Prey Demo Performance Tested

[Ratchet]

I've just published a quick, short article where I benchmarked a bunch of different graphics cards using the recently released Prey demo.

I tested both ATI and NVIDIA cards over five resolutions including 2048x1536 and with and without anti-aliasing and anisotopic filtering. All told there are 16 charts representing the performance results gathered from more than 1,000 benchmark runs.

You can check it out here.

[cvearl]

My guess is the framerate lead with nVidia is because this is an OpenGL engine?

Thanks for the full monty Ratchet!

And ya. The DEMO completely gauranteed my purchase! Looks like a winner!

[Ratchet]

I would think that the lead for NVIDIA has something to do with it being based on the Doom 3 engine, yes. Note, however, that it has an ATI logo on the spalsh screen and is not an NVIDIA TWIMTBP game.

[bigtabs]

Elite Bastards claim that at settings close to what I game at - 1280x1024 4xAA 8xAF (I would choose 16xAF as in your review), that the ATI 1900XT is faster than the 7900GTX

I wonder if the 6.6 drivers you used are slowing it down at all, as they are using the 6.5s and the same 91.31 for nVidia.

Another alternative is that there is a bigger hit when enabling 16xAF in the ATI cards than their nVidia counterparts. This is certainly reflected in your benchmarks, where the high end nVidia parts lose only about 1 FPS going from 0xAF to 16xAF in this game. It is still odd though that EB would have us believe that ATI perform better with 8XAF. One would think that it would scale.

Well, if nothing else, it goes to show the lack of bias with the reviews round here. Well written and laid out once again Ratchet.

[cvearl]

Originally Posted by Ratchet

I would think that the lead for NVIDIA has something to do with it being based on the Doom 3 engine, yes. Note, however, that it has an ATI logo on the spalsh screen and is not an NVIDIA TWIMTBP game.

I noticed the DOOM3 ish look to a few of the items. But really this game is the most innovative FPS I have seen since HL and HL2. DOOM3 is not even close to this game. From the demo I would have to say fwe are. Not now. And not when it was released. It was tired before it went gold IMO. I am very much looking forward to this game.

The look of the game is incredible! So is the feel. The style and the sound.

C.

[cvearl]

Originally Posted by Ratchet

I would think that the lead for NVIDIA has something to do with it being based on the Doom 3 engine, yes. Note, however, that it has an ATI logo on the spalsh screen and is not an NVIDIA TWIMTBP game.

Sorry one more question if I may...

On my 7800GT it is crisp and amazing. The colors and the textures are awesome! I have not tried it yet on my X800XL rig as I have not had access to that box in a few days.

How's it look on the ATI hardware? As good as on the nVidia?

C.

[CR2500]

It seems that most reviewers put in 7900GTX/OC/TOP/MAX/XXX, when no one uses the OC'd Radeons like the Sapphire Toxic (675/1600) or HIS IceQ3 (700/1700)
Overclocked Cards do make a difference.

[snookums]

im actually pretty impressed by the performance here... it was only a year ago when the x1800xt could barely pull out playable frames from the D3 engine. This is also a demo, and demo's rarely show the true potential of the final game's performance.

Also, the x1600 stacks up well against the 7600... I wonder how the x1800gto does?

Rachet, did you use HQAF onm the ati cards and TAA/AAA for the AA options?

[-n7-]

Great review.

[Emilee]

What driver settings were used?

Another top notch review. Really like your layout. Do wish there was some sort of conclusion though.

[Ratchet]

Originally Posted by bigtabs

Elite Bastards claim that at settings close to what I game at - 1280x1024 4xAA 8xAF (I would choose 16xAF as in your review), that the ATI 1900XT is faster than the 7900GTX

I wonder if the 6.6 drivers you used are slowing it down at all, as they are using the 6.5s and the same 91.31 for nVidia.

Another alternative is that there is a bigger hit when enabling 16xAF in the ATI cards than their nVidia counterparts. This is certainly reflected in your benchmarks, where the high end nVidia parts lose only about 1 FPS going from 0xAF to 16xAF in this game. It is still odd though that EB would have us believe that ATI perform better with 8XAF. One would think that it would scale.

Well, if nothing else, it goes to show the lack of bias with the reviews round here. Well written and laid out once again Ratchet.

Andy's (Hanners at EB) Prey benchmarks were taken using Fraps which doesn't disable the forced 60fps lock (that's why all his benchmarks are capped at 60 fps). So all his higher end cards there are basically tied at 60 fps, there are no performance "winners". For our benchmarks we figured out how to run timedemos that automatically disable the fps cap, so our results are "untethered" and the graphics cards are free to render as fast as they can.

I did check out the 6.5s in Prey before moving on to the 6.6s (I still have some results, I can maybe make a few charts if you want), and there are some definite performance improvements, not only in Crossfire, but in single cards as well.

[Ratchet]

Originally Posted by cvearl

I noticed the DOOM3 ish look to a few of the items. But really this game is the most innovative FPS I have seen since HL and HL2. DOOM3 is not even close to this game. From the demo I would have to say fwe are. Not now. And not when it was released. It was tired before it went gold IMO. I am very much looking forward to this game.

The look of the game is incredible! So is the feel. The style and the sound.

C.

I agree, it's going to be an awesome game. It is still based on the Doom 3 rendering engine though, which NVIDIA cards have notoriously good performance in.

[Ratchet]

Originally Posted by cvearl

Sorry one more question if I may...

On my 7800GT it is crisp and amazing. The colors and the textures are awesome! I have not tried it yet on my X800XL rig as I have not had access to that box in a few days.

How's it look on the ATI hardware? As good as on the nVidia?

C.

Honestly I couldn't really see any difference between ATI or NVIDIA. Both look amazing.

I'm sure that if I took some screenshots and zoomed in there would be slight differences, but from past experiences with Doom3 image quality tests I would expect them to be very very subtle.

[Ratchet]

Originally Posted by CR2500

It seems that most reviewers put in 7900GTX/OC/TOP/MAX/XXX, when no one uses the OC'd Radeons like the Sapphire Toxic (675/1600) or HIS IceQ3 (700/1700)
Overclocked Cards do make a difference.

I agree, and I would have definitely used those cards if I had access to them.

[Ratchet]

Originally Posted by snookums

im actually pretty impressed by the performance here... it was only a year ago when the x1800xt could barely pull out playable frames from the D3 engine. This is also a demo, and demo's rarely show the true potential of the final game's performance.

Also, the x1600 stacks up well against the 7600... I wonder how the x1800gto does?

Rachet, did you use HQAF onm the ati cards and TAA/AAA for the AA options?

I was pleasently surprised by the X1600 as well. Unfortunately, the cards you see are pretty much the only SM3 cards I have access to (besides NVIDIA's 6-series.. I had to draw the line somewhere!).

I used the default driver settings for both cards (the only exception was to force off vsync). No AAA/TAA or HQAF was used.

[Ratchet]

Originally Posted by Emilee

What driver settings were used?

Another top notch review. Really like your layout. Do wish there was some sort of conclusion though.

Default driver settings, as mentioned above.

I purposly avoided making a conclusion or providing any real commentary. I just wanted to provide the numbers. When the full game appears I will hopefully get some time to go really in-depth.

[Emilee]

Thanks for the response. Is that all the way up on quality for the mipmap for ATi, and Quality for NV then? I havent installed a NV driver in a few months, but if memory serves, its set to Quality, and not High Quality in the drivers. I thought ATi used to be the same way, but since coming back to them, the mipmap is maxed out on the default settings.

[Ratchet]

yep that's right. NVIDIA's default is Quality, one tick below High Quality. ATI defaults to the highest setting.

[Emilee]

Thanks for the reply. I understand that keeping stock settings in drivers is the norm for benchmarking, but do you think its "fair"? I have doubts NV would have kept its lead with the drivers set to HQ, as they take a pretty large hit when doing so.

[AluminumHaste]

Originally Posted by Ratchet

yep that's right. NVIDIA's default is Quality, one tick below High Quality. ATI defaults to the highest setting.

Ummm, and why is that?? In all fairness, shouldn't both card's settings be set at highest possible?? Especially Quality wise??

[a12ctic]

in all honestly, the average user cant tell the difference between quality and highest quality, so why default it to highest quality if 98% of the people cant tell a difference?

[SIrPauly]

It would be nice to see some Super-AA Vs. SLi-AA comparisons since you're using Sli and CrossFire. It's not like performance with these modes are so low that they're not a viable option for the end-user. One X1900 is benching around 74 FPS at 1920 x 1200 -- so x8 and x10 would bench around 74 FPS with CrossFire.

Also, check out both and see how well the mixed modes handle specular aliasing - which this title has a lot of it.

[Ratchet]

Sorry, I've got that machine took apart already and have moved on to my next article, so I won't be adding anything more to this particular article.

It was pretty easy article to do though (only took a couple days), so when the full game comes out I will try and put aside some time to benchmark the cards again, using the default and the highest non-default quality settings, and do some IQ comprisons as well. I should also have a few more graphics cards (X1900GTs at least, and I'm hoping to get a couple 7900GTs as well) along with a faster test system and a SLIx16 NVIDIA board. My timedemos will also be longer and more accurate then as well.

[hobs0n]

Wasnt prey that game that have been developed like 5-7 years and they stopped developing a few years ago? Anyway, downloading the demo, spanks for the article Ratchet

[a12ctic]

Originally Posted by hobs0n

Wasnt prey that game that have been developed like 5-7 years and they stopped developing a few years ago? Anyway, downloading the demo, spanks for the article Ratchet

yes, if not longer, i think they started development in late 1996 and planned to release in 97 or 98

[ShaidarHaran]

You finally upgrading your system, Ratchet?

[Ratchet]

yes, a few months ago. NCIX has been supporting me with processors and other equipment in exchange for ads (hence the NCIX ads). Right now (as of yesterday), my benchmark machine base is an FX-62 and 2GB SuperTalent DDR2 @ 800MHz 4-4-4-11 with a currently NDAed AM2 board on the ATI side and a as yet undetermined AM2 nforce board on the NV side.

[ShaidarHaran]

Kickass

[AluminumHaste]

Originally Posted by a12ctic

in all honestly, the average user cant tell the difference between quality and highest quality, so why default it to highest quality if 98% of the people cant tell a difference?

Because comparison benchmarks should ALWAYS be done apples to apples, to half an apple to apple.

EDIT: Also, if you're going to leave one card with just "Quality" settings, then the other card should be changed to just "Quality" setting.

I ALWAYS run my card at highest possible quality with playable framerates. I've done everything I can to turn off "optimizations" and all the "tricks" that get +5fps but look like crap (I always notice artifacts).

And here I was all prepared to give NVidia the thumbs up for doing so well in the benchmarks, until Ratchet pointed out that NVidia was a step below ATI in quality settings.

[Ratchet]

I also said they both looked the same to me. What would you trust, some obscure slider that can't possibly be a representation of like-for-like image quality, or your own eyes?