it's very jarring to adjust to the different contrast level(s) between LCD/CRT on a regular basis (at least for me) so why not stick with a similar one to what you have?
I personally have two aperture grille CRTs (19" and 17") and run them one at 1600 and the other at 1280 to have roughly the same DPI and calibrated them so they have very very similar color rendition, it's much better to move a window between the two and have it NOT change apparent size or colour.
and just put in any old PCI video card (I used a matrox millennium), remember to set the BIOS to AGP-primary and voila', dual display without many hassles: works perfectly in 2000 and Linux.
about what exactly caused the roll last time? Given that now they had time go to through the telemetry data one assumes they would know for sure exactly what happened: did they make the info public?
that lasts for 3 hours on 'light' usage (if you use everything and turn on WiFi the article seems to imply it lasts much less)? My good old Palm IIIc lasts for many, many hours for example: the only reason why I'd like to upgrade it is to get a 320x480 screen (and maybe a little more RAM) but that's about it: I don't care about playing DVDs on my pilot, I just want to use it as a PIM and to read some e-books (where having a higher resolution screen and more memory would help).
Why doesn't anybody come out with a reasonably priced PDA with a *slow* CPU that doesn't suck a lot of juice, 32MB of RAM (it's not like RAM size impacts battery life), and a reasonable 320x480 color screen that maybe sacrifices super-sharpness for battery life? (I think a PDA should last 20-30 hours on a battery charge to be really useful)
Are there any palm-compatible PDAs like that around? The T3 would be the ticket (in terms of features) but I heard that it has horrible battery life for example.
IMO, Karma was first gen. Havok was second gen. Meqon is third gen. It's way way faster than Karma ever was, and it does more, and it's cleaner to use. I know Havok is working on next gen stuff now, but it's not available yet.
emphasis mine: I guess this means that they're going to play with Meqon for the time being and when the new Havok engine comes out they'll switch once again:)
The rendering has been done a long time.
aka, when it finally comes out it's going to look like crap compared to games like Doom3 and HL2
Why did we do it? Because Karma was too slow to do anything we were trying to do.
so, they had a ton of physics gameplay code and at the same time he says the physics engine was too slow to do anything we were trying to do, hmmmmm, this does not compute...
Printers are unreliable. You don't want poll workers (who are volunteers, not technicians) having to spend all their time clearing paper jams, etc.
what kind of printers are you using? Heck, even my old, old, old Laserjet 4L would be able to handle the job of printing maybe 600 sheets of paper in one day.
Scanning the bar codes is going to be a lot of work, and will probably have some error rate.
oh yeah, because scanning 'pregnant chads' is obviously very fast instead. Have you shopped at a supermarket lately? Bar code readers at the checkout counters seem to be able to cope with semi smudged bar codes wrapped around lettuce, a flat sheet with a recently printed bar code doesn't seem that hard
t makes vote buying possible, because the person walks out of the booth with a piece of paper showing how he voted
huh? I suggest you reread my post: you get a token, go in the booth, vote, get a ballot, come out, put the ballot in one box and the token in the other: the elector doesn't take anything with them once they get out of the poll station.
rather than going 'all electronic' there are not more efforts to have a hybrid paper-computer model, off the top of my head:
- the voter comes to the poll, is identified and is given a paper token with a barcode that contains the polling ID station ID and a sequential number (note that the ID is not humanly readable, important for privacy)
- the voter goes in the box, which has a touch screen and an 'easy' UI, voter inserts the paper token in the box which scans it
- voter votes on the touch screen (make it really easy, BIG buttons, BIG text, whatever)
- machine prints out a ballot with the voter's vote in humanly readable form (say, prints out a 'real' ballot with blackened out rectangles on the relevant candidate(s)) and a 2D barcode at the bottom with the vote in machine readable form including the ID on the 'paper token'
- voter looks at the ballot to make sure it's ok, folds it, comes out, puts the ballot in one box and the paper token in the other. If the ballot is not ok there is a shredder right there inside the poll station and the voter votes again.
========= election over ===========
the paper token are shipped to the central office, scanned (should be very fast via the 2d barcodes) and votes tabulated accordingly; for an additional level of security you can always count the votes via the 'human readable' part of the ballot before shipping them.
If a recount or anything is necessary there are several safeguards with this system:
- you can't have ballot box stuffing, because 1 'token' = 1 vote and if those ID are generated 'well' you could even double check that all IDs make sense, sort of like a 'there are only so many valid serial numbers' there. Multiple votes with the same 'ID' will be discarded.
- you can't have doubts on the voter intent, they'll vote on the screen *AND* look at the paper copy before putting it in the ballot box later on
- if there is really no trust in the computers no problem, you can just look at the 'human readable' portion of the ballot as many times as you want: no nonsense about hanging chads or anything.
this (or something like it) would cover all the bases in terms of fast results (via scanning ballots, ship them all to a central location and do it), paper trail and so on. I really can't understand who in their right mind would consider putting the fate of the election in the hands of MS Access, for crying out loud!
So, if 500$ is too much for a videocard, why these things seem to be selling like hotcakes?
they aren't, dealers are basically not willing to stock them and find themselves with white elephants once the new latest-and-greatest model is announced 6 months before availability (who buys a $500 card when its successor has been announced and tested everywhere? let alone the fact that you can't buy it anywhere!) so they basically sell them 'on demand' (you order, it gets shipped from China, you get it 2 months later).
I wouldn't mind getting a previous-generation top-of-the-line card for $200, but guess what, nobody has any: only things you can find everywhere are low-low-end radeons or low-low-low-end 5200s etc.
Personally I'd like to go with a 6800GT, but if the stores I've called are any indication I might not be able to get one (reputable brand name) for a few months at least. Interestingly it seems one of the local stores got a bad batch of 6800GTs so several people shelled out their $$$ and are now SOL until the new shipment arrives (when? who knows!)
Instead of continuing with the stupid 'announce vaporware 6 months in advance' mantra, I wish NVidia focused more on *AVAILABILITY* for a change: heck, some years back I was able to buy a card within 2-3 weeks of its announcement, now it's more like 3 months, if you can find any at all:(
it says it has the same current draw as QVGA but this one is full VGA.
You can also find out for yourself by doing some simple math: if this is approximately 2.2 inches with a 4:3 ratio it means it's going to be approx. 1.76in wide and 1.32in tall, which means that it has an area of around 2.3 square inches, which means that (at 368ppi, 135424 pixels per square inch) it would have 311475 pixels, which confirms full-VGA resolution (640x480 = 307200) due to probable slight measurement differences (I don't think it's going to be *exactly* 368ppi).
- what kind of CPU can currently run above 3.7GHz?
- unless I completely missed something it does seem that there are very few (if none) nvidia 6800-class (standard, GT or ultra) cards around (my local hw shops have been backordered on these for MONTHS, practically impossible to buy). Also given how many 'older' video cards are around no wonder very few games are willing to push the video HW requirements
- what processor vendor is named 'ValveIsGreat'? what about the Euro symbol?:)
- why do 33 people run with an horizontal resolution of 832 pixels ?!?
- what video card can run in 49bpp? and what about 6bpp (EGA?)?
- there are two versions of the Korean language?!?!? (Adult and Teen)
- wow, 288 people have more than 250GB free hd space (!)
as it'd really mess up the pace of the end of the film.
you mean as it'd wake people up?
With all the 'it's over, no, this time it's over, no, really, THIS time it's over' I swear I was tempted to walk out the theatre... and I loved the first two movies and the third up to that point. I just hope they'll have an alternate ending on the DVD when the whole enchilada is cut down to more manageable proportions.
as environments who have radios tuned to music, usually don't have computers connected to the net (think cars, lots of different workplaces), and people in environments with access to net connected computers don't listen to over-the-air radio anyways (di.fm, mp3s, etc. etc.)
So the Gimp hackers work on the functionality that they need to compete with PS (like adjustment layers),
given that the BZ for adjustment layers was created in early 2002, and has had no activity since late 2002 (set to Milestone 'Future' in early 2003) I'm not sure just how interested the Gimp developers are. Given just how majorly important this functionality is for any serious Gimp/PS user this is very disenheartening. At times it feels as if the Gimp developers are 'just' developers and don't really have that many heavy/pro PS users to consult with.
I really wish that an experienced PS person or two drew up a list of the most important PS features missing in Gimp (my guess: CYMK, Adjustment Layers, PS-like UI) and the developers actually followed through implementing them regardless of how they felt about them.
I guess this is the main problem with Free Software, a lot of the time it's written by developers for developers and not for average users: note the tutorial on how to draw straight lines in Gimp referenced later in the thread, if the Gimp was developed from a user-centric perspective, a 'straight line' tool would've been added to the UI. I don't care that having a shift modifier is 'more elegant' or 'more in tune with the dev's vision', if so many people find it troublesome, it should be changed.
At least this is the way things work in the 'commercial software' world, where the customer most of the time is king: often it's the only reason why developers end up spending a lot of time on complex low-reward (for devs) tasks that in the end, though, make life that much easier for the users. I totally understand how if I was coding for fun in my spare time I wouldn't want to deal with a lot of things that I have to deal with in my job, but if we want Gimp to become a good alternative to PS, well, somebody will have to make the calls about what to work on next regardless of how appealing it is.
I think it would be a bad thing - one more item for somebody to maintain, more developer resources consumed.
in general UI developers tend to be UI developers, not people that would be interested in writing CYMK support or implementing adjustment layers. You are falsely assuming that the pool of Gimp developers is finite, I am sure that if there was a clear componentization of the Gimp (clear, defined boundaries between UI and back-end) it wouldn't be too hard to allow an 'UI useability group' to operate in parallel with the traditional 'backend' development.
he irony is that if they cloned the PS interface, people like you would be lambasting them for being unoriginal,
People like me would adopt the 'copycat' software and be productive instantly. I am long past the age where installing a new program and learning it is an exciting adventure: I want software that works the way software I already know works (ok, I am not against improvements, I am talking about paradigms here), and if this means that Gimp on Linux will look =exactly= the same as PS on Windows, hey, by all means count me in. If PS worked under Wine as well as on native win32 (it's much slower and currently it seems tablet support doesn't work very well, as in, no pressure info seems to get to PS) I wouldn't even bother with dual booting at all.
The GIMP UI is good. It's not a barrier to productivity with the GIMP - the only barrier is people's refusal to let go of something familiar in PS.
that is a barrier for a lot of experienced people. If Gimp folks want to target a user-base that has no PS experience ('newbies' in a lot of ways), hey, fine, but I really can't see why the Gimp folks wouldn't WANT more people to switch away from PS (championing the open source movement and all that), a little investment in UI would go a long way towards that shift (that and more PS-comparable functionality of course).
PS is the standard for image manipulation programs, so I don't understand the reluctance of gimp developers to provide a 'ps emulator' mode for Gimp so people familiar with PS could feel more at home. Heck, even emacs has vi modes for crying out loud! It's not like actually getting more users for Gimp would be a bad thing, right?
Personally I don't mind as much the Gimp UI (in 2.0, in 1.3 I minded it very much) despite the fact that I am more used to the PS keyboard shortcuts, but can't really use it as my primary app until adjustment layers will finally be supported (people have been asking for this feature for years and years, yeah, I know, if we want it so bad, why don't we code it)
The link for pebble bed reactors is a bit slow...
on
China Goes Nuclear
·
· Score: 3, Insightful
it's just they're anticipating the gun, sometimes they get it right and most of the time they get it wrong: given the new rule that any false starter (after the first start) will be DQ'd I'm sure you won't see 0.1sec reaction times the second time around: to the naked eye the reaction times of the 2nd start they did the other day in the 100m semi-final seemed slower than the 1st for example.
It'd also be interesting to know how far from the athletes the gun is located and if sound travel speed can have an impact on things (how is the electronic system synchronized to the gun? via sound? some other way?)
'full speed'?
on
3D Monitor
·
· Score: 2, Insightful
hmmmm, I fail to see how this could run at 'full speed' given that for every 3d technology (save volumetric displays obv) you need to compute twice as many scenes (left eye & right eye independently) in order to be able to display things correctly, which will end up possibly halving your framerate... although, given that UT2004 is much more CPU limited than video-limited, maybe in this particular case the performance hit is much less than that.
yeah, but the first pass (the one to decide who renders what) takes only a fraction of the time of the second as it's only basic vertex transforms: NO shaders, AA, AF, etc. etc. which in the latest generation of games are really what causes slowdowns.
I mean, a 6800 ultra does something like 400+fps in quake3 right? now, imagine how even faster it could be if you removed all the dynamic lights, textures and so on (since this is coarse grained you could even have the first pass at 640x480 or less regardless of the screen resolution): in the grand scheme of things this 'redundant' first pass would cost maybe a 1-2fps hit in a game like Doom3, which is dwarfed by the speedup achieved by having a 2nd card to split the work with.
hmmmmmm, I really don't know how this is supposed to work (haven't read the articles) but it shouldn't be too hard to do the following:
- both cards to the 'flat surface' pass on the full scene (aka, vertex transform, visibility testing etc. etc., the 'T' part of T&L), this means a (supposedly) negligible slowdown (as you're doing only very basic transforms to get to a flat shaded view of the world)
- based on this result (and maybe some other 'hints' like number of shaders associated to things) the drivers decide on a scene split (say, top 60% of the screen to card #1 and bottom 40% to card #2): I'd assume you'd always want the 'bigger' portion to go to the screen-connected card to minimize the amount of data to be copied later on.
- each card now 'throws away' the scene data it's not interested in.
- card #1 (connected to the screen) starts rendering normally to a back buffer
- card #2 (not connected) starts rendering to a back buffer.
- once card #2 is finished its back buffer is copied to card #1 and probably there is a very small anti-aliasing pass applied to the 4-5 pixel-wide 'seam' to make it invisible. After this card #1 flips the back buffer to front and everything starts again.
- this could obviously be generalized to 3+ cards if there were more available PCIe slots.
- Note that all of this would not require a physical connection between the two cards per-se, I'm assuming that since this connection exists at least the buffer copying will be done over it rather than on the PCIe bus (which would make a lot of sense).
In the end, the more shader stuff you have going on, the closer to a 100% speedup you'll likely get. Again, this is all off the top of my head, I haven't been following how NVIDIA said their 'SLI' solution is going to work.
yeah, but where can you find SMAC/SMACX for linux? I've been trying to track it down for a while now (preferably from a reputable store/seller) to no avail...
unfortunately due to the insane hw requirements there will probably be a maximum of 2-3 monsters on the screen at the same time, which really blows as in the original doom/doom2 part of the fun was also mowing down hordes of low-level monsters (imps, marines, flaming skulls) rather than doing the sneak-kill-1-monster-repeat routine.
Doom3 won't run on my box so I'm waiting for some 'real' reviews before deciding if it's worth upgrading now or later for HL2/CS-source. By 'real' I mean 'written by people who have played the whole game two or three times', as I'm sure EVERYBODY will be totally insanely blown away for the first hour, but not if they will be just as much when the novelty of the great graphics wears off.
It also worries me how most reviews seem to say that there are only 10-15 hours of gameplay in the game, which, considering that you probably have to creep around a lot, doesn't seem to point to the levels being that big and/or that many.
thanks for the tip, although I already have SC3000 I guess it's a pretty good deal to get smac+smacx for 20 bucks (not interested in RA2 and golf): I wish they had the linux executables for download somewhere though...
Slashdot Mirror
it's very jarring to adjust to the different contrast level(s) between LCD/CRT on a regular basis (at least for me) so why not stick with a similar one to what you have?
I personally have two aperture grille CRTs (19" and 17") and run them one at 1600 and the other at 1280 to have roughly the same DPI and calibrated them so they have very very similar color rendition, it's much better to move a window between the two and have it NOT change apparent size or colour.
and just put in any old PCI video card (I used a matrox millennium), remember to set the BIOS to AGP-primary and voila', dual display without many hassles: works perfectly in 2000 and Linux.
about what exactly caused the roll last time? Given that now they had time go to through the telemetry data one assumes they would know for sure exactly what happened: did they make the info public?
that lasts for 3 hours on 'light' usage (if you use everything and turn on WiFi the article seems to imply it lasts much less)? My good old Palm IIIc lasts for many, many hours for example: the only reason why I'd like to upgrade it is to get a 320x480 screen (and maybe a little more RAM) but that's about it: I don't care about playing DVDs on my pilot, I just want to use it as a PIM and to read some e-books (where having a higher resolution screen and more memory would help).
Why doesn't anybody come out with a reasonably priced PDA with a *slow* CPU that doesn't suck a lot of juice, 32MB of RAM (it's not like RAM size impacts battery life), and a reasonable 320x480 color screen that maybe sacrifices super-sharpness for battery life? (I think a PDA should last 20-30 hours on a battery charge to be really useful)
Are there any palm-compatible PDAs like that around? The T3 would be the ticket (in terms of features) but I heard that it has horrible battery life for example.
I think it's more likely that this will degenerate into some sort of c3po 'language' after a few passes: beep BOOP beep beep blip
IMO, Karma was first gen. Havok was second gen. Meqon is third gen. It's way way faster than Karma ever was, and it does more, and it's cleaner to use. I know Havok is working on next gen stuff now, but it's not available yet.
:)
emphasis mine: I guess this means that they're going to play with Meqon for the time being and when the new Havok engine comes out they'll switch once again
The rendering has been done a long time.
aka, when it finally comes out it's going to look like crap compared to games like Doom3 and HL2
Why did we do it? Because Karma was too slow to do anything we were trying to do.
so, they had a ton of physics gameplay code and at the same time he says the physics engine was too slow to do anything we were trying to do, hmmmmm, this does not compute...
Printers are expensive.
and diebold machines are not?
Printers are unreliable. You don't want poll workers (who are volunteers, not technicians) having to spend all their time clearing paper jams, etc.
what kind of printers are you using? Heck, even my old, old, old Laserjet 4L would be able to handle the job of printing maybe 600 sheets of paper in one day.
Scanning the bar codes is going to be a lot of work, and will probably have some error rate.
oh yeah, because scanning 'pregnant chads' is obviously very fast instead. Have you shopped at a supermarket lately? Bar code readers at the checkout counters seem to be able to cope with semi smudged bar codes wrapped around lettuce, a flat sheet with a recently printed bar code doesn't seem that hard
t makes vote buying possible, because the person walks out of the booth with a piece of paper showing how he voted
huh? I suggest you reread my post: you get a token, go in the booth, vote, get a ballot, come out, put the ballot in one box and the token in the other: the elector doesn't take anything with them once they get out of the poll station.
rather than going 'all electronic' there are not more efforts to have a hybrid paper-computer model, off the top of my head:
- the voter comes to the poll, is identified and is given a paper token with a barcode that contains the polling ID station ID and a sequential number (note that the ID is not humanly readable, important for privacy)
- the voter goes in the box, which has a touch screen and an 'easy' UI, voter inserts the paper token in the box which scans it
- voter votes on the touch screen (make it really easy, BIG buttons, BIG text, whatever)
- machine prints out a ballot with the voter's vote in humanly readable form (say, prints out a 'real' ballot with blackened out rectangles on the relevant candidate(s)) and a 2D barcode at the bottom with the vote in machine readable form including the ID on the 'paper token'
- voter looks at the ballot to make sure it's ok, folds it, comes out, puts the ballot in one box and the paper token in the other. If the ballot is not ok there is a shredder right there inside the poll station and the voter votes again.
========= election over ===========
the paper token are shipped to the central office, scanned (should be very fast via the 2d barcodes) and votes tabulated accordingly; for an additional level of security you can always count the votes via the 'human readable' part of the ballot before shipping them.
If a recount or anything is necessary there are several safeguards with this system:
- you can't have ballot box stuffing, because 1 'token' = 1 vote and if those ID are generated 'well' you could even double check that all IDs make sense, sort of like a 'there are only so many valid serial numbers' there. Multiple votes with the same 'ID' will be discarded.
- you can't have doubts on the voter intent, they'll vote on the screen *AND* look at the paper copy before putting it in the ballot box later on
- if there is really no trust in the computers no problem, you can just look at the 'human readable' portion of the ballot as many times as you want: no nonsense about hanging chads or anything.
this (or something like it) would cover all the bases in terms of fast results (via scanning ballots, ship them all to a central location and do it), paper trail and so on. I really can't understand who in their right mind would consider putting the fate of the election in the hands of MS Access, for crying out loud!
So, if 500$ is too much for a videocard, why these things seem to be selling like hotcakes?
:(
they aren't, dealers are basically not willing to stock them and find themselves with white elephants once the new latest-and-greatest model is announced 6 months before availability (who buys a $500 card when its successor has been announced and tested everywhere? let alone the fact that you can't buy it anywhere!) so they basically sell them 'on demand' (you order, it gets shipped from China, you get it 2 months later).
I wouldn't mind getting a previous-generation top-of-the-line card for $200, but guess what, nobody has any: only things you can find everywhere are low-low-end radeons or low-low-low-end 5200s etc.
Personally I'd like to go with a 6800GT, but if the stores I've called are any indication I might not be able to get one (reputable brand name) for a few months at least. Interestingly it seems one of the local stores got a bad batch of 6800GTs so several people shelled out their $$$ and are now SOL until the new shipment arrives (when? who knows!)
Instead of continuing with the stupid 'announce vaporware 6 months in advance' mantra, I wish NVidia focused more on *AVAILABILITY* for a change: heck, some years back I was able to buy a card within 2-3 weeks of its announcement, now it's more like 3 months, if you can find any at all
it says it has the same current draw as QVGA but this one is full VGA.
You can also find out for yourself by doing some simple math: if this is approximately 2.2 inches with a 4:3 ratio it means it's going to be approx. 1.76in wide and 1.32in tall, which means that it has an area of around 2.3 square inches, which means that (at 368ppi, 135424 pixels per square inch) it would have 311475 pixels, which confirms full-VGA resolution (640x480 = 307200) due to probable slight measurement differences (I don't think it's going to be *exactly* 368ppi).
... a few things that struck me:
:)
- what kind of CPU can currently run above 3.7GHz?
- unless I completely missed something it does seem that there are very few (if none) nvidia 6800-class (standard, GT or ultra) cards around (my local hw shops have been backordered on these for MONTHS, practically impossible to buy). Also given how many 'older' video cards are around no wonder very few games are willing to push the video HW requirements
- what processor vendor is named 'ValveIsGreat'? what about the Euro symbol?
- why do 33 people run with an horizontal resolution of 832 pixels ?!?
- what video card can run in 49bpp? and what about 6bpp (EGA?)?
- there are two versions of the Korean language?!?!? (Adult and Teen)
- wow, 288 people have more than 250GB free hd space (!)
as it'd really mess up the pace of the end of the film.
you mean as it'd wake people up?
With all the 'it's over, no, this time it's over, no, really, THIS time it's over' I swear I was tempted to walk out the theatre... and I loved the first two movies and the third up to that point. I just hope they'll have an alternate ending on the DVD when the whole enchilada is cut down to more manageable proportions.
as environments who have radios tuned to music, usually don't have computers connected to the net (think cars, lots of different workplaces), and people in environments with access to net connected computers don't listen to over-the-air radio anyways (di.fm, mp3s, etc. etc.)
Now, if I could magically get di.fm in my car...
but does an aircraft carrier even pitch that much? I'd assume that given its size and the stabilizers it surely has it wouldn't move very much...
So the Gimp hackers work on the functionality that they need to compete with PS (like adjustment layers),
given that the BZ for adjustment layers was created in early 2002, and has had no activity since late 2002 (set to Milestone 'Future' in early 2003) I'm not sure just how interested the Gimp developers are. Given just how majorly important this functionality is for any serious Gimp/PS user this is very disenheartening. At times it feels as if the Gimp developers are 'just' developers and don't really have that many heavy/pro PS users to consult with.
I really wish that an experienced PS person or two drew up a list of the most important PS features missing in Gimp (my guess: CYMK, Adjustment Layers, PS-like UI) and the developers actually followed through implementing them regardless of how they felt about them.
I guess this is the main problem with Free Software, a lot of the time it's written by developers for developers and not for average users: note the tutorial on how to draw straight lines in Gimp referenced later in the thread, if the Gimp was developed from a user-centric perspective, a 'straight line' tool would've been added to the UI. I don't care that having a shift modifier is 'more elegant' or 'more in tune with the dev's vision', if so many people find it troublesome, it should be changed.
At least this is the way things work in the 'commercial software' world, where the customer most of the time is king: often it's the only reason why developers end up spending a lot of time on complex low-reward (for devs) tasks that in the end, though, make life that much easier for the users. I totally understand how if I was coding for fun in my spare time I wouldn't want to deal with a lot of things that I have to deal with in my job, but if we want Gimp to become a good alternative to PS, well, somebody will have to make the calls about what to work on next regardless of how appealing it is.
I think it would be a bad thing - one more item for somebody to maintain, more developer resources consumed.
in general UI developers tend to be UI developers, not people that would be interested in writing CYMK support or implementing adjustment layers. You are falsely assuming that the pool of Gimp developers is finite, I am sure that if there was a clear componentization of the Gimp (clear, defined boundaries between UI and back-end) it wouldn't be too hard to allow an 'UI useability group' to operate in parallel with the traditional 'backend' development.
he irony is that if they cloned the PS interface, people like you would be lambasting them for being unoriginal,
People like me would adopt the 'copycat' software and be productive instantly. I am long past the age where installing a new program and learning it is an exciting adventure: I want software that works the way software I already know works (ok, I am not against improvements, I am talking about paradigms here), and if this means that Gimp on Linux will look =exactly= the same as PS on Windows, hey, by all means count me in. If PS worked under Wine as well as on native win32 (it's much slower and currently it seems tablet support doesn't work very well, as in, no pressure info seems to get to PS) I wouldn't even bother with dual booting at all.
The GIMP UI is good. It's not a barrier to productivity with the GIMP - the only barrier is people's refusal to let go of something familiar in PS.
that is a barrier for a lot of experienced people. If Gimp folks want to target a user-base that has no PS experience ('newbies' in a lot of ways), hey, fine, but I really can't see why the Gimp folks wouldn't WANT more people to switch away from PS (championing the open source movement and all that), a little investment in UI would go a long way towards that shift (that and more PS-comparable functionality of course).
PS is the standard for image manipulation programs, so I don't understand the reluctance of gimp developers to provide a 'ps emulator' mode for Gimp so people familiar with PS could feel more at home. Heck, even emacs has vi modes for crying out loud! It's not like actually getting more users for Gimp would be a bad thing, right?
Personally I don't mind as much the Gimp UI (in 2.0, in 1.3 I minded it very much) despite the fact that I am more used to the PS keyboard shortcuts, but can't really use it as my primary app until adjustment layers will finally be supported (people have been asking for this feature for years and years, yeah, I know, if we want it so bad, why don't we code it)
There is a good writeup as well on wikipedia
it's just they're anticipating the gun, sometimes they get it right and most of the time they get it wrong: given the new rule that any false starter (after the first start) will be DQ'd I'm sure you won't see 0.1sec reaction times the second time around: to the naked eye the reaction times of the 2nd start they did the other day in the 100m semi-final seemed slower than the 1st for example.
It'd also be interesting to know how far from the athletes the gun is located and if sound travel speed can have an impact on things (how is the electronic system synchronized to the gun? via sound? some other way?)
hmmmm, I fail to see how this could run at 'full speed' given that for every 3d technology (save volumetric displays obv) you need to compute twice as many scenes (left eye & right eye independently) in order to be able to display things correctly, which will end up possibly halving your framerate... although, given that UT2004 is much more CPU limited than video-limited, maybe in this particular case the performance hit is much less than that.
yeah, but the first pass (the one to decide who renders what) takes only a fraction of the time of the second as it's only basic vertex transforms: NO shaders, AA, AF, etc. etc. which in the latest generation of games are really what causes slowdowns.
I mean, a 6800 ultra does something like 400+fps in quake3 right? now, imagine how even faster it could be if you removed all the dynamic lights, textures and so on (since this is coarse grained you could even have the first pass at 640x480 or less regardless of the screen resolution): in the grand scheme of things this 'redundant' first pass would cost maybe a 1-2fps hit in a game like Doom3, which is dwarfed by the speedup achieved by having a 2nd card to split the work with.
hmmmmmm, I really don't know how this is supposed to work (haven't read the articles) but it shouldn't be too hard to do the following:
- both cards to the 'flat surface' pass on the full scene (aka, vertex transform, visibility testing etc. etc., the 'T' part of T&L), this means a (supposedly) negligible slowdown (as you're doing only very basic transforms to get to a flat shaded view of the world)
- based on this result (and maybe some other 'hints' like number of shaders associated to things) the drivers decide on a scene split (say, top 60% of the screen to card #1 and bottom 40% to card #2): I'd assume you'd always want the 'bigger' portion to go to the screen-connected card to minimize the amount of data to be copied later on.
- each card now 'throws away' the scene data it's not interested in.
- card #1 (connected to the screen) starts rendering normally to a back buffer
- card #2 (not connected) starts rendering to a back buffer.
- once card #2 is finished its back buffer is copied to card #1 and probably there is a very small anti-aliasing pass applied to the 4-5 pixel-wide 'seam' to make it invisible. After this card #1 flips the back buffer to front and everything starts again.
- this could obviously be generalized to 3+ cards if there were more available PCIe slots.
- Note that all of this would not require a physical connection between the two cards per-se, I'm assuming that since this connection exists at least the buffer copying will be done over it rather than on the PCIe bus (which would make a lot of sense).
In the end, the more shader stuff you have going on, the closer to a 100% speedup you'll likely get. Again, this is all off the top of my head, I haven't been following how NVIDIA said their 'SLI' solution is going to work.
yeah, but where can you find SMAC/SMACX for linux? I've been trying to track it down for a while now (preferably from a reputable store/seller) to no avail...
unfortunately due to the insane hw requirements there will probably be a maximum of 2-3 monsters on the screen at the same time, which really blows as in the original doom/doom2 part of the fun was also mowing down hordes of low-level monsters (imps, marines, flaming skulls) rather than doing the sneak-kill-1-monster-repeat routine.
Doom3 won't run on my box so I'm waiting for some 'real' reviews before deciding if it's worth upgrading now or later for HL2/CS-source. By 'real' I mean 'written by people who have played the whole game two or three times', as I'm sure EVERYBODY will be totally insanely blown away for the first hour, but not if they will be just as much when the novelty of the great graphics wears off.
It also worries me how most reviews seem to say that there are only 10-15 hours of gameplay in the game, which, considering that you probably have to creep around a lot, doesn't seem to point to the levels being that big and/or that many.
thanks for the tip, although I already have SC3000 I guess it's a pretty good deal to get smac+smacx for 20 bucks (not interested in RA2 and golf): I wish they had the linux executables for download somewhere though...