290X Vs 780 in BF3 Bench results.

Gregster · 24 Oct 2013 at 16:14

So reading this forum, we seem to have a bit of a divide on which card is faster and people want game bench results for the reason of buying to play games

Anyways, this got me to thinking. BF3 lauched and was played by many. The 7970 was nailing the game till the 680 came along. The 680 was faster out the box till the drivers/GE edition 7970 came along blah blah blah...Anyways, this game has been fairly good for Nvidia and AMD with both having a good crack at it. So I thought I would go round the various sites and grab their 1080P bench results for Ultra settings and as you can probably guess, there is a bit of difference between them. Anyways, here is the results:

Anandtech GTX 780 101 fps - 290X 94.3fps - 780 wins

Guru3D GTX 780 81fps - 290X 85fps - 290X wins

Toms GTX 780 102fps - 290X 114fps - 290X wins

HWinfo GTX 780 95fps - 290X 108fps - 290X wins

Legitreviews GTX 780 102fps - 290X 105fps - 290X wins

Techpowerup GTX 780 104 fps - 290X 115fps - 290X wins

Sweclockers GTX 780 104 fps - 290X 94 fps - 780 wins

Techspot GTX 780 102.7 fps - 290X 105.2 fps - 290X wins

Vortez GTX 780 104.6 fps - 290X 113.3 fps - 290X wins

As you can see, quite differing results but so far we have 7 wins for the 290X and 2 wins for the GTX 780. Please feel free to add any other BF3 Vs results and I will update the OP.

alex_123_fra · 24 Oct 2013 at 16:20

Good threat Greg, will be nice to compare some real world performance.

Gregster · 24 Oct 2013 at 16:25

alex_123_fra said:
Good threat Greg, will be nice to compare some real world performance.

Cheers and when you look at the results, it is massive frames regardless of what GPU you are using.

The 290X at that price is nice and the 780 with the free games from the 28th is also nice. Win win for either either.

shadow_boxer · 24 Oct 2013 at 16:29

I think it's just nice to have gpu prices back to relatively normal for a top end gpu after the recent inflations. Hopefully nvidia will drop the 780 to a similar price and we can get the ball rolling with more competition.

MichaelM · 24 Oct 2013 at 16:34

As you said they are close and capable in BF3 but what people would be buying either of them for, at this point in time, is BF4 which Mantle will be having a huge influence on.
It's rather academic to debate this when BF3 could become a graveyard soon.

Gregster · 24 Oct 2013 at 16:38

MichaelM said:
As you said they are close and capable in BF3 but what people would be buying either of them for, at this point in time, is BF4 which Mantle will be having a huge influence on.
It's rather academic to debate this when BF3 could become a graveyard soon.

Any chance of some BF4 bench results with the 290X and the 290X running Mantle, as I couldn't find any?

I think that as many of us still play BF3 a lot and it has generally been the game that has been fair to both AMD and Nvidia, I considered this a worthy game for benching.

w3bbo · 24 Oct 2013 at 16:42

For reference, here's mine Gregster.

http://www.vortez.net/articles_pages/amd_radeon_r9_290x_review,16.html

For me the 290 is the winner in BF3, especially against a stock 780. An overclocked 780 however starts to get the upper hand when AA is applied despite having less memory which surprised me.

A lot will depend on where the benchmark is taken from imo. This will account for much of the inconsistency. This of course is without considering other variables like test setup, temperatures etc which will also effect the results in one form or another. For reference my benchies are run in the car park section (sniper/rpg). I make a point of running the benchmark as consistently as possible, blowing the back two cars up with a couple of grenades which then take out the camper van so lots of smoke effects, fire etc for the cards to deal with. After running the same section hundreds and hundreds of times you can pretty much get every run consistent as you know where the men are coming from, which cars blow up etc etc. The consistency can be proven if you look at the frame time analysis I also add in my reviews which shows very similar peaks and troughs throughout each 60 second run although BF3 is not the best example here as the 290X has some large fluctuations in frame times.

I've got some time to kill tonight so if you need a quick test of something give me a shout. Hope this helps

teppic · 24 Oct 2013 at 16:46

Here's techspot's benchmark:

http://www.techspot.com/review/727-radeon-r9-290x/page3.html

They have the 290X a little ahead of the 780.

Hardwarecanucks don't have BF3 results - they tend to do decent benchmarking.

Avenged7Fold · 24 Oct 2013 at 16:54

BF3 seems like one of the fairest games to compare given past performance. Good thread.

Wonder what sort of performance increase we get when drivers will update.

Gregster · 24 Oct 2013 at 16:54

Cheers Webbo/Teppic and I will add that to the OP.

I am scratching my head on this AA (I am going to say issue) on the 290X, as I would assume that the 512 bit bus would start to hammer the 384 bit bus but at high AA usage, it seems to have an advesre affect. It seems harsh to call it an issue but anomaly may be a better word but anyone have any ideas as to why?

You run the same bench as what a few of us did here many many many times on BF3. I honestly can't stand that section but if you are up for it, I wouldn't mind seeing how your 290X fairs with my Titan at clock for clock on ultra settings and 1440P.

I have my Titan modded to not boost and sit at what clock I set. So I can set 1050Mhz and will leave the ram as is. This I would like to test and if someone has a 780 that is modded, if they could bench it at 1050Mhz also, this would be sweet.

I normaly start the bench when they get out of the truck (get out on the road shouted) and finish it when you go back to the garage.

sieze2 · 24 Oct 2013 at 16:55

Great thread. BF3 / 4 performance is all I really care about at the moment !

MrMarvelous · 24 Oct 2013 at 16:55

Wonder if the 290X driver will mature and chuck out even better FPS.

~>Dg<~ · 24 Oct 2013 at 16:56

the 780 gtx prices went up on ocuk thats all ill say on this matter . that alone speaks volumes tbh.

pgi947 · 24 Oct 2013 at 16:57

w3bbo said:
For reference, here's mine Gregster.

http://www.vortez.net/articles_pages/amd_radeon_r9_290x_review,16.html

For me the 290 is the winner in BF3, especially against a stock 780. An overclocked 780 however starts to get the upper hand when AA is applied despite having less memory which surprised me.

A lot will depend on where the benchmark is taken from imo. This will account for much of the inconsistency. This of course is without considering other variables like test setup, temperatures etc which will also effect the results in one form or another. For reference my benchies are run in the car park section (sniper/rpg). I make a point of running the benchmark as consistently as possible, blowing the back two cars up with a couple of grenades which then take out the camper van so lots of smoke effects, fire etc for the cards to deal with. After running the same section hundreds and hundreds of times you can pretty much get every run consistent as you know where the men are coming from, which cars blow up etc etc. The consistency can be proven if you look at the frame time analysis I also add in my reviews which shows very similar peaks and troughs throughout each 60 second run although BF3 is not the best example here as the 290X has some large fluctuations in frame times.

I've got some time to kill tonight so if you need a quick test of something give me a shout. Hope this helps

You poor poor man, having to play sword breaker that many times I'm surprised you haven't developed an illness. Have a man hug, I (and probably greg and matt) feel your pain

flopper · 24 Oct 2013 at 17:05

Avenged7Fold said:
Wonder what sort of performance increase we get when drivers will update.

hard to say depends on how much the GCN update changed between the 7970 and 290. I would assume any anomaly currently is driver dependent.
takes 3 to 5 drivers before stability is there and then some conclusion can be drawn. you could see a 30%+ increase with driver development with 7970 due to a brand new arch and tech. if this applies here is hard to say yet.

normally a driver is created with stability first so it works.
later come optimizations for titles and speed.

LtMatt · 24 Oct 2013 at 17:10

Gregster said:
So reading this forum, we seem to have a bit of a divide on which card is faster and people want game bench results for the reason of buying to play games Anyways, this got me to thinking. BF3 lauched and was played by many. The 7970 was nailing the game till the 680 came along. The 680 was faster out the box till the drivers/GE edition 7970 came along blah blah blah...Anyways, this game has been fairly good for Nvidia and AMD with both having a good crack at it. So I thought I would go round the various sites and grab their 1080P bench results for Ultra settings and as you can probably guess, there is a bit of difference between them. Anyways, here is the results:

Anandtech GTX 780 101 fps - 290X 94.3fps - 780 wins

Guru3D GTX 780 81fps - 290X 85fps - 290X wins

Toms GTX 780 102fps - 290X 114fps - 290X wins

HWinfo GTX 780 95fps - 290X 108fps - 290X wins

Legitreviews GTX 780 113fps - 290X 105fps - 780 wins

Techpowerup GTX 780 104 fps - 290X 115fps - 290X wins

Sweclockers GTX 780 104 fps - 290X 94 fps - 780 wins

Techspot GTX 780 102.7 fps - 290X 105.2 fps - 290X wins

Vortez GTX 780 104.6 fps - 290X 113.3 fps - 290X wins

As you can see, quite differing results but so far we have 6 wins for the 290X and 3 wins for the GTX 780. Please feel free to add any other BF3 Vs results and I will update the OP.

Actually Greg the Legit reviews shows the 290X winning. You're comparing it to a pre-overclocked 780. Double check your results old chap.

The 290X wins in every single review except two. Both of those reviews (swe clockers+anand) use the same part of the game, thunder run. It is a very poor reflection of real world bf3 performance and all this bench does is over inflate Nvidia scores vs AMD ones. Its why when you look elsewhere the 290X wins consistently, even in quiet mode. The part of the bench they use is very light on the gpu, so like 1080p it always favours Nvidia cards. You look elsewhere and you see the bigger picture as most sites use more demanding parts of the game.

I've done digging before about this when i tried to understand how anand always show Nvidia cards miles ahead in this bench when every other site shows something different. Well i found my answer.

Dave Bauman - AMD comments on the anand review

Ryan @ anand replies about it

Source
http://www.anandtech.com/show/5261/amd-radeon-hd-7970-review/22

TL : DR If you want accurate representation of battlefield 3 performance probably best to look elsewhere.

[/QUOTE]

And just in case you doubted my claims ive taken results from all the reviews. Every single one aside from Thunder run will show a 290X victory vs a 780 and often a titan.

compare to Nvidia Geforce 780 3GB

~>Dg<~ · 24 Oct 2013 at 17:31

7950 boost crossfire damn that value

Gregster · 24 Oct 2013 at 17:42

I have corrected the results from Legit reviews, as you was quite right with me using the SC version and not the stock version. I went for the stock version on all of them but didn't notice that one. I will leave those other 2 results in, as they still bench parts of the game and we play parts of the game, regardless if we have a AMD GPU or an Nvidia GPU

I did that run for curiosity and at 1440P. I guess I should do another at 1080P, as this is what we game at and will do in a bit

Anyways, here is the results with my clocks locked at 1050Mhz and 3005Mhz memory. CPU was a 3930K @ 4.625 and memory was 2333Mhz.

Got the call to get out of the truck and started the bench. Cleared the car park and got back in the garage and stopped the bench.

Frames, Time (ms), Min, Max, Avg
25244, 328959, 51, 114, 76.739

w3bbo · 24 Oct 2013 at 17:46

Gregster said:
Cheers Webbo/Teppic and I will add that to the OP.

I am scratching my head on this AA (I am going to say issue) on the 290X, as I would assume that the 512 bit bus would start to hammer the 384 bit bus but at high AA usage, it seems to have an advesre affect. It seems harsh to call it an issue but anomaly may be a better word but anyone have any ideas as to why?

You run the same bench as what a few of us did here many many many times on BF3. I honestly can't stand that section but if you are up for it, I wouldn't mind seeing how your 290X fairs with my Titan at clock for clock on ultra settings and 1440P.

I have my Titan modded to not boost and sit at what clock I set. So I can set 1050Mhz and will leave the ram as is. This I would like to test and if someone has a 780 that is modded, if they could bench it at 1050Mhz also, this would be sweet.

I normaly start the bench when they get out of the truck (get out on the road shouted) and finish it when you go back to the garage.

Yeah I'm up for it. I have a Titan and 780 too although I wont be benching them again as I was benching for the last two nights getting the review finished and as pgi947 said, theres only so much sword breaker a man can take lol.

I start the clock (fraps) running from the moment you shoot the first of the two RPGs in the building, turn and run left until you are almost opposite the camper van. Lob two grenades toward the first car (if done correctly you will blow both cars and van up, if not the van will blow up with a few shots. By this time the enemy will have picked them selves up from the blast and pick them off one at a time

.

Give me a full run down of the settings you want me to run and I'll go load it up.

CPU=3960X @ 4.5ghz
16gb 1866MHz 9-10-9

LtMatt · 24 Oct 2013 at 17:47

Gregster said:
I have corrected the results from Legit reviews, as you was quite right with me using the SC version and not the stock version. I went for the stock version on all of them but didn't notice that one. I will leave those other 2 results in, as they still bench parts of the game and we play parts of the game, regardless if we have a AMD GPU or an Nvidia GPU

Furry muff mate. You know this one is a sticking point for me though, regarding that anand bench. Mentioned it a few times before and even earlier to Rusty.