Slashdot Mirror
ATI R300 and R250V
Chuu writes "The ATI R300 (Radeon 9700) and R250V (Radeon 9000/Radeon
9000 Pro) reviews are out, at all the
usual suspects, but the one you want to
pay attention to is over at anandtech.com, since somehow Anand got permission to publish his benchmark results for the R300
while the other sites were stuck with whitepapers. The results? The R250V is a GF4MX killer, which is not saying much. On the other hand, the R300 absolutely trounces the GeForce4 Ti4600, running
54% faster in
Unreal Tournament 2003 and 37% faster in Quake 3 at 1600x1200x32 on a Pentium4 2.4ghz."
The reason Anand got to post his benchmarks is because he doesn't have actual numbers... just correlated data. The R300 didn't post 256 fps in UT2k3, it just did x% better than a GF4 Ti4600.
Of course, as he points out, the GF4 numbers are available, and it only takes some simple math to extrapolate from there.
The card looks very impressive. It's out 4 months before the NV30. Maybe by then ATI will have drivers worth a crap too.
Am I the only one who read that story and had eyes glazing over at all the captial letters an numbers?
Not that ATI doesn't make some good shit - I loved their ATI-TV for the longest time (until I got a Mac, but that's another story).
But they've been known to, um, "help" their drivers along with specific applications. When I see one plugged into my PowerMac while I'm playing Medal of Honor or Warcraft III and I see better performance, then I'll believe it.
Now, see what happens to the boy who cried "framerates", kids?
52 Weeks, 52 Religions with John Hummel
No big surprise, it SHOULD trounce the GF4... after all, it's the first of the next generation cards - it's interesting to compare to previous generation cards, but it really should more appropriately be put head-to-head against the GF4 successor NV30. Trouncing current cards is a big yawn, but if it can go toe-to-toe with the big boys in 3 or 4 months they'll have something.
That said, congrats to ATI - I love competition in the marketplace. Now if only they could write some decent drivers for once in their lives.
"So on one hand, honey is an amazingly sophisticated and efficient food source. On the other hand it's bee backwash."
Now the Radeon 8500 will come down in price enough for me to afford one. Sweet!
Carpe Cerevisi - Seize the Beer
The page you really need to go is this. It talks about not just raw FPS, but about running UT2003 with 4X Anti Aliasing enabled at 1600x1200x32. This is where ATI trounces Nvidia with a whopping 251% faster performance.
Though the framerates at 1600x1200 on UT2003 are not exactly playable (there goes my hopes of running DoomIII at 1600x1200 on this baby) ATI has finally produced a card worthy of their name.
Nvidia has atleast six months to go before they can have something to show. And running the 927 leaked build of UT2003 on a GF4 Ti 4600, you dont get playable framerates beyond 1024x768 with every detailed notched up.
Rapid Nirvana
Insert a comment here where I talk about technology which I do not understand and use phrases such as vertex shaders, fog effects and texels per second, and attempt to make it look like any of this is actually of any relevence what so ever. Make a reference to Doom III.
Insert seven comments below this comment which all do the same thing.
Yes it does. 3d games are becoming more and more complex, and the amount of GPU horsepower they need also increases. Using 'quake 3' as a benchmark showing it can do 200 FPS is merely an indication of the processing power of the unit. When you benchmark on next generation games (such as UT2003) it will run somewhere on the order of 60 FPS.
If you play Quake 2 still.. stick to your Voodoo 2. Otherwise , upgrade if your interested in next gen gaming..
You bastards!!! I was trying to look up some motherboard reviews and now Toms is slashdotted!
...And when they came for me, there was no one left to speak out for me." - Martin Niemoeller (1892-1984)
Have you played a game recently? Does it look real to you? How about even CG movies? do they look real? If so I think you need some of those new ocular implants...
"Save me jebus!" - Homer Simpson (btw, I'm probably talkin out of me arse)
The Raven.
The Raven
It is not that we need more FPS, it is that we need stable FPS, droping from 250 to 84 in a seciond and you can feel it (not see it) cus you have to re ajust the way you were to move your mosue, We have to get faster befor we can ssettle at a nice 60 fps, I would trade a 250 fps that drops alot when the GFX get high, for a card that will nto drop below 60fps and no higher than 62 fps. todays cards (R8500) i play games and it still drops to 40 fps in some gammes when around the corner it is blazing at 120 fps.
Okay, I'll bite.
Playing current games at 120fps isn't any gain over 60fps. It's the next generation of games that will push these cards to perform at 60fps and last generations cards will start to show their age.
-Spyky
Will there be Linux drivers for these cards supporting 3D acceleration/OpenGL under XFree86, so I can play RtCW and Flightgear on my favorite platform?
If so, count me in. Otherwise, I'll stick to NVidia.
If you think the Radeon 9700 is amazing, just wait till ATI produces the R300 chipset in the 0.13 micron process version and cranks up the graphics card clock speed to likely way over 400 MHz.
:-)
Slow it definitely won't be.
It's not just about speed, it's also about quality. I want to be able to run my games at 1280x1024 in 32-bit color with FFA and bump-mapping. And I don't want it to be slow. I'm not looking for 256fps at 1024x768 on medium quality, I'm looking for 85fps at 1280x1024 on high quality, and I think that's what the GPU manufacturers are counting on.
What we need to clue into is that due to marketing reasons ATI wanted to get the chip running at 300mhz. They didn't care about the possible performance loss, all the marketing assholes want is a high MHZ and they had to take a "different approach" (meaning, shittier design) to reach the 300mhz mark.
The sooner the average joe can accept that MHZ no longer equals performance... the better off chip design will be.
The Pentium 4 basically is less efficient than a pentium 3, however 2ghz makes morons happy. So 2ghz whatever the cost!
Noodle.
Are we going to see Linux drivers for these cards that will perform as well as their Windows counterparts? As much as I hate the closed-source nature of NVidia's Linux drivers, I have to admit they work, and work well.
Additionally, ATI cards have always been renowned for their DVD playback capabilities. I think they could increase their user-base by porting their DVD playback software to Linux, and make it run just a pretty as it does under Windows.
It really does matter, it's a reserve for one, for when you get into more intense sequences, throw in some gibs, a blast radius or two, the quad damage aura lighting ect, it drops quickly. furthermore, they're using quake 3 for most of the bench marks... quake 3 is a bit out of date on its graphics technology, they should use rtcw, or wait until doom 3 comes out, i'd like to see if you can maintain that 230 average with a new game designed to push your graphics card to the limits. i'll be more than impressed if you can still maintain the famed 60 fps that you're knocking
--fetch daddy's blue fright wig, i must be handsome when i release my rage
http://www.anandtech.com/video/showdoc.html?i=1656 &p=1
No, you're mistaken. That's a direct URL, not a link. Maybe you should learn basic HTML during recess.
AnandTech went on record as saying that the NV30 WILL be faster than the Radeon 9700 (which is expected due to a smaller manufacturing process of 0.13u) but this gives ATI the performance crown for pretty well the rest of this year. Nvidia plan on having the NV30 out November but this may be just wishful thinking.... could be early 2003 before we see a 9700 killer.
:)
Even when NV30 does come out and if it does beat the 9700 in performance, ATI will simply respond with their planned Radeon 10000 which will use the 0.13u process that the NV30 will release at - this should get ATI back up near or at the top. The you have the fact that the R300 is meant to work with DDR-II memory chips..... and Samsung just sampled 1GHz DDR-II chips so the R300 is going to rock the boat for the NV35 too.
The NV30 is rumoured to be multi-chip, I hope it can scale up well so Nvidia win the "numbers war" at least (unless ATI want to go R300 MAXX which is possible for them).... I would be happy with either of these two cards
- HeXa
Yea, so when previously I had to replace a bad stick of RAM I now will have to purchase an entire computer. Hrmm, will the poor college student want to spend $30 on new ram or $300+ for a new board...?
"The power requirements will be more than what ATI wants to run through the AGP port, so the card will have an extra floppy-drive sized power connection."
That's very interesting. For one thing, I don't know of many cases which come with two floppy power connections any more. Other than that, it sounds like a good idea. Finally use the legacy floppy crap for a modern purporse...
Its the waiting game as usual. Do I buy a GF4TI 4600 or the new R300 when it comes out... or do I wait for the new NVidia card that is sure to come out in the next 6 months. My current card is being pushed to its limits (gf2mx32mb) with the new games that are coming out and I'm willing to spend the big bucks this time round to get a high end card. What to do!?
An optimist believes we live in the best world possible; a pessimist fears this is true.
Now we can all play games at 3x the refresh rate of the monitor.
Sure you can. Hell, you've been able to for years now. Just drop down to the lowest resolution available, turn off all the effects, textures on lowest quality, and look at the not-so-pretty pictures.
What? You want all that? You want fog, bump mapping, realistic lighting, high quality textures, and anti aliased to boot? At 1280x1024? Well, then keep upgrading. Because while the R300 comes closest to that of any card to date, I doubt it'll be able to handle it for long. Real time graphics still aren't even approaching the level of Toy Story, much less that of Final Fantasy (the movie), or (*gasp*) photorealism.
When we can do realtime 3D effects that are indistinguishible from reality, we might be there. We aren't even close yet.
Oh, and you contradict yourself - you ask for disposable form factors and then ask for an open laptop standard. Hint - if it's disposable, it's not going to be open. Unless you're talking about something as silly and trivial as alkaline batteries.
It's my understanding that the human eye sees at approximately 60fps, not that it can't tell the difference between 60fps and 120fps. It COULD be smooth if the monitor was synced with your brain but we all know its not. I don't know about the rest of you but I have an easy time telling between 90 FPS and 120 FPS.
Who came up with these numbers anyways? They could easily be referring to movie or television framerates (which IIRC are approximately 60 FPS). The diffence here is that television is blurry as all hell with quite a crapload of noise. It becomes quite difficult to tell any theoretical difference in framerates while watching television. With video games we see solid colors and high resolutions. Which makes movement easier to distinguish, and therefore, easier to gauge with the naked eye.
That depends on what your definition of drivers is....</voice>
If by "drivers" you mean "closed source drivers for the FireGL card based on this chip that support all the card's rendering features, but none of the video capture or tuner functions of the inevitable AIW version", then I would guess about 8 months.
If you mean "closed source drivers that support all the rendering, video capture, tuner, etc. functions of this card" from ATI, then I suggest you monitor Mr. Andy Krist's credit cards for purchases of cold weather gear - this will happen about the same time the MBA selects Dr. Hawking as a star player.
If you mean "Open source drivers that support some of the rendering, none of the video capture, and none of the tuner", then I would guess about 18 months.
Sad but true. A pity - were there to be good drivers for this card (good = open source, all features supported by the standard APIs (Xv, Video4Linux2, DRI)) then I would pay up to $500 for one.
Now, the question is, what about all the Mac owners?
www.eFax.com are spammers
If I change the name from "QUAKE" to "QUACK", will the performance drop by 20%?
Fascism starts when the efficiency of the government becomes more important than the rights of the people.
Congratulations ATI, on your wonderful new video card. But you lost me back when your "new and improved" drivers for the AIW-128Pro effing freaked my system out. It took me a week to get it back to the way it was, and I am still not convinced it is as stable as it used to be. Until I hear people proclaiming "ATI finally got the drivers right", I will not consider purchasing another ATI video card. And even then, probably not, because there isn't that much of a gap between you and your competition. I don't need the latest and greatest card out there, I just want something stable.
My beliefs do not require that you agree with them.
Unless ATI has changed it's strategy regarding linux users, I'll ignore their new whizz-bang card.
Regular Quake3 benchies are not really of much importance anymore imho. Noone cares if they get either 200 or 300 fps in q3 :)
Things start to get interesting once you enable FSAA and AF....
I thought this might help those people who dont want to take the time to calculate the FPS from the earlier article Anandtech did.
Unreal Tournament 2003 (DM-Antalus) 1024x768x32 High Detail Settings
Radeon 9700: 130.4
GF4 Ti4600: 94.5
Parhelia: 54.4
Radeon 8500: 57.6
Unreal Tournament 2003 (DM-Antalus) 1280x1024x32 High Detail Settings
Radeon 9700: 87.8
GF4 Ti4600: 59.3
Parhelia: 35.1
Radeon 8500: 37.9
Unreal Tournament 2003 (DM-Antalus) 1600x1200x32 High Detail Settings
Radeon 9700: 63.3
GF4 Ti4600: 41.1
Parhelia: 24.6
Radeon 8500: 25.2
Unreal Tournament 2003 (DM-Asbestos)1024x768x32 High Detail Settings
Radeon 9700: 210.3
GF4 Ti4600: 178.2
Parhelia: 100.4
Radeon 8500: 91.1
Unreal Tournament 2003 (DM-Asbestos)1280x1024 High Detail Settings
Radeon 9700: 144.3
GF4 Ti4600: 115.4
Parhelia: 65.5
Radeon 8500: 58.9
Unreal Tournament 2003 (DM-Asbestos)1600x1200 High Detail Settings
Radeon 9700: 104.1
GF4 Ti4600: 82.0
Parhelia: 46.9
Radeon 8500: 42.0
Jedi Knight 2 'demo jk2ffa' @ 1600x1200
Radeon 9700: 124.3
GF4 Ti4600: 113.0
Parhelia: 65.9
Radeon 8500: 93.2
Serious Sam 2: The Second Encounter 'Little Trouble' 1024x768x32
Radeon 9700: 115.2
GF4 Ti4600: 100.2
Parhelia: 67.4
Radeon 8500: 58.2
Serious Sam 2: The Second Encounter 'Little Trouble' 1280x1024x32
Radeon 9700: 102.6
GF4 Ti4600: 72.9
Parhelia: 49.5
Radeon 8500: 45.3
Serious Sam 2: The Second Encounter 'Little Trouble' 1600x1200x32
Radeon 9700: 77.6
GF4 Ti4600: 51.7
Parhelia: 37.3
Radeon 8500: 32.1
"Im drowning here, and you're describing the water!"
I really don't think slashdot was the one that killed Tom's. Every hardware site on the planet probably has a link to their article.
Carpe Cerevisi - Seize the Beer
Every time a new card is announced, some uninformed consumer feels the need to point out that Quake 3 runs at 3x the refresh rate of the monitor. And, he always gets several intelligent resposnes explaining how Quake 3 is a benchmark, and that some games, like UT3 at full quality setting, still can't run very fast with the current cards.
And, yet, the original post always gets to "5, Insightful".
It's almost as if the hardware review sites need to explain the situation in the intro of their review, so that the masses could understand what's going on.
We still need cards to push three TIMES that to be able to play DOOM 3 by next year. PLEASE hurry up. It will really suck if Doom 3 is released before any cards can play it decently. (The demo shown was on an unreleased card running at low resolution..) ..
No, what we need is people to make games that will take advantage of the hardware we have now...
"Save me jebus!" - Homer Simpson (btw, I'm probably talkin out of me arse)
You don't seem to understand the direct relationship between MHz and pixel/texual fill rates in video cards.
The Parhelia gets beaten in DX8 bechmarks by the GF4Ti because of the difference in clock speeds. If the 9700/10000 wants to compete with the NV30 at the top then raw MHz is needed - the DX9 specs state that both cards need 8 pixel pipelines for compliance (and both ATI and Nvidia say their next-gen cards do). This means whoever has the highest clock rate will have the highest pixel fill rate. Need to wait and see if the NV30 has more than one texture unit per pipeline, the R300 only has one (for a total of 8 texture units) which means if the NV30 has two or more for each pipeline, then it will beat the R300 in texel filrates by architecture alone (though for more than one texture unit per pipeline will need HUGE amount of memory bandwidth which is not likely to happen until DDR-II is utilized).
So MHz does equal performance... which can mean a marketing success or failure. Parhelia is not grat for gamers (other than TripleHead) due to low clock speeds. Nvidia have delayed the NV30 to make use of the 0.13u process to get higher clock speeds (rumoured to be 400+) and ATI plan on releasing the Radeon 10000 next year based on a 0.13u process as well to try and beat the NV30 if it proves to outperform the 9700.... which is expected to happen.
- HeXa
No; considering that it's games that are driving the graphics market forwards; we need proformance to make Doom 3 run at 100 fps; MS word on the other hand......
This new card is still not fast enough to play Doom3 were it to come out today. Buying these generation of video cards are worthless because they will need to be replaced when Doom3 is released anyways. Just like Doom and Quake made everyone upgrade their systems, Doom3 will make you do both. Technology isn't keeping up with the masterminds. Help.
Boy, I sure hope people give ATI a ton of shit for breaking their own naming convention like they
did NVIDIA for calling NV17 GeForce4 MX.
ATI gave tons of presentations about how clear their naming scheme was and that the first digit
represents the DirectX version. Now they have screwed that up by calling R250 Radeon 9000 when
it is a DirectX 8 part and not a Direct X 9 part like the Radeon 9700.
The difference between 0 and 7 is no better or worse than the difference between MX and Ti.
I was just looking at Toms review and his 9000Pro makes due with only a passive heatsink. Anand otoh has a 9000pro with a fan. I'm a quiet pc kind of guy and am really interested in a modern, performant Graphics Card that does not generate noise. What is the story on this?
Now, with no claims of being objective, go buy a 9700 for yourself, and show your family how much you care by buying one for all your family members.
If the stock triples, I might be able to afford a 1987 Honda Civic with only 200,000 kilometres on the engine.
Hello? Anyone listening?
:)
Yes, an excellent troll; you've sounded passionate enough and invoked "cheaper!" enough to confuse some moderators into giving you points. Let me go over your completely misleading rants:
Now we can all play games at 3x the refresh rate of the monitor.
We don't PLAY games at that rate; we benchmark them there for comparison with other cards. I have a 15" LCD that I game with. I run my refresh at 60Hz, which I am imaging you are translating to "a maximum displayable 60 frames per second." I'll let someone more technical than myself debunk that one.
I just got a very nice MSI GF4 Ti4200 (for $145 from GameVE.com, free shipping, only because newegg doesn't carry them and they are extremely overclockable cards with a great software bundle). If I ran this card in my LCD's native resolution of 1024x768, with the basic graphics settings, I pull approximately 180 frames per second in Quake 3. If, however, I go to the driver settings, crank up Aniso filtering, and turn on 4x FSAA (anti-aliasing is beautiful, btw), and set all settings to max quality in the game, I get about 85 frames per second in Quake 3. That is what my GF4 MX440 card was pulling with no options on.
We need cheaper and more integrated. Get rid of the DIMM and PCI slots and all the legacy hardware. Put the memory on the motherboard and create a disposable form factor and an open laptop standard.
Again, very nice karma troll. Cheaper is nice, and integrated has its place, but we do not need it. We don't want to put memory, cpu, and all peripherals on the board, for a variety of reasons. The two bigs ones are 1) repair/replacement after failure, and 2) CHOICE. If you want to buy a $20 video card to put in that AGP slot, you can! If you want to buy a $400 Matrox Parhelia to run 3 monitors in Quake 3, you can! Anything and everything between, as well! Everybody has a different budget and a different set of needs. Let the consumer decide.
Disposable form factor? Is that tongue in cheek? Do we want disposable PC cases? Or just good standards like ATX? I know plenty of people who have had ATX cases for 5 years and have housed 4 different generations of hardware in them.
Open laptop standard? Good idea, but many OEMs already work from something similar. The problem is the high cost of development on miniturized, highly integrated systems like laptops, especially when they need extremely tight cooling systems. Someday there will be a standard, where you can go into a store, buy a chassis (for 12, 14, 15, 17 inch LCDs), assemble the mobile parts, and walk out the door... but why bother? There are tons of cost-effective, and vendor-serviced laptops available in any conceivable configuration RIGHT NOW. Just because you can't get it for $1.99 at 7-Eleven does NOT mean the market is broken.
So my summary is that we don't need more integrated, and cheaper is good, but we have cheap already. You were trolling, and I was feeding you. Any questions?
SlashSigTheorem: Humorous, Political, Critical, Constructive- If you have a
hmmm point taken... but (heh, yes a but) did Carmack not say that a GF4TI4600 would play D3? I'm getting in the range of 30-40 fps in RTCW now so any improvement would be nice :)
I wish Mr. Carmack would comment on the R300. I guess he will when he gets his hands on one. ATI give Mr.C a card!!
An optimist believes we live in the best world possible; a pessimist fears this is true.
Many pages only have 10 or 11 sentences of text on them. Is the high number of page/banner views that Anand gets really worth the extra traffic (and annoyance) that this causes?
Slashdot: come for the pedantry, stay for the condescension.
My question is: will there be decent Linux drivers for this (R300) sucker? Otherwise, I'm just limited to nVIDIAs for good gaming graphics. (or am i missing something?)
We /.'ed Anandtech. We're bastards.
I thought AT had the hardware goods to handle this crowd.
I am pro-lifechoice.
1) miquoted.
"Linux is only free if your time has no value"
2) misattributed.
Jamie Zawinski
Read the whole thing here:
http://www.jwz.org/doc/linux.html
--
E_NOSIG
> Now we can all play games at 3x the refresh rate of the monitor.
:)
> We don't need faster anymore.
Nonsense.
A *lot* of people want movie quality graphics in real-time. Imagine playing a game with the visual quality of "The Matrix", but completely interactive!
There are a ton of physical effects that still can't be done in real-time (yet), due to the memory bandwidth and geometry complexity.
e.g.
High resolution (4Kx2K) color/z/stencil buffers used for ray-tracing, radiosity, and displaying thousands of models each with a million+ vertices (used by CAD/Medical/Games), etc, come to mind.
Then when you add in compositing / blending multiple alpha layers you just burnt all your left-over speed (if you had any). DOH!
There is a reason that Pixar and other CG studios render scenes at the *sub* pixel level @ 64 bits/pixel. We're talking about 100+ triangles PER pixel. Because detail, such as hair which is less then a pixel wide, needs to be "super-sampled". Right now, games show hair by approximating the surface of it which makes it look "blocky". UGH.
So if you want graphics to stagnate, and never look more "realistic", then sure, stick with your GeForce4 (or below.) It will continue to be usefull for the years to come.
The rest of us will be trying to dazzle the world with new visual FX making people go "Wow!".
He thinks the R300 is sweet (not his exact words but similar meaning). He also stated that the ATI 8500 and GF4Ti cards would play DOOM 3 fine, just at medium quality levels or so.
The big difference is the memory your card has, 64MB cards won't be great performers for high texture settings due to a statement about DOOM 3 having about 80MB worth of textures to be loaded onto the card at any one time.... 128MB cards will be needed for maximum eye candy.
- HeXa
Wow. Impressive specs on the R300. For anyone in the know ("hint" ATI Employees "hint"), was the design of the R300 done primarily by the folks ATI got from buying out ArtX? For folks who don't know, ArtX was the company that did the GPU in the N64 and Nintendo Game Cube.
Like a big mouthful of marketing! It makes terrible business sense to release a game that only runs on the NEWEST cards. Almost nobody would play it. That would drive the cost of the game to roughly $450; far more than 99% of people are willing to pay. Believe it or not, but game companies are in it for the money, and though they work hard to satisfy the hardcore gamers, I'll bet they get most of thier cash from more casual players, like myself. I'm interested in playing Doom3, but if I'm expected to shell out $450 for the priveledge, I'll pass. Frankly, if you want a new video card now, get a GF4Ti4200 or Radeon8500 based card. You'll save $200 and will have most of the performance of the highest level. The idea that Doom3 will not run with a GF4 comes from one place, card marketers. Sure, you may have to tone down the detail and/or the resolution, but honestly, that's ok. I for one am not willing to shell out $400 every 6 months so that I can see the whiskers burn off my opponent's face when I hit him with a flame thrower.
The drivers for the ATI 7500AIW aren't working very well with the latest DRI code now, and there are problems with the I2C bus on the ATI - it's not documented where the bits for it are. ATI has been asked to provide this information, and last I'd heard it has not been provided.
www.eFax.com are spammers
What if I don't run linux? This card sounds pretty sweet.
That being said, I'm not buying a new videocard until Doom 3 is actually out.
What, is ATI paying slashdot kickbacks? It seems like there's a story about how great the new ATI cards at least twice a week.
/bin/fortune | slashdotsig.sh
According to the horse's mouth (Carmack) current cards will play Doom3 "fine". Naturally this means that you can't do 1600x1400 maximum graphics but with reasonable settings it will play.
He warns of the GF4MX though, but I think everyone knows that that is not a good card for a gamer anyways.
Cheaper, yes; more integrated, no. Cheaper computer junk is always nice! but I disagree that we need more integration, that does make it easier to improve performance, but we have got so much performance already that I would rather have the ability to easily upgrade components. I say keep things seperate, and make them more afforadable.
Comment removed based on user account deletion
I am personally waiting to see benchmarks from retail off the shelf cards using the supplied drivers. ATI always seems to have good lab numbers but comes up short when the released product is tested under everday conditions. Also as others have pointed out it's not too impressive comparing it to video cards that have been out for awhile now. I hope that ATI can bring a truly competitive final product including stable drivers and I look forward to seeing the benchmarks against a released NV30 in the next several months. Christmas could be expensive this year.
http://www.kubuntu.org/
The "Video Shader" in the new chips is meant to offer great video playback quality...
1. Yes - ATI is one of the best for the consumer PC market
2. Probably - need to hear more feedback
3. YES - ATI actually do work on their drivers for the last year.... I download leaks every 2-3 weeks on average for my 7500 Mobility
4. Eventually.... Windows comes first for ATI due to the fact that makes up most of the target OS market.... Linux support will via binaries will happen later most likely.
- HeXa
about your LCD, I've been thinking of buying one of those for awhile but was always afraid that it would "smear" the screen in CS, Quake, etc. Do you get any of those effects or is it clear? What kind do you have?
Who cares if ATI is top dog untill the next chip comes out? Who cares if ATI will be second place next month? Who cares who is the current kick ass product?
/. releases a story about the security hole big enough to drive a semi-trailer through anymore? Or do we read it, and patch it as needed (or mutter something to the effect of thank god I have Linux)?
Personally I don't, and you know why? Competition. Good clean, healthy, product innovating competition.
Something that is sadly lacking in the DeskTop OS market. Not to name any names *cough*microsoft*cough* but there is a very good example of what having one and only one player in the field. Poor quality product that we keep seeing so many bugs that we've become desensitized. Really, who falls over stunned when
So ATI is ahead today, so nVidia will be ahead tomorrow, so what?
Be glad that there is more than one dog fighting over the bone
Phoenix
-- Wiccan Army, 13th Airborne Division "We will not fly silently into the night"
So, Nvidia has at least competitors. As an 3dfx owner, I don't want to become alone.
We, 3dfx owners know how a company they are... Since we are without new drivers since nvidia bought 3dfx.
No need to describe, I guess 3dfx owners with a clue understood what kind of a company they are... In hard way...
Oh me? When it ships (or shipped already), I am buying it... I won't buy from a company which left me in "digital cold" just because they bought my card/chip maker...
mod me as you wish, I couldn't stand not saying this stuff...
Looking at these benchmarks it appears that ATIs advantage is in their memory bandwidth, by using state of the art memory and better pipelining they are able to beat Nvidia by significant amounts only in benchmarks that require extreme amounts of memory bandwidth. At normal resolutions their actual '3D' throughput using their next generation chip is not that much better than nvidias current generation chipset. This does not look good for ATI. Give Nvidia equal memory bandwidth and they would be within spitting distance on all the benchmarks even with the current level of GeForce.
They haven't released the actual card specs, this is just a sample.
The released card could run at different speeds.
How hard would I have to overclock a Ti4600 to meet these specs?
not Apple the marketing freeks, no way..
/. beets Apple!!!
Either Apple has turned from a Jornalist/ graphics designer platform for a Geek one or those apple boys have got there marketing team in full swing!!..
Sections..
Apachie 1
Apple 8 !!!!
askslashdot 9
books 1
bsd ZERO
developers 8
features 2
interviews 1
radio ZERO
science 7
yro 4
Only ask
thank God the internet isn't a human right.
When we can do realtime 3D effects that are indistinguishible from reality, we might be there.
I find it interesting that untouched realism is frequently just not fun. There are aspects to games that require tweaking or simplifying the environment so it isn't frustrating or impossible to make progress.
Masters of game-making understand that fun isn't derived purely from realism and that some unrealistic elements are the only way to make a game interesting and playable. For example, do you really want a football to get lost in the sun, so the receiver screws up and you lose the game? Or do you want clues in a mystery game to be so well hidden that you have to have take the hours of a real forensic investigation to find that triply-ricocheted bullet embedded in the neighbors compost pile?
I really think that super-realism in games is a pipe-dream, and the only way to achieve it is in a Star Trek-style immersive holodeck...or, perhaps, just going outside.
It also seems to be harder to find the basic time-waster games, since, I guess, it is a waste to put classics like Tetris or Solitaire on gigaflop-class consoles. In a way, this really is not progress at all.
Healthcare article at Kuro5hin
Seems to me that if ATI is going to be making a new video card, perhaps first they should make some drivers for their old cards that actually work. I switched awhile back from a Voodoo3 3500TV to a Radeon AIW. It's a really nice card, it's got lots of features, it's fast... it's a nice card... but the drivers are such crap that it almost never works like it's supposed to. The win98 drivers are swell... but in XP it doesn't run anything that runs fullscreen (that used to run on the 3500... XP was a fresh install so it's not a residual driver issue) and 90% of the directX stuff doesn't work (including Warcraft III or other new games). Note to ATI: If you make a good card, make drivers that make it work!
Anand states: We didn't have much time with the Radeon 9700 so we couldn't run a full suite of AA tests nor take screen shots
Come on, this has got to be bullshit. All it takes in most games to take a screenshot is to push PRINT-SCREEN on the keyboard. All these tests, different games, different settings, it must have taken at least half a day to complete, and not once did they have the time to push print-screen?
Just say 'ATI wouldn't let us publish screenshots' instead of lying about it. Oh, maybe you weren't allowed to say that either? Bah.
I think it's great we'll soon have some competition in the arena, but these are really previews of things to come, previews tightly controlled by ATI, I'll wager.
Belief is the currency of delusion.
I bought an all in wonder Radeon for TV tuning and video capture, but it wont work with 2000 or XP so I had to put ME (ick) on my Asus K7 for it to capture.
Nvidia has a very good track record of putting out quality drivers. ATI has an even longer track record, but its in producing shit drivers for the hardware.
So a new card runs an old game faster than another card? I would hope the new card is optimized for more than just running old games faster. The benchmarks are "totally" meaningless. AS anyone that has read a Carmack plan file can attest.
I have a NEC MultiSync LCD 1530V - 15" diagonal, 1024x768 @ 75hz native.
It works beautifully for everything I've played thus far. As long as it stays at the native resolution. Try and run it at anything outside that (nevermind the fact that you can't go higher), and it begins to look horrible. Another thing you might want to try is playing CS in 1024x768 cloned onto a TV screen. You can't read the chat text for shit, but most of that is just insults anyhow. The game itself looks pretty good, in my opinion. Worth doing just for humor's sake, at the least.
I think that he's saying that graphics cards can't yet do real time photorealistic rendering. When they can, that's when we need to stop developing faster boards and make them more intergrated.
Games will never need to be photorealistic to be fun, the two things are totatly unrelated. I think that most graphics in games today are wasted on the player anyway, I never realized how good looking many games were until I could watch instead of play them. GT3 is a good example, amazing to watch, fun to play but you really can't enjoy the graphics when you are playing it. Same with many games for the X-Box. Or in MGS2, all the time, I'm checking out the little radar in the corner of the screen to see where guys are, not seeing any of the awesome special effects that they spent so much time on.
I'll post numbers from my Word benchmarks*
SIS6326 video card:
Loads Word in 12.0003712 seconds
Nvidia NV30 pre-production overclocked:
Loads Word in 12.0003101 seconds
As you can see here, the Nvidia card blows the SIS card out of the water. Our base-line TNT2 Ultra card falls somewhere in the middle of these at 12.0003442 seconds. As you can see this is a very good benchmark that is better than those stupid game benchmarks.
*All benchmark numbers are figments of my imagination and totally worthless
Only problem with this comparison is that you are comparing the new ATI numbers on a P4 2.4B platform to the older GeForce 4 numbers on an Athlon XP 2100 platform. I believe the difference wouldn't be too much in most games tho.
I bet that Anand had to use the P4 platform in order to make it as different/diffcult as possible to find exact real performance numbers.
-T
It should be noted that the sample they sent out is not necessarily what the final clocks will be. The obviously didn't pick the slowest sample they got back from the fab. If the production boards ship at 325 MHz then these numbers mean something. I am betting on 300-315 MHz but I could be wrong.
Now we can all play games at 3x the refresh rate of the monitor. We don't need faster anymore.
:P Anyway, you're obviously really clueless on this, or a troll.
I wasn't sure whether I should just mod you down, or reply. Obviously I'm replying.
First, most benchmarks measure average framerate. That's fine, but what's really more important is the LOWEST framerate in a benchmark. A card might average 80fps in a certain game, but dip down into the 30's, 40's, or lower during really intense action... which is when you need high framerates the most.
Second, PC games evolve. Sure, nobody needs to run Quake3 at 250fps, but future (and even current games... Q3 is several years old) games will be much more demanding. So when people drool over ridiculously-high Q3 benchmarks, they're really drooling over the raw power of the card, and how it's going to run games a year or two from now. Some people like to purchase for the future.
Third, while I don't know of any open laptop standards, there's plenty of motherboards with integrated video and other components. While it still has SDRAM slots and 1 PCI slot, the motherboard has a built-in NIC, modem, video, and sound. And I don't see a downside to non-integrated memory... why would you want it soldered onto the motherboard?
OtakuBooty.com: Smart, funny, sexy nerds.
I bought mine about a year ago (wise spending of my "tax relief" check from George W.) at WalMart; the KDS RAD-15. They still sell around $370 at most Wal-Marts, and many SAM'S wholesale stores.
I have zero complaints. The cost was good, I got zero dead pixels, the fit and finish of the unit is very tight, and the refresh runs up to 85Hz, depending on the video card you have it hooked up to. The brightness and contrast are superior; I have hooked up a 17" tritiron and a 19" tube to test at 1600x1200 to see what I am missing, and I REALLY miss the flicker-free brightness of the LCD. (PS - No magnetic interference from things like fans, either! Just solid picture.)
The only nags that I have are it's not big enough pixel-wise (because I am too cheap to buy a nice 17-19" unit), it only has the standard 15-pin VGA connector (most newer ATI/Matrox/Nvidia/SIS cards have DVI connectors), and when I am in a fast round of capture the flag in Q3:TA with Scout powerup, I can see a little bit of clipping during power running and jumping. Just a hint of it, but enough that the purists would complain.
There really isn't any "blurring" as you might have seen on dual-scan LCDs, but sometimes you can see some pieces clip as the screen does fast color-changes during colorful terrain areas. Granted I have a GF4 Ti4200 cranked up to 300 core and 600 RAM, which is capable of running at many more frames per sec than an LCD monitor is capable of displaying cleanly (because of the alluded-to activation/deactivation time on the pixels), but it is a good value for $370. At least it is for me, and I have been dealing with monitors for a long time. I plan on staying with my 15" until the 17s are available for under $300... in other words, a long time.
SlashSigTheorem: Humorous, Political, Critical, Constructive- If you have a
You obviously dont play video games, and never had. When QUAKE came out, everyone had 486 DX4 100s. It WOULD NOT RUN acceptably on those machines, P60's were JUST released, and it BARELY ran on those. People weren't getting GOOD performance for QUAKE till nearly 6 months after it was released! The comment came from the PROGRAMMER HIMSELF. Doom3 will NOT run acceptably on these generation of cards NOR the ones posted about here. He told everyone this at E3 recently when he demo'd Doom3 on an UNRELEASED *new* card from either ATI or NV and it WAS STILL TOO SLOW because it was in LOW RESOLUTION. You are so entirely wrong it isnt funny.
With all respect to the benchmarks, ATI was "cought" making drivers with game specific optimizations (Quake / Quack etc).
Check if this is the case here. Might not be, but it just might, which should be taken into consideration.
^_^
He doesn't say "fine". He says they will run it. Much like how he said a high end would run Quake1. (Ie. 320x200 with the screen shrunk 90%)At e3 they had Doom3 running on one of the soon to be released cards in low res and it was just barely performing fluidly on a totally buffed out system.
Does anyone else get the feeling there's a director of marketing at the GPU companies who just pulls a Dr. Evil inspired number out of the air and declares to the engineers that that's the FPS for Q3 they need for the next card?
"I want our next card to not just give me a frame per second but TEN frames per second in Quake. We'll see how they like that, huh?!"
"Uh, Sir? We already do one hundred and ninety two frames per second at maximum resolution."
"I see. Then I want..." finger curls against his bottom lip, "....One Beeeeeeeellion frames per second."
"But Sir, that's insane! No one can tell after about 50fps anyway!"
"Mini-Marketing-Me, keeeeeell him!"
The article at anand's says that ATI has abandoned their previous 'unified' drivers and that the drivers for R300 are 'brand new' and that this could create compatibility issues with older games and so on...
Now, if there's somebody here that knows more, what exactly does 'unified' mean in technical terms (not marketspeak)? How can a 'unified' driver work for vastly different cards like GeForce1 (basic T&L) GeForce3/4 (shaders and so on) and NV30 (FP pipeline)?
I don't see much that can be 'unified' in those architectures, and even less between the DX_8.1_and_lower and DX_9_and_higher parts, given the jump from integer to FP pipelines.
So, is the claim of 'unified' drivers purely marketspeak (and maybe it's just a collection of 'workarounds' for specific game problems) or is there a technical case to be made for them?
-- the cake is a lie
Of course it matters!
Just think, how many bragging rights does 60 f/s get you these days?! NONE! But if your card gets over 200 f/s then you get ALL the bragging rights in the chat rooms! You then become really 'l33t!!!
Ohh, other then bragging rights though, you're mostly correct. The human eye tends to consider motion to be fluid at somewhere between 16 and 25 f/s, depending on who you ask. The eye also stops noticing any difference between frame rates at around 60 f/s, and it stops noticing even complete picture changes slightly higher then that (hence the reason why 60Hz refresh rates are a bit too low for most people, but 75 or 85Hz is fine).
Another thing to remember is that most games play at a refresh rate of 75 or 85Hz, so beyond that, the difference is totally none-existant (the extra frames are never even displayed), so all those people who say "I can tell the difference between 75 f/s and 120 f/s" are completely full of shit and are seeing things that very litterally are NOT there.
I know that a lot of people all talk about how these are only "average" frame rates and that when things get really complicated in games your frame rate drops, however most of these people have never bothered to qualify this. If they did, they would find that in most typical games, the worst-case frame rate is more then 50% of your average frame rate quoted in most of these tests. In other words, anything beyond ~150 f/s makes absolutely NO difference at all because both cards ALWAYS display exactly 75 f/s (or 85 f/s, depending on the refresh rate). What's more, most of the frame-rate tests used these days are designed as an absolute worst-case scenario to begin with, so in typical play the numbers shown in this tests are closer to your minimum frame rate then an average.
However, that still doesn't change the bragging-rights requirement..
Just a bit of food for thought...
Intel had to REALLY stretch to get the PIII core up to 1.1GHz on a 180nm fab process, including having to recall their first attempt entirely. With the P4, they had little trouble releasing a 2.0GHz core on the exact same 180nm fab process.
Which do you think is faster, a 1.1GHz PIII or a 2.0GHz P4? Intel's design strategy for the P4 wasn't all about marketing, being a little bit less efficient but clocking a LOT higher isn't entirely a bad thing.
Of course, I don't know just how well this correlates to ATI's newest video card, we'll just have to wait and see.
I don't really even care about staying at 85 fps, but going below 30 is irritating.
Free Mac Mini Yeah, it's
Dumb-utt or Dum-butt? I don't know what either of those would mean.
--
E_NOSIG
If you cast your mind back to around the end of 2001, ATI shot themselves in the foot by placing a few sneaky modifications in the driver to produce faster benchmarks in quake3. Here's John's views on the ATI Quake3 driver trickery.
Personally, that sort of behaviour is just low. I'm quite happy with my GeForce3 Ti200, and I shall remain so until I cannot run the games I want to at 1024x768 with good framerate and visuals.
Ali
Ph33r m3!!!
Small nitpick: More MHz at less 'work' per MHz, while possibly driven by marketing concerns, does not equal a "shittier" design. It still works (and the fastest P4's do outperform the fastest Athlons, in general, I believe).
Also, making things run that fast is still difficult. Any design that runs that fast and can still actually do reasonable amounts of work is not especially bad.
Ok so it depends on your definition of "fine". I'd expect 800x600 with low-medium effects or something like that.
Now if you're a serious gamer that isn't really "fine" but for a lot of people it is.
Not a troll, and not off topic, or at least, not intended to be.
Guess I've just been burned too many times by crappy drivers that don't do everything that was promised. And I'm talking Windows drivers here. Not quite complete OpenGL support, games not working correctly when they come out, games still not working correctly months after they come out because they are not big enough titles to get ATI's attention...
Now I'm not the world's greatest NVidia fan either. I'll complaign about their lack of innovation, the way they seem to just want to throw more hardware at the problem, rather than find a more elegant solution, whatever. But as long as their drivers manage to play the new games, and they keep the new drivers comming out that I can play the new games when they come out, I'll take their cards, even if they are slower.
A faster video card that doesn't play what I want it to, when I want it to, is of no use to me...
And to bring this a bit back on topic... This is basically a warning to prospective buyers, check out ATI's track record for drivers before making a purchase. I haven't heard too many problems with the Radeon card, so they may have finally turned their policies of not caring about you after you give them your money around. But lets be honest here, current correctly working drivers are more important than the gap from 120 fps to 150 fps...
This is absolutely not true. At 120fps, you get half the latency that you do at 60fps, and that is very important (even in a single player game).
- A game developer who actually knows what he's talking about.
That's because the pixel recharge/cycle delay is too great... I have a Dell 2000FP 21" LCD (around $1,400), and there's no such "smearing" or blurring that I've noticed. If you buy a cheap LCD, you're most likely going to hate it for gaming/video/etc. The cheapies are really meant for "2d" type stuff.
Not All Who Wander Are Lost
ATI's drivers suck badly. I made the mistake of upgrading my Radeon 7200's drivers from the 4-13-7075 version to whatever happens to be current at the moment and now Visual Basic 5sp3 crashes when I try to compile (VB, yes, I know... Please don't go there.) and only half of my DIVX movies play with overlay enabled. On the plus side, now I can actually play GTA3 with a VERY THICK FOG, which is much better than the old drivers which just caused it to crash.
As much as I like playing GTA3 slowly with extra fog, I found a patch to fix the fog problem on ATI's site. I reported the other issues I was having to ATI's technical support; however, since ATI doesn't even have a checkbox for reporting application or video related issues (I'm not kidding, see for yourself on their support page!), I have my doubts about the issues being resolved. My only recourse is to go back to the old drivers and get a Playstation2 if I want to play GTA3... Course, there's also Nivida.
---
Siggy, siggy, siggy, can't you see? Sometimes your puns just irritate me.
No really. I NEVER use floppies. CD's are cheaper,faster, and more reliable
No, that's not 51% faster. If the GF4 is normalized to 1, and the ATI card is at 1.38, that means it's running at 138% of the GF4 capabilities, or 38% faster.
Honestly, if they'd used an fps base and you'd had to do this with Gnome's calc, I could see you screwing up. But screwing up (1.38 - 1) * 10? Ouch. You should be able to do that in your head.
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
We don't PLAY games at that rate; we benchmark them there for comparison with other cards. I have a 15" LCD that I game with. I run my refresh at 60Hz, which I am imaging you are translating to "a maximum displayable 60 frames per second." I'll let someone more technical than myself debunk that one.
What's there to debunk? If your refresh rate is 60hz, then how is that not "a maximum displayable 60 frames per second."??
Maybe you've discovered a way to overclock your monitor? If so, please use a decent heat sink.
How do you know it meant a shittier design? Were you in on any of the design meetings? Didn't think so.
Now, reread the section you misquote so heavily. The reviewer was specifically discussing an Intel-like design in that they hand picked transistors to go in specific places, rather than letting a VLSI design program layout the chip according to how it thinks is best. Intel and others have shown that this strategy - hand tuning critical junctures - can pay off in performance and manufacturing.
Intel's chip designs have been pretty damn amazing for the past two decades. They've frequently been the ones pushing Moore's law (yeah, go ahead and take the obvious whine - "because they needed to, their chips are so inefficient"), and they've eeked a helluva lot more features and performance out of designs than anyone in their right mind expected. Their fabbing is second to none and their processes are emulated industry wide. A 35% yield for a first run Intel design is godawful, but considered spectacular for other companies.
Have they made missteps? Yup. And I largely attribute those to upper management sticking its head deeply up its ass rather than to the engineers. Intel's brass stopped listening to their engineers 4 or 5 years ago. And it's been biting them since. Remains to be seen if they've figured this one out yet.
that MHZ no longer equals performance
No, it doesn't. But it's often a damn good indicator of performance, particularly in the GPU world. Frankly, the only people who know what the clock speeds on the chips are are the geeks who are into this thing. They're not advertised on the packaging.
Yep, he's talking about photorealism, not superrealism. Photorealism would be a troop of goblins marching through Times Square - no way it could be real, but it's rendered so well that you can't even tell it's not real without knowing goblins don't exist. Same for the FF movie, none of it's real, but lots if it you would be hard pressed to tell it's not real without knowing it came from a computer.
we are building a religion
a limited edition
we are now accepting callers
for these pendant key chains
If you're working with a standard (tube) monitor, the refresh rate is just how frequently the gun is making it back to the same point on the screen. While in practice this means that the screen redraws 60 times a second at 60 Hz, that does not mean that it drew 60 frames generated by the video card in that second.
:)
If the video card is generating 90 frames per second, then during that second, the 60 Hz monitor has displayed bits and pieces of the majority of those 90 frames, depending on the gun position when the frame was displayed.
If the video card is generating 35 frames per second, then during that second, the 60 Hz monitor has displayed (probably) all 35 of those frames, sometimes two or three times as the gun scanned the screen several times before another frame was pushed.
At least I hope I'm thinking this out right, or MAN am I going to look like an idiot.
SlashSigTheorem: Humorous, Political, Critical, Constructive- If you have a
Or how about (1.38-1)*100 instead of *10? Or 1.38 1 - 100 * in RPN.
If you had a clue you'd realize that nobody buys graphics cards based on clock speed. They're not advertised on the box! The only people who know the clock speed are geeks who read up on it, and those people depend on benchmarks anyway!
A deep unwavering belief is a sure sign you're missing something...
Bullshit. D3 is being demoed on R300s.
A deep unwavering belief is a sure sign you're missing something...
The thing that Intel did with the P4 is stupid though, a 2.0Ghz P4 may be faster then a 1.1 Ghz PIII, but is it almost twice as fast? I don't think so.
I agree that they do know what they are doing to get more performance, but it seems like they prefer to always increase the MHz rating, which as all geeks know isn't always the way to get more performance out of a processor.
With VSYNC enabled, it also means that any FPS above what the monitor can display are lost frames. This can actually be detrimental to gameplay when the game speed relies on the frames rendered, as you can miss a frame you "need".
Nah, it'll never happen. Apple doesn't have, nor will they ever have support for DirectX, the most commonly used API for games. Sucks for Apple users, but thats the cost of the crusade against MS.
You can use the sub-system clustering to shove the workload back on to the requesting utility if it has itsown processing tool, under the One Seat Couch additions to the abjournouis kernel.
"Yeah...it was the numbers that were irrational, not the murderous cult of vegetarians...." -- Hippasus of Metapontum
If you properly read that section, you'd see that they qualified the aspect of 'Intel like' approach.
Here is the actual paragraph:
>An admittedly very "Intel-like" approach, ATI didn't go as far as to hand pick transistors but they did a considerable amount of the R300 design by hand thus enabling them to reach decent clock speeds at profitable yields.
There is nothing there about sacrificing efficiency for clock speed, but about improving implementation so as to allow for higher clock speed.
Also the Pentium 4 is not 'less efficient' than a pentium 3, it just does things differently.
As you point out, clock speed doesn't mean anything, so why does it matter if a P4 with the same clock speed of a P3 doesn't perform better at everything?
Especially since the P3 simply cannot reach the speeds a P4 can.
Also, faster clock speed means lower latency, which can be a performance benefit on it's own in certain circumstances, even with a lower IPC count.
Advanced users are users too!
Remember that refresh rates on LCD monitors don't really matter. With the way LCD technology works, the pixel will just 'stay on' untill it needs to change color or go black. Thats why you dont get annoying flicker, even with 60Hz on LCD monitors.
It's easier to fight for one's principles than to live up to them.
I smell a know-it-all that needs whackin'.
Modern graphics API's use time, not frames, to determine the speed of a game. The graphics rendering is completely independant of actual gameplay. It's the only way a game can be expected to run 'at the same speed' on the wide range of hardware.
Ever use a boot disk to load DOS and then play an old game? The original, unadapted Wing Commander (1989 version) is completely unplayable because hardware speeds were so close back then, that the programmers could get away with using frame rate to regulate gameplay.
A modern game doesn't care if the rocket impact is rendered. The game registers 'impact' by the vertex's position, which is computed seperately. When the graphics card does the vertex handling, the game still keeps a (much smaller) set to calculate object positions. In other words:
The graphics card computes thousands of vertices, and renders the entire scene once.
The CPU will compute a few hundred vertices. (the collision boundaries, which is generally a bunch of cubes the model fits inside) There is all kinds of time for the CPU to compute a few hundred intermediate steps before the graphics card asks for the next 'snapshot' to render.
No, you don't miss the frames at all. What is this so-called 'need'? First, there is a very big difference between keeping track of the objects (Poly boundaries/collisions, positioning the vertices, etc), and actually rendering them. Vertex calculations (including physics and animation) is much less computationally-intensive. That's why the first 3D cards really only handled rendering. The CPU still did all the vertex operations-- the 3D card did the (exponentially) more intensive rendering of the frame.
The way it usually works is as follows:
Frame Buffer A is displayed on screen
Graphics card renders to Frame Buffer B
Graphics Card renders to Frame Buffer C
When all of Buffer A has been displayed, flip display pages (or use a blit) to Buffer B.
Frame Buffer B is displayed on screen
Graphics Card renders to Frame Buffer A
IF Frame Buffer A finishes rendering before B finishes drawing, flip pages (or blit) to C.
Begin rendering B
If A is being displayed, render C.
If C is being displayed, render A.
If the buffer isn't being displayed, render the next frame. Show frames in order, but drop frames when a more recent one is available.
And so on. This is 'triple buffering', which not all games support (although it is becoming much more common). Double-buffering is almost always used, where there are only 'A' and 'B' buffers.
Which means, that even with vsync enabled, the card is capable of rendering 120 (double) or 180 (triple) buffered. (And that's at an eye straining 60 Hz. With a better monitor that refreshes at, say 85 Hz, the card renders 170 (double) to 255 frames per second.
-- Sometimes you have to turn the lights off in order to see.
It has been shown, however, that even though it's impossible to tell the difference in frame rate. However, in real life (as in games) there are things that happen too fast to see the motion.
Games are full of explosions, etc. Very high-speed motion. Most people have watched too many movies; they're used to 'slow' explosions where debris & effected objects are visible on screen. Movie makers know we like eye candy, so they give it to us.
Reality is quite different. A TOW missile explodes before it hits its target. The expanding gas forms a 2-4" hole in the targets armor in microseconds. A person watching it can't see the transition. Bullet wounds take 2-3 frames to fully appear in a movie. Reality is more like 1e-6 frames. Explosives can lift an entire car feet into the air so fast that a human thinks it's instantaneous.
Video games use fairly real physics, as it both makes animating easier, as well as having a more realistic 'feel'. The frames rendered follow the model. With even moderately real physics, an object can move large distances in between frames.
And, of course, there's the ultimate trump: Online gaming. Where the object boundaries (often simplified/compressed) must be transmitted over a low-bandwidth link, with a latency of hundreds of milliseconds. It doesn't matter how fast the graphics card renders, or how well the game keeps track of positions interally.
Updates of 30/sec is pretty optimistic, with 10-20/sec more typical. Other players can 'pop' locations in between frames simply because, in between location updates, the opponent's 'actual' location(s) end up being different than the one the CPU guessed it would be.
Which can mean 4-5 frames were rendered with incorrect locations, the update is recieved, and the 'real' frame is rendered. Next the game guesses where the opponent will be by the next update, and renders the frames necessary to make things look smooth.
The guessing is an imperfect way to make up for the large difference in frame rates and multiplayer location updates. However, there simply isn't any option; there are 4-5 frames that must be rendered before the next update. Simply 'stitting still' looks awful, and lends itself to the perception of a lower framerate than actually exists.
Programmers try to close the gap by making an educated guess. Since they use a realistic motion model (inertia, gravity, etc.), nearly all the possibilities for the 'next frame' can be eliminated immediately. Then it just chooses a 'middle road' that is close enough that us humans don't notice.
Any high-speed, unexpected changes (such as an explosion) can foil the system:
The player thinks they've killed someone (that's what was rendered/displayed on their screen, after all)
But the estimate was wrong. The 'someone' was actually in a safe place when the explosion changed things.
There is no 'backing up', so the next frame shows the person alive and well, and in a completely different place.
The gamer gets upset because they want a perfectly synchronized game
The much lower frequency of positional updates is unacceptably 'chooppy' when such synchronization is used.
The programmers use a 'physics' trick to try and smooth out the picture, but the trick sacrifices accuracy.
-- Sometimes you have to turn the lights off in order to see.
I missed a zero, and you missed that the point for my reply wasn't even correct in the first place because there's a 1.54 if you scroll down on the link.
My fault for that.
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
You apparently haven't read that the ti4600 can render final fantasy in real time.
You didn't read that it was a fraction of the movie's size with a subset of the functions they performed to generate the movie.
And I've read a good bit of Carmack, et. al. - and sorry, even the next generation of cards (R300/NV30) aren't up to cinematography level effects in real time. Digital effects are done at resolutions of roughly 4000x2000 with 48-bit or 64-bit color depths. The top end effects houses do rendering at a sub-pixel level so when things are super-sampled they actually start looking realistic.
They're getting better, but have a long way to go yet.
Not to be contentious, but this is Slashdot ;-)
At 60fps each frame lasts for 16.7 ms
At 120fps each frame lasts for 8.3 ms
Human response time is on the order of 100 to 200 ms
Therefore at best (best case scenario for improvment: fast human and longest delay to screen update) The response time improvement goes from 116.7ms to 108.3 ms, a mere 7% improvement.
I'm not arguing against your point. Clearly there is a difference, however the improvement is very small. It's not like it doubles your response time like some people may mistakenly believe.
-Spyky
Having written hardware-level non-accelerated graphics programs in the past (involving implemented software based double buffering), I am well aware of all of this.
Despite quoting what I said, you seem to have missed the critical part of it: "when the game speed relies on the frames rendered"
I never said game speed had to rely on them. And in fact, you yourself give examples of where I am correct, so while you see me as needing "whackin'", I think you need your glasses checked.
Also, while I know that all recent games rely on time, I have seen some semi-recent titles in which some animations or other non-physics related portions still seem to use frames. (I recall reading that some games had 1 frame delays between bullets being fired and them hitting their targets for example)
And what masterpieces of good design are these?
-- Sometimes you have to turn the lights off in order to see.
LOL, knew you'd ask, can't remember though, sorry.
Well, I do want to try & stay away from poorly designed games ;) I mean-- if they don't bother to get a somewhat trivial issue like that right, it makes me wonder what other things they fudged.
-- Sometimes you have to turn the lights off in order to see.
I've been in lots of places (went back to re-reg the place, also got Aron's for amusement sakes)
Currently, I'm almost entirely ignored in Aussie-Land, though I thought about playing with DRicci and telling him I'm only an hour's drive away.
If you're in IRC space i'm at #abiword; learning, understanding, well, mostly, making jokes about Dom's system right now.
CC's out there, but unresponsive and Steven is poking around just to see what I know. Saturn's site is giving me trouble (pop over to roogroup to see YARenovation) I haven't even gotten my email...
Anywho, harrassing DRicci, trying to find a one seat couch, attempting to contact CC, and for an interesting side note, I was in Jail...AGAIN!
Send Cookies.
"Yeah...it was the numbers that were irrational, not the murderous cult of vegetarians...." -- Hippasus of Metapontum