Ti4600 vs 9700 Pro: who's the image quality king in FS

[Noel 4,504]

I usually run at 1792 x 1344 x 32. Which card would look best in FS? I'm guessing they will perform similarly. And when does NV30 debut?Noel

[Guest flightpro08]

The 9700 will probably look better, but that's if you can get it working. I've heard quite a few stories of users having trouble with the 9700 and FS2k2. While some people have got it to work great, I wouldn't take the risk. Get yourself a Ti4600 and you'll love it.The NV30 will most likely be out in small quantities before the holidays with a larger release sometime in the 1st quarter of next year. Ryan-Flightpro08 :-cool VATSIM Pilot/ControllerZLA ARTCC Senior Controller (C-3)SAN TRACON Lead [link:www.taxiwaysigns.com|Taxiwaysigns.com] Scenery Designer-----------------------------My "Home Made" System Specs:Intel Pentium 4 2.2GHz ProcessorTurbo Gamer ATX Mid-Tower with 420W Power SupplyEPoX 4G4A Motherboard with Intel 845G ChipsetVisiontek XTASY GeForce4 128MB Ti4600 (Det 30.30 Drivers)512MB PC2100 DDR RAM40GB Matrox 7200RPM Hard DriveWindows XP Home Edition SP1*No CPU or GPU Overclocking*3dMark2001SE Score: 11298-----------------------------Click [link:ftp.avsim.com/library/esearch.php?DLID=&Name=&FileName=&Author=Ryan+Fretwell&CatID=Root]Here to Download my New American Eagle POSKY CRJ-200!

[Guest JonP01]

Yes I certainly seem to have spent an inordinate amount of time over the last few weeks reading about user's woes with the 9700 Pro. I don't quite recall a time where I have read about so many problems with a new piece of hardware. That is not to say the majority of purchasers have not got their cards working without difficulties because I am sure they have. But the posting rate in the various forums regarding issues with this card is higher than I have ever seen. Perhaps if you insist on a 1.3 revision 9700 Pro card you might stand a chance of getting it to work flawlessly. Personally I would sit tight for another few months and re-assess the market then.But let's face it Noel - you are the type of person who wants the very best it is possible to get. Seeing as you have been so careful up to now in picking the right bits and pieces, a few more months isn't going to hurt (too much). In a few months time, I very much doubt the current Radeon 9700 will be the most coveted gamer's graphics device - and that is not to suggest that the NV30 neccessarily will be either. But it is definitely worth the wait to see if the NV30 does live up to the hype. Personally, the only caveat I have regarding the NV30 being the best card by end of year is that it does not have any significant bugs or compatability issues. I think the mind blowing performance of NV30 is, however, something we can take as a given right now.

[JimmiG 244]

NV30 is expected in December in limited quantities, and it will be widely available early next year.If you have a good videocard now, and don't absolutely need to play all your games at max detail with full FSAA and AF, wait for the NV30. Even if it is actually slower than the R9700, at least the price of the 9700 will have dropped by then.. Also, chances are that by the time the NV30 is released, ATI will have something new as well, maybe a faster version of the 9700 with faster memory and higher clockspeeds.If you really need to upgrade now, the Radeon9700 is a much better choice than the GF4. The advantages it offers today are better performance and higher image quality. It also supports DX9 (GF4 is only a DX8 card) and a number of advanced shader features not available on the GF4, meaning that it will be able to run all the next-generation games comming out next year, with full detail and effects.I wouldn't worry too much about the drivers. ATI drivers require some knowledge to install and tend to be a bit more sensitive to user-error. Make sure you uninstall any old videocard drivers. When updating ATI drivers, always make sure you uninstall the old ones before you install the new updated driver. Also, the card is power hungry, so if you get a 9700, make sure you have a good power supply and a quality mainboard.

[Guest]

There were benchmarks out of Tawain recently on either Tomshardware.com or Anandtech.com that showed the NV30 having twice the performance of the 9700. No kidding. It was a small article presented from another publication on either of those two websites. One of the things that they noted was the possibility of FSAAX4 @ 1600x1200 with no loss of framerate performance but it was using something like QuakeIII. I'll try and look for it and post it.

[JimmiG 244]

Sorry, but those are just wild rumours made up by someone looking for attention "Hey look at me! I've got NV30 benchmarks!!"Looking at the leaked hardware specs of the NV30, it will most likely be slightly faster than the Radeon9700, but not even close to twice as fast. I doubt anyone has even got a working NV30. There may be a couple of very early and semi-working boards inside the Nvidia headquarters, but those are for testing the chip for bugs and trying things, they would not give them away to some hardware site in Taiwan.

[Guest]

Image quality. The never ending question.Image quality can best be defined as what ? That is the question.1. No jagged lines;2. Sharp textures on the ground;3. True 32 bit colour;And all of that on a good monitor (it counts too) at the highestresolution and fps possible.This is my personal opinion on image quality in FS. What the 9700 does on that matter is worth mentioning: it doubles, triples the performance when you use antialiasing and anisotropic (depending on your setting of course). And antialiasing reduces jagged lines and anisotropic sharpens textures. So, if you can't go any higher than 16x12x32, the Radeon9700 is definitely the best choice for image quality (as per the abv definition). Without those features (antia or aniso) I doubt there is any discernable difference under fs2002. I don't have either card.I have a GF3 200 which I run at 2048x1536x32 (without AA or aniso).Pierre

[Guest]

I understand your perspective, and my gut reaction would have been the same as yours if I had seen a post like mine. I usually check out both Anandtech.com and Tomshardware.com at least twice a day for hardware updates. Both of those sites are the most highly regarded sites on the web in terms of computer hareware. I don't think they would accept "wild rumors" from "someone looking for attention." Anyway, here is the link I obtained from Anandech.com. There is also one available on Tomshardware.com.Again, overall the NV30 appears twice as fast as the ATL 9700Pro. It is three times as fast as the current 4600ti. http://www.pcpop.com.cn/view.asp?type=news&id=7722Have a good day,Robbie

[JimmiG 244]

First, that page is very old. It was put out at about the same time that the NV30 first taped out several months ago. Back then, there were not even any functional boards out, so it was completely impossible for them to get any board to run benchmarks on, not even Nvidia had working boards. There simply did not exist a working and functional NV30 chip at the time that those "benchmark results" were put out.It was later found out that the results were not from actual benchmark results. Rather, those scores were based off of the compiled (and false) rumors that the site ReactorCritical gathered some time before that. For example, those rumours included DDR-II memory on a 256-Bit bus, giving the card a bandwidth of over 30GB/Sec. The latest rumours now are a 128-bit bus and DDR-II, giving the card about 16-18GB/s of memory bandwidth. I'd like to know where on Anandtech you found that link. In the main news section? or in the Community forum? HardOCP is pretty much the only big hardware site that would throw up a link to something like this on the main site.Also, just by looking at the results, you can see that they are not realistic. There's no CPU out today that would be able to score 27000 3DMarks, no matter how fast the videocard.Same with their Quake3 results. Quake3 relies on the CPU for much of the geometry calculations. 600FPS is not possible on current CPUs. If there truly was a card that performed three times better than the Radeon9700, you would still not get much higher results than with a 9700, because the card would mostly sit and wait for the CPU.

[Guest]

The article I referred to was post two months ago. How do you know there were no functional boards out? If you want to play the rumor-mill card, I have heard rumors that several functional boards have been out for quite awhile. However, due to oversupply of current Nvidia chips, it would not have been wise to release the NV30 because to do so would mean lowering the value of the current Nvidia chips ever further. But rumors are rumors, so I will stick to information obtained from legitimate sites:I see no direct connection between the benchmarks I presented and information from ReactorCritical, rumor, or otherwise. Please post the link referring to the benchmarks I presented and any direct source of information you have discounting them. Thanks,Robbie

[Guest flightpro08]

>Looking at the leaked hardware specs of the NV30, it will >most likely be slightly faster than the Radeon9700, but not >even close to twice as fast.Sorry, but that's just not correct. The NV30 might not be twice as fast as the 9700, but it will be far more that "just slightly faster." Obviosuly, we won't know until it's released, but my money is on it being far better than any GPU we've seen. Ryan-Flightpro08 :-cool VATSIM Pilot/ControllerZLA ARTCC Senior Controller (C-3)SAN TRACON Lead [link:www.taxiwaysigns.com|Taxiwaysigns.com] Scenery Designer-----------------------------My "Home Made" System Specs:Intel Pentium 4 2.2GHz ProcessorTurbo Gamer ATX Mid-Tower with 420W Power SupplyEPoX 4G4A Motherboard with Intel 845G ChipsetVisiontek XTASY GeForce4 128MB Ti4600 (Det 30.30 Drivers)512MB PC2100 DDR RAM40GB Matrox 7200RPM Hard DriveWindows XP Home Edition SP1*No CPU or GPU Overclocking*3dMark2001SE Score: 11298-----------------------------Click [link:ftp.avsim.com/library/esearch.php?DLID=&Name=&FileName=&Author=Ryan+Fretwell&CatID=Root]Here to Download my New American Eagle POSKY CRJ-200!

[JimmiG 244]

Two months is a long time. The article was posted on the 20th of Aug.Just because Anandtech linked to it does not mean it's legit.See this article, posted one week earlier:http://www.anandtech.com/video/showdoc.html?i=1678&p=1It sais the NV30 had just only taped out one week earlier, which would mean the 7th. Do you seriously think there would be working NV30 boards just two weeks later? If there were, NV30 would have been released in September. There were no working NV30 boards on the 20th of august.Also, see the final words section of the Anandtech article. The NV30 will be SIMILAR to the R300, with faster clockspeeds and some more features basically. Not twice as fast. It will be slightly faster, and offer more features, which is what you'd expect since it's comming out later. But if you think the NV30 will be twice as fast as the Radeon9700, you will be very dissapointed.

[JimmiG 244]

It's funny how everyone thinks the NV30 will be some kind of magic chip and be much faster than everything else...just because Nvidia's chips have always been faster in the past the have to be the fastest now too, right?What everyone forgets is that the design of the NV30 started almost two years ago. Originally, David Kirk said that the NV30 would perform 2.5-3x faster than the GF4, with about double the vertex throughput. That, is a card that would be about as powerful as the Radeon9700. Any major change in the design would cause a 6-9 months delay. The Radeon9700 design started at about the same time, which means that the NV30 is DELAYED.NVidia is about .5 generations behind right now, As Carmack has stated.The chip will have about 120m transistors. The R9700 has around 110 milllion. There's only so much you can do with an additional 10m transistors. Just adding a second TMU to each pipe would easily use up all of those 10m transistors. Say they went with a 128-bit memory bus, that would give them room for the more advanced shaders that the NV30 is rumoured to have.Let's take a look at what we've got:R300: 8x1 pipes/tmu's, 19+GB/s memory bandwidth, fully DX9 compliant pixel and vertex shadersNV30: 8x2 architecture (recently speculation about it being a 4pipes x 4 TMU's has been going on), ~16GB/s memory bandwidth, slightly more advanced pixel and vertex shaders.To all those who think the NV30 will be some sort of miracle-chip, all I can say is, wait and see.

[Guest]

Again, you are making leaps in this argument based on your beliefs rather than hard data. The article you posted was composed PRIOR to the benchmarks I posted, so I suppose direct evidence, suggesting that the subsequent NV30 benchmarks are not legitimate, does not exist.Based on you line of reasoning, the AMD Clawhammer should have been released on time this Fall since Anantech and Tomshareware had articles and pictures of a that chip up and running in February of 2002. However, that chip won't be released until after March of 2003, over a year later. That is an example of a piece of hardware that was up and running months before the NV30 and will not be released months after the NV30. But, rational debate won't win out on this sort of argument, only hard data will - because this kind of discussion could go on and on. I have presented hard data, and you have presented nothing but supposition.The Anantech article and discussion was made PRIOR to the NV30 benchmarks is simply and educated guess. Obviously, the NV30 benchmarks posted AFTER the article you referenced do not support either your or their supposition. Talk and rumor are just that, talk and rumor Benchmarks are benchmarks and are the litmus test of the computing hardware world.I'll state it again in clear language:Please find one piece of DIRECT evidence suggesting that the NV30 benchmarks posted on Anandtech.com are not legitimate. Thanks,Robbie

[JimmiG 244]

http://www.anandtech.com/video/showdoc.html?i=1711&p=3about tapeout:"After the verification process is complete and the layout is done then comes the elusive tape-out. This is the preparation of everything necessary to be sent out to the foundry for actual production of the chip. About 4 weeks later you get your first chip or A0 silicon as it is often referred to as. "It takes FOUR WEEKS from tapeout, until you have a chip that possibly would work. Your benchmarks were posted only two weeks after the chip taped out.Also in that article, dated September 23rd, 2002, they were not shown a working NV30. Only one runing under simulation.ALSO, the benchmarks are not posted ON Anandtech. You may have found the link there, but the benchmarks were posted on PCPop.