[Reverend]

"Popular games" = Games frequently played, as reported by surveys, which usually means great gameplay
"Games with great graphics" = games with prolly lousy gameplay but has great graphics or utilizes latest 3D tech and prolly doesn't sell too well

Which do you prefer in vid card reviews at this site?

edit/NOTE : Keyword is "General", i.e. "generally acknowledged as having good gameplay" and/or "generally acknowledged as having great graphics but has lousy gameplay"

And make sure you cast a vote instead of just replying to this thread. It will help greatly towards the purpose of this site. Yes, yes, I know that this should be more appropriate on the main page but that won't happen due to the unavailability of a poll option on the main page (as yet) as well as not knowing if the majority of you guys even visit the main page .

[Rodéric]

What about Future Popular games, I can think of DOOM III, Morrowind, Neverwinternights, UnrealII, UT2...

of course you maybe don't have access to those titles, but when I buy a product I expect it to run perfectly current games, and I want it to run fine upcoming games.

Of course I could wait for my 'game of the year' release before buying a new board but... I don't even buy a card every year, barely every two years.

[demalion]

I don't see the point of doing anything besides a spectrum of games that stress the graphics card in different ways...it isn't a game review after all. A lot of others voted otherwise (I interpreted 2 as fitting my preference)...I'd be interested if anyone would care to give their reasoning?

[Randell]

I voted for 3, because I dont just play the latest and greatest. You can go back to old games, especially with multiplayer, eg LAN gaming with friends.


Just because drivers may be optimised for the latest games doesnt mean to say that there are no issues with old games.

Testing/reviewing shouldnt just be raw performance - overall stability, compatibility and IQ need to count as well.

[Rodéric]

My point, is that I think most people buy the cards in order to play the games they like, so the popular games, so that should be those tested, but people don't plan to change the card every now and then, and such the card should be 'future proof' to some extent.

Some people 'hardcore gamers' can afford to replace their cards every now and then but they are really few, others want the best card for their money.

I think you get the idea.

[Rookie]

I d prefer a combi of both preferably but I know it's far and few between...

as for most gamers,the great gaming experienece will be paramount,more specificly,a fps game,speed will be No.1 factor,that's reasonable,we r just play game,not enjoy IQ with 2-3 fps.ofcoz,plus better IQ will much better. but if my sys cant run a game,such as quake,I ll low the IQ to persue the extreme speed.

3dfx rule still effective:fps is the king! 8)

[jb]

Good question. In an ideal world we would like to have both. But in today's age that is not really gonna happen. I guess a mix is needed. Q3 is great as its one of the best benchmarks of all times. However Counter Strike is the most popular game played (at least on-line) for the last 18 months. And believe it or not CS is very simple to benchmark. It provides min, max avg fps as well as a list of the avg FPS for each sec which can be thrown into excel to spit out a nice histogram. Problem is CS is old and seems to be gated at times by your CPU and other times by your video. Add to that lack of support for 32bit natively (yea I know you can force that into 32bit mode but its not the same) and limited res and you wind up with a average benchmark tool. So what's a reviewer to do?

[Randell]

how do you run the CS bench, is that only for the Retail version or the free download as well?

[LittlePenny]

Originally Posted by Ingenu

My point, is that I think most people buy the cards in order to play the games they like, so the popular games

I am in complete agreement with this statement.

[pcchen]

But on the other hand, people also buy cards to last for at least a while. Therefore, some "pretty game benchmarks" may provide some insights for possible future pretty games with nice gameplay.

[jb]

I was ask to write hardware reviews for Cryoeniac (http://www.cryoeniac.com/) but the sight has stopped while the owner looks for a new host. Which I dont think is gonna happen. I have written a review for the 8500 but its old now. In this review I eplained how to do this:

http://www.pc-gamers.net/jb/reviews/...0/ati8500.html

(forgive the above as it was not the final review...it has some typos and other html errors but for the most part is. I dont have the final one on-line as they did not find a host yet..so I left it for now)

Counter Strike has been the most popular On-line game played for almost 2 years now. Even though its based off an old engine and does not stress 3D cards, it still can bog down your FPS with the infamous smoke grenades. Like we said before its does not support 32bit color natively (can force it to use 32bit textures with a command line switch or registry edit) and it does not run at resolution higher than 1280 x 960. When we tried to run at the highest resolution, we keep crashing to the desktop. We used the Fever demo record by Dave B and which was used in his KyroII review. It showcases a typical Counter Strike game play on De_dust with lots of smoke grenades. To run Counter Strike in benchmark mode (or sometimes called Game Gauge mode) you have to set client variable cl_gg = 1 and then type into the console window gg demename The Game Gauge mode creates a text file in your main Valve/Serria directory call FPS.txt This file contains the Avg, Min and Max FPS as well as break down of the FPS for each second. We recorded the Average, Min and Max.

You can see from the above link what the older drivers did as far a CS. The demo that DaveB did can be found here:

http://www.pc-gamers.net/jb/reviews/ati8500/fever.zip

Sorry but I dont have a copy of the text file it creates or I would show that

Enjoy

[Randell]

Hey thanks for that, and nice review BTW. I agree about CS, when I saw Dust2 at 2xQuality SV wih 16x aniso I decided I wouldnt miss my V5 AA that much

[tobbe]

I didnt vote. I dont think a game being 'popular' or 'having good graphics' by itself makes it a good benchmark. What IS important for a benchmark is that it scales, and that you can extrapolate the results of it to other apps. That is what made Q3 a great benchmark: it scaled great with both cpu, memory, and rasterizer; since it scaled so well it could give a hint of performance for almost everything, and sharing engine with several games made extrapolation to those games possible. Saying eg "I dont care for Q3 as benchmark cos I dont play it" is a bit missing the point of a benchmark.

NB: Just using some properties of Q3 as an example

[Sharkfood]

Quite a provactive poll there Rev.

I think my vote would be obvious to anyone that knows my posts- definately a combination of both, although few and far between.

I'm more interested in how videocards perform in *games* versus how well optimized they are for the latest benchmarks that sell for $49.00 masquerading as games.

Let's face it- how many people actually play the heck out of Aquanox or DroneZ? Of what value is it to read how well a videocard runs a particular game when that particular game stands little chance of being run much at all after purchasing said videocard?

In the past, reviews featuring Giants, MDK2, Sacrifice, Max Payne, Quake3 and the like- these were all good since the audience reading the review was likely going to playing these very games once they got the hardware in their hot little hands. This was very applicable benchmarking and didn't leave much to extrapolation... which is rarely ever correct (i.e. what scores the best in 3DMark isn't necessarily going to be worth a hill of beans in Nascar2002 or AvP2).

Just my $0.02.. from the standpoint of a *gamer*

[msan_ewu]

I voted for "both" as well... I wish there was a choice for "all of the above" too.

I don't really care as I read the last page of most reviews and make a purchase decision.

I'm just an average joe who reads a lot of reviews. I think I represent the average joe pretty well too, lol.

I read the first page, the conclusion, then view any screenshots... if I have the time and I don't see any major flaws or biases I'll look through the review at the 1024x768x32 benches, + AA benches, +aniso benches, and then the new features that are being shown (smoothvision or HRAA/Lightspeed).

Looking at the current MadOnion poll I think it's safe to say that any review for "gamers" vs. what you would write for tech folks should concentrate at that res....My support for the above for that rationale from MO:

At what resolution & color depth do you play games?

Less than 800x600 [2.1%]

800x600 16bit [5.5%]

800x600 32bit [8.6%]

1024x768 16bit [12.2%]

1024x768 32bit [46.6%]

1152x864 16bit [0.6%]

1152x864 32bit [3.5%]

1280x960 16bit [0.4%]

1280x960 32bit [1.6%]

1280x1024 16bit [2.1%]

1280x1024 32bit [9.9%]

1600x1200 16bit [0.4%]

1600x1200 32bit [6.5%]

*note: Total votes: 19583

For reviews that are focused on gamers I'd write a great introduction that pulls a reader in, a comprehensive conclusion that *actually* sums up what you want to say without bias, and has screens and benches that focus at the above res (but not limited too). For the love of God.. Please (!) don't bring the history of Nvidia or ATI in during the conclusion then base your "card" or "game" conclusion on that. It undermines and sabotages your entire review of the actual hardware.

I skip a lot of what I consider to be fluff in reviews and then the author gives me the "damn man, I put so much effort into ~this other stuff~ and it isn't even appreciated." Well, yes... of course not. It's nice to have some fluff in there (like the differences between 6x4 and 16x12 for tech info) but that don't mean squat for me when I just want to buy a game or a chunk of hardware.

Flame me all you want but I think I really do represent the average joe reader I know that this really isn't the place to bring this up either because B3D doesn't cater to the average joe The above is what flew through my brain when I was deciding which choice to pic in the poll.

[LeStoffer]

Hmmmm....

I would much rather have some review of the graphics/technology in interesting games so we can learn more about the engines that's out there. The upcoming Unreal Tournament 2003 is my case in point.

Edit: I misunderstood something

[Randell]

since when was anyone talknig about game reviews? You need a combination of games in a grpahics card review for 2 reasons;

i) Latest graphical showpieces to show how current how the graphics card handles state of the art, and a hint of future performance. Take Tribes 2, it really hurt systems when it first came out to have everything turned up to the max. Now it would fall into 2. ie.

ii) Games people are actually are playing. It doesnt matter how old these games are. the point about CS is one, ie how fast AA that can handle all the alpha's and aniso can really improve the look of a game. I didnt need to upgrade my V5 to play CS on my 8500, but it sure looks better and of course Giants and Tribes 2 do as well because of features the V5 didnt have.

[ram]

My point is than one should benchmark a lot of games, not just 3 or 4. This gives a much more realistic view of the performance you can expect.

Benchmarking Counterstrike is a cool idea for FSAA / anisotropic filtering benches, because the game is widly used (almost every month in the top 10 game charts). You can find a very good timedemo http://www.3dcenter.de/downloads/half-life-counterstrike-vga.php

[ben6]

I think most everyone knows how I voted

[ben6]

I think most everyone knows where I stand

[BenSkywalker]

I voted for games with great graphics because none of the choices matched up exactly with what I was thinking.

I'd like to see the games that have popular engines being utilized for benches. Quake3 or RtCW may not be very popular any longer, but there are so many games on the market that use the Q3 engine it is still nice to see how the boards hold up. Likewise, the 3DMark/MaxPayne engine appears to be the foundation for a number of upcoming games making it a good platform to test with IMO. I don't find the graphics terribly impressive in any of those, but overall they are better then the gameplay at least. It would also be very nice to see the Unreal2 engine tested if you guys could get a hold of a copy of the benchmark. I think if you focus on the latest revision of the most popular engines being used you'll find a good balance between what people play and what looks good overall.

[Freon]

Another vote for adding CS in to the mix. RTCW would probably be a better pick than Q3 these days, too. Q3 doesn't have the outdoor areas and absolutely abusive use of alpha blended effect like RTCW.

The problem is HL's demo system still kinda bites. In my experience the first run was up to 25% slower than subsequent runs, with the output only becoming consistant after the 4th or 5th run. Even then it's a bit flaky. The playback speed relies incredibly on the record speed, which is of course not conductive to getting a good demo in the first place. Just make sure the demo is recorded with vysnc off, fps_max X, 512x384, and on a ridiculously powerful (CPU) machine. fps_max x (where X is the framerate cap) is useful so the record doesn't speed up and slow down according to the speed of the host record machine. Set it to the minimum framerate the host machine can handle, perhaps 40-60.