something to try
This is OT but whaddever, noticed that win (I run server 2k3) seemed a bit keen to force stuff onto disk even when there buckets of ram free so I disabled the pagefile and it made it noticeably snappier.
Worth a try.
Posts by BlueGreen
1205 publicly visible posts • joined 15 Jul 2008
Intel: Just 3,000 employees run Windows 7
MS and Oracle's big dev tools - who needs 'em?
Monday 1st March 2010 11:05 GMT
~in lieu of informative title~
FTR I prefer emacs.
> more structured code editing
what does this mean? What is unstructured code editing?
> ...and navigation
now this to me is the big win since I started using IDEs, that and, surprisingly trivially, having syntax coloured comments. I guess some of the smallest things of the most significant.
> Code quality improves with semantic analysis, which can detect uninitialized and unused variable errors.
I'd say it's the job of the compiler rather than the IDE to detect uninitialised/unused variables, but in any event if this is the level of semantic checking we're talking about then we haven't really moved on since the 80s
> ...differentiate between static, instance, and local variables
how this is supposed to be a significant step forward I can't imagine
> Here are some tips..
from experience with the opposite approach, I have to completely endorse this
I like emacs because I can use it anywhere but it does have its problems -- I have been able to find exactly one piece of halfway useful information about skeleton.el and it look like somebody dashed it off in a couple of hours.
I've been using the SQL Server 'management studio' for SQL 2008 and it is bloody awful. It can't even cut and paste correctly and I've lost hours of work from this alone, it has its lousy IntelliSense feature which I've just discovered how to disable. I don't think they did any user testing because it just gets on the way of the most basic tasks -- like having a new line happen when you press the return key (seriously), and it keeps crashing when memory gets tight (sometimes losing work). I've had to resort to OSQL for large batches.
The help is all screwed up -- you can't easily open two help topics in the same help window so you end up opening two help windows, each of which is a 50MB app. I tried to download BOL for 2008 so I could install it outside the VM where I run the server and -- for real -- it wanted me to install .NET! So I could read *the help*! All this crap and a whole lot more. What happens when you let the marketing droids drive, I guess.
They got a lot right in SQL server 2000 query analyser and they screwed up a whole lot by trying to get all big and corporate for 2008.
So it looks like it's back to emacs, again.
Experts reboot list of 25 most dangerous coding errors
Tuesday 23rd February 2010 21:03 GMT
@sed gawk, @jake
@sed gawk, as I'm not fond of beer and there's no whiskey icon it'll have to be the good lady . If you're interested in reliability, check this <http://en.wikipedia.org/wiki/SPIN_model_checker> which totally fails to explain what it can do, so try this <http://www.albertolluch.com/research/promelamodels> and look for 'deadlock'. One day I'll actually read the book I bought on it.
@jake: cheeky bastard, it's paris hilton for you too.
Monday 22nd February 2010 10:34 GMT
@sed gawk
> Thank you for your reasoned and well thought out critique, you have a genuine flair for language and the sheer variety of your response was most refreshing. ;)
Positively Wildean, I grant
> Well since you ask java doesnt' do deterministic finalization because [reasons]
It's more complicated than you made out. Generational, Mark-sweep, compacting -- none of these can immediately pick up all dead objects. The only thing that can approach this is reference counting which has other problems (speed, overhead), and still can't make immediacy guarantees (consider cycles).
More: <http://msdn.microsoft.com/hi-in/magazine/bb985010%28en-us%29.aspx>, look for "There are several reasons for this" for a summary, but this doesn't do justice. Conflating garbage collection and object finalisation was recognised as a bad idea back in the 80s by Modula 3's designers, but Java's creators were too witless to learn the lesson (like so many others they fail to learn), so Microsoft had to follow on and we all move backwards, again.
> think peer nodes associations
I would if I knew what they were.
> 5) Strategies exist in languages with determinstic finalization e.g. allowing resource management to be implicitly bound to object lifetime with the language doing the work rather than the developer.
I think you are thinking of C/C++ where object lifetimes are explicitly and sharply delimited by free() or implicitly by subroutine returns. It is reasonable, I suppose, to ask that Java provide some kind of equivalent to smart pointers so things happen automatically on subroutine returns, but there's not much you can do about other objects you intend to have longer lives. If you want deterministic deallocation for "local" variables, you have to wrap a try/finally around the routine body.
> 6) A data point, there are plenty of garbage collectors for C/C++ yet GC is not that widely used for what ever reason, perhaps because of point five).
These are conservative garbage collectors (I'm sure wiki has an article. Read up on them and have nightmares) and they don't provide the behaviour you are asking for.
> 7) These stategies are examples of designing out problems rather than coding round them, using say explicit calls to synchonization primitives like mutexes.
I don't know what you're saying that I have a tendency to build frameworks to hide grubby detail in the same way that you suggested wrapping realloc(), but typically on a bigger scale. I guess that's easy to say though.
13)I don't think resource/memory management too big a deal,
then you're a better person than I.
> or here's a thought write a C/C++ extensionn for perl/ruby/python whatever and get the best of both worlds, memory managed access to all the C/C++ libraries through a thin shim layer.
I don't see how this would get you the kind of deterministic memory management you want.
> (Stuff about concurrency)
I'm not saying that concurrency was easy or irrelevant, only that you were speaking too generally, and if you like Erlang's model, perhaps you should recognise that it is a model and not a language, and see if there is a framework to support what you want. Perhaps this is of interest <http://www.google.co.uk/search?hl=en&source=hp&q=actor+framework+%22c%2B%2B%22&btnG=Google+Search&meta=>
> (re. erlang) I don't have memory leaks,
Hmm. Is that because Erlang has a garbage collector?
> erlang solves that for me by inconnecting C/C++/whatever consumer/producers and letting them ignore concurrency completely
if it's that simple, why are you using Erlang? Just for producers/consumers? I'm missing something.
If you want deterministic finalisation, here's how to do it: work out the exact rules that fulfil the requirements, then slap a preprocessor over Java to provide it. How's that?
Castleford locals storm Tickle Cock bridge
Dell servers block un-Dell HDDs
Thursday 11th February 2010 10:27 GMT
Short story
I bought a small server as I wanted something reliable. HD died after 14 months (shortest time of any disk I've ever had. One in another machine still running after >10 years). The vendor's support said I should be able to plug in any sata drive (they didn't sound enthusiastic but they said I could) so got a caviar green (nearest thing to hand, shopwise) and it ran slowly, and steadily got worse (weirdly, over couple of months, slowly) until machine was unusable.
Could not work it out, assumed it was software problem, drilled down with perfmon, assorted sysinternals tools etc. Checked for rootkits etc. Got nowhere. No errors, just increasing propensity to thrash madly. Finally sussed that it was likely the firmware was incompatible. The caviar green is not made for speed and spins down. Replaced with a caviar black, now it's usable. Not outstanding but so much better. Lesson learned, painfully.
Why didn't I get a vendor-recommended drive? Well, new I was quoted £250 for a 250 gig drive. When I asked what I was buying for that price they said reliability, and when I asked them what the guarantee was, they said 3 months - WTF?? What sort of reliability does a 3 month promise imply? says I. They didn't answer.
The original drive was a barracuda ES.2, available from the web for about £80 for same capacity.
Researchers penetrate last bastion of Windows security
Thursday 4th February 2010 13:39 GMT
defeating DEP is worrying
It's not clear from the article how they did it unless it really was as simple as it sounds, that the flash compiler + environment can't be depped because that would negate the compiler, so a hole is left. Can anyone clarify?
@gimbal: I can only speak for myself but I don't allow java in the browser either. I've noticed when it comes preinstalled it seems to stick itself everywhere and nag you repeatedly for upgrades. Also ask yourself if these complex in-browser UIs are needed at all (Hint: 97% of the time, no).
Ingres' VectorWise rises to answer Microsoft
Tuesday 2nd February 2010 21:17 GMT 
say.... what?
> ...faster than rival computational, client products that run on a single, multi-core machine.
what?
> VectorWise achieves this by allowing standard SQL applications to access the compute power on multi-core chips while serving up huge blocks of data, Ingres said
err... what?
> In some cases, people are writing hand-coded programs against flat files.
eh?
> Instead of managing data buffers in random access memory, VectorWise uses the cache on the underlying chip, while all instructions are loaded from cache onto a multi-core chip
??
> VectorWise also transfers a large amount of data to get around the fact that while processors are speeding up, discs can only go so fast: VectorWise will read up to 500kilobytes of data compared to the usual rate of between four and eight kilobytes
bloody what?
I've had a good giggle, now for some clarification please. Regfail logo seemed right at first but I think this is a windup so here's one better
Zoo man uses virtualisation to tame his server herd
Saturday 30th January 2010 00:32 GMT
@Destroy All Monsters
Maybe it matters if you have a mobile browser (phone, wireless dongle etc) which typically has limited monthly capacity.
Or trying to get info from within another country with restrictive govt policies, where smaller is easier to smuggle.
Or live in an undeveloped country that has limited bandwidth (not everyone lives in the west).
Or still use dialup, like our jake (an he's not unique IME).
Or feel that text + static pics can often put things over better.
Or feel that just because there's an apparently abundant resource doesn't mean we are required to burn it as fast as possible.
If none of that's an issue, there's security to consider.
Swiss computing lab offers free bug-immunity tool
Wednesday 27th January 2010 22:03 GMT
@Ken Hagan
It is very misleading. Paper they provide is interesting (skimming it now) but does not back up the snake oil garbage complete with cloud reference. It's not an external addon but 'Avoidance code can be directly instrumented into the target binary or can reside in a thread library'. It builds resource graphs and looks for cycles, in essence, and adds some sniffing code after cycle is detected. Pretty skanky solution. It should be possible to provide this stuff anyway, at lower resource cost, directly in the code. There are model checkers that can detect these things statically I understand (spin). Better than nothing I suppose.
Oil companies hit by 'state' cyber attacks, says report
Tuesday 26th January 2010 16:39 GMT
@Charles 9
I really should not be arguing as I don't have the time and certainly don't have the knowledge, but that's me I guess.
>Basic security isn't as effective against an adversary with near-unlimited funding.
No, I guess, but it's about pushing up the costs to the adversary till it's not worth it. Cost may be more than money, it may be criminal charges, risk to reputation (could destroy a company, severely damage a country financially) etc.
> Whitelists won't work when they'll simply find a colleague's e-mail address already in the list.
Idea is that in combination with digital signatures + a totally separate network (vpn? backed up by large on-time pad?) for business emails makes that much harder. (WTF are they not using encryption here at all. Fuck. They need mincing down into dog food.).
> ...some social engineering tactics ...
okay, there's that but social engineering around major technical blocks is going to be much harder. But point taken anyway.
> Digital signatures [...] create a collision.
That's fine. Crack one layer with great effort others remain. At each point they have to probe which leads to risk of detection.
> Some browser exploits attack the browser DIRECTLY ...
But they have to then 1) know the browser type 2) gain access to it 3) compromise it 4) have it reach out over a whitelist-only proxy. etc. Break one layer, another remains, and they start becoming very noticeable. A true zero-day vuln is one crack in several layers.
Dunno about privilege escalations but I'm sure you can increase the cost so much it's not worth it to them. And there's plenty more basic stuff I could suggest before we get to anything fancy like rewriting stuff in 'safe' languages.
Now's the time to mention you're a security consultant with 30 years experience.
@AC 10:52: point well made but anticipated. look for the phrase 'get it' and the word 'shoot'. The technical is the easy stuff I grant. People are hard.
Tuesday 26th January 2010 03:02 GMT
Basic security should cover it.
Whitelist email senders. Dissociate business emails from home, pref. with separate accounts. This stops the emails getting in.
Use digital signatures? Ensures it comes from who it claims.
Install a web proxy with whitelisted web access only. Speaks for itself.
Turn of java, javascript & flash (this is getting boring). This closes a major hole. And remove/IE totally if we're talking windows.
Run the accounts as user rather than admin. Blocks things further.
If necessary prescan emails for proper structure & a subset of functionality. Block those that don't conform.
Use a completely internal system for internal work/emails etc. Reject HTML emails.
Store critical data completely isolated from elsewhere. Provide business-only laptops with USB ports etc physically disabled, with encryption etc and no external access except whitelisted. Or don't let them take this work home with them at all - physical theft is a fine means of hacking and one that would be much higher if data management wasn't so inept.
Install anti-virus stuff, maybe.
Provide security training and shoot execs too stupid to learn.
I'm not an expert and some of the above may be misguided but just simple changes should make it orders of magnitude harder for hopeful intruders. How much more of this is going to happen before they 'get it'. I'm not sure they will.
@Matthew 13: I guess security is more to do with admin and good practices than with the OS. I've run windows for a long time and never had virii, malware or anything nasty. That's for the last 2+ years without any security software (norton etc.) at all.
Police arrest MD of dowsing-rod 'bomb detector' firm
Sunday 24th January 2010 05:56 GMT
@Geoffrey Summerhayes
Scientific testing is where it's at, and the bloody great ringing chasm between what science finds and what others find (q.v. Steve Gill, Trygve Henriksen and perhaps myself) is the most curious part of it, and your cynicism is understandable. I gave my experience only as that, not evidence for or against. If I try it again and it works reliably, I'll put it to better test.
As a kid I had some idea about how pipes ran so that doesn't bother me. The rest of your criticisms do stand up rather well though. Fair post.
Friday 22nd January 2010 16:55 GMT
@various
first up, typo: s/was not so surprising/was so surprising/
@Rich 11: I was probably 10 or 11, it was a trashy joke thing. About your little brother, you should have told him it works better when rubbed with cow poo & let him get busy. It's what little brothers are for.
@Peter2: does a coin constitute a large metal object? Assuming they are magnetic it would take a bloody strong magnet to affect hand-held metal rods held three+ feet above. It more plausible for the pipe though, but still a stretch. Possible... Bear in mind that the rods were not pulled towards the object, but together, to cross in the stereotyped dowsing way over it IIRC.
@Steve Gill: thanks, very interesting.
Friday 22nd January 2010 13:06 GMT
Well, my experiences with dowsing
Putting aside the utter crap described in this article, when 'ah were a sprog I was given a dowsing kit comprising two metal rods and two placcy holders wherein they could swivel freely. It was only meant as a joke but trying it out, and without any expectation of them working, they did start to move near metal. This was not so surprising that me + bros experimented by them hiding a coin (say, under one of three cups) when I was elsewhere and me finding them quite reliably.
I say quite reliably because I started getting false positives - but consistently in the same place. Wandering around and marking where I got these we found they formed a straight line. Pulling up the carpet, there was a faint but definite scar in the concrete below (the house was concrete, hence no nails. Just realised that). We assumed this was a pipe. Said scar crossed the room at a right angle, as a pipe would, and seemed to fit in with where the kitchen would be.
So, I'm not going to dismiss dowsing outright. It needs scientific testing even if it looks ridiculous, and what we I found was a phenomenon that was testable, repeatable, reliable and with which I could gain new data (assuming the pipe I found meant anything). NB. the range was very short; you had to be almost over it.
One day I'll try that experiment again.
In the meantime, I mention this experience here without posting anon in the hope that it could be read with interest and due scepticism, but without some enfilade of twattery posts following.
NNB I'm aware of the clever hans effect. I does not explain the (presumed) pipe we 'found'.
I'm going to regret this... <hits submit>
Is Mandy right to cut science funding?
Monday 18th January 2010 13:55 GMT
Actually there's something in this but it works both ways
I've had experience with both academia and industry (more in the latter) and there's an attitude in academia often of being indifferent to the applications of their work, sometimes even scornful (yes, really). So they don't push their work over onto the dirty, real-world, side of the fence where it would be stressed, mangled and generally tempered until it was either improved and used, or finally shown to be nonviable. There's too often too much theory without practice there.
And to our the business side - they are so busy getting stuff done that they are ignorant of useful prior work and do it again badly, or are outright scornful of theory (yes, really, I've heard the disparagement from this side too). Sometimes theory is just scarily hard and would involve, you know, doing some reading etc.
There are plenty of exceptions but still comprise a small minority. Both sides need to build bridges. Basic research matters of course, but not as much as getting them co-operating and using what's already been done, and in this I have to blame business more than universities.
Firefox 3.6: I am more than my Monkey
Friday 15th January 2010 11:32 GMT
@heyrick, @Trixr, @Bob Gateaux
agree with 1, 6 & 7 very much. Regarding 2, FF does this. tools:options:privacy and under the 'keep until' dropdown, select 'ask me every time' (provided you've picked 'firefox will: use customsettings for history' first). Should be what you want. 3 & 5 are typically dealt with using blocklists (some addon will do this, use squid myself so don't know). Yu ask for more flxibility that usual (damn this keyboard!) but that's rarely needed; you tend to block everything from a given domain, not selectively. 4 isn't fair - that's the job of an addon.
@Trixr: teh OS will cache read files anyway so if FFjust adds caching that's redundant. If it's any faster then they will almost certainly have done something else. The bookmarks/history/etc db is pretty small anyway.
@Bob Gateaux: stop yer trolling
What I don't like about FF (well, one gripe) is the lack of docs. There's a lot of functionality that I can't use e.g most people will know that if you type the forward slash / the you get a quick find box, but if you type the apostrophe you get the same but restricted to URLs - try it. There about:config stuff & loads more - just write it down somewhere, developers! It's too good to waste.
Thursday 14th January 2010 14:42 GMT
Never happened to me, mate
I've used it for years (with and without squid, which I basically use as a blocklister). And regarding other's complaints of memory, I've regularly had 700 tabs open at once and it used 1.2 gig.
May I suggest now as I've done before that it may not be the browser proper - I run without jscript or flash. Try disabling each then both of those and try reproducing it the problem.
Also, when it does start cpu-banging, open the task manager (sorry, I use windows - I mean poke around with the mac/ubuntu equivalent) & have a look.
Browsers could host a (simple) database
Tuesday 12th January 2010 15:33 GMT
@Russell Howe
I believe that cookies are no more than 4KB. Also some people deny cookies by default so a different mechanism is necessary (could fudge it perhaps but messy. Dunno). DB fits the requirements.
However, I do hope they consider the abuse-potential - tracking records instead of cookies. SQL injection on the client if they go down that road. etc. Yuck.
Boris bares data bunker
Thursday 7th January 2010 14:31 GMT
extract of animal attacks db in xml from site
<ROW>
<Ward_Code>00BJGG</Ward_Code>
<Ward_Name>Latchmere</Ward_Name>
<District_Code>00BJ</District_Code>
<District_Name>Wandsworth</District_Name>
<Count_12-2007/>
<Count_1-2008/>
<Count_2-2008/>
<Count_3-2008>1</Count_3-2008>
<Count_4-2008/>
<Count_5-2008/>
<Count_6-2008/>
<Count_7-2008/>
<Count_8-2008/>
<Count_9-2008>1</Count_9-2008>
<Count_10-2008/>
<Count_11-2008/>
<Count_12-2008/>
<Count_1-2009>1</Count_1-2009>
<Count_2-2009/>
<Count_3-2009/>
<Count_4-2009>2</Count_4-2009>
<Count_5-2009/>
<Count_6-2009/>
<Count_7-2009/>
<Count_8-2009/>
<Count_9-2009/>
<Count_10-2009/>
<Count_11-2009/>
<Count_12-2007_to_11-2009>5</Count_12-2007_to_11-2009>
</ROW>
.
Fold the date metadata right into the tags why don't you. Include non-events. Add pointless summary.
<sigh/>
Firefox 3.7 to feel need for speed with multicore boost
Thursday 7th January 2010 09:32 GMT
@Neil Stansbury
My criticism is the execution of arbitrary JS fetched from web pages. Have you noticed how many vulnerabilities there are with it? Try this <http://www.google.co.uk/search?hl=en&source=hp&q=site%3Atheregister.co.uk+javascript+vulnerability>
I'm not complaining about speed but security. Again: security, not speed.
I am vaguely aware that FF uses jscript with xul internally, I don't know what the chrome stuff is nor its security implications (if any). I'll take a look.
If you feel I don't "understand the web today" then please be more precise - what exactly don't I get? That unnecessary scripting must be used? That security comes last?
Wednesday 6th January 2010 18:04 GMT
I care not about javascript
I disable it. It never runs except in a vm on rare occasions. JS is necessary sometimes but mostly it's just a security risk. Crank up the speed and more idiot web managers/designers/commissioners will let it in. If mozilla issued guidelines on JS (non-)use the web would be a better and safer place.
Outrageous new means of megastar demise spotted
Tuesday 5th January 2010 21:09 GMT
cause of sock-knock-pop
was very curious about the antimatter claims here as I didn't know how it could possibly be produced nor how it could accumulate to cause a significant explosion so, as ever, wiki to the rescue. It seems that antimatter can be produced thusly <http://en.wikipedia.org/wiki/Pair_production>, this article leading on to the actual suspected mechanism in <http://en.wikipedia.org/wiki/Pair-instability_supernova>, which says something fractionally more prosaic may have happened:
" For very high mass stars, with mass at least 130 and up to perhaps roughly 250 solar masses, a true pair-instability supernova can occur. In these stars, the first time that conditions support pair creation instability, the situation runs out of control. The collapse [caused by the redirection of energy production from photons, pressure from which keeps the star 'inflated', into the production of antiparticles, which presumably somehow don't] proceeds to efficiently compress the star's core; the overpressure is sufficient to allow runaway nuclear fusion to burn it in a few seconds, creating a thermonuclear explosion. With more thermal energy released than the stars' gravitational binding energy, it is completely disrupted; no black hole or other remnant is left behind."
So by my reading it's not directly antimatter driven and I really must must must stop getting distracted by Lewis' articles and do some bloody work instead.
@Chris W: about the size of the turtle at the bottom. P'raps.
English language falls to the Slashdot effect
It's the end of TV as we know it
Tuesday 29th December 2009 17:00 GMT
If it is true on-demand unicast then what's the bandwidth requirements going to be?
How about some figures, and a comparison with these to what current pipes can provide, and the financial implications of adding more infrastructure if necessary.
And while I'm here, hope everyone had a good xmas.
US warships to get plane-snatching robot arms?
AMD cuts to the core with 'Bulldozer' Opterons
Wednesday 16th December 2009 00:58 GMT
@John Savard
Not an expert but it's too easy an assumption that one can increase the number of instruction stages. Pipelines can already be quite long and each stage would have to take about the same time to execute or it would stall but for example the execution can be very variable, from a simple shift to an integer sqrt (some PPCs had this I think), and a mem fetch can be getting on two orders of magnitude difference depending whether it hits L0 cache or misses everything and reads ram. Also longer pipelines = longer bubbles. I don't think AMD techs are would miss something so obvious.
German shoppers slug it out with salami
Monday 14th December 2009 18:54 GMT
@Well ...
What was a cathedral with buttresses rising from the ancient cobbles outside, built from weathered blocks the size of a car with gargoyles gurning down at you from so far up you have to squint - and within, silence and peace but for a distant monk's murmuring on his knees by a single candle, and a old, old reliquary with bones of Charlemagne on show next to the quarter-price jaffa cakes. That's how I like to imagine an Aachen Lidl.
Mozilla to open - gasp! - Firefox add-on store
Saturday 12th December 2009 23:32 GMT
problem is plenty of selfish users
read the comments on, say, NoScript. There's some nasty, demanding feedback from petty sods who feel Giorgio Maone owes them for the indulgence of putting his free and Most Excellent addon into their browser. *Dammit* he owes them! If I were him I'd have issued a two word press release and walked. They don't recognise something for nothing, conceiving of something for real money is beyond them.
If this payments scheme works, we'll just have to live with a lot of parasites, but I feel it'll collapse under their combined leeching.
@Notas Badoff: fershur fershur!
A Deadlock Holiday
Thursday 10th December 2009 15:18 GMT
% Exploit the fact that ∞! = ∞;
inf with nan behaviour?? Aaargh, you actually got me checking it!
Some outstanding stobbery <http://james-iry.blogspot.com/2009/05/brief-incomplete-and-mostly-wrong.html>.
Also thanks soooo much for the mention of desktop tower defence. I had to find out what it was, lost weeks.
Brown launches 'Zip it, Block it, Flag it' net code for children
Tuesday 8th December 2009 13:06 GMT
Another braindead initiative.
witty, catchy phrases that would make a kid cringe (per @Ken Hagan) and no forethought will emasculate this one too.
> But we must ensure that the virtual world is as safe for them as this one
The virtual world is a part of the real one and neither are 'safe', and the pretense of safety should not be attempted or the sprogs' sense of danger will never be tested and so remain underdeveloped into adulthood. And maybe more kids get damaged by their parents than by strangers, how about some stats Mr. Brown, just a thought.
> as well as get help and advice on issues such as [...] viruses and hacking
or just have an account where the kid doesn't have the rights to download & run executables? Just for once an actual existing technical solution to a problem.
> Highlight any suspicious individuals, activities or websites to the relevant authority, including site admins, teachers or even police.
You haven't even the guts to deal with criminal spam, now you want to do this.
Have some balls and fix the underlying problems.
Combat walker machines: $3m for new studies
Mozilla lights fire under Thunderbird
Monday 7th December 2009 09:53 GMT
alternative email client
Never been happy with tbird; just installed it & very first thing it did wasn't what I expected. Bug report filed. sigh
For normal pop browsing I really like this: <http://www.ultrafunk.com/popcorn/>. It's technically unsupported and it isn't totally bug-free but it's pretty good and too simple to be an way in for virii etc. Pure POP3, *very* simple, only every used it to browse stuff on the server (it can download mails but I've never used that facility).
Used it for a long time, recommended.
Intel puts cloud on single megachip
Thursday 3rd December 2009 12:59 GMT
@Steven Jones, @Ken Hagan
> "Guessing" that throughput can be estimated from power output is exactly that - guessing.
Yes, and moreso without info on cache, clock, architectural details etc. It was a guess. I was trying to analyse their implication (not claim!) of how speedy it is. I'm aware of clock/power tradeoff (though not the precise tradeoff; need to get hennesey & patterson - the 20%/double-the-cores ratio is higher than I'd expected)
> Throughput per watt has increased enormously over the last few years, and as a new chip, then I'd expect this one to have gained some benefit.
pretty much the same benefits as an equivalent process, hence my comparison with a modern quad-core - I *assumed* they were the same process. They didn't provide info on it and I'm aware it could have been an older process. I still suspect that the wattage/heat is the most reliable indicator of speed if I can interpret it properly...
> especially if it supports single consistent memory models
by my reading of the article ("The cores communicate by means of a software-configurable message-passing scheme using 384KB of on-die shared memory."), it doesn't. More load on the programmer, less on the hardware, and good thing too IMO. We need to go down that road. Also see my comment on microsecond comms overheads (I'll save you the effort: "ouch")
> (the AI crap) - we're agreed.
@Ken Hagan:
> If you are willing to accept (say) half the per-core performance, the per-core power consumption drops by an order of magnitude.
This confirms how little I know about this.
> And the programming model...
I'm not disputing that, just performance
> finally proves Sun were on the right track with Niagara.
I'm damn sure that general bulk-SMT architecture ***was*** right for the majority of computing. It seemed obvious to me that we were going down a wrong alley just by the incredible complexity of modern cores.
But I was purely trying to inject some reality into intel's marketing.
Thursday 3rd December 2009 01:27 GMT
actual computing speed
it's clearly labelled a research project but I'd like to know what it's best current performance is. They say the cores are simplified to in-order instruction issue and probably fewer instruction dispatches (wiki says atom can do 2 issues per core, I'll bet top end processors can do 3 or more) so that's a significant constant slice taken off, then there's undoubtedly smaller caches (can't find spec for this) which will be significant hit, depending on workload of course, and clock speed is also unspecified, probably for good reason. And from their press release "Application software can use this network to quickly pass information directly between cooperating cores in a matter of a few microseconds..." a few microseconds??? ouch!
My guess is that performance can be estimated from the power output and separately, number of transistors. 125 watts corresponds to a decent modern quad-core IIRC, and the number of trannies in my dual core is ~300 million, so I'd guess real world performance is equivalent to a decent spec quad core, very roughly. So while it's a laudable and very valuable piece of research, I doubt it's the rocket they're implying.
@Louis Savain: back again, troll?
Should you lose your religion on your CV?
Monday 30th November 2009 14:35 GMT
My middle-eastern name + my experiences
my name is almost unique in the UK. It's not immediately apparent as a 'moslem' name but I suppose given my background it is.
Experiences? It hasn't made a whit of difference jobwise anywhere. The two criteria that have mattered are, can I do the job and how well I get on with people.
Pentagon world-sim tool making good progress, say profs
How blogger kept Belle de Jour's ID secret
New 'reversible' paralysis-ray turns victims blue, flaccid
Intel: 3D Web to save HPC
US boffins build, test working 2-qubit quantum processor
Monday 16th November 2009 13:42 GMT
@Michael Shaw: computation V accuracy
I think you're asking a question I've been wondering about ever since this quantum crap started, namely that they were promising too much to be credible. Quantum computation apparently grows (from what I've read) non-linearly as qbits are added [*], yet accuracy is always the problem. I've long wondered if there's a fundamental link between computation and accuracy in that there's an upper limit to one which, if exceeded, starts to eat into the second. But no-one's ever raised this point that I've seen. Anyone here know any better?
[*] for a given unit of computing power, you can expand this by a cube power in 3 dimensions ie. a box 1 foot per side holds 1 unit of computation, double the boxes in each direction & you get a box 2 foot per side & holding 8 units of computation. You can't do better in 3d space, but quantum promises a much higher scaling, so there's a conflict.
Biting the hand that feeds IT
