Matthew Gast, the author of 802.11 Wireless Networks: The Definitive Guide, has written a good article for the O'Reilly Network describing the future of 802.11 authentication:
Anyone who bothers to set up a triangulation system would probably already be using MAC restriction or other security measures.
Calling MAC address filtering a "security measure" is really pushing it. Any decent driver will allow MAC address spoofing (even commercial xDSL gateways have that feature).
However, what would cause more than 3*25 instructions to be thrown away?
I've been told the Pentium 4 sports a 128-entry reorder buffer. Imagine a long latency branch (e.g. an indirect branch which misses in the cache) sitting at the top of the reorder buffer while the entire instruction window fills up. When we find out the branch was mispredicted, we'd have to squash 128 instructions. I think the CPU stalls when the reorder buffer is full, thus 128 is probably the upper bound. BTW, where did you get 25 from? I think the Pentium 4's pipeline is 20 stages long.
[...] that is 25 pipe stages times three instructions (max) for the P-III or 75 instructions worth of work [...]
The Pentium III was based on the P6 micro-architecture which featured a *14-stage* pipeline. Perhaps you were thinking of the P4? Another point: you are oversimplifying when you say that we'd need to squash 3*25 instructions on a misprediction, it could be more, it could be less.
Maybe more appropriate to suggest Sun and some people at HP (though not those working on McKinley) are jumping up and down with joy.
I don't think HP's thrilled, since they bet the farm on IA-64 and will retire PA-RISC ASAP.
Re:Intel relies on compiler, Turing says it's fool
on
Itanium Problems
·
· Score: 1
You make interesting points.
Do you have pointers regarding the amount of dynamic optimization in the IA-64? In other words, if the compiler in only run-of-the-mill, can the IA-64 still perform?
I haven't read it myself but you might want to give a look at the Reference Manual for Software Development and Optimization which "describes microarchitectural details of the Intel Itanium 2 processor, including cache hierarchies, memory management, and instruction execution latencies. It is targeted for developers of compilers and performance software".
I know a little bit about the Itanium because I wrote a paper arguing that we could cut down on the cost of context switches thanks to the Itanium's large address space, large TLB and page-level protection keys. It might bring microkernel research back to life.
Re:Intel relies on compiler, Turing says it's fool
on
Itanium Problems
·
· Score: 1
The Itanium relies heavily on exceedingly good compilers that will perform for the IA64 the same level of optimization that regular, on-the-fly predictive optimization do in RISC chips.
Take a look here. The original RISC paradigm relied heavily on efficient compilers. As a matter of fact, some people say RISC means Reject Important Stuff into Compiler.
He recognized that an appropriately defined set of machine instructions, program controls, and programs produced by a compiler--carefully designed to exploit the instruction set--could realize a very high performance processor with relatively few circuits.
One could say the Itanic is a true RISC architecture! The Pentium IV and its all-but-the-kitchen-sink approach is very far from the RISC philosophy.
While Vorbis supports true stereo encoding, it fakes 5.1-channel audio using a "joint-joint-stereo" method, where the left-back/left-front and right-back/right-front channels and joined together into the two stereo tracks in a similar fashion. Not very good at all.
Lossless channel coupling kicks in when you encode at quality 6 and above. Graham Mitchell covers the topic nicely in his Vorbis introduction.
Also, be advised that normal Ogg Vorbis files use lossy channel coupling, meaning that redundancies between the left and right channels are combined to save space. This does keep the files smaller, but also means that technically the stereo image of an Ogg Vorbis file might not always be identical to the original stereo image. If this concerns you, you'll want to encode at quality 6 or higher, which is where the lossy channel coupling is turned off and all channel coupling is lossless. Most can't tell the difference, but maybe you can.
Also, Xiph published a stereo-specific channel coupling discussion.
The truth is RCE is still in a 'testing' phase. The only studio which is really 'using' it is Columbia Tri-Star, and even they are only using it on a very small number of titles. The 2 Warner Bros titles on the list (both South Park) have RCE only because it was put on by original company producing the DVD and not WB and subsequent South Park DVDs have NOT had RCE. For now Columbia Tri-Star seems to have put RCE on hold, a number of recent high profile releases have NOT had it, but there has been no formal announcement on their future plans.
I recall the leaked MS documents. ISTR they were called the October papers or something like that where Bill gates and his cohorts saw the open source communal development projects as a serious threat.
The work was done for a paper in the Communications of the ACM (the paper itself is not online).
I doubted that statement as I'd read the paper online several days ago. I think it was linked to from Bruce Schneier's Crypto-Gram Newsletter. Anyway a simple Google search with homographic attack dns yields one and only one result:
Slashdot Mirror
A Technical Comparison of TTLS and PEAP
ZDNet also has a good overview of the proposed solutions.
Itanium Manuals
Itanium 2 Manuals
I haven't read it myself but you might want to give a look at the Reference Manual for Software Development and Optimization which "describes microarchitectural details of the Intel Itanium 2 processor, including cache hierarchies, memory management, and instruction execution latencies. It is targeted for developers of compilers and performance software".
I know a little bit about the Itanium because I wrote a paper arguing that we could cut down on the cost of context switches thanks to the Itanium's large address space, large TLB and page-level protection keys. It might bring microkernel research back to life.
The webpage linked to in the submission mentions you can forward your 419 spam to the US Secret Service Task Force. It wouldn't hurt to sent it to the Federal Trade Commission as well.
The link to the Trusted Computing Platform Alliance on the left is ominous. When Intel speaks of "building block" do they have Palladium in mind?
The truth is RCE is still in a 'testing' phase. The only studio which is really 'using' it is Columbia Tri-Star, and even they are only using it on a very small number of titles. The 2 Warner Bros titles on the list (both South Park) have RCE only because it was put on by original company producing the DVD and not WB and subsequent South Park DVDs have NOT had RCE. For now Columbia Tri-Star seems to have put RCE on hold, a number of recent high profile releases have NOT had it, but there has been no formal announcement on their future plans.
There are a few free BIOS projects. See Google's Web Directory.
Version 1008 was released on 22/02/2002. Version 1009 was released on 30/04/2002. The latest versions often appear first on the German FTP server.
The Homograph Attack