
from tomshardware (on Networked Storage Devices that do AES encryption in SOFTWARE instead of HARDWARE):
Intel’s addition AES-NI to its 32 nm Clarkdale-based Core i5 desktop CPUs, six-core Gulftown processors, and second-gen Core i5 and Core i7 chips impressively demonstrates how much dedicated acceleration hardware can increase the speed of the encryption/decryption process.
-----
Nevertheless, it must be said that the encryption performance [of these NAS devices] leaves a lot of room for improvement. The implementation of a dedicated hardware cryptography unit would affect the data transfer rates very positively. Intel’s dual-core Atom D510 offers modest performance in everyday use, but for this type of encryption task, it is simply underwhelming, in turn affecting the data transfer rates. Maybe AES-NI has value in the embedded market; hopefully Intel has something planned there.
more likely this "AES Acceleration" is actually introduced to help 3-rd party to exploit vulnerabilities in [intentional or not. but expected anyway]flaws of implementation.
just watch on [laughable]silicon amount waste on such "acceleration" and look at full-scale CPU performance in such tasks. there is no miracles, yet in IT.
my conclusion: stay with software-based AES/SHA.
just watch on [laughable]silicon amount waste on such "acceleration" and look at full-scale CPU performance in such tasks. there is no miracles, yet in IT.
my conclusion: stay with software-based AES/SHA.
Is the AES encryption that's built into the latest Sandy Bridge chips similar enough to the SHA-256 that our miners are performing, that someone could write a miner specifically for the new intel chips?
No, but there is a shift left double instruction that is supposed to be good for SHA. (CPU mining is still going to be terrible.)
Is the AES encryption that's built into the latest Sandy Bridge chips similar enough to the SHA-256 that our miners are performing, that someone could write a miner specifically for the new intel chips?
Jump to: