Overclock.net banner

1 - 19 of 19 Posts

·
Registered
Joined
·
15 Posts
Discussion Starter #1
Whew, that was a lot of work. I've entered all the info from the reviews listed in the first post of the 5970 review roundup thread. I ignored 1680x1050 and below, I'm simply not interested in low resolution results, and noone who is considering one of these cards should even think of buying one until they have a real monitor

Some of the data is sketchy, but I'm not here to judge, and just entered it as-is. I ignored a few games that were only tested on one or two sites, but the data for those games was very similar to what is shown. I also took liberties with things like lumping together COD4:MW/COD4:MW2 data and Crysis/Crysis Warhead data since most sites didn't test both and the results were similar.

Why, oh why.. Well, I was seeing results all over the place in various reviews. There's always someone in a thread who links to a specific review and says 'See, it's only 10% faster!' when there are other reviews showing a 40% difference in that same game with the same cards. This, I think, is a better way of doing it. You can look at the chart and see exactly what's happening with each game, and also see how each site did overall. The site-specific percentages are the overall difference among games tested at that resolution, barring the games I did not include due to small sample size. The FPS numbers are just an average of the FPS gotten in each game across all sites. Ditto for the difference percents. The number is how much faster the 5970 is compared to the GTX 295.

Big thanks to MattNo5ss for help with the spreadsheet stuff. Also thanks Earthdog for putting up the review list... Before any of you spam and say this is copy/pasted, Ratbuddy is actually me on OCF




The overclocking percents only apply to the 5970, the 725/1000 is the stock for that card, not the GTX 295. I just happened to stick those numbers under the headings for 2560x1600 results. The percent is just how much over stock that particular 5970 review sample went. The power numbers do apply to both cards, I'm just too lazy to figure out how to make things more clear, so folks will just have to read this to know what's going on

Basically you can expect roughly 22% core and 24% memory OC on these things, based on the very small sample size here.

I don't know who this Monst3r jerk is, but it isn't me, and he ripped off my spreadsheet wholesale.

edit: BTW, the original thread is at http://www.overclockers.com/forums/s...d.php?t=625408 if anyone cares to see the numbers calculated slightly differently - averaged FPS then did the %, per request of someone who also noticed the math seemed off for ET:QW.
 

·
Banned
Joined
·
8,014 Posts
Nice Nice. Good gain over the 295 but not the earth shattering gains I was expecting to see as some made it out to seem
 

·
Registered
Joined
·
478 Posts
The 300 series as said by Nvidia is worth waiting for let them launch than Ati will lower the prices
So that i can buy this beast
 

·
Registered
Joined
·
2,894 Posts
i think the total % shown by those sites is an idication of that sites bias towards one camp or the other. so anand is in the green camp and techpowerup is in the red, not a shocker really knowing those sites but this does show it to a T


one of the bests threads i have seen in a while this bro, you deserve many reps
 

·
Gaming Enthusiast
Joined
·
2,174 Posts
Good numbers here, may I ask for one more?

Does the 5970 saturate a PCI 1.1 slot? 295 doesn't but comes close, meaning I could use it on my aging motherboard without having to upgrade. Was wondering if the same was true for the 5970.
 

·
 
Joined
·
13,023 Posts
Quote:


Originally Posted by tweaker123
View Post

very nice!good work, man is the 295 a beast still

It should be, it's the 2nd most fastest card on the planet.
 

·
Registered
Joined
·
650 Posts
Quote:

Originally Posted by MrTOOSHORT View Post
It should be, it's the 2nd most fastest card on the planet.
It looks like Canadian schools are as bad as American schools
 

·
Team Red Lobbyist
Joined
·
1,595 Posts
Please change the title to your thread - it's very misleading as this has nothing to do with number crunching (besides how you come to the results)..

I thought this was going to compare BOINC/folding numbers for both cards...(which would be VERY interesting)....

I know for a fact that 5870 crushes the 295 on BOINC apps, let alone what a 5970 would do...but 295 would crush the 5800/5900 series in folding... so a chart showing both would be very cool....just a thought (or - it could be 2 line items in the existing chart....if you add that plus decoding/encoding times - it would even match the title (not really, but it would be closer))
 

·
Registered
Joined
·
1,074 Posts
Quote:

Originally Posted by kyleax1 View Post
It looks like Canadian schools are as bad as American schools
Hehehe. Good stuff.


Nice job on the list! +rep fosho!
 

·
Registered
Joined
·
2,458 Posts
Quote:

Originally Posted by Gill.. View Post
Please change the title to your thread - it's very misleading as this has nothing to do with number crunching (besides how you come to the results)..

I thought this was going to compare BOINC/folding numbers for both cards...(which would be VERY interesting)....

I know for a fact that 5870 crushes the 295 on BOINC apps, let alone what a 5970 would do...but 295 would crush the 5800/5900 series in folding... so a chart showing both would be very cool....just a thought (or - it could be 2 line items in the existing chart....if you add that plus decoding/encoding times - it would even match the title (not really, but it would be closer))
You just wait till they port the [email protected] client to OpenCL. So both ATi and nVidia clients can work on the same work units. Not only that, but I'm sure it will run much faster.
 

·
Premium Member
Joined
·
2,656 Posts
I have a very simple question: we play on large and/or high quality screens, so why are all of these benchmarks run at 2x or 4x Anti-Aliasing. I have yet to find a single review that ran tests at 16x or even more than one or two tests at 8x.

I want to know if ATI has continued their long history of poor AA support or if they finally got with the program, because I don't know about you guys, but there is a marked difference between 4x AA and 16x AA in games and I remember a lot of 4870x2 reviews being skewed because all the tests were run with low AA settings, but when you ramped them up NVIDIA took the crown (nearly triple the FPS in certain instances).

*EDIT* I don't mean this as ATI bashing, I'm merely asking for due diligence in benchmarking. I like my 16x AA and if I have to stick with NVIDIA to get it, then I want to know before I buy a card */EDIT*
 

·
Registered
Joined
·
15 Posts
Discussion Starter #19
Quote:


Originally Posted by grunion
View Post

Rightful ownership has been transferred to OCN/OCF ratbuddy.

Bump, nice to have info from the legitimate creator.

Actually I'd like to know how much of an improvement you've seen since the release drivers.

Thanks for setting things right. I can't believe someone would not just repost my work but actually pretend to be me. Weird.

If anyone is curious to see the raw data from each review, you can view the spreadsheet directly at http://spreadsheets.google.com/ccc?k...oYmItLUE&hl=en

I haven't seen any reviews since launch, but I strongly suspect when GF100 launches, there will be a round of reviews featuring the 5970 with updated drivers. I'll be incorporating those results, assuming they are directly comparable - same site, same hardware, new drivers. I'll probably be waiting until the GF100 launch to add 5870 results as well.

Until then...


edit: Oh yeah.. I should explain how the numbers were done, in case anyone doesn't feel like clicking through to the OCF thread. The % difference columns are calculated by site, then the %'s from each site are averaged. The difference columns are not calculated from the averaged FPS numbers across sites, they're calculated from the average difference across sites. Hopefully that makes sense. The FPS columns are just an average across all sites, and aren't directly comparable, they're just there to show just how fast both cards are - basically everything was playable.
 
1 - 19 of 19 Posts
Top