The Official Graphics Card and PC gaming Thread

GTX285 is about 25% better than a GTX260-216. The HD5830 is about 30% better than an HD4870, so in a fair fight, the HD5830 is marginally superior. I don't like the thought of a GPU with a TDP as enormous as the GTX285 in a microATX system (204W is a big jump from the 150W of a 4870, and indeed the 175W of an HD5830), remember that overheating nvidia cards shortens their lifespan far more than it does ATI cards. Also consider the GTX285 doesn't have DX11 or any of that, and since you're upgrading, you may as well do it properly. If upgrading means choosing a geforce, no matter, but don't buy into outdated, inefficient old tech, especially when there's a relatively minimal increase in cost to get something new.

 

I'd be tempted for this badboy. It's getting very good reviews. Apparently for one review/user, he saturated the Windows Experience Index. At least most of the scores And several people can max all their games. Sounds like a winner to me. Low heat, low power. Even if Nvidias best is ~5% quicker, this one isn't far behind, and very nice where Wattage and heat are an issue.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814150476

 

Those XFX cards are meant to be pretty quiet. Only complaints against them really are that XFX don't offer any kind of support, which is true of essentially all partner companies, Sapphire included.

 

Did you uninstall the old nvidia drivers if any?

 

what psu do you have?

 

I didn't have any video drivers. I did a clean install of windows and then used the disk provided with the card. It seemed to install alrite.

I have a coolerMaster silent pro 600W.

 

The 5770 isn't exactly a pig when it comes to watts. Surely that 600W CM can handle it. Though I wouldn't go with a CM from what I've heard

 

The SilentPro units are thankfully much better than the stuff CM used to put out, and as you say, the 5770 is quite a forgiving card power-wise.

 

What would you recommend. I have since asked for a refund foe the PSU however I wont kno if its the PSU until about a week tho.

 

Either of the following I would not hesitate to buy
http://www.newegg.com/Product/Produ...-004^17-139-004-S01,17-139-003^17-139-003-S01

 

Totally with you there Omega. Would not hesitate for a second.

 

Any of them will do, the 400W is adequate for the system.

 

I wasn't positive. Thanks for clearing that up. Myself, I don't like taxing a PSU. I don't even like putting more than 60% load on it. That just comes too close to the full potential, where components begin to sweat

 

An HD5770 with a 95W CPU is actually only about 45-50% load on a 400W PSU, so no worries
Edit: It's a 130W CPU isn't it, OK then 55-60%

 

I keep underestimating the 400W beauty There really is nothing to worry about

 

Remember I ran an X1900XT (at least as much power as an HD5770, probably a little more), an E4300 and nine hard drives off an EarthWatts 380W unit without incident, which is the same as the 400W CX.

 

Wowie Zowie! Good to know. That's what I like to hear LOL!

 

MafiaII (nvidia bias: 25% for DX10 cards, 35% for DX11 cards).
Frame rate limit enforced on ATI hardware via PhysX crippling if APEX mode set to anything but off (biasing up to 320%)
1.655 1.607
non-PhysX biases considered in comparison. AA levels not included in testing

Minimal: Radeon HD2900 Pro/HD3850/HD4670/HD5570 or above, Geforce 8800GS/9600 series/GT220 or above
Reduced: Radeon HD4700 series/HD5750 or above, Geforce 8800GT/9800GT/GT240 or above
Moderate: Radeon HD5830 or above, Geforce GTX260-216 or above
Good: Radeon HD5870 or above, Geforce GTX460 1GB or above
Optimal: Radeon HD5970 or above (80%), Geforce GTX275 SLI (93%)/GTX285 SLI (80%)/GTX460 SLI (75%)/GTX465 SLI (69%)/GTX460 1GB SLI (59%)/GTX470 SLI (35%)
Extreme: 3xHD5870 or above (97%), Geforce GTX480 SLI or above (96%)

One of the more biased titles I've seen recently, with cards like the HD5870 (and therefore the HD5970) easily getting handed their asses by cards like the 8800GTS with the settings at max. PhysX crippling at work.
With PhysX disabled, bias is less extreme but still significant, with HD5770 vs GTS250 and HD4890 vs GTX260-216 being winning fights for nvidia, despite being in the wrong classes. With the current generation things get worse, with the HD5870 on level pegging with the GTX460 1GB, and the HD5850 gaining no ground over the 768MB version.
For anything up to 1680x1050, assuming no AA is used, a single card will handle MAfia 2, from either side, but you'll be paying 50% more for an ATI to manage it.
At 1920x1200 things get tricky, 80% scaling in crossfire is unlikely for a game with such an nvidia-bias, if it ever gains a profile at all. A pair of two of last gen's geforces will take on this resolution if scaling is good, and £300 worth of DX11 Geforces will have no trouble at all, as long as SLI is being used. Even if the HD5970 can pull off the unlikely 80%, that's a 65% markup.
At 2560x1600 where results are always extrapolated, it's effectively an nvidia-only game. Even four HD5870s would be unlikely to meet the 73% scaling required to pull extreme off (and this is before AA, assuming the game has it), that's with £1200 of hardware, and it will likely fall short. Meanwhile, you're likely to get pretty close with two GTX480s, as 96% in a heavily-biased game is not unusual, and this is, with OEM cards at least, only £750, so again, ATI are looking at a 60% markup, and far worse chances of success.

Enabling APEX of course completely defeats any ATI card. Even with a 45nm Core 2 Quad at over 4Ghz, mid-quality APEX limits you to a minimum of 23fps and an average of 30, meaning that at 1280x1024 everything HD4870 and above is on equals, and at 1920x1080 all the HD5800s are equal (results comparable to a GTX260-216 and GTX465 respectively, though the 465 is a little faster and the 260 a little slower in these cases)
With max APEX (which gameGPU state has little visual effect, surprise surprise) this is limited to a minimum of 9fps and an average of 16, which at 1280x1024 places everything from an HD3850 upwards on equals with a 9800GT. At 1920x1080 this means everything from the HD4700 series onwards is equal to a GTS250.

One other test mentioned is the use of an additional card for PhysX. Presumably they were told off for testing this with ATI cards as primary as it's illegal in nvidia law, so the benefits of using a Radeon with a secondary geforce as the APEX card aren't immediately obvious (though due to the primary performance of ATIs here, it's not likely to be that impressive)
The primary card was a GTX480. At 1920x1080, the addition of a card as weak as the 9800GTX severely reduced frame rates - this is an interesting effect of tandemming PhysX processors, you need to get the balancing act right it seems. When a GTX260 was added, the minimum frame rate was relatively unchanged but the average rose. All the figures slowly rose with increasing performance up to the GTX460, where minimum fps had risen from 31 to 47, and average from 50 to 95, at 1920x1080. This would seem to be the limit, for this resolution, as the GTX470 results were identical.
Ultimately then, to get total performance beyond 1920x1200, it takes three cards, and they have to all be geforces. You'd need two for SLI to get the fps up, and a third card to get APEX to work well. The best you can do with any number of Radeons is likely to be a laggy showing at 1920x1200, and that with APEX completely disabled. The game may as well come with an nvidia only sticker.

Starcraft II (nvidia bias: -5-5%)

Minimal: Radeon X850XT/X1800 series/HD2900 series/HD3670/HD4650/HD5570 or above, Geforce 7800 series/8800 series/9600 series/GT220 or above
Reduced: Radeon X1900XT/HD2900 Pro/HD3850/HD4670/HD5570 or above, Geforce 8800GS/9600GSO G92/GT220 or above
Moderate: Radeon HD3870/HD4700 series/HD5750 or above, Geforce 8800GT/9800GT/GT240 or above
Good: Radeon HD4860/HD5770 or above, Geforce GTX260-192 or above
Optimal: Radeon HD5850 or above, Geforce GTX460 1GB/GTX465 or above
Extreme: 2xRadeon HD5870 or above (96%), Geforce GTX470 SLI (90%)/GTX480 SLI (72%) or above

Relatively neutral game here, ATI have a slight lead at low res, nvidia a slight lead at high res (though GameGPU have artifically inflated the metric for the GTX480 for some reason, so the 480 scores from these tests are a little overblown, treat them more as a fully overclocked 480 rather than a stock one)
DX9 cards still being evident in the first two categories highlights the game isn't inherently hugely demanding, though things do steepen up quickly, with DX11 cards being required for 1920x1200 with 4xAA.
ATI cards fare slightly less well at high res than nvidias, correcting the earlier slight imbalance. As a result two 5870s may find the game slightly more difficult to max at continuous 60fps than a pair of GTX470s. They certainly would over a pair of 480s, but that's hardly surprising. The recent price cuts to the GTX470 place it as, albeit a horrendously efficient, but a reasonable rival to the HD5870. It's a lot cheaper, and it's not hugely slower.

Arma II: Operation Arrowhead (nvidia bias: 20% for DX10, -7-5% for DX11) (AA not included)

Minimal: Radeon HD3870/HD4700 series/HD5750 or above, Geforce 8800GS/9600GSO G92/GT240 or above
Reduced: Radeon HD4870/HD5770 or above, Geforce 9800GTX+/GTS250 or above
Moderate: Radeon HD5870 or above, Geforce GTX470 or above
Good: 2xRadeon HD5830 (90%)/1xHD5970 (60%) or above, 2xGeforce GTX460/GTX280 (100%)/GTX465 (90%)/GTX460 1GB/GTX275(82%)/GTX285(74%)/GTX470(54%) or above
Optimal: 3xRadeon HD5850 (157%)/HD5870 (125%) or above, 2xGeforce GTX480 (100%)/3xGTX280 (200%)/GTX285 (157%)/GTX470 (145%) or above
Extreme: N/A

Curiously this game is only pro-nvidia with older DX10 cards, and this is not due to the HD5870 being better than usual versus the HD4870, so the problem here stems from nvidia's DX11 products.
Two Asus Ares cards overclocked to 1Ghz per-core, or four GTX480s overclocked would not be sufficient to earn the extreme preset for Operation Arrowhead. Ouch.
1920x1200 is looking at nearly £700 of ATI hardware (and only if scaling is good, else £900). If SLI scaling is absolutely perfect (doubtful) nvidia could do it for £750. Unlikely though, it's more likely to take a triple-slot board and £800, or an old DX10 Triple-SLI setup may do, though that of course would have been £1100+ in its day for the cards alone.
Even at 1680x1050, ATI cards will run you nearly £350 if scaling is excellent, otherwise in excess of £400. High-end DX10 SLI pairs will do from nvidia, or for newer cards, £360 of GTX460s, £300 if you get away with the 768MB versions.
In order to max the game on less than £300 of graphics hardware and using a single card, you'd have to go down to 1280x1024. Arma II belongs in the chart of needlessly demanding games.

 

I have a slimline S3420f and i need a new graphics card. But i don't know if there is any out there i can get. I've heard that the power supply pretty much makes upgrading hard. I don't know how many watts it is though. I'm using a nvidia 6150 SE . I want to get starcraft 2, but can only play with like 10 fps on lowest possible setting

 

Starcraft 2 isn't a horrendously demanding game. You won't be able to max it out, but a Radeon HD5670 should be enough to see you play it with a reasonably fluid frame rate. The HD5670 is the most powerful card you can use in a standard HP system without needing to change the PSU. However, you can only get HD5450s for standard width PCs. If your slimline S3420f is an actual slimline case, you may as well forget it, you'll need a new PC to play games like Starcraft 2.

To establish whether your case is slimline or not, check the width of the expansion slots in the back panel at the bottom.

This is slimline


This is regular size