The Official Graphics Card and PC gaming Thread

Worst console Ports? You mean... its even worse on PS3/Xbox 360? Wow...sounds like Rockstar may be cutting too many corners huh.

 

No, he means that it was slapped together from the xbox/ps3 version instead of being made from scratch for the pc. It is not very well optimized.

I agree that keyboards are best for most games but I had to go and buy a 360 controller this afternoon to play grid. When I installed it I tried playing with my keyboard and it did not work at all. But it worked great with the controller.

 

Thanks for that.

Keyboards are the best huh? Perhaps im doing something wrong, or am simply too use to joysticks. Because, when in a car, or even on foot, im REALLY fumbling with the keyboard LOL! I can type pretty quick, but when it comes to playing a game, im all thumbs LOL! Though I must admit, if I had the money at my disposal, I would have a steering wheel at my side when hopping in a car ROFL!

Dont worry guys. What ever I buy, you'll be the first to know LOL! I must admit, Saphire has me a bit nervous. HIS and Evga, their reviews are looking pretty strong! Unfortunately, Evga doesn't do Ati chips I really wanna try Ati, but I wanna have a look at Evga too.

 

The 8600GT can play GTA4. Just. 640x480 absolute minimum detail, 20-25fps ish. The main hit is the CPU though. Only Core 2 Duos or higher end AMDs can do it. A Socket 939 4200+ fell flat on its face with a CPU ceiling frame rate of 10-15 at absolute minimum graphic detail.
One thing's for sure though, an 8600GT does not meet the minimum requirements. The 8600GTS is what compares to those cards.

The installation for GTA4 was a bit of a pain, but as per usual, I found that owning the game is more difficult than not owning the game, if you get what I mean...
The stuttering is probably you running out of video memory, GTA4 uses a lot. The ATI HD3300 would not fare better than the 8600GT, it'd fare worse. The HD3300 is more like the 8500GT at best.
I don't know about controllers because I've only tried the 360 one, as that's my preferred PC gamepad.
Ah, Estuansis and his 9600GT fanboyism. To be fair, it was a lot more stable than the HD3870 I owned...

I prefer keybord and mouse for a lot of things, but racing games are stupid, especially GRiD, it's literally unplayable without a pad. However, it redeems itself by being very fun.
I haven't played DiRT since being disgusted at only being able to run it at 800x600 on my at the time high end system (X2 4200+, X1900XT). I will try it again later I suppose.

Agreed on GTA4 being a diabolical console port. A shame, since it's so fun, but rendering the graphics with the processor instead of the graphics card? What were they thinking?

Ultimately, if GTA4 is the first game you care about, nVidia all the way. ATI have a hard time with this game, which is evidenced by the lack of Quad CF support. Your best bet for GTA4 is probably a GTX275. You would also be wise to try and squeeze a bit more out of your Phenom, as most of the graphics come from that, not your GPU.
The game that runs GTA4 on the highest detail though is the special 2GB HD4870 singular, as it's the only 2GB GPU out there, and thus the only one that lets you max the settings out. Performance-wise though, the GTX275 will flatten it.

 

I've invented a new spreadsheet template which should make it even easier to make comparisons in the GPU sector as new games and cards are released.
What I've done is arrange the cards and resolution settings with average and minimum frame rate in a grid. The cards go along the top, the settings down the side in sequence (e.g. 1280x1024 4xAA average, 1280x1024 4xAA minimum, 1280x1024 8xAA average, 1280x1024 8xAA minimum, 1680x1050 4xAA average, 1680x1050 4xAA minimum, etc.)
I have then drawn a secondary grid which automatically multiplies the frame rate by the pixel count of the resolution to give a 'Megapixels per second' figure, where all figures in the 1280x1024 section of the chart are multiplied by 1.31072, all 1680x1050s by 1.764, and so on.

At the base of the chart I have produced two totals rows, one which sums all the figures vertically, a 'total score', and another which does the same, but weighting the average frame rates at 1x and the minimums at 2x, since minimum frame rate is more important.
Now the template is done, this is a stuningly easy way to statistically analyse benchmark data.

Here are some sample results (rounded to nearest 5 MPPS) from the Bit-Tech review:

Crysis (DX10 High) (Test Base 14.85, 1.43x)
HD4870 1GB: 500
GTX260 216: 530
HD4890: 590
GTX275: 700
GTX275 (New Drivers): 720
GTX285: 740

Weighted
HD4870 1GB: 680
GTX260 216: 725
HD4890: 800
GTX275: 960
GTX275 (New Drivers): 990
GTX285: 1020

HD4870 vs GTX260: GTX260 wins, 6.3%
HD4890 vs GTX275: GTX275 wins, 19.3% / 22.9%

Crysis Very High (Test Base 10.76, 1.67x)
GTX260 216: 355
HD4870 1GB: 365
HD4890: 415
GTX275: 460
GTX275 (New Drivers): 485
GTX285: 495

Weighted
GTX260 216: 480
HD4870 1GB: 490
HD4890: 555
GTX275: 630
GTX275 (New Drivers): 660
GTX285: 680

HD4870 vs GTX260: HD4870 wins, 2.5%
HD4890 vs GTX275: GTX275 wins, 12.2% / 17.9%

Fallout 3 (Test Base 16.33, 4.67x)
GTX260 216: 1325
HD4870 1GB: 1400
GTX275: 1500
GTX275 (New Drivers): 1555
GTX285: 1620
HD4890: 1625

Weighted
GTX260 216: 1730
HD4870 1GB: 1830
GTX275: 2020
GTX275 (New Drivers): 2100
GTX285: 2160
HD4890: 2160

HD4870 vs GTX260: HD4870 wins, 5.8%
HD4890 vs GTX275: HD4890 wins, 7.6% / 3.7%


Far Cry 2 DX10 (Test Base 20.42, 2x)
GTX260+: 1165
HD4870 1GB: 1280
GTX275: 1405
HD4890: 1435
GTX275 New Drivers: 1480
GTX285: 1525

Weighted
GTX260+: 1665
HD4870 1GB: 1850
GTX275: 2005
HD4890: 2075
GTX275 New Drivers: 2120
GTX285: 2185

HD4870 vs GTX260: HD4870 wins, 10.5%
HD4890 vs GTX275: Old Drivers, HD4890 wins, 2.8%. New Drivers, GTX275 wins, 2.7%

Call of Duty: World At War (Test Base 14.56, 2.4x)
HD4870 1GB: 1080
GTX260 216: 1115
HD4890: 1240
GTX275: 1380
GTX275 New Drivers: 1455
GTX285: 1500

Weighted
HD4870: 1515
GTX260 216: 1560
HD4890: 1735
GTX275: 1945
GTX275 New Drivers: 2040
GTX285: 2120

HD4870 vs GTX260: GTX260 wins, 3.1%
HD4890 vs GTX275: GTX275 wins, 11.7% / 17.5%

S.T.A.L.K.E.R. Clear Sky High (Test Base 9.47, 0x)
GTX260 216: 370
HD4870 1GB: 385
GTX275: 475
GTX275 New Drivers: 470
HD4890: 495
GTX285: 500

Weighted
GTX260 216: 530
HD4870 1GB: 540
GTX275: 680
GTX275 New Drivers: 675
HD4890: 700
GTX285: 720

HD4870 vs GTX260: HD4870 wins, 3.0%
HD4890 vs GTX275: HD4890 wins, 3.6% / 4.5%

Race Driver GRID (Test Base 16.33, 4.67x)
GTX260 216: 1580
HD4870 1GB: 1865
GTX275: 1955
GTX275 New Drivers: 2050
GTX285: 2080
HD4890: 2165

Weighted
GTX260 216: 2290
HD4870 1GB: 2675
GTX275: 2800
GTX275 New Drivers: 2940
GTX285: 2985
HD4890: 3110

HD4870 vs GTX260: HD4870 wins, 17.4%
HD4890 vs GTX275: HD4890 wins, 10.9% / 5.7%

HD4870 vs GTX260 MODAL WINNER: HD4870, 5/7 wins
HD4870 vs GTX260 SUM WINNER: HD4870, 39.2% vs. 9.4%
WINNER: HD4870

HD4890 vs GTX275 MODAL WINNER: TIE, 3.5 wins each
HD4890 vs GTX275 SUM WINNER: GTX275, 53.5% vs. 20.8%
WINNER: GTX275

The Test Base numbers indicate the total pixel count of the resolutions tested, so the scores can be roughly compared on a 'how demanding is this game compared to this game' basis. The multiplier figure is the average level of AA used across the tests.

 

I was playing it(GTA IV) last night at 1680 Res. I even overclocked it a little. But when the card rapidly hit 65C, I was like, OK! The actual gameplay ran OK. In my limited understanding of FPs, Hz, etc. It was the video playback that suffered.

According to the packaging im hitting the recommended limit. The Requirement(minimum) was a mere 7900 256mb nvidia, or x1900 ati! So...were they trying to make it more appealing to the public. LYING?

Nvidia huh. Figures. Evga's GTX275 is out of stock!!! Thats my luck

By the way, my CPU didnt seem to be trying that hard. But I ask again. Is there a monitor that can show during game play, a little tiny number as to CPU GPU temperatures?

 

Why would you want the crummy EVGA version? XFX, Palit and BFG are much better.
Also, 65ºC is ice cold for a graphics card. The GTX275, HD4870 and HD4870X2 run at 90ºC without complaint.
What's happening in your situation is your CPU is rendering most of the graphics, as that's how GTA4 works. It's mainly draw distance that the graphics card working on.

 

The evga version has the step up program and a lifetime warranty. I have not had any problems with my evga gtx 260. You should try one of their cards before bashing them any more.

 

I would, but with as many reports of failures as there have been with EVGA, I fail to see why I should risk it.

 

I mentioned Evga, because they have some of the strongest, if not the strongest reviews for GPU's. Obviously, since Ive never bought them, Thats all I can say about them.

Where in riva tuner is the setting for Background monitoring.

 

I just checked newegg and evga has the top 7 best rated cards in the gtx 200 series. They also have a lifetime warranty so there is not real risk. Although it would be annoying to have to rma a card and be without for a while.

http://www.newegg.com/Product/Produ...E&N=2010380048 106792634&bop=And&Order=RATING

 

I found the background monitoring setting

 

They appear to be out of the 275 Evga. Which is out of my price range anyway. The 260 however is looking mighty tempting. Shows a similar pricing to the 4870! Would I be right in assuming, that their performance/capabilities are VERY similar?
http://www.newegg.com/Product/Produ...mpareItemList=N82E16814130370,N82E16814130475

Left 4 Dead looks less demanding! VERY interesting. I noticed that at Walmart last night LOL! I could be interested in that one as well!
http://www.left4dead411.com/left-4-dead-system-requirements

 

The 192 core is as cheap, but slower than the HD4870. The 216 core is the HD4870's equal, well, almost equal. As for EVGA, perhaps they've improved a lot since their rather dire Geforce 8 & 9 days.

 

So you would say that nvidia is overpriced? Would the 260 not impress a lowly gamer like me LOL?

 

Anything would compared to the 8600, but for the same price you can get an HD4870 1GB, which is noticeably better:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814125256

 

I could swear you just said that the Nvidia chip would fair better with GTA IV (MORE OR LESS). At this point though, I just want a card, Ati or Nvidia, that will play most games close to effortlessly. And since 4870 is all the talk of the web...I guess I have my answer huh? Neweggs got several 4890's now. So that should mean that the 4870's should start dropping a little. Probably not much though

 

In GTA4 it probably would. In most other games, however, the situation is different. The tradeoff is better performance in all games, versus better performance in GTA4.
The HD4870 price drop came shortly before the HD4890 was released, they were never anywhere near as cheap as this.

 

Here's a question thats been concerning me.

How are ati's drivers? I heard both you and others complaining about the 9.2 I think it was...Which has me rather nervous!

 

Grand Theft Auto Maximum Playable (Texture, Render, View Distance, Detail Distance, Vehicle Desity, Shadow Density)

2560x1600
4870X2: Hi,Hi,35,50,45,5 19-32fps
GTX280: Hi,Hi,21,50,50,8 21-35fps
GTX260 216: Med,Hi,30,45,40,5 23-33fps
GTX260 192: Med,Hi,30,45,40,5 18-31fps
4870 1GB: Lo,Lo,1,1,1,0 12-27fps
HD4870: Lo,Lo,1,1,1,0 12-28fps
HD4850: Lo,Lo,1,1,1,0 9-22fps

1920x1200
4870X2: Hi,Hi,50,50,50,6 22-32fps
GTX280: Hi,Hi,30,71,65,12 20-33fps
GTX260 216 Hi,Hi,20,60,60,5 23-34fps
GTX260 192: Hi,Hi,20,60,50,5 22-34fps
4870 1GB: Med,Hi,22,40,45,5 21-31fps
HD4870: Lo,Hi,20,45,45,8 21-35fps
HD4850: Lo,Lo,15,20,20,1 16-31fps

1680x1050
4870X2: Hi,Hi,40,50,46,8 18-29fps
GTX280: Hi,Hi,40,78,80,11 15-29fps
GTX260 216: Hi,Hi,20,60,60,5 24-35fps
4870 1GB: Hi,Hi,45,45,50,5 22-33fps
GTX260 192: Hi,Hi,20,50,50,5 23-34fps
HD4870: Med,Hi,35,45,45,8 24-37fps
HD4850: Med,Hi,25,40,25,5 23-36fps


You will note that even at 1680x1050, no card can max the settings out. The GTX280 does relatively well, and the GTX275 and GTX285 would fare better still, but not by much. It's down to SLi setups to try and max the game (Since the HD4870X2 has scaled and two HD4890s would be unlikely to beat the GTX280. There is no Quad crossfire support for the game) However, to do so would require hacking the game to remove the memory restrictions. Numerous benchmarks have proven that the memory usage calculations in the options menu are nearly double the amount of memory the game actually uses. Other than maximum detail at 2560x1600, video memory would not be an issue for 1GB cards, and the same for 896MB cards at 1920x1200.

ATI's drivers were poor between 8.11 and 9.2 but 9.3 is a much better driver. It still has bugged PowerPlay for dual GPU systems but apart from that, fares well. The only truly unusable drivers were 9.1 and 9.2 due to game CTDs (9.1) and VPU recovers (9.2)