What's with the "Gran Turismo(?)"... what bleeding question mark? Gran Turismo is a *great* game, and when played properly encompasses many of the skillsets that are needed when playing any game well:
* The optimal distribution of finite resources. * The creation of short, medium and long term strategies. * The absolute, positive *need* to enter a mental state of heightened concentration while remaining relaxed. Much like Rocket Arena on a good day, actually.
The only thing I can't do is rail my mate through the back of the head.
I guess it's the phrasing, and not really the content that keeps dragging me back to this. The content's good, but I don't think my reply was in any way utopian.
I didn't, for instance, advocate ditching.net altogether and going for a mod_perl solution. Neither did I (and this will probably always make me laugh) advocate mono. All I suggested was taking the most expensive, slowest performing and most easily ported bit and moving *that* to OSS.
So, while this isn't advice with scaling.net, per se. It *is* advice regarding scaling.net for those on a limited budget.
let me apologize for all the nutheads who say "drop MS - use Linux"
And why, exactly, is this a nuthead reaction? Our original poster has hit some major problems with a lump of technology that are, essentially, entirely financial in nature. Basically he's saying "we've developed in.net but have discovered that we're going to have to spend $big PER CLIENT to roll the bloody thing out".
Yes, they've been bait'n'switched - and he'll probably do better technology assessments next time, and as you point out that's not really the point. They need to sit down, have a look at how much it would cost to move it to some kind of hybrid approach and the long term gains they stand to make.
Personally, I'd drop MSSQL and put Postgres on a dedicated box. Leave the.net application more or less as is and scale it across a couple of application servers via an el cheapo load balancer such as the coyote point product.
Holy shit! What a monster opportunity for a scam: Randomly IM MSN users saying "you've won $1000, put your credit card details here http://microsoft.mydomain.com/easyscam.pl"
I don't understand what the problem is here. You're saying that you like XML, but it's slow. Fine, don't use it. It's not like it's the only tool in existence, is it?
[disclaimer: the following is based on research, not personal experience]
So, I've been looking at using 900MHz for some low bandwidth stuff. Primarily because 2.4GHz can be a PITA to deploy - near line of sight (over distance), trees, rain etc.
Anyway, it appears that the products available round 900MHz have taken a completely different, and much more "agricultural" approach to the problem. Basically they're 900MHz FM radios with a modem bolted on. No clever multipath resolution, no time division, no orthogonal frequency division (outside that which the modem does). Consequently the amount of bandwith available is about seven eigths of fuck all - a claimed 115Kbit (being the limit of the RS232 connectors they use) - with something rather worse being the practicality of situation.
Security is kinda interesting though. Quite a lot frequency hop across a range of channels within the spectrum. It appears that you need to get both radios, from the same manufacturer, and put them on the same hopping scheme before they'll talk.
you could spend another month or so hand vectorizing code for a tenfold increase in speed
You could do, or you could spend a month being scared of it, then actually get it done in a couple of days (when push comes to shove), which is what I did.
I recently managed to get MMX code into some software I'm wirting using GCC instrinsics and, y'know, it wasn't that hard. The hard part is architecting the application as such that the data just happens to be hanging around in exactly the right structures in the first place.
Someone please explain to me how 5177 MFLOPS and ~300 MFLOPS are even comparable.
They're not, which is what makes this whole benchmark so entirely useless.
Look at it: The conclusion, basically, is that there's no point in running CFD code using scalar FP. So why didn't they port their code to SSE2? P4's, and particularly the new 800MHz FSB P4 get data through SSE2 code like there's no tomorrow.
Nah, I'll listen when someone compares SSE2 and AltiVec properly. Until then it's just more blah. Don't get me wrong, I'm rapidly turning into the biggest Mac fanboy you've ever seen (Cocoa, since you ask) but the G5's are not the quantum leap Apple are making them out to be. Back in contention? Sure, but I promise you a dual Opteron 2GHz will blow the doors off a dual G5.
At least from a local (New Zealand) perspective, this is totally wrong. Let me elaborate.
Here the ISP and the phone company are completely different people. Telecom scamper around plumbing in DSLAM's etc. etc. and wire them all together to make a grande country wide ATM network called (ironically) IPnet. The point with IPnet is that it gets packets from your subscribers ADSL routers to some boxen in your datacentre, then you deal with it from there, including international backhaul. Telecom charge you, basically, $60/month/subscriber for this service.
Therefore, fixed costs of setting up an ISP in New Zealand = fuck all. Marginal costs = lots.
Under these circumstances, the speakeasy plan would work a treat. Imagine I own an ISP and charge $100/month for an ADSL connection, but you can connect your neighbours via WiFi and get $50/month back. So, each ADSL connection I get $40 (because $60 goes to Telecom), and each WiFi piggybacked connection I get $50. Woohoo! Quids in, with Telecom footing the worst of the bill!
Unfortunately it doesn't quite work like this. The $60/month buys you a very limited quantity of throughput - 512MB/month in it's most basic form. Yes, you did read that right, less than one iso. Per month. After this it's $0.20/MB, enough to kill the connection sharing thing outright.
But... shit. 512MB/month, and IPnet craps itself, like, all the time. The funny thing is that Telecom have no idea why their broadband takeup is so low. Really. NO idea. They also have no idea why competitiors are startingto appearbloody everywhere.
I don't wear this "we used the SSE flags" shit for one second. All you're saying (give or take a bit) when you set an SSE flag is that it may be used. It's not like it magically parallises your code, or aligns the data on the right boundaries.
That being said I would like to see a G5/AltiVec vs P4/SSE2 number crunching bake off. Arguably these are the only benchmarks that matter in this day and age since the applications that need that much speed should be using the parallel instructions anyway.
The PC demo of splinter cell was completely unplayable. After the noises the XBox fanboys were making I was hugely keen to give the game a go. I downloaded the demo, it fell over. Some googling revealed a collection of hacks to.ini files that made it work, mostly. In a window. And then it was just a bit crap.
Note to self: XBox fanboys are just that. Save money and effort for GT4.
simple, sealed box that could browse the web, read email, play audio and video, keep track of appointments and todos, and maybe do simple spreadsheets, word processing, and sell it to first-time consumers, you'd make a fortune.
It's called an iMac. And, yeah, they did OK out of it.
I doubt that Apple will be dropping G3's any time soon. IBM apparently has an AltiVec enabled G3 in the wings, and my take is that these will remain in iBooks while PowerBooks will keep getting G4's and potentially these new SOI G4's from Motorola. The 970's will be PowerMac territory only, and they'll cost a ton.
Mind you, the rapidly closing performance gap between iBooks and PowerBooks has got to be a cause for some concern.
It's really, horribly complicated. Basically the router has to build as little of the TCP stack as possible in order to look at the actual, data contents of the packets to decide what application is being tunnelled.
How the hell was Unreal Tournament an original game? The thing that made UT such a big hit was that the quake series were (and still are) complete bastards to set up complex servers with... as anyone who has seen the insane map rotation scrips will attest to, let alone the lunacy associated with getting a rocket arena dedicated server going.
No, UT was common sense. Let's get this thing that people want to do and make it easy.
Because they have a horrifying amount of legacy code to maintain, on what is basically a legacy architecture - i.e. the PC.
Apple got to throw away all their mistakes when they started making OS X. They don't need to support nearly so many hardware experiments - ISA, VLB, MCA, assorted stupid methods of getting to "high" memory, fifty different ways of using large hard drives etc. etc. They also don't need to support a wide assortment of "good idea at the time" legacy technologies, DCOM and others of their ilk not to mention Netware drivers and similar amusement.
Given how large it is, that windows works at all is a bit of a miracle. Doesn't make me want to use it any more.
Altivec would have been nice, certainly. I develop video compression algorithms and they get a nice bit of extra wallop when you use MMX. I have a more than slight suspicion that Altivec would have an even larger benefit.
Bluetooth is the other one I wanted. I'm particularly keen on the battery life savings associated with putting a bluetooth access point in the living room instead of an 802.11 one. Buy hey, no loss.
The difference in New Zealand is about two thousand dollars. The average household income is something like 35. Give you some idea?
I really considered the iBook for my wife before I got her the 12"
Slashdot Mirror
Ditto, except I'd go for an iBook - they're much cooler (temperature wise).
And, yes, OS X rules. I don't see why you'd ever want to run Linux on Apple hardware to be honest.
Dave
What's with the "Gran Turismo(?)" ... what bleeding question mark? Gran Turismo is a *great* game, and when played properly encompasses many of the skillsets that are needed when playing any game well:
* The optimal distribution of finite resources.
* The creation of short, medium and long term strategies.
* The absolute, positive *need* to enter a mental state of heightened concentration while remaining relaxed. Much like Rocket Arena on a good day, actually.
The only thing I can't do is rail my mate through the back of the head.
Dave
Bigger computer.
I guess it's the phrasing, and not really the content that keeps dragging me back to this. The content's good, but I don't think my reply was in any way utopian.
.net altogether and going for a mod_perl solution. Neither did I (and this will probably always make me laugh) advocate mono. All I suggested was taking the most expensive, slowest performing and most easily ported bit and moving *that* to OSS.
.net, per se. It *is* advice regarding scaling .net for those on a limited budget.
:)
I didn't, for instance, advocate ditching
So, while this isn't advice with scaling
And, yes, the utopian shit pisses me off too
Dave
let me apologize for all the nutheads who say "drop MS - use Linux"
.net but have discovered that we're going to have to spend $big PER CLIENT to roll the bloody thing out".
.net application more or less as is and scale it across a couple of application servers via an el cheapo load balancer such as the coyote point product.
And why, exactly, is this a nuthead reaction? Our original poster has hit some major problems with a lump of technology that are, essentially, entirely financial in nature. Basically he's saying "we've developed in
Yes, they've been bait'n'switched - and he'll probably do better technology assessments next time, and as you point out that's not really the point. They need to sit down, have a look at how much it would cost to move it to some kind of hybrid approach and the long term gains they stand to make.
Personally, I'd drop MSSQL and put Postgres on a dedicated box. Leave the
But hey, that'd be a dustmop, right?
Dave
Holy shit! What a monster opportunity for a scam: Randomly IM MSN users saying "you've won $1000, put your credit card details here http://microsoft.mydomain.com/easyscam.pl"
With any luck this'll bite them in the arse.
Dave
I don't understand what the problem is here. You're saying that you like XML, but it's slow. Fine, don't use it. It's not like it's the only tool in existence, is it?
Dave
[disclaimer: the following is based on research, not personal experience]
So, I've been looking at using 900MHz for some low bandwidth stuff. Primarily because 2.4GHz can be a PITA to deploy - near line of sight (over distance), trees, rain etc.
Anyway, it appears that the products available round 900MHz have taken a completely different, and much more "agricultural" approach to the problem. Basically they're 900MHz FM radios with a modem bolted on. No clever multipath resolution, no time division, no orthogonal frequency division (outside that which the modem does). Consequently the amount of bandwith available is about seven eigths of fuck all - a claimed 115Kbit (being the limit of the RS232 connectors they use) - with something rather worse being the practicality of situation.
Security is kinda interesting though. Quite a lot frequency hop across a range of channels within the spectrum. It appears that you need to get both radios, from the same manufacturer, and put them on the same hopping scheme before they'll talk.
Dave
you could spend another month or so hand vectorizing code for a tenfold increase in speed
You could do, or you could spend a month being scared of it, then actually get it done in a couple of days (when push comes to shove), which is what I did.
I recently managed to get MMX code into some software I'm wirting using GCC instrinsics and, y'know, it wasn't that hard. The hard part is architecting the application as such that the data just happens to be hanging around in exactly the right structures in the first place.
Dave
Someone please explain to me how 5177 MFLOPS and ~300 MFLOPS are even comparable.
They're not, which is what makes this whole benchmark so entirely useless.
Look at it: The conclusion, basically, is that there's no point in running CFD code using scalar FP. So why didn't they port their code to SSE2? P4's, and particularly the new 800MHz FSB P4 get data through SSE2 code like there's no tomorrow.
Nah, I'll listen when someone compares SSE2 and AltiVec properly. Until then it's just more blah. Don't get me wrong, I'm rapidly turning into the biggest Mac fanboy you've ever seen (Cocoa, since you ask) but the G5's are not the quantum leap Apple are making them out to be. Back in contention? Sure, but I promise you a dual Opteron 2GHz will blow the doors off a dual G5.
Dave
At least from a local (New Zealand) perspective, this is totally wrong. Let me elaborate.
Here the ISP and the phone company are completely different people. Telecom scamper around plumbing in DSLAM's etc. etc. and wire them all together to make a grande country wide ATM network called (ironically) IPnet. The point with IPnet is that it gets packets from your subscribers ADSL routers to some boxen in your datacentre, then you deal with it from there, including international backhaul. Telecom charge you, basically, $60/month/subscriber for this service.
Therefore, fixed costs of setting up an ISP in New Zealand = fuck all. Marginal costs = lots.
Under these circumstances, the speakeasy plan would work a treat. Imagine I own an ISP and charge $100/month for an ADSL connection, but you can connect your neighbours via WiFi and get $50/month back. So, each ADSL connection I get $40 (because $60 goes to Telecom), and each WiFi piggybacked connection I get $50. Woohoo! Quids in, with Telecom footing the worst of the bill!
Unfortunately it doesn't quite work like this. The $60/month buys you a very limited quantity of throughput - 512MB/month in it's most basic form. Yes, you did read that right, less than one iso. Per month. After this it's $0.20/MB, enough to kill the connection sharing thing outright.
But... shit. 512MB/month, and IPnet craps itself, like, all the time. The funny thing is that Telecom have no idea why their broadband takeup is so low. Really. NO idea. They also have no idea why competitiors are starting to appear bloody everywhere.
Dave
I don't wear this "we used the SSE flags" shit for one second. All you're saying (give or take a bit) when you set an SSE flag is that it may be used. It's not like it magically parallises your code, or aligns the data on the right boundaries.
:(
That being said I would like to see a G5/AltiVec vs P4/SSE2 number crunching bake off. Arguably these are the only benchmarks that matter in this day and age since the applications that need that much speed should be using the parallel instructions anyway.
Not like I can afford *either* chip though
Dave
To the PS2, you mean?
.ini files that made it work, mostly. In a window. And then it was just a bit crap.
The PC demo of splinter cell was completely unplayable. After the noises the XBox fanboys were making I was hugely keen to give the game a go. I downloaded the demo, it fell over. Some googling revealed a collection of hacks to
Note to self: XBox fanboys are just that. Save money and effort for GT4.
Dave
Gee, well I'd like to see the OS an integrated dictionary that could be used by all applications
OS X.
Dave
My thoughts exactly. Perhaps I should write an article "Don't fuck with it! A guide to leaving OS X on an iBook"?
Dave
simple, sealed box that could browse the web, read email, play audio and video, keep track of appointments and todos, and maybe do simple spreadsheets, word processing, and sell it to first-time consumers, you'd make a fortune.
It's called an iMac. And, yeah, they did OK out of it.
Dave
I doubt that Apple will be dropping G3's any time soon. IBM apparently has an AltiVec enabled G3 in the wings, and my take is that these will remain in iBooks while PowerBooks will keep getting G4's and potentially these new SOI G4's from Motorola. The 970's will be PowerMac territory only, and they'll cost a ton.
Mind you, the rapidly closing performance gap between iBooks and PowerBooks has got to be a cause for some concern.
Dave
It's really, horribly complicated. Basically the router has to build as little of the TCP stack as possible in order to look at the actual, data contents of the packets to decide what application is being tunnelled.
Dave
My OsX only panicked twice in 4 months!
Use the "big arse download" combi updater instead of software update. It's generally safer.
Dave
How the hell was Unreal Tournament an original game? The thing that made UT such a big hit was that the quake series were (and still are) complete bastards to set up complex servers with ... as anyone who has seen the insane map rotation scrips will attest to, let alone the lunacy associated with getting a rocket arena dedicated server going.
No, UT was common sense. Let's get this thing that people want to do and make it easy.
Dave
Oh, sir, nice troll.
Dave
Because they have a horrifying amount of legacy code to maintain, on what is basically a legacy architecture - i.e. the PC.
Apple got to throw away all their mistakes when they started making OS X. They don't need to support nearly so many hardware experiments - ISA, VLB, MCA, assorted stupid methods of getting to "high" memory, fifty different ways of using large hard drives etc. etc. They also don't need to support a wide assortment of "good idea at the time" legacy technologies, DCOM and others of their ilk not to mention Netware drivers and similar amusement.
Given how large it is, that windows works at all is a bit of a miracle. Doesn't make me want to use it any more.
Dave
None that I've heard of, although I know the Radeon equipped iMacs (17" only, IIRC) can do it. The linked webpage says:
:)
I don't have any info about the eMac with Radeon 7500 yet but I expect no problems with this machine either (the old eMac did not work though).
There is also a temporary version of the hack, I'd go for it if I were you
Dave
Altivec would have been nice, certainly. I develop video compression algorithms and they get a nice bit of extra wallop when you use MMX. I have a more than slight suspicion that Altivec would have an even larger benefit.
:)
Bluetooth is the other one I wanted. I'm particularly keen on the battery life savings associated with putting a bluetooth access point in the living room instead of an 802.11 one. Buy hey, no loss.
The difference in New Zealand is about two thousand dollars. The average household income is something like 35. Give you some idea?
I really considered the iBook for my wife before I got her the 12"
Yeah yeah, rub it in
Dave
Tried a LART?