From Andrew.Rogers@cantab.net Sat, 1 Jun 2002 10:15:35 +0100 (BST) Date: Sat, 1 Jun 2002 10:15:35 +0100 (BST) From: Andrew Rogers Andrew.Rogers [at] cantab [dot] net Subject: A conclusion? [CST-2] Types: Ex 4.3.2 On Fri, 31 May 2002, Matej Pfajfar wrote: > > let x = ((\x.x) (\x.x)) in (x 5) > > > > Since A != {} and M1 is an application... > > I think this types OK in both? Maybe it bears explanation why this doesn't work and the example I gave does. The (letv) rule may be applied when either: i) M1 is a value, or ii) A = {} Thus we may _not_ use (letv) when: i) M1 is not a value, _and_ ii) M1's type is not a ground type (i.e. A != {}) But |- (\x.x)(\x.x) : int -> int in our extended ML system, so that A = {} and |- let x = ((\x.x)(\x.x) in (x 5) : int. So to generate a counter-example, we need both M1 not to be a value (e.g. an application as above) and we need M1 not to be a ground type, i.e. it has to be used polymorphically. The only way to do this is for M1 to be a function as was suggested, but it must be applied to (at least) two different types, otherwise M1 may be simply typed at a ground type using (var >), thus giving A = {}. I was looking through my Types notes when I got back in yesterday evening, and as Pitts says, the introduction of (letv) is not a significant hindrance to the use of ML, because we can turn an application into a lambda-extraction by visiting our favourite takeaway...*, e.g. ((\x.x)(\x.x)) => \y.((\x.x)(\x.x))y so that my example: let x = ((\x.x)(\x.x)) in (let y = x(\y.y) in x true) becomes: let x = \a.((\x.x)(\x.x))a in (let y = \b.(x(\y.y))b in x true) Which is typeable in both ML and ML+ref. I also happened to note down Pitt's own example, as he seems to love bool lists... let f = (\x.x)(\x.x) in (f true)::(f nil) which is of course typeable as bool list in his ML but not ML+ref. The fix is, as above: let f = \y.(\x.x)(\x.x)y in (f true)::(f nil) so that M1 is a value. I hope this may help some people understand at least some of the threads on this list... |- cst2 : long list Yours, Andrew *: I'm a CompSci, I have to make jokes about Curry... -- Andrew D. Rogers | | "For it is by grace you have 126 Huntingdon Road | Phone: 01223 523198 | been saved, through faith.. Cambridge | ICQ: 65332411 | it is the gift of God" CB3 OHL | | Eph 2:8-9 From na238@hermes.cam.ac.uk Sat, 1 Jun 2002 11:06:44 +0100 (BST) Date: Sat, 1 Jun 2002 11:06:44 +0100 (BST) From: N. Abedin na238@hermes.cam.ac.uk Subject: [CST-2] Advanced Graphics Was the section on radiosity covered in lectures?(handwritten notes) How much detail did he go into regarding tensor products for bezier/bspline surfaces, and vector transformations for ray tracing arbitrarily positioned objects? Cheers for any help Naz From jrs53@cam.ac.uk Sat, 1 Jun 2002 12:05:40 +0100 Date: Sat, 1 Jun 2002 12:05:40 +0100 From: James Srinivasan jrs53@cam.ac.uk Subject: [CST-2] Advanced Graphics > Was the section on radiosity covered in lectures?(handwritten notes) Yes, a whole lecture > How much detail did he go into regarding tensor products for > bezier/bspline surfaces, Not much beyond mentioning they existed and giving the formulae > and vector transformations for ray tracing > arbitrarily positioned objects? Covered that in some detail - all in the notes I think, James From jdjs2@cam.ac.uk Sat, 1 Jun 2002 15:26:24 +0100 Date: Sat, 1 Jun 2002 15:26:24 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] Advanced Graphics In http://www.cl.cam.ac.uk/tripos/y2001p7q9.pdf part (b) he asks for the intersection points of a ray with a cone. I would like to check that my understanding is correct: 1) substitute ray equation into x^2 = z^2 + y^2 (eq of a cone) 2) simplify to quadratic, solve quadratic for t 3) gives 0 real solutions => no intersections 1 repeated real soln => ray scrapes the edge of the cone 2 real solns => two intersection points if denominator = 0 then ray is along x-axis so no intersections (** in my lecture notes I've got something written about possibly having up to four intersection points if you are allowed the two sided cone - how is this possible if you are solving a quadratic?? **) 4) check values of t are >=0 else discard (otherwise behind eye) 5) check that the x values are within the range, if not, discard. Then for part (ii) he asks for closed cone stuff: 4) if you have 2 intersection points t_i and t_j, if either straddle the x values of either of the two end caps (total of four cases) then intersect with end-cap(s) instead. If you have 1 intersection point only then it can't have cross through an end-cap ??? 5) discard negative t values. So I can't see what the special cases are...? Finally, does the normal vector work out to be (-(y^2+x^2), x, x) normalised of course? Cheers, Jamie From cmn26@cam.ac.uk Sat, 1 Jun 2002 15:34:24 +0100 Date: Sat, 1 Jun 2002 15:34:24 +0100 From: Chris North cmn26@cam.ac.uk Subject: [CST-2] Advanced Graphics You have the normal wrong, although I agree with most of the other stuff you've written. The normal is actually: (-2x,2y,2z) (calculate del(y^2+z^2-x^2)) I got a big tick for that anyway..... -- Chris North This address will soon be terminated, please use norfus@cantab.net instead :) From cmn26@cam.ac.uk Sat, 1 Jun 2002 15:35:47 +0100 Date: Sat, 1 Jun 2002 15:35:47 +0100 From: Chris North cmn26@cam.ac.uk Subject: [CST-2] Advanced Graphics Oh yeah, and for some special cases, imagine what happens if you start inside the cone...... -- Chris North This address will soon be terminated, please use norfus@cantab.net instead :) From cia20@cam.ac.uk Sat, 1 Jun 2002 15:43:59 +0100 Date: Sat, 1 Jun 2002 15:43:59 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics > You have the normal wrong, although I agree with most of the > other stuff > you've written. > > The normal is actually: > > (-2x,2y,2z) The x-component normal of the surface of the cone does not depend on x at all, surely - it's constant...something like root 2 from what i remember. Also you should consider the normals at the endcaps, but these are trivial: (-1,0,0) and (1,0,0) Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request From tp225@cam.ac.uk Sat, 1 Jun 2002 15:58:57 +0100 Date: Sat, 1 Jun 2002 15:58:57 +0100 From: Tom Puverle tp225@cam.ac.uk Subject: [CST-2] Neural Computing I got myself a bit confused about Exercise 9 in the worked examples for NC. What is the step that takes you from the second equation to the third? (i.e.) as it says: normalising and cancelling out... Thanks, Tom From cmn26@cam.ac.uk Sat, 1 Jun 2002 15:57:41 +0100 Date: Sat, 1 Jun 2002 15:57:41 +0100 From: Chris North cmn26@cam.ac.uk Subject: [CST-2] Advanced Graphics No, the normal DOES depend on x. For instance, what do you think happens if the intersection point changes from x=1 to x=-1?? It will point in opposite directions with respect to x is what. Therefore there has to be a dependence. HTH -- Chris North This address will soon be terminated, please use norfus@cantab.net instead :) From mp292@cam.ac.uk Sat, 1 Jun 2002 16:03:31 +0100 (BST) Date: Sat, 1 Jun 2002 16:03:31 +0100 (BST) From: Matej Pfajfar mp292@cam.ac.uk Subject: [CST-2] Advanced Graphics > No, the normal DOES depend on x. For instance, what do you think happens if > the intersection point changes from x=1 to x=-1?? It will point in opposite > directions with respect to x is what. Therefore there has to be a dependence. Yeah but x is given as positive in the question ... > > HTH > > -- > Chris North > This address will soon be terminated, please > use norfus@cantab.net instead :) > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > -- Matej Pfajfar St John's College, University of Cambridge, UK GPG Public Keys @ http://matejpfajfar.co.uk/keys WARNING: THIS E-MAIL ACCOUNT WILL BE DELETED ON 15/07/2002. PLEASE USE mp@cantab.net. From cia20@cam.ac.uk Sat, 1 Jun 2002 16:05:41 +0100 Date: Sat, 1 Jun 2002 16:05:41 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics Good point, but there is no difference between the x-component of the normal at x=1 and x=2 (which is what I was trying to get). Which side of x=0 the point lies will change the sign of the component, but not its magnitude. Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Chris North > Sent: 1 June 2002 15:58 > To: cst-2@srcf.ucam.org > Subject: RE: [CST-2] Advanced Graphics > > > No, the normal DOES depend on x. For instance, what do you > think happens if > the intersection point changes from x=1 to x=-1?? It will > point in opposite > directions with respect to x is what. Therefore there has to > be a dependence. > > HTH > > -- > Chris North > This address will soon be terminated, please > use norfus@cantab.net instead :) > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From jdjs2@cam.ac.uk Sat, 1 Jun 2002 16:07:18 +0100 Date: Sat, 1 Jun 2002 16:07:18 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] Advanced Graphics Are you sure: Imagine a normal vector close the origin and the direction it is pointing. Then move out to say 100 * the original x. If x doesn't change, but the normal does depend on y and z then the normal vector will point in a much more outwards direction than before, and that is clearly wrong as it needs to point the same way regardless of x. J > > Good point, but there is no difference between the > x-component of the normal at x=1 and x=2 (which is what I was > trying to get). > > Which side of x=0 the point lies will change the sign of the > component, but not its magnitude. > > Cheerio, > > Chris > do something lastminute.work > > Chris Applegate > Room X6, Corpus Christi College, Cambridge, CB2 1RH > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > ICQ 41706821 PGP key available on request > > > > > -----Original Message----- > > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > > Behalf Of Chris North > > Sent: 1 June 2002 15:58 > > To: cst-2@srcf.ucam.org > > Subject: RE: [CST-2] Advanced Graphics > > > > > > No, the normal DOES depend on x. For instance, what do you think > > happens if the intersection point changes from x=1 to > x=-1?? It will > > point in opposite > > directions with respect to x is what. Therefore there has to > > be a dependence. > > > > HTH > > > > -- > > Chris North > > This address will soon be terminated, please > > use norfus@cantab.net instead :) > > > > > > _______________________________________________ > > CST-2 mailing list > > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From cia20@cam.ac.uk Sat, 1 Jun 2002 16:19:21 +0100 Date: Sat, 1 Jun 2002 16:19:21 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics The big perpendicular vector from the surface will be something like (-x, y ,z), so that does depend on x. But when you normalise it to unit size by dividing by root(x^2 + y^2 + z^2) = root(2x^2) then the x term cancels to become 1/root(2)...thus constant...and the others become y/x*root(2) and z/x*root(2). Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Jamie Shotton > Sent: 1 June 2002 16:07 > To: cst-2@srcf.ucam.org > Subject: RE: [CST-2] Advanced Graphics > > > > > Are you sure: > Imagine a normal vector close the origin and the direction it is > pointing. Then move out to say 100 * the original x. If x doesn't > change, but the normal does depend on y and z then the normal vector > will point in a much more outwards direction than before, and that is > clearly wrong as it needs to point the same way regardless of x. > > J > > > > > > Good point, but there is no difference between the > > x-component of the normal at x=1 and x=2 (which is what I was > > trying to get). > > > > Which side of x=0 the point lies will change the sign of the > > component, but not its magnitude. > > > > Cheerio, > > > > Chris > > do something lastminute.work > > > > Chris Applegate > > Room X6, Corpus Christi College, Cambridge, CB2 1RH > > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > > ICQ 41706821 PGP key available on request > > > > > > > > > -----Original Message----- > > > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > > Behalf Of Chris North > > Sent: 1 June 2002 15:58 > > To: cst-2@srcf.ucam.org > > Subject: RE: [CST-2] Advanced Graphics > > > > > > No, the normal DOES depend on x. For instance, what do you think > > happens if the intersection point changes from x=1 to > x=-1?? It will > > point in opposite > > directions with respect to x is what. Therefore there has to > > be a dependence. > > > > HTH > > > > -- > > Chris North > > This address will soon be terminated, please > > use norfus@cantab.net instead :) > > > > > > _______________________________________________ > > CST-2 mailing list > > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > _______________________________________________ CST-2 mailing list CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 From cia20@cam.ac.uk Sat, 1 Jun 2002 16:27:45 +0100 Date: Sat, 1 Jun 2002 16:27:45 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics So Northus' earlier solution was quite correct, and I was just getting confused by remembering my solution had a constant in it as I'd normalised it - sorry. Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Chris Applegate > Sent: 1 June 2002 16:19 > To: cst-2@srcf.ucam.org > Subject: RE: [CST-2] Advanced Graphics > > > The big perpendicular vector from the surface will be something like > (-x, y ,z), so that does depend on x. > > But when you normalise it to unit size by dividing by root(x^2 + y^2 + > z^2) = root(2x^2) then the x term cancels to become 1/root(2)...thus > constant...and the others become y/x*root(2) and z/x*root(2). > > Cheerio, > > Chris > do something lastminute.work > > Chris Applegate > Room X6, Corpus Christi College, Cambridge, CB2 1RH > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > ICQ 41706821 PGP key available on request > > > -----Original Message----- > > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > > Behalf Of Jamie Shotton > > Sent: 1 June 2002 16:07 > > To: cst-2@srcf.ucam.org > > Subject: RE: [CST-2] Advanced Graphics > > > > > > > > > > Are you sure: > > Imagine a normal vector close the origin and the direction it is > > pointing. Then move out to say 100 * the original x. If x doesn't > > change, but the normal does depend on y and z then the normal vector > > will point in a much more outwards direction than before, > and that is > > clearly wrong as it needs to point the same way regardless of x. > > > > J > > > > > > > > > > Good point, but there is no difference between the > > > x-component of the normal at x=1 and x=2 (which is what I was > > > trying to get). > > > > > > Which side of x=0 the point lies will change the sign of the > > > component, but not its magnitude. > > > > > > Cheerio, > > > > > > Chris > > > do something lastminute.work > > > > > > Chris Applegate > > > Room X6, Corpus Christi College, Cambridge, CB2 1RH > > > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > > > ICQ 41706821 PGP key available on request > > > > > > > > > > > > > -----Original Message----- > > > > From: cst-2-admin@srcf.ucam.org > [mailto:cst-2-admin@srcf.ucam.org]On > > > Behalf Of Chris North > > > Sent: 1 June 2002 15:58 > > > To: cst-2@srcf.ucam.org > > > Subject: RE: [CST-2] Advanced Graphics > > > > > > > > > No, the normal DOES depend on x. For instance, what do you think > > > happens if the intersection point changes from x=1 to > > x=-1?? It will > > > point in opposite > > > directions with respect to x is what. Therefore there has to > > > be a dependence. > > > > > > HTH > > > > > > -- > > > Chris North > > > This address will soon be terminated, please > > > use norfus@cantab.net instead :) > > > > > > > > > _______________________________________________ > > > CST-2 mailing list > > > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 > _______________________________________________ CST-2 mailing list CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 _______________________________________________ CST-2 mailing list CST-2@srcf.ucam.org http://www.srcf.ucam.org/mailman/listinfo/cst-2 From pjt33@cam.ac.uk Sat, 01 Jun 2002 16:50:41 +0100 Date: Sat, 01 Jun 2002 16:50:41 +0100 From: pjt33@cam.ac.uk pjt33@cam.ac.uk Subject: [CST-2] Advanced Graphics > 1) substitute ray equation into x^2 = z^2 + y^2 (eq of a cone) I think it should be x^2 = k(y^2 + z^2) because he doesn't specify "unit cone" (or whatever you call it). > (** in my lecture notes I've got something written about possibly having > up to four intersection points if you are allowed the two sided cone - > how is this possible if you are solving a quadratic?? **) End-caps. > 4) if you have 2 intersection points t_i and t_j, if either straddle the > x values of either of the two end caps (total of four cases) then > intersect with end-cap(s) instead. > If you have 1 intersection point only then it can't have cross > through an end-cap ??? > > So I can't see what the special cases are...? Wrong. Consider (here goes with the ASCII art!) \ |/ \ / \/| /\| / \ / |\ /----+-\ / | \ / | \ /-------+----\ / | \ The intersection points straddle neither end-cap, but the ray intersects both end-caps. The only special case is that you _don't_ need to check the end-caps when both intersections are within the range. > Finally, does the normal vector work out to be (-(y^2+x^2), x, x) > normalised of course? I made it (-sqrt(y^2+z^2), y, z) for k=1 (which agrees with consensus), and argued that you scale homogenous vector by matrix [1 0 0 0] -1 [0 1/sqrt(k) 0 0] [0 0 1/sqrt(k) 0] [0 0 0 1] for general case. Plus normals for end-caps, of course. Peter ===== Peter Taylor pjt33@cam.ac.uk From jdjs2@cam.ac.uk Sat, 1 Jun 2002 17:17:20 +0100 Date: Sat, 1 Jun 2002 17:17:20 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] Advanced Graphics > Wrong. Consider (here goes with the ASCII art!) > [snip] Yes but in that example, it actually DOES have two intersection points, one of which happens to be negative, but at that stage I've not removed this. But is this necessarily the case for all other examples? J From pjt33@cam.ac.uk Sat, 01 Jun 2002 17:42:46 +0100 Date: Sat, 01 Jun 2002 17:42:46 +0100 From: pjt33@cam.ac.uk pjt33@cam.ac.uk Subject: [CST-2] Advanced Graphics > Yes but in that example, it actually DOES have two intersection points, > one of which happens to be negative, but at that stage I've not removed > this. But is this necessarily the case for all other examples? I'm not sure what you mean by "all other examples". To take one, > If you have 1 intersection point only then it can't have cross > through an end-cap ??? \ / / \ / / \/ / /\ / / \/ / /\ /---/--\ / / \ / / \ /---/--------\ / / \ Peter ===== Peter Taylor pjt33@cam.ac.uk From gbmv2@cam.ac.uk Sat, 1 Jun 2002 22:28:30 +0100 Date: Sat, 1 Jun 2002 22:28:30 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] Comp. Arch 1998.7.3 - backwards compatibility problems "Older computer architectures often contain features that hamper high-performance implementation using modern techniques. Describe several such features, explaining how they affect implementation; describe what techniques (if any) have been developed to help overcome these difficulties." [14 marks] So far, I have the following: o Small register file (x86): introduces extra data dependencies into code, causing more pipeline stalls. Superscalar architectures cannot work at all with a small number of registers, so CPI is limited to >1. Overcome by register renaming: undo register colouring, spreading data among many registers, to exploit ILP and enable OoO execution. o Specialised instructions (VAX POLY, memory indirect load/store): complicate pipeline stages, limiting clock speeds. May have to handle exceptions mid-execution, adding further complication. Also requires higher transistor count, of course. Overcome by emulation, either in u-code or by throwing an exception and letting the OS deal with it. o Unaligned memory accesses: loading/storing a register from an address that is not a multiple of the word size. Requires two loads, bytemasking, ORing and then maybe a store, instead of one load/store. Complicates pipeline stages again, increases transistor count, limits clock speed. Can also throw exceptions partway through. Overcome by: emulation in u-code, or raising exception to OS (!). o Special registers for instructions (x86 MUL): keep instruction encoding small, but give compiler a hard time allocating registers, so that false dependencies arise (as in first point). How does one overcome this? o Does support for self-modifying code count? It means the I-cache has to snoop the D-cache, which I suppose can be a little irritating, especially if you're trying to do OoO execution. How does one overcome this? Does anyone have any further points to add, criticism/corrections to make, or answers to my questions? Cheers, George Soon to be gbmvdd@cantab.net From bgrg2@cam.ac.uk Sat, 1 Jun 2002 23:41:41 +0100 Date: Sat, 1 Jun 2002 23:41:41 +0100 From: Barnaby Gray bgrg2@cam.ac.uk Subject: [CST-2] Comp. Arch 1998.7.3 - backwards compatibility problems On Sat, Jun 01, 2002 at 10:28:30PM +0100, George wrote: > "Older computer architectures often contain features that hamper > high-performance implementation using modern techniques. Describe several > such features, explaining how they affect implementation; describe what > techniques (if any) have been developed to help overcome these > difficulties." [14 marks] > > So far, I have the following: > o Unaligned memory accesses: loading/storing a register from an address > that is not a multiple of the word size. Requires two loads, bytemasking, > ORing and then maybe a store, instead of one load/store. Complicates > pipeline stages again, increases transistor count, limits clock speed. Can > also throw exceptions partway through. Overcome by: emulation in u-code, or > raising exception to OS (!). One small addition - support for subword loads/stores are a (very) similar problem, but maybe another point. Barnaby From jlg33@cam.ac.uk Sun, 02 Jun 2002 00:10:46 +0100 Date: Sun, 02 Jun 2002 00:10:46 +0100 From: Jonny Graham jlg33@cam.ac.uk Subject: [CST-2] Comp. Arch 1998.7.3 - backwards compatibility problems > Does anyone have any further points to add, criticism/corrections to make, > or answers to my questions? The ARM exposes the pc, such that it always points 3 (or is it 4, I don't remember...) instructions ahead of the one currently being executed. Extending the pipeline to improve performance means that the pc would naturally point even further ahead which would break existing code. The solution is to hide the real pc and expose a 'psuedo-pc', pointing the expected distance ahead of the currently executing instruction. Jonny > > Cheers, > George > > Soon to be gbmvdd@cantab.net > > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From sn249@cam.ac.uk Sun, 2 Jun 2002 01:35:20 +0100 Date: Sun, 2 Jun 2002 01:35:20 +0100 From: Shahrum Gilani sn249@cam.ac.uk Subject: [CST-2] HCI: Hick's Law & visual search Hi, I was going through the HCI notes and I've pencilled down from the lectures that finding a letter in a list of N letters takes kN time. But surely this contradicts what's said in Hick's Law? Have I missed something, or have I just written down the wrong thing? Thanks, Shahrum From mkm25@cam.ac.uk Sun, 2 Jun 2002 01:59:33 +0100 (BST) Date: Sun, 2 Jun 2002 01:59:33 +0100 (BST) From: Matthew Monaghan mkm25@cam.ac.uk Subject: [CST-2] HCI: Hick's Law & visual search On Sun, 2 Jun 2002, Shahrum Gilani wrote: > Hi, > > I was going through the HCI notes and I've pencilled down from the lectures > that finding a letter in a list of N letters takes kN time. But surely this > contradicts what's said in Hick's Law? Have I missed something, or have I > just written down the wrong thing? Hick's law applies to a field of objects, ie. a 2d set of them. The kN thingy is for a 1d list. cya -- Matt From mkm25@cam.ac.uk Sun, 2 Jun 2002 02:08:05 +0100 (BST) Date: Sun, 2 Jun 2002 02:08:05 +0100 (BST) From: Matthew Monaghan mkm25@cam.ac.uk Subject: [CST-2] HCI: Hick's Law & visual search I said: > Hick's law applies to a field of objects, ie. a 2d set of them. > The kN thingy is for a 1d list. I'm sorry there's no basis for that at all, I've just made it up to fit what I think is true. www.usabilityfirst.com says that Hick's law is to give you the time it takes to make a descion based on the number of possibilities. That might be closer to the truth, but I really don't know at all. cya -- Matt From ah296@cam.ac.uk Sun, 02 Jun 2002 02:34:05 +0100 Date: Sun, 02 Jun 2002 02:34:05 +0100 From: Alvin ah296@cam.ac.uk Subject: [CST-2] HCI: Hick's Law & visual search Hick's law is based on gun aiming, and I think the way a person searching for a letter in a string (scanning linearly) and targeting an item amongst "similar" objects is somewhat different - it's unlikely that you'll be just scanning the objects from left to right until you find what you're looking for. Alvin > I said: > >> Hick's law applies to a field of objects, ie. a 2d set of them. >> The kN thingy is for a 1d list. > > I'm sorry there's no basis for that at all, I've just made it up to fit > what I think is true. > > www.usabilityfirst.com says that Hick's law is to give you the time it > takes to make a descion based on the number of possibilities. > That might be closer to the truth, but I really don't know at all. > > cya From acl33@hermes.cam.ac.uk Sun, 2 Jun 2002 03:01:18 +0100 Date: Sun, 2 Jun 2002 03:01:18 +0100 From: Alan Lawrence acl33@hermes.cam.ac.uk Subject: [CST-2] HCI: Hick's Law & visual search This is what our supervisor told us - essentially, by taking the things in the list and scattering them into a 2d field, you'll find 'em faster....!! --Alan On Sunday, June 2, 2002, at 01:35 am, Shahrum Gilani wrote: > Hi, > > I was going through the HCI notes and I've pencilled down from the > lectures > that finding a letter in a list of N letters takes kN time. But surely > this > contradicts what's said in Hick's Law? Have I missed something, or > have I > just written down the wrong thing? > > Thanks, > > Shahrum > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From dps29@hermes.cam.ac.uk Sun, 2 Jun 2002 11:44:00 +0100 (BST) Date: Sun, 2 Jun 2002 11:44:00 +0100 (BST) From: David Singleton dps29@hermes.cam.ac.uk Subject: [CST-2] Types: y2000p9q13 Has anyone done this question, and could they tell me how to do the proof? I presume that we need to show that if phi holds of a judgement, phi still holds when the axiom and rules are applied. I think I've managed this for the axiom. How do we proceed - if by structural induction, on what? I tried substituting the rules into each \Phi (since \Phi holds of judgements and the rules hold of the judgements), but this didn't seem to get me anywhere. Thanks, David -- David Singleton _________________ St John's College, Cambridge C5 3rd Crt +44(0)7980 641608 http://www.srcf.ucam.org/~dps29 From gbmv2@cam.ac.uk Sun, 2 Jun 2002 11:51:11 +0100 Date: Sun, 2 Jun 2002 11:51:11 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] Types: y2000p9q13 ----- Original Message ----- From: "David Singleton" To: Sent: Sunday, June 02, 2002 11:44 AM Subject: [CST-2] Types: y2000p9q13 > > Has anyone done this question, and could they tell me how to do the > proof? > > I presume that we need to show that if phi holds of a judgement, > phi still holds when the axiom and rules are applied. > > I think I've managed this for the axiom. > > How do we proceed - if by structural induction, on what? Rule induction. Prove that phi holds for the axioms, and also that for each rule, if it holds for the premises then it holds for the conclusion. You need to cope with several different implications to get the rules to work, because there's implication internal to phi, and also from premises to conclusion. It gets a bit confusing. Once you've done that, the rest falls into place. George From jdjs2@cam.ac.uk Sun, 2 Jun 2002 12:33:59 +0100 Date: Sun, 2 Jun 2002 12:33:59 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] Adv Graphics Can anyone please explain an answer to http://www.cl.cam.ac.uk/tripos/y2001p8q4.pdf on B-splines. I can do parts c and d, and I intuitively can see what is going on in parts a and b, but I can't manage to get the numbers to come out. Any help much appreciated. Cheers, Jamie From adw28@cam.ac.uk Sun, 2 Jun 2002 12:44:23 +0100 (BST) Date: Sun, 2 Jun 2002 12:44:23 +0100 (BST) From: Andy Wilson adw28@cam.ac.uk Subject: [CST-2] Types: y2000p9q13 On Sun, 2 Jun 2002, George van den Driessche wrote: > ----- Original Message ----- > From: "David Singleton" > To: > Sent: Sunday, June 02, 2002 11:44 AM > Subject: [CST-2] Types: y2000p9q13 > > > > > > Has anyone done this question, and could they tell me how to do the > > proof? > > Rule induction. Prove that phi holds for the axioms, and also that for each > rule, if it holds for the premises then it holds for the conclusion. You > need to cope with several different implications to get the rules to work, > because there's implication internal to phi, and also from premises to > conclusion. It gets a bit confusing. See the section 2.4 starting on page 18 of the IB Semantics notes (http://www.cl.cam.ac.uk/Teaching/2001/Semantics/sempl.ps.gz) for an explanation of Rule Induction... Andy. From mp292@cam.ac.uk Sun, 2 Jun 2002 12:50:49 +0100 (BST) Date: Sun, 2 Jun 2002 12:50:49 +0100 (BST) From: Matej Pfajfar mp292@cam.ac.uk Subject: [CST-2] Adv Graphics There is a typo the bound should be n >= 2k-2. I think. On Sun, 2 Jun 2002, Jamie Shotton wrote: > > Can anyone please explain an answer to > http://www.cl.cam.ac.uk/tripos/y2001p8q4.pdf on B-splines. I can do > parts c and d, and I intuitively can see what is going on in parts a and > b, but I can't manage to get the numbers to come out. Any help much > appreciated. > > Cheers, > Jamie > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > -- Matej Pfajfar St John's College, University of Cambridge, UK GPG Public Keys @ http://matejpfajfar.co.uk/keys WARNING: THIS E-MAIL ACCOUNT WILL BE DELETED ON 15/07/2002. PLEASE USE mp@cantab.net. From mirw2@cam.ac.uk Sun, 2 Jun 2002 13:03:24 +0100 (BST) Date: Sun, 2 Jun 2002 13:03:24 +0100 (BST) From: Matt Williams mirw2@cam.ac.uk Subject: [CST-2] Adv Graphics On Sun, 2 Jun 2002, Matej Pfajfar wrote: > There is a typo the bound should be n >= 2k-2. > I think. I don't think so. Wasn't this quite heavily discussed last week? Matt From sjc86@cam.ac.uk Sun, 2 Jun 2002 17:13:10 +0100 Date: Sun, 2 Jun 2002 17:13:10 +0100 From: Simon Crumpton sjc86@cam.ac.uk Subject: [CST-2] Advanced Graphics - CSG Can somebody explain an answer to (c) in 2000 P9 Q4? "Describe how an object built using CSG can be represented using a binary tree. Given the intersection points of a ray with each primitive in the tree, show how to calculate the first intersection point of the ray with the entire CSG object." [6] Cheers Simon From mike@tropic.org.uk Sun, 2 Jun 2002 17:28:23 +0100 (BST) Date: Sun, 2 Jun 2002 17:28:23 +0100 (BST) From: Mike Pinna mike@tropic.org.uk Subject: [CST-2] Advanced Graphics - CSG On Sun, 2 Jun 2002, Simon Crumpton wrote: > Can somebody explain an answer to (c) in 2000 P9 Q4? > > "Describe how an object built using CSG can be represented using a binary > tree. Given the intersection points of a ray with each primitive in the > tree, show how to calculate the first intersection point of the ray with the > entire CSG object." [6] Constructive Solid Geometry allows us to perform three operations on primitives and previously constructed objects: intersection, union and subtraction. We can therefore regard any object as a binary tree, where each node stores which of the construction operations it corresponds to and pointers to the two objects being combined (these may have been constructed recursively in the same way. Each leaf corresponds to a primitive object. To compute the intersection of a ray with a CSG object, we produce an algorithm for finding out whether a ray is inside or outside an object: For a node, merge the list of intersections with the two objects (which we compute recursively). From this it is trivial to work out when the ray goes in and out of our combined object. For a leaf, we are effectively given this by the question. The first intersection point of the ray with the final object will just be the first element of the list returned here. HTH, Mike -- Plese note my new email address: mike@tropic.org.uk All my addresses in cam.ac.uk will stop working on 15 July. From gbmv2@cam.ac.uk Sun, 2 Jun 2002 17:33:17 +0100 Date: Sun, 2 Jun 2002 17:33:17 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] Advanced Graphics - CSG > Can somebody explain an answer to (c) in 2000 P9 Q4? > > "Describe how an object built using CSG can be represented using a binary > tree. If each primitive in the object represents the set of points within it, then the object is just an expression combining those sets using conjunction, disjunction, subtraction. The binary tree corresponds to that expression. > Given the intersection points of a ray with each primitive in the > tree, show how to calculate the first intersection point of the ray with the > entire CSG object." [6] A node in the binary tree has two children. If the sorted (by distance along ray) list of intersections for each child is known, then the list for the parent can be calculated by merging the lists to create a new sorted list, and then passing a FSA over the new list, which removes items as necessary according to the operator of the node. The FSA works by keeping track of whether the ray is currently inside each child object, so for example if you're calculating (A and B) and the ray is inside both A and B, and encounters an "exit B" item, you would just delete it, but if it next met an "exit A" item, you would keep it. Of course, the FSA could be engineered to have two input streams and select items based on lower t, thus merging the two lists as it goes along. Thus each node including that of the whole object can be calculated in terms of its descendents. From ah296@cam.ac.uk Sun, 02 Jun 2002 18:32:53 +0100 Date: Sun, 02 Jun 2002 18:32:53 +0100 From: Alvin ah296@cam.ac.uk Subject: [CST-2] Information theory and coding 1999.9.11 Hmmm The third part asks for the capacity in terms of the average power, power spectral density and bandwidth... And gives 10 marks. So do we get ten marks just for quoting the theorem? Or do we have to prove it from first principles... 10 marks doesn't seem right for either of them! Alvin From gbmv2@cam.ac.uk Sun, 2 Jun 2002 18:58:52 +0100 Date: Sun, 2 Jun 2002 18:58:52 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] Information theory and coding 1999.9.11 I wondered about that. Then I noticed that the exact same question turns up in the Learning Guide (albeit without mark allocation), and Daugman's answer is just to quote the theorem. So either the proof used to be required knowledge and now isn't, or he screwed up allocating marks. George ----- Original Message ----- From: "Alvin" To: Sent: Sunday, June 02, 2002 6:32 PM Subject: [CST-2] Information theory and coding 1999.9.11 > Hmmm > > The third part asks for the capacity in terms of the average power, power > spectral density and bandwidth... And gives 10 marks. So do we get ten > marks just for quoting the theorem? Or do we have to prove it from first > principles... 10 marks doesn't seem right for either of them! > > Alvin > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From cia20@cam.ac.uk Sun, 2 Jun 2002 20:05:09 +0100 Date: Sun, 2 Jun 2002 20:05:09 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Information theory and coding 1999.9.11 Either that, or he just didn't think it necessary to write out the entire proof when answering the learning guide questions, as it would just be a carbon copy of what's in the notes. It still might be required to learn the derivation for an examination. I'm taking the model answers in the LG with a pinch of salt, as later on he gives very short answers to questions worth 5 marks on the corresponding exam paper - I think they're 'model for learning/supervision' rather than 'model for exams' answers... Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of George van den Driessche > Sent: 2 June 2002 18:59 > To: cst-2@srcf.ucam.org > Subject: Re: [CST-2] Information theory and coding 1999.9.11 > > > I wondered about that. Then I noticed that the exact same > question turns up > in the Learning Guide (albeit without mark allocation), and > Daugman's answer > is just to quote the theorem. So either the proof used to be required > knowledge and now isn't, or he screwed up allocating marks. > > George > > ----- Original Message ----- > From: "Alvin" > To: > Sent: Sunday, June 02, 2002 6:32 PM > Subject: [CST-2] Information theory and coding 1999.9.11 > > > > Hmmm > > > > The third part asks for the capacity in terms of the > average power, power > > spectral density and bandwidth... And gives 10 marks. So > do we get ten > > marks just for quoting the theorem? Or do we have to prove > it from first > > principles... 10 marks doesn't seem right for either of them! > > > > Alvin > > > > > > > > _______________________________________________ > > CST-2 mailing list > > CST-2@srcf.ucam.org > > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From cia20@cam.ac.uk Sun, 2 Jun 2002 20:34:48 +0100 Date: Sun, 2 Jun 2002 20:34:48 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics - Radiosity Apart from one part of 1999P9Q4 there doesn't seem to be many past exam questions on radiosity (The learning guide asks two others only two more) - anyone know of a source of good practice questions? Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request From jrs53@cam.ac.uk Sun, 2 Jun 2002 21:57:06 +0100 Date: Sun, 2 Jun 2002 21:57:06 +0100 From: James Srinivasan jrs53@cam.ac.uk Subject: [CST-2] E-commerce - individualisation vs personalisation My notes about the definitions of individualisation and personalisation from Tim King's lecture and the examples class seem to exactly contradict each other. Does anyone have a non-contradictory definition for which is which? Many Thanks, James From ss368@cam.ac.uk Sun, 2 Jun 2002 22:34:16 +0100 (BST) Date: Sun, 2 Jun 2002 22:34:16 +0100 (BST) From: Sam Staton ss368@cam.ac.uk Subject: [CST-2] Info Theory Just running through the info theory examples sheet again. Set 8, part d. I reckon C' = C/8 (approx) by the taylor expansion of lg, (some brief experiments with gnuplot confirm this). Have I misunderstood the question -- I read it as C' = W lg (1 + P/(8NW)) ? Thanks -- sam From tp225@cam.ac.uk Sun, 2 Jun 2002 22:48:40 +0100 Date: Sun, 2 Jun 2002 22:48:40 +0100 From: Tom Puverle tp225@cam.ac.uk Subject: [CST-2] Advanced Graphics - Radiosity There is a learning guide for advanced graphics? Tom > Apart from one part of 1999P9Q4 there doesn't seem to be many past exam > questions on radiosity (The learning guide asks two others only two > more) - anyone know of a source of good practice questions? > > Cheerio, > > Chris > do something lastminute.work > > Chris Applegate > Room X6, Corpus Christi College, Cambridge, CB2 1RH > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > ICQ 41706821 PGP key available on request > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From cia20@cam.ac.uk Sun, 2 Jun 2002 22:55:32 +0100 Date: Sun, 2 Jun 2002 22:55:32 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Advanced Graphics - Radiosity I mean the handout of lecture notes & photocopied reference text...sorry, wasn't very clear. Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Tom Puverle > Sent: 2 June 2002 22:49 > To: cst-2@srcf.ucam.org > Subject: RE: [CST-2] Advanced Graphics - Radiosity > > > There is a learning guide for advanced graphics? > Tom > > > Apart from one part of 1999P9Q4 there doesn't seem to be > many past exam > > questions on radiosity (The learning guide asks two others only two > > more) - anyone know of a source of good practice questions? > > > > Cheerio, > > > > Chris > > do something lastminute.work > > > > Chris Applegate > > Room X6, Corpus Christi College, Cambridge, CB2 1RH > > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > > ICQ 41706821 PGP key available on request > > > > > > > > _______________________________________________ > > CST-2 mailing list > > CST-2@srcf.ucam.org > > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From ah296@cam.ac.uk Sun, 02 Jun 2002 23:13:41 +0100 Date: Sun, 02 Jun 2002 23:13:41 +0100 From: Alvin ah296@cam.ac.uk Subject: [CST-2] Info Theory I think Daugman's working goes along the lines of: C = Wlog(1 + P/NW) C' = Wlog(1 + P/8NW) ~= W(log(1 + P/NW) + log(1/8) = W(log(1 + P/NW) - 3) = C - 3W Which is a close approximation if P/NW was a lot larger than 1... What is the norm for SNR anyway? Alvin > Just running through the info theory examples sheet again. > > Set 8, part d. I reckon C' = C/8 (approx) by the taylor expansion of lg, > (some brief experiments with gnuplot confirm this). Have I misunderstood > the question -- I read it as C' = W lg (1 + P/(8NW)) > > ? > > Thanks From cia20@cam.ac.uk Sun, 2 Jun 2002 23:17:39 +0100 Date: Sun, 2 Jun 2002 23:17:39 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Info Theory Yeah, that was my working too. A quick example of W=1, SNR = 99 (which can be thought of as a 0.01 error rate, can't it?) gives 2.90 as the difference. The model answer does say 'about' when it says bandwidth, so presumably he has done some sort of approximation somewhere. Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Alvin > Sent: 2 June 2002 23:14 > To: cst-2@srcf.ucam.org > Subject: Re: [CST-2] Info Theory > > > I think Daugman's working goes along the lines of: > > C = Wlog(1 + P/NW) > > C' = Wlog(1 + P/8NW) > > ~= W(log(1 + P/NW) + log(1/8) > > = W(log(1 + P/NW) - 3) > > = C - 3W > > Which is a close approximation if P/NW was a lot larger than > 1... What is > the norm for SNR anyway? > > Alvin > > > Just running through the info theory examples sheet again. > > > > Set 8, part d. I reckon C' = C/8 (approx) by the taylor > expansion of lg, > > (some brief experiments with gnuplot confirm this). Have I > misunderstood > > the question -- I read it as C' = W lg (1 + P/(8NW)) > > > > ? > > > > Thanks > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From chvc2@hermes.cam.ac.uk Sun, 2 Jun 2002 23:51:22 +0100 (BST) Date: Sun, 2 Jun 2002 23:51:22 +0100 (BST) From: Crispin chvc2@hermes.cam.ac.uk Subject: [CST-2] Opt Comp quick question y2001p8q7 part b It seems to me that you can guarentee a_t+1 will not terminate when f does not terminate; and as the question emphasizes `safety' it's probably best to say no more. So do you really get 5 marks for writing down not f# => not a_{t+1}# or have I missed something big? -- Crispin This email account will die soon - please use crispin@cantab.net instead! From sdm29@hermes.cam.ac.uk Mon, 3 Jun 2002 01:09:35 +0100 (BST) Date: Mon, 3 Jun 2002 01:09:35 +0100 (BST) From: Stephen McIntosh sdm29@hermes.cam.ac.uk Subject: [CST-2] Opt Comp quick question > y2001p8q7 > > part b > > It seems to me that you can guarentee a_t+1 will not terminate when f does > not terminate; and as the question emphasizes `safety' it's probably best > to say no more. > > So do you really get 5 marks for writing down > > not f# => not a_{t+1}# > > or have I missed something big? I think you've missed something... b) (n.b. % is "for all", <= is "less than or equal to" and e is "a member of") f# -> SAFE but not PRECISE a_t+1# -> PRECISE - defined by compiler designer, hence no computability problems. %x1,...,xn e {0,1} a_t+1#(x1,...,xn) <= f#(x1,...,xn) where 0 <= 0 0 <= 1 <- a_t+1# and f# give different answers in this case. 1 <= 1 c) Consider f with abstract strictness funvtion f# f#(0) = 1 f#1 = 0 i.e. f#(x) = not(x) >From definition of 0,1 this is a contradiction (in terms of Turing machines + the halting problem) => If argument doesn't HALT, then f may HALT and if argument may HALT, then f doesn't HALT. => f# is not computable => Standard interpretation f is not computable. Any fn. including negation can be reduced to the form above => No strictness fn. including negation can be computed. ========================= Stephen McIntosh. www.stephenmcintosh.com From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 01:41:41 +0100 (BST) Date: Mon, 3 Jun 2002 01:41:41 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Opt Comp quick question > So do you really get 5 marks for writing down > > not f# => not a_{t+1}# > > or have I missed something big? I thought it prudent to write down also a_{t+1}#(d_1, ..., d_n) = 0 <==> ((forall i. d_i = 0 <==> x_i = _|_) ==> a_{n+1}(x_1, ..., x_n) = _|_) and a similar expression for f# with ==> for the first <==>. So I don't think you missed anything big, but perhaps should play it a bit safer. Peter ===== Peter Taylor pjt33@cam.ac.uk From tp225@cam.ac.uk Mon, 3 Jun 2002 01:55:58 +0100 Date: Mon, 3 Jun 2002 01:55:58 +0100 From: Tom Puverle tp225@cam.ac.uk Subject: [CST-2] Information thry & neural nets There are two things I am not clear on, any help would be very appreciated: 1) In the learning guide on Neural Networks: Exercise 9 How do you get from the first equation to the second? (There's one line between the two equations, starting "and similarly for P~(N|X)..." 2) Information thry, lecture notes, page 42. He deduces that h(Y|X) = h(N) "since N is independent of X"... I don't understand. Anyone point me in the right direction? Thanks. From chvc2@hermes.cam.ac.uk Mon, 3 Jun 2002 02:10:30 +0100 (BST) Date: Mon, 3 Jun 2002 02:10:30 +0100 (BST) From: Crispin chvc2@hermes.cam.ac.uk Subject: [CST-2] Information thry & neural nets > 2) Information thry, lecture notes, page 42. > He deduces that h(Y|X) = h(N) "since N is independent of X"... I don't > understand. Anyone point me in the right direction? In a noiseless channel, h(Y|X) = 0 as there is no uncertainty about X when we have measured Y. (When calculating h the term log 1/p(y|x) is 0 as p(y|x) is 1). As Entropy is additive, the entropy of the noisy channel is 0 + the entropy of the noise, i.e. H(N) If the noise wasn't independent of X then H(N) would depend on P(X) as well. I hope. IANApersonwithmuchofaclue. -- Crispin This email account will die soon - please use crispin@cantab.net instead! From tp225@cam.ac.uk Mon, 3 Jun 2002 02:20:45 +0100 Date: Mon, 3 Jun 2002 02:20:45 +0100 From: Tom Puverle tp225@cam.ac.uk Subject: [CST-2] Information thry & neural nets > In a noiseless channel, h(Y|X) = 0 as there is no uncertainty about X when > we have measured Y. Now that was a bit obvious, wasn't it! D'oh! Thanks very much. From ss368@cam.ac.uk Mon, 3 Jun 2002 08:38:25 +0100 (BST) Date: Mon, 3 Jun 2002 08:38:25 +0100 (BST) From: Sam Staton ss368@cam.ac.uk Subject: [CST-2] Info Theory Thanks -- my (stupid) mistake. When W is big (when trying to prove the shannon limit), then the x in lg(1+x) is small, so the Taylor expansion is sensible. However in this case the x is big, the Taylor cannot be so truncated since the increasing terms are quite important. -- sam On Sun, 2 Jun 2002, Chris Applegate wrote: > Yeah, that was my working too. > > A quick example of W=1, SNR = 99 (which can be thought of as a 0.01 > error rate, can't it?) gives 2.90 as the difference. The model answer > does say 'about' when it says bandwidth, so presumably he has done some > sort of approximation somewhere. > > Cheerio, > > Chris > do something lastminute.work > > Chris Applegate > Room X6, Corpus Christi College, Cambridge, CB2 1RH > chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 > ICQ 41706821 PGP key available on request > > > > > > -----Original Message----- > > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > > Behalf Of Alvin > > Sent: 2 June 2002 23:14 > > To: cst-2@srcf.ucam.org > > Subject: Re: [CST-2] Info Theory > > > > > > I think Daugman's working goes along the lines of: > > > > C = Wlog(1 + P/NW) > > > > C' = Wlog(1 + P/8NW) > > > > ~= W(log(1 + P/NW) + log(1/8) > > > > = W(log(1 + P/NW) - 3) > > > > = C - 3W > > > > Which is a close approximation if P/NW was a lot larger than > > 1... What is > > the norm for SNR anyway? > > > > Alvin > > > > > Just running through the info theory examples sheet again. > > > > > > Set 8, part d. I reckon C' = C/8 (approx) by the taylor > > expansion of lg, > > > (some brief experiments with gnuplot confirm this). Have I > > misunderstood > > > the question -- I read it as C' = W lg (1 + P/(8NW)) > > > > > > ? > > > > > > Thanks > > > > > > > > _______________________________________________ > > CST-2 mailing list > > CST-2@srcf.ucam.org > > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From cia20@cam.ac.uk Mon, 3 Jun 2002 14:18:15 +0100 Date: Mon, 3 Jun 2002 14:18:15 +0100 From: Chris Applegate cia20@cam.ac.uk Subject: [CST-2] Digi Comm 2 Can anyone point to a decent explanation of how DHCP and NAT work? My notes for the IPv6 lecture are incomprehensible, and none of the books I have describe either in any detail. Ta in advance, Cheerio, Chris do something lastminute.work Chris Applegate Room X6, Corpus Christi College, Cambridge, CB2 1RH chris@qwghlm.co.uk / www.qwghlm.co.uk / 07779 592 845 ICQ 41706821 PGP key available on request From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 16:00:57 +0100 (BST) Date: Mon, 3 Jun 2002 16:00:57 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Fano's inequality p19 of InfoTh notes. Pe >= ((H(X|Y) - 1) / |A|) where A is the number of possible outcomes. Does this assume X and Y have the same set of outcomes, or is it the number of possible outcomes of X? Peter ===== Peter Taylor pjt33@cam.ac.uk From mp292@cam.ac.uk Mon, 3 Jun 2002 16:04:34 +0100 (BST) Date: Mon, 3 Jun 2002 16:04:34 +0100 (BST) From: Matej Pfajfar mp292@cam.ac.uk Subject: [CST-2] Fano's inequality Same set of outcomes. On Mon, 3 Jun 2002, Peter Taylor wrote: > p19 of InfoTh notes. Pe >= ((H(X|Y) - 1) / |A|) where A is the number of > possible outcomes. Does this assume X and Y have the same set of outcomes, > or is it the number of possible outcomes of X? > > Peter > ===== > Peter Taylor > pjt33@cam.ac.uk > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > -- Matej Pfajfar St John's College, University of Cambridge, UK GPG Public Keys @ http://matejpfajfar.co.uk/keys WARNING: THIS E-MAIL ACCOUNT WILL BE DELETED ON 15/07/2002. PLEASE USE mp@cantab.net. From mike@tropic.org.uk Mon, 3 Jun 2002 16:09:05 +0100 (BST) Date: Mon, 3 Jun 2002 16:09:05 +0100 (BST) From: Mike Pinna mike@tropic.org.uk Subject: [CST-2] Fano's inequality On Mon, 3 Jun 2002, Peter Taylor wrote: > p19 of InfoTh notes. Pe >= ((H(X|Y) - 1) / |A|) where A is the number of > possible outcomes. Does this assume X and Y have the same set of outcomes, > or is it the number of possible outcomes of X? Surely we don't care how many possible outcomes there are for Y, given that we know what value it has taken? Mike -- Plese note my new email address: mike@tropic.org.uk All my addresses in cam.ac.uk will stop working on 15 July. From jas88@cam.ac.uk Mon, 3 Jun 2002 16:10:41 +0100 (BST) Date: Mon, 3 Jun 2002 16:10:41 +0100 (BST) From: James Sutherland jas88@cam.ac.uk Subject: [CST-2] Digi Comm 2 On Mon, 3 Jun 2002, Chris Applegate wrote: > Can anyone point to a decent explanation of how DHCP and NAT work? > My notes for the IPv6 lecture are incomprehensible, Since it's IPv4 not IPv6, that shouldn't be an issue :-) > and none of the books I have describe either in any detail. DHCP (Dynamic Host Configuration Protocol): When first booted, the only information available to a DHCP-configured host is its own MAC address. It uses this to form an Ethernet broadcast frame, to which a DHCP server will (hopefully!) respond by offering the host an IP address (and other configuration information). The full protocol is defined by RFCs 2131 and 2132. DHCP is backwards compatible with, and typically replaces, BOOTP (Boot Protocol); RARP (Reverse ARP) has also been used to fulfil a similar role. For RARP, the host will issue a RARP Ethernet frame requesting the IP address corresponding to its own MAC address, and then use this IP address. BOOTP and RARP both lack facilities to share IP addresses: hosts assume an IP address obtained in this way is permanent. DHCP eliminates this flaw by including a "lease duration" field: the IP address may only used until the specified time has elapsed. The host should then attempt to "renew" the lease *before* it expires, obtaining an extension on the lease. This may fail, in which case the host will be forced to change IP addresses. NAT (Network Address Translation) A mechanism typically used for a group of machines to share a single "public" IP address for Internet access. A NAT gateway functions as a normal router, except that it modifies outbound IP packets to reflect the public address, rather than the original machine's address (typically a "private" address, per RFC 1918, such as 10.* or 192.168.*), then modifies inbound packets so they reach the intended destination. The biggest advantage (and disadvantage!) is that it prevents the "internal" machines from acting as servers to the Internet at large. This may be welcome from a security point of view - even if you leave wu-ftpd running on an old Linux box, it can't be compromised from outside - but less welcome for users. It also prevents facilities such as IRC DCC file transfers and chats working, unless the NAT gateway has additional logic to handle such connections. HTH! James. From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 16:48:11 +0100 (BST) Date: Mon, 3 Jun 2002 16:48:11 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Discrete memoryless channel error prob p30 of InfoTh notes. Pe = sum_{k != j} P(Y = y_k | X = x_j) = sum_{k != j}p(y_k | x_j) p(x_j). Clearly there's some difference between P(Y = y_k | X = x_j) and p(y_k | x_j), but what is it? Thanks, Peter ===== Peter Taylor pjt33@cam.ac.uk From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 16:51:35 +0100 (BST) Date: Mon, 3 Jun 2002 16:51:35 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Discrete memoryless channel error prob Scratch that - the previous page defines them as the same. It's clearly a typo, and the first | should be a , That's good, because it's what I thought it should be. Peter ===== Peter Taylor pjt33@cam.ac.uk From jarvis_dan@hotmail.com Mon, 3 Jun 2002 16:56:12 +0100 Date: Mon, 3 Jun 2002 16:56:12 +0100 From: jarvis_dan jarvis_dan@hotmail.com Subject: [CST-2] Advanced Algorithms - Big number arithmetic I was looking through my lecture notes for the big-number arithmetic section, and I think I can follow as far as converting between coefficient and point value representations of polynomials via the DFT. But then my notes say the following after introducing the FFT: let w=2 choose modulus M such that w^n = 1 (mod M) but no value k can have w^k = 1 (mod M) for k What's 1+1? 110. Now work out the base! Peter ===== Peter Taylor pjt33@cam.ac.uk From mike@tropic.org.uk Mon, 3 Jun 2002 17:39:57 +0100 (BST) Date: Mon, 3 Jun 2002 17:39:57 +0100 (BST) From: Mike Pinna mike@tropic.org.uk Subject: [CST-2] Arithmetic On Mon, 3 Jun 2002, Peter Taylor wrote: > > What's 1+1? > > 110. Now work out the base! -2 Mike -- Plese note my new email address: mike@tropic.org.uk All my addresses in cam.ac.uk will stop working on 15 July. From ss368@cam.ac.uk Mon, 3 Jun 2002 17:48:50 +0100 (BST) Date: Mon, 3 Jun 2002 17:48:50 +0100 (BST) From: Sam Staton ss368@cam.ac.uk Subject: [CST-2] Arithmetic On Mon, 3 Jun 2002, Mike Pinna wrote: > On Mon, 3 Jun 2002, Peter Taylor wrote: > > > What's 1+1? > > > > 110. Now work out the base! > > -2 > or 1 (especially since it is usual for the digits to be less than the base) -- sam From amj30@cam.ac.uk Mon, 3 Jun 2002 17:51:17 +0100 Date: Mon, 3 Jun 2002 17:51:17 +0100 From: Anthony Jones amj30@cam.ac.uk Subject: [CST-2] Arithmetic On Mon, Jun 03, 2002 at 05:48:50PM +0100, Sam Staton wrote: > > > 110. Now work out the base! > > > > -2 > > > > or 1 (especially since it is usual for the digits to be less than the > base) But then 1 (the digit) isn't less than 1 (the proposed base)? Ant -- The obligatory bit about my email address: This account will be closed on July 15th 2002. Please use ant@cantab.net From gbmv2@cam.ac.uk Mon, 3 Jun 2002 17:58:35 +0100 Date: Mon, 3 Jun 2002 17:58:35 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] Arithmetic I'm expecting Bjarne Puverle to come up with something along the lines of: 364. Now work out how I've overloaded the + operator! Well, all the best, everyone. See you tomorrow. George ----- Original Message ----- From: "Peter Taylor" To: Sent: Monday, June 03, 2002 5:37 PM Subject: Re: [CST-2] Arithmetic > > What's 1+1? > > 110. Now work out the base! > > Peter > ===== > Peter Taylor > pjt33@cam.ac.uk > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 From mike@tropic.org.uk Mon, 3 Jun 2002 17:58:55 +0100 (BST) Date: Mon, 3 Jun 2002 17:58:55 +0100 (BST) From: Mike Pinna mike@tropic.org.uk Subject: [CST-2] Arithmetic On Mon, 3 Jun 2002, Anthony Jones wrote: > On Mon, Jun 03, 2002 at 05:48:50PM +0100, Sam Staton wrote: > > > > 110. Now work out the base! > > > > > > -2 > > > > > > > or 1 (especially since it is usual for the digits to be less than the > > base) > > But then 1 (the digit) isn't less than 1 (the proposed base)? Indeed. Anyway, Peter's answer seemed to indicate his solution was unique for the base he was considering, but if the base is 1, 1000100 is just as valid an answer, as is any other string of ones and zeros containing exactly two ones. I think my answer is better, but it seems unlikely that it matters anyway :) Mike -- Plese note my new email address: mike@tropic.org.uk All my addresses in cam.ac.uk will stop working on 15 July. From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 18:11:27 +0100 (BST) Date: Mon, 3 Jun 2002 18:11:27 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Arithmetic > -2 Correct. Peter ===== Peter Taylor pjt33@cam.ac.uk From mp292@cam.ac.uk Mon, 3 Jun 2002 19:47:37 +0100 (BST) Date: Mon, 3 Jun 2002 19:47:37 +0100 (BST) From: Matej Pfajfar mp292@cam.ac.uk Subject: [CST-2] butterfly division Doing some last minute revision - could someone please tell me what the weights are for butterfly subdivision? Thanks! On Mon, 3 Jun 2002, Peter Taylor wrote: > > -2 > > Correct. > > Peter > ===== > Peter Taylor > pjt33@cam.ac.uk > > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > -- Matej Pfajfar St John's College, University of Cambridge, UK GPG Public Keys @ http://matejpfajfar.co.uk/keys WARNING: THIS E-MAIL ACCOUNT WILL BE DELETED ON 15/07/2002. PLEASE USE mp@cantab.net. From tmh31@cam.ac.uk Mon, 3 Jun 2002 20:15:10 +0100 Date: Mon, 3 Jun 2002 20:15:10 +0100 From: Timothy Hospedales tmh31@cam.ac.uk Subject: [CST-2] Neural Computing / Vision Hi, The NC notes have a bullet point about Zero-Crossings in visual codes being a counter example to Marr's theory. Does anyone know/remember why this is so? IF your zero-crossings are edge information then it seems in line with Marr's 2D primal sketch contains edge information. Thanks, Tim From pjt33@hermes.cam.ac.uk Mon, 3 Jun 2002 20:47:23 +0100 (BST) Date: Mon, 3 Jun 2002 20:47:23 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] butterfly division > Doing some last minute revision - could someone please tell me what the > weights are for butterfly subdivision? Thanks! According to http://graphics.stanford.edu/~dzorin/multires/butterfly/sig96_noimg.ps.gz they are b c ________________ c \ /\ / \ / \ / \ / \ / a \/______\/ a /\ /\ / \ / \ / \ / \ /______\/______\ c b c Weights: a = 1/2, b = 1/8, c = -1/16 Peter ===== Peter Taylor pjt33@cam.ac.uk From tp225@cam.ac.uk Mon, 3 Jun 2002 20:58:11 +0100 Date: Mon, 3 Jun 2002 20:58:11 +0100 From: Tom Puverle tp225@cam.ac.uk Subject: [CST-2] Arithmetic > What's 1+1? > > Best of luck tomorrow everybody! > > Alvin I take it you have some insider info about the contents of the paper tomorrow? Tom From acl33@hermes.cam.ac.uk Mon, 3 Jun 2002 23:42:57 +0100 Date: Mon, 3 Jun 2002 23:42:57 +0100 From: Alan Lawrence acl33@hermes.cam.ac.uk Subject: [CST-2] Graphics: Subdivision Schemes Does anyone have any notes on the interpolating subdivision schemes (for 2d or 3d)? i.e. what coefficients etc.....I remember him lecturing it but the notes don't have any info on these.... Cheers, Alan ----------------- This address may soon be terminated, please use acl@cantab.net instead. From pjt33@hermes.cam.ac.uk Tue, 4 Jun 2002 00:49:24 +0100 (BST) Date: Tue, 4 Jun 2002 00:49:24 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] Graphics: Subdivision Schemes > Does anyone have any notes on the interpolating subdivision schemes (for > 2d or 3d)? i.e. what coefficients etc.....I remember him lecturing it > but the notes don't have any info on these.... There is a 2D subdivision scheme which he called "interpolation" - this has weights (1) and (-1, 9, 9, -1). The tensor product of this is Kobbelt. Butterfly is also interpolating. HTH, Peter ===== Peter Taylor pjt33@cam.ac.uk From jdjs2@cam.ac.uk Tue, 4 Jun 2002 09:16:57 +0100 Date: Tue, 4 Jun 2002 09:16:57 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] Graphics: Subdivision Schemes > > > Does anyone have any notes on the interpolating subdivision schemes > > (for 2d or 3d)? i.e. what coefficients etc.....I remember him > > lecturing it but the notes don't have any info on these.... > > There is a 2D subdivision scheme which he called > "interpolation" - this has weights (1) and (-1, 9, 9, -1). > The tensor product of this is Kobbelt. Butterfly is also > interpolating. In the 1D case of the Kobbelt scheme (i.e. the "4 point method"), how do you get a point between the first two vertices - do you have to assume that p_(j-1) is zero? Good luck everyone, Jamie From pjt33@hermes.cam.ac.uk Tue, 4 Jun 2002 17:33:32 +0100 (BST) Date: Tue, 4 Jun 2002 17:33:32 +0100 (BST) From: Peter Taylor pjt33@hermes.cam.ac.uk Subject: [CST-2] As a matter of interest... Those who attempted the Security question - how did you do the last part? Implement the queue as two stacks using the technique Larry Paulson taught us in IA? Peter ===== Peter Taylor pjt33@cam.ac.uk From mrs35@cam.ac.uk Tue, 04 Jun 2002 18:05:13 +0100 Date: Tue, 04 Jun 2002 18:05:13 +0100 From: Mark Seaborn mrs35@cam.ac.uk Subject: [CST-2] As a matter of interest... Peter Taylor wrote: > Those who attempted the Security question - how did you do the last > part? Implement the queue as two stacks using the technique Larry > Paulson taught us in IA? No, because the question already gives half the answer, as it suggests signing each record with a MAC. I suggested adding a sequence number i (which you increment for each message) to each record X to make it (X, i), and signing that. When you receive records from the terminal, just check that its sequence number is the one after that of the last record you received. If you implemented the queue as a pair of lists, you'd need a burst of traffic between the smartcard and the terminal to reverse one of the lists when it becomes empty. -- Mark Seaborn - mseaborn@bigfoot.com - http://www.srcf.ucam.org/~mrs35/ - ``We are quite lucky really this year because Christmas falls on Christmas day'' -- Bobby Gould From mp292@cam.ac.uk Tue, 4 Jun 2002 21:17:49 +0100 (BST) Date: Tue, 4 Jun 2002 21:17:49 +0100 (BST) From: Matej Pfajfar mp292@cam.ac.uk Subject: [CST-2] topics in concurrency Could someone please remind me what the definition of deadlock is, with regards to CSP/CCS? I got myself a bit confused ... that's what the night before exams does to you .. -- Matej Pfajfar St John's College, University of Cambridge, UK GPG Public Keys @ http://matejpfajfar.co.uk/keys WARNING: THIS E-MAIL ACCOUNT WILL BE DELETED ON 15/07/2002. PLEASE USE mp@cantab.net. From amj30@cam.ac.uk Tue, 4 Jun 2002 22:06:00 +0100 Date: Tue, 4 Jun 2002 22:06:00 +0100 From: Anthony Jones amj30@cam.ac.uk Subject: [CST-2] VLSI Page 16 of the VLSI notes, the CMOS gate matrix: At the top there's a p-type diffusion, with breaks in it where there are p-type transistors with their gates tied high: MMMMMMMMMMMMMMMMMMMM <- Metal M P M DDDDMDDPDDMDDDDDDDDD <- Diffusion (p-type) P ^ | \---------------- Polysilicon (causing break in diffusion) But then the diffusion is connected to the power line on both sides of the p-type transistors, meaning that the break caused by the p-type transistor is effectively pointless, and you could do away with it entirely... Or am I missing something? Ant -- The obligatory bit about my email address: This account will be closed on July 15th 2002. Please use ant@cantab.net From jmw74@cam.ac.uk Tue, 4 Jun 2002 22:14:13 +0100 Date: Tue, 4 Jun 2002 22:14:13 +0100 From: John Wyllie jmw74@cam.ac.uk Subject: [CST-2] VLSI The p-type transistors are there because they are a smaller way to make a gap between each array in the matrix. That's what I've got scribbled in my notes anyway - that diagram is well complicated. John > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Anthony Jones > Sent: 04 June 2002 22:06 > To: cst-2@srcf.ucam.org > Subject: [CST-2] VLSI > > > Page 16 of the VLSI notes, the CMOS gate matrix: At the top there's a > p-type diffusion, with breaks in it where there are p-type > transistors with > their gates tied high: > > MMMMMMMMMMMMMMMMMMMM <- Metal > M P M > DDDDMDDPDDMDDDDDDDDD <- Diffusion (p-type) > P > ^ > | > \---------------- Polysilicon (causing break in diffusion) > > But then the diffusion is connected to the power line on both sides of the > p-type transistors, meaning that the break caused by the p-type transistor > is effectively pointless, and you could do away with it entirely... > > Or am I missing something? > > Ant > > -- > The obligatory bit about my email address: > This account will be closed on July 15th 2002. > Please use ant@cantab.net > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From amj30@cam.ac.uk Tue, 4 Jun 2002 22:19:40 +0100 Date: Tue, 4 Jun 2002 22:19:40 +0100 From: Anthony Jones amj30@cam.ac.uk Subject: [CST-2] VLSI On Tue, Jun 04, 2002 at 10:14:13PM +0100, John Wyllie wrote: > The p-type transistors are there because they are a smaller way to make a > gap between each array in the matrix. Yes, I know they make a gap -- but what's the point in having a gap if you're going to connect both sides to the same thing? Ant -- The obligatory bit about my email address: This account will be closed on July 15th 2002. Please use ant@cantab.net From jmw74@cam.ac.uk Tue, 4 Jun 2002 22:29:06 +0100 Date: Tue, 4 Jun 2002 22:29:06 +0100 From: John Wyllie jmw74@cam.ac.uk Subject: [CST-2] VLSI err. Good point - I don't know. Let's hope it won't come up tomorrow. John > -----Original Message----- > From: cst-2-admin@srcf.ucam.org [mailto:cst-2-admin@srcf.ucam.org]On > Behalf Of Anthony Jones > Sent: 04 June 2002 22:20 > To: cst-2@srcf.ucam.org > Subject: Re: [CST-2] VLSI > > > On Tue, Jun 04, 2002 at 10:14:13PM +0100, John Wyllie wrote: > > The p-type transistors are there because they are a smaller way > to make a > > gap between each array in the matrix. > > Yes, I know they make a gap -- but what's the point in having a gap if > you're going to connect both sides to the same thing? > > Ant > > -- > The obligatory bit about my email address: > This account will be closed on July 15th 2002. > Please use ant@cantab.net > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From jdjs2@cam.ac.uk Tue, 4 Jun 2002 23:27:11 +0100 Date: Tue, 4 Jun 2002 23:27:11 +0100 From: Jamie Shotton jdjs2@cam.ac.uk Subject: [CST-2] VLSI > On Tue, Jun 04, 2002 at 10:14:13PM +0100, John Wyllie wrote: > > The p-type transistors are there because they are a smaller way to > > make a gap between each array in the matrix. > > Yes, I know they make a gap -- but what's the point in having > a gap if you're going to connect both sides to the same thing? I think it's cos it's got two p-type transistors in *parallel*. HTH, Jamie From amj30@cam.ac.uk Tue, 4 Jun 2002 23:35:20 +0100 Date: Tue, 4 Jun 2002 23:35:20 +0100 From: Anthony Jones amj30@cam.ac.uk Subject: [CST-2] VLSI On Tue, Jun 04, 2002 at 11:27:11PM +0100, Jamie Shotton wrote: > I think it's cos it's got two p-type transistors in *parallel*. Hmmmm... After Churchill bar based discussion (we have a high proportion of compscis so we can get away with that kind of thing), we reached the conclusion that it was probably just that the example that was a bit crappy, and in some situations you wouldn't have both sides of the break in the diffusion tied high... Ant -- The obligatory bit about my email address: This account will be closed on July 15th 2002. Please use ant@cantab.net From hl236@cam.ac.uk Thu, 06 Jun 2002 20:58:54 +0100 Date: Thu, 06 Jun 2002 20:58:54 +0100 From: hl236@cam.ac.uk hl236@cam.ac.uk Subject: [CST-2] Advanced do nothing Good work everyone, thanks for the discussions, really helped. Now we can do nothing. Celebration time = sleep Good night, and hope we will meet again. (Sad isn't it, today will be the last time we will see some of us, probably ever...) Hui p.s I've heard something about the existance of computer science alumini, anyone knows anything about it? p.p.s sorry about the mass email, will be my first and last time. From targetemailextractor@btamail.net.cn Mon, 10 Jun 2002 03:32:11 +0800 Date: Mon, 10 Jun 2002 03:32:11 +0800 From: targetemailextractor@btamail.net.cn targetemailextractor@btamail.net.cn Subject: [CST-2] ADV: Direct Email Blaster, email addresses extractor,maillist verify, maillist manager ........... =3Chtml=3E =3Chead=3E =3C=2Fhead=3E =3Cbody=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =A1=A1 =3C=2Fdiv=3E =3Cdiv=3E =3Cstrong=3E=3Cfont color=3D=22#ff0080=22 size=3D=224=22 face=3D=22Arial=22=3EDirect Email Blaster=3C=2Ffont=3E=3C=2Fstrong=3E =3C=2Fdiv=3E =3Cfont size=3D=222=22=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cb=3E=3Cfont color=3D=22#006600=22=3E=3Ci=3EThe program will send mail at the rate of over 1=2C 000 e-mails per minute=2E =3B=3C=2Fi=3E=3C=2Ffont=3E=3Cbr=3E Legal and Fast sending bulk emails =3B=3Cbr=3E =3Cfont color=3D=22#006600=22=3E=3Ci=3EBuilt in SMTP server =3B=3C=2Fi=3E=3C=2Ffont=3E=3Cbr=3E Have Return Path =3B=3Cbr=3E Can Check Mail Address =3B=3Cbr=3E =3Cfont color=3D=22#006600=22=3E=3Ci=3EMake Error Send Address List=28 Remove or Send Again=29 =3B=3C=2Fi=3E=3C=2Ffont=3E=3Cbr=3E Support multi-threads=2E =3B=3Cbr=3E Support multi-smtp servers=2E =3B=3Cbr=3E Manages your opt-in E-Mail Lists =3B=3Cbr=3E Offers an easy-to-use interface! =3B=3Cbr=3E Easy to configure and use =3B=3C=2Fb=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Fedeb=5Fset=2Ezip=22=3E=3Cstrong=3EDownload Now=3C=2Fstrong=3E=3C=2Fa=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cstrong=3E=3Cfont color=3D=22#ff0080=22 size=3D=224=22 face=3D=22Arial=22=3EMaillist Verify=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3EMaillist Verify is intended for e-mail addresses and mail lists verifying=2E The main task is to determine which of addresses in the mail list are dead=2E The program is oriented=2C basically=2C on programmers which have their own mail lists to inform their users about new versions of their programs=2E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3EThe program works on the same algorithm as ISP mail systems do=2E Mail servers addresses for specified address are extracted from DNS=2E The program tries to connect with found SMTP-servers and simulates the sending of message=2E It does not come to the message =3Cnobr=3Esending ‿=3B=2FNOBR>=3B EMV disconnect as soon as mail server informs does this address exist or not=2E EMV can find=3C=2Fnobr=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E =3Cnobr=3E =3Babout 90% of dead addresses ‿=3B=2FNOBR>=3B some mail systems receive all messages and only then see their =3B=3C=2Fnobr=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E =3Cnobr=3E addresses and if the address is dead send the message back with remark about it=2E=3C=2Fnobr=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cnobr=3E =3Cp=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Fbemv=5Fset=2Ezip=22=3E=3Cstrong=3E=3Cfont size=3D=223=22 face=3D=22Arial=22=3EDownload Now=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Fa=3E=3C=2Fp=3E =3Cp=3E=3Cstrong=3E=3Cfont color=3D=22#ff0080=22 size=3D=224=22 face=3D=22Arial=22=3EExpress Email Blaster =3B=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3EExpress Email Blaster =3B is a very fast=2C powerful yet simple to use email sender=2E Utilizing multiple threads=2Fconnections=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E =3Band multiple SMTP servers your emails will be sent out fast and easily=2E There are User Information=2C Attach Files=2C =3B=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E Address and Mail Logs four tabbed area for the E-mails details for sending=2E About 25 SMTP servers come with the =3B=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E demo version=2C and users may Add and Delete SMTP servers=2E About =3Cfont color=3D=22#008000=22=3E=3Cb=3E60=2C000=3C=2Fb=3E=3C=2Ffont=3E E-mails will be sent out per hour=2E"=3B=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cstrong=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Fbeeb=5Fset=2Ezip=22=3E=3Cfont size=3D=223=22 face=3D=22Arial=22=3EDownload Now=3C=2Ffont=3E=3C=2Fa=3E=3C=2Fstrong=3E=3C=2Fp=3E =3Cp=3E=3Cstrong=3E=3Cfont color=3D=22#ff0080=22 size=3D=224=22 face=3D=22Arial=22=3EExpress Email Address Extractor=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Fp=3E =3Cfont size=3D=224=22=3E =3Cp=3E=3Cfont size=3D=223=22 color=3D=22#008000=22=3EThis program is the most efficient=2C easy to use email address collector available on the =3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cfont size=3D=223=22 color=3D=22#008000=22=3E =3Binternet! =3C=2Ffont=3E=3Cfont color=3D=22#000000=22 size=3D=223=22=3EBeijing Express Email Address Extractor =28ExpressEAE=29 is designed to extract=3C=2Ffont=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont color=3D=22#000000=22 size=3D=223=22=3E =3Be-mail addresses from web-pages on the Internet =28using HTTP protocols=29 =2EExpressEAE=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont color=3D=22#000000=22 size=3D=223=22=3E =3Bsupports operation through many proxy-server and works very fast=2C as it is able of =3B=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont color=3D=22#000000=22 size=3D=223=22=3E loading several pages simultaneously=2C and requires very few resources=2E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont color=3D=22#000000=22 face=3D=22Arial=22=3E=3Cfont size=3D=223=22=3EWith it=2C you will be able to=3C=2Ffont=3E=3Cfont size=3D=222=22=3E =3C=2Ffont=3E=3Cfont size=3D=223=22=3Euse targeted searches to crawl the world wide web=2C extracting =3B=3C=2Ffont=3E=3C=2Ffont=3E =3Cp=3E=3Cfont size=3D=223=22 color=3D=22#000000=22 face=3D=22Arial=22=3E thousands of clean=2C fresh email addresses=2E Ably Email address Extractor is unlike other =3B=3C=2Ffont=3E =3Cp=3E=3Cfont size=3D=223=22 color=3D=22#000000=22 face=3D=22Arial=22=3E address collecting programs=2C which limit you to one or two search engines and are unable=3C=2Ffont=3E =3Cp=3E=3Cfont size=3D=223=22 color=3D=22#000000=22 face=3D=22Arial=22=3E =3Bto do auto searches HUGE address=2E Most of them collect a high percentage of incomplete=2C =3B=3C=2Ffont=3E =3Cp=3E=3Cfont size=3D=223=22 color=3D=22#000000=22 face=3D=22Arial=22=3E unusable addresses which will cause you serious problems when using them in a mailing=2E =3B=3C=2Ffont=3E =3Cul=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EEasier to learn and use than any other email address collector program available=2E=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EAccesses eight search engines =3B=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EAdd your own URLs to the list to be searched=3C=2Ffont=3E =3Cli=3E=3Cfont size=3D=223=22 face=3D=22Arial=22=3E=3Cfont color=3D=22#008000=22=3ESupports operation through =3C=2Ffont=3E=3Cfont color=3D=22#ff00ff=22=3Ea lot of=3C=2Ffont=3E=3Cfont color=3D=22#008000=22=3E proxy-server and works very fast =28HTTP Proxy=29=3C=2Ffont=3E=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EAble of loading several pages simultaneously=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3ERequires very few resources=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3ETimeout feature allows user to limit the amount of time crawling in dead sites and traps=2E=3C=2Ffont=3E =3Cli=3E=3Cfont size=3D=223=22 face=3D=22Arial=22=3E=3Cfont color=3D=22#008000=22=3EEasy to make =3C=2Ffont=3E=3Cfont color=3D=22#ff00ff=22=3EHuge=3C=2Ffont=3E=3Cfont color=3D=22#008000=22=3E address list=3C=2Ffont=3E=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EPause=2Fcontinue extraction at any time=2E=3C=2Ffont=3E =3Cli=3E=3Cfont color=3D=22#008000=22 size=3D=223=22 face=3D=22Arial=22=3EAuto connection to the Internet=3C=2Ffont=3E=3C=2Fli=3E =3C=2Ful=3E =3Cdiv=3E =3Cstrong=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Feeae=5Fset=2Ezip=22=3E=3Cfont size=3D=223=22 color=3D=22#008000=22 face=3D=22Arial=22=3EDownload Now=3C=2Ffont=3E=3C=2Fa=3E=3C=2Fstrong=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cstrong=3E=3Cfont color=3D=22#ff0080=22 face=3D=22Arial=22=3EExpress Email Address Downloader=3C=2Ffont=3E=3C=2Fstrong=3E =3C=2Fdiv=3E =3Cul=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#006600=22 size=3D=222=22=3EExpressEAD =3B =3C=2Ffont=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3Eis a 32 bit Windows Program for e-mail marketing=2E It is intended for easy and convenient=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3E =3Bsearch large e-mail address lists from mail servers=2E The program can be operated on =3B=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3E Windows 95=2F98=2FME=2F2000 and NT=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#006600=22 size=3D=222=22=3EExpressEAD =3B =3C=2Ffont=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3Esupport multi-threads =28up to 1024 connections=29=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#006600=22 size=3D=222=22=3EExpressEAD =3B =3C=2Ffont=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3Ehas the ability =3B to reconnect to the mail server if the server has disconnected and =3B=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3E continue the searching at the point where it has been interrupted=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#006600=22 size=3D=222=22=3EExpressEAD =3B =3C=2Ffont=3E=3Cfont color=3D=22#000000=22 size=3D=222=22=3Ehas an ergonomic interface that is easy to set up and simple to use=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E=3C=2Fli=3E =3C=2Ful=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont color=3D=22#008000=22 size=3D=224=22 face=3D=22Arial=22=3EFeatures=3A=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cul type=3D=22disc=22=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont size=3D=222=22 face=3D=22Arial=22=3Esupport multi-threads=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont size=3D=222=22 face=3D=22Arial=22=3Eauto get smtp server address=2Csupport multi-smtp servers=2E=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont size=3D=222=22 face=3D=22Arial=22=3Eauto save =3B E-Mail Lists=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E =3Cli=3E=3Cfont face=3D=22Arial=22=3E=3Cstrong=3E=3Cfont size=3D=222=22 face=3D=22Arial=22=3Eoffers an easy-to-use interface!=3C=2Ffont=3E=3C=2Fstrong=3E=3C=2Ffont=3E=3C=2Fli=3E =3C=2Ful=3E =3Cdiv=3E =3Cstrong=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Feead=5Fset=2Ezip=22=3E=3Cfont size=3D=223=22 color=3D=22#008000=22 face=3D=22Arial=22=3EDownload Now=3C=2Ffont=3E=3C=2Fa=3E=3C=2Fstrong=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cstrong=3E=3Cfont color=3D=22#ff0080=22 face=3D=22Arial=22=3EExpress Maillist Manager=3C=2Ffont=3E=3C=2Fstrong=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cfont size=3D=222=22=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cfont color=3D=22black=22 size=3D=223=22=3EThis program was designed to be a complement to the =3C=2Ffont=3E=3Cfont color=3D=22#800080=22 size=3D=223=22=3EDirect Email Blaster =3B =3C=2Ffont=3E=3Cfont color=3D=22black=22 size=3D=223=22=3Eand =3C=2Ffont=3E=3Cfont color=3D=22#800080=22 size=3D=223=22=3EEmail Blaster =3C=2Ffont=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cfont color=3D=22black=22 size=3D=223=22=3Esuite of bulk email software programs=2E Its purpose is to organize your email lists in order to be more =3B=3C=2Ffont=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E=3Cfont color=3D=22black=22 size=3D=223=22=3E effective with your email marketing campaign=2E Some of its features include=3A=3C=2Ffont=3E=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=3Cb=3E=3Cfont color=3D=22#008000=22 face=3D=22Arial=22=3E=3Cfont size=3D=223=22=3E‿=3BCombine several lists into one file=2E=3C=2Ffont=3E=3Cbr=3E =3Cfont size=3D=223=22=3E‿=3BSplit up larger lists to make them more manageable=2E=3C=2Ffont=3E=3Cbr=3E =3Cfont size=3D=223=22=3E‿=3BRemove addresses from file=2E=3C=2Ffont=3E=3Cbr=3E =3Cfont size=3D=223=22=3E‿=3BManual editing=2C adding=2C and deleting of addresses=2E=3C=2Ffont=3E=3Cbr=3E =3Cfont size=3D=223=22=3E‿=3BAbility to auto clean lists=2C that is=2C remove any duplicate or unwanted addresses=2E=3C=2Ffont=3E=3Cbr=3E =3Cfont size=3D=223=22=3E‿=3BMaintain all your address lists within the program so you no =3B longer need to keep all your=3C=2Ffont=3E=3C=2Ffont=3E=3C=2Fb=3E=3C=2Fp=3E =3Cp=3E=3Cb=3E =3Cfont size=3D=223=22 color=3D=22#008000=22 face=3D=22Arial=22=3E =3Blists saved as separate text files=2E=3C=2Ffont=3E=3C=2Fb=3E=3C=2Fp=3E =3Cp=3E=3Cstrong=3E=3Ca href=3D=22http=3A=2F=2Fwww=2Ewldinfo=2Ecom=2Fbj=5Fdownload=2Fbemm=5Fset=2Ezip=22=3E=3Cfont size=3D=223=22 color=3D=22#008000=22 face=3D=22Arial=22=3EDownload Now=3C=2Ffont=3E=3C=2Fa=3E=3C=2Fstrong=3E=3C=2Fp=3E =3Cp=3E=3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E=3C=2Fp=3E =3Cp=3E=A1=A1=3C=2Fp=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3Eif you want to remove your email=2C please send email to =3Ca href=3D=22mailto=3Atargetemailremoval=40btamail=2Enet=2Ecn=22=3Etargetemailremoval=40btamail=2Enet=2Ecn=3C=2Fa=3E=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =3Cfont face=3D=22Arial=22=3E =3B=3C=2Ffont=3E =3C=2Fdiv=3E =3Cdiv=3E =A1=A1 =3C=2Fdiv=3E =3C=2Ffont=3E =3C=2Fdiv=3E =3C=2Ffont=3E=3C=2Fnobr=3E=3C=2Ffont=3E =3C=2Fbody=3E =3C=2Fhtml=3E From pjt33@cam.ac.uk Sun, 09 Jun 2002 22:48:39 +0100 Date: Sun, 09 Jun 2002 22:48:39 +0100 From: pjt33@cam.ac.uk pjt33@cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc To save you some research... btamail.net.cn is a genuine site, apparently webmail. I have e-mailed abuse@btamail to complain. The apparent source IP address is 61.174.204.130, which is registered to Chinanet. Their contact address as given in the whois registration is hostmaster@ns.chinanet.cn.net With a bit of luck, this will turn out to be the culprit's ISP, and capable of cutting them off. I have also e-mailed this address to complain. Perhaps if they get lots of complaints they will decide it's easier to take the spammer offline? Worth a try? Peter ===== Peter Taylor pjt33@cam.ac.uk From hz211@cam.ac.uk Sun, 9 Jun 2002 23:11:40 +0100 (BST) Date: Sun, 9 Jun 2002 23:11:40 +0100 (BST) From: H. Zheng hz211@cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc Unfortunately I don't think Chinanet will act to such spams. They are the biggest and sole ISP that owns the fibres going out of China, and they are too bureaucratic usually. I suspect this is actually somebody in the US abusing the btamail.net.cn to send out spams and pretend the source is in China. Cheers, Han On Sun, 9 Jun 2002 pjt33@cam.ac.uk wrote: > To save you some research... > > btamail.net.cn is a genuine site, apparently webmail. I have e-mailed > abuse@btamail to complain. > > The apparent source IP address is 61.174.204.130, which is registered to > Chinanet. Their contact address as given in the whois registration is > hostmaster@ns.chinanet.cn.net With a bit of luck, this will turn out to be > the culprit's ISP, and capable of cutting them off. I have also e-mailed > this address to complain. > > Perhaps if they get lots of complaints they will decide it's easier to take > the spammer offline? Worth a try? > > Peter > ===== > Peter Taylor > pjt33@cam.ac.uk > > > _______________________________________________ > CST-2 mailing list > CST-2@srcf.ucam.org > http://www.srcf.ucam.org/mailman/listinfo/cst-2 > From jas88@cam.ac.uk Mon, 10 Jun 2002 07:52:49 +0100 (BST) Date: Mon, 10 Jun 2002 07:52:49 +0100 (BST) From: James Sutherland jas88@cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc On Sun, 9 Jun 2002 pjt33@cam.ac.uk wrote: > To save you some research... > > btamail.net.cn is a genuine site, apparently webmail. I have e-mailed > abuse@btamail to complain. > > The apparent source IP address is 61.174.204.130, which is registered to > Chinanet. Their contact address as given in the whois registration is > hostmaster@ns.chinanet.cn.net With a bit of luck, this will turn out to be > the culprit's ISP, and capable of cutting them off. I have also e-mailed > this address to complain. > > Perhaps if they get lots of complaints they will decide it's easier to take > the spammer offline? Worth a try? Nope. They are a NOTORIOUS spamhaus; the day they stop shovelling spam into people's inboxes, there will be a cool breeze in hell... It's actually bad enough that one of my mail filtering rules junks ANY mail which has passed through any PRC or South Korea mail server; I've virtually abandoned my Hermes account, it just gets too much foreign-language spam I can't even read! James. From gbmv2@cam.ac.uk Mon, 10 Jun 2002 16:22:01 +0100 Date: Mon, 10 Jun 2002 16:22:01 +0100 From: George van den Driessche gbmv2@cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc ----- Original Message ----- From: "James Sutherland" To: Sent: Monday, June 10, 2002 7:52 AM Subject: Re: [CST-2] ADV: Direct Email Blaster, etc [snip] > Nope. They are a NOTORIOUS spamhaus; the day they stop shovelling spam > into people's inboxes, there will be a cool breeze in hell... > > It's actually bad enough that one of my mail filtering rules junks ANY > mail which has passed through any PRC or South Korea mail server; I've > virtually abandoned my Hermes account, it just gets too much > foreign-language spam I can't even read! > > > James. If you're willing to tolerate a small delay in mail delivery, and also either a cap on the number of messages you receive a day, or having to pay for the service, then consider forwarding through a bigfoot.com address. The only spam I get this way is from bigfoot themselves, once every few months. But then, I have hardly any friends, so the rate cap doesn't bother me :) George From amj30@cam.ac.uk Mon, 10 Jun 2002 16:25:48 +0100 Date: Mon, 10 Jun 2002 16:25:48 +0100 From: Anthony Jones amj30@cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc On Mon, Jun 10, 2002 at 04:22:01PM +0100, George van den Driessche wrote: > If you're willing to tolerate a small delay in mail delivery, and also > either a cap on the number of messages you receive a day, or having to pay > for the service, then consider forwarding through a bigfoot.com address. The > only spam I get this way is from bigfoot themselves, once every few months. > But then, I have hardly any friends, so the rate cap doesn't bother me :) I've already had spam through my bigfoot.com forwarding address, despite never ever using it! Ant -- The obligatory bit about my email address: This account will be closed on July 15th 2002. Please use ant@cantab.net From sdm29@hermes.cam.ac.uk Mon, 10 Jun 2002 22:03:55 +0100 (BST) Date: Mon, 10 Jun 2002 22:03:55 +0100 (BST) From: Stephen McIntosh sdm29@hermes.cam.ac.uk Subject: [CST-2] ADV: Direct Email Blaster, etc > If you're willing to tolerate a small delay in mail delivery, and also > either a cap on the number of messages you receive a day, or having to pay > for the service, then consider forwarding through a bigfoot.com address. The > only spam I get this way is from bigfoot themselves, once every few months. > But then, I have hardly any friends, so the rate cap doesn't bother me :) > > George There's a free isp called connectfree (www.connectfree.co.uk). They give you many free email addresses when you sign up for their isp service, that has a (rather simple and featureless) web interface. They don't change you a penny, the site has no adverts, I assume there is file size limit but I haven't found it (and I was over 25Mb for a while) and I've been using the email for a year and I've never actually used the isp, so they apparentlt dont check. So its just a nice free email, if you dont want advanced features (it can view and send and that's about the limit). They send you a service update once ever month or two that is mainly useful info about their services and that's about the spam limit. ========================= Stephen McIntosh. www.stephenmcintosh.com