Sign in with
Sign up | Sign in
Your question

4870X2+ 4870 1GB Tri-Fire Benchmarks! Now with Quad Core results!

Last response: in Graphics & Displays
Share
a b U Graphics card
February 16, 2009 9:14:19 PM

EDIT:



Scroll down for Quad Core results...................








Ok guys I added a 48701GB to my 4870X2 rig, I was a little anxious to see the results, there are not many Tri-fire rigs around, and even less benchmarks on them.

I tested alot of games, most post increases in fps, some, which are either cpu limited or multi-gpu unsupported show no increase.

I am specifically comparing the X2 vs. Tri-fire, to see what kind of increases you could look forward to by adding another gpu.



Here is the test configuration:

Asus P5E Bios ver. 704

Asus 4870X2, + Sapphire 4870 1GB Catalyst 9.1

4GB Mushkin DDR2 800 Ram @ 844Mhz 5-6-6-13 timings

E8400 @ 3.9 Ghz, Zalman 9700 HSF

X-Fi Fatality Pro Series Audio

Zalman 850W Power Supply

Lite-On Blu Ray Drive

WD Raptor 150 HDD

Cosmos S Case ( 7X 120 mm fans)





For my testing I used 1920X1200 Res, 4X AA in every title, and 16AF as well.
All settings are at max unless otherwise specified.




Far Cry 2: (Ranch Small 3 loops)

X2: min: 43 --- X3: min: 43
X2: avg: 61 --- X3: avg: 62
X2: max: 95 --- X3: max: 98

Very small increases here, possible CPU starved ? Maybe an update to a Quad core would unleash some better results.


Crysis: (Ambush 3 loops)

X2: min:1 --- X3: min: 19
X2: avg:31 --- X3: avg: 33
X2: max:40 --- X3: max: 55

Not bad increases across the board, the gains in minimum fps are substantial.

Lost Planet DX9 benchmark

X2: Snow avg: 74 ---- X3: Snow avg: 86
X2: Cave avg: 63 ---- X3: Cave avg: 66

I am honestly kind of surprised to see increases in this game, I know it's kind of a weird benchmark, but it shows improvemnets none-the-less.


World in Conflict benchmark

X2: min: 20 ---- X3: min: 22
X2: avg: 46 ---- X3: avg: 48
X2: Max: 100 ---- X3: max: 118

This is another title that is CPU dependent, but there are still increases to be seen.


Counter Strike Source Video Stress Test

X2: avg: 276 ----- X3: avg: 292

What can I say ? Who doesn't want 292 frames a second while playing Counterstrike? = P


Call of Duty World at War ( 2 minutes Fraps recorded )

X2: min: 26 ----- X3: min: 28
X2: avg: 70 ----- X3: avg: 82
X2: max: 93 ----- X3: max: 93

Absolutely fantastic results here, for the record the X3 setup felt so much more smoother than the X2, and seemed to dip down much less frequently.


Medieval Total War 2 ( 2 minutes Fraps recorded )

X2: min: 12 ----- X3: min: 13
X2: avg: 26 ----- X3: avg: 26
X2: max: 82 ----- X3: max: 78

I really didn't expect increases in this game, and I didn't get any.


Stalker Clear Sky Benchmark ( DX10.1 used)

X2: Day: min: 18 ----- X3: Day: min: 18
X2: Night: min: 13 ----- X3: Night: min: 17
X2: Rain: min: 18 ---- X3: Rain: min: 15
X2 Sun Shafts min: 14 ------ X3: Sun Shafts: min: 17

X2: Day: avg: 34 ----- X3: Day: avg: 47
X2: Night: avg: 31 ------ X3: Night: avg: 44
X2: Rain: avg: 34 ------ X3: Rain: avg: 47
X2: Sun Shafts avg: 24 ------ X3: Sun Shafts: avg: 33

X2: Day: max: 61 ------ X3: Day: max: 95
X2: Night: max: 56 ------ X3: Night: max: 87
X2: Rain: max:58 ------- X3: Rain: max: 90
X2: Sun Shafts max: 39 ------ X3: Sun Shafts: max: 60



All I can say is 'wow', take a look at the max frame increases, they are unbelievable, and help bring the avg. up across the board.
The Stalker benchmark is an excellent tool, is free to download, and does not require the game to use.



Conclusion: I don't specifically recommend a Tri-Fire setup over another multi-gpu compilation, but there is no doubt that there is alot of power in this rig.
I would like to say that installation of the 1GB card was a breeze, I ran into no dreaded driver issues, and only one reboot was required. It seems like ATI has finally got Tri-fire 'supported' in their Catalyst drivers. I wonder what kind of increases I would see by switching to an overclocked i7 platform etc. ?

Some games like Stalker, COD WAW and Crysis showed excellent gains, these games are largely GPU taxing and benefited most from a third GPU.

Other games like WIC and Far Cry 2, seem to restricted by my dual core processor, or are not multi-gpu optimized. It has been a treat to test this setup and I hope some people can use it to determine what kind of multi-gpu config is right for them.

I appreciate any comments or questions from anybody, a special thanks to L1QuiD for his help and encouragment regarding this build.
February 16, 2009 9:30:44 PM

Nice system. Its unfortunate that Tri and quad video card systems are pointless though.
February 16, 2009 11:40:53 PM

They aren't pointless...not if you want to crank up the AA...now quad sli might not scale well, but it does still help at high resolutions, and tri sli/Crossfire do well with 3 GPUs. Though most games only use 2, 3 helps add that extra push when the fps might dip low...

I mean look at his Crysis benchmarks, it pushed his min from 1 to 19....if thats accurate and not a typo, thats pretty impressvie.

I wouldn't say tri is pointless if running 1920x1200 + resolutions ....with tri I could crank up the AA to 16X @ 1920x1080:) 

Though I don't like multi GPUs, they do work well sometimes:)  I play @ 1680x1050, and to be honest, some games don't do well at that resolution with 1 285 GTX:) 
Related resources
Can't find your answer ? Ask !
a b U Graphics card
February 17, 2009 12:41:59 AM

Liquid is right, the point of Tri/Quad solutions is to achieve that extra 'oomph' at high resolutions, and those Crysis scores are not a typo either. All 3 runs with just the X2 yielded a minimum fps of '1' and all 3 runs with the tri-fire yielded minimums of 19.

Obviously I don't recommend this for those on a budget either.
February 17, 2009 9:42:34 AM

That system needs some quadcore action! I might go out an buy a 4870, even though my dualcore likely wont push it. Then again I can max every game except Crysis and oddly enough burnout paradise (can you say terrible port action?).
a b U Graphics card
February 17, 2009 6:01:33 PM

Anyone think I need to go to i7 for improvements ? Or would a Q9650 overclocked to say 3.8-4.0Ghz get me bid improvements as well ? It's either a 400$ upgrade, or an 800$ upgrade.
February 17, 2009 6:26:34 PM

if your going from a dual core, in my honest opinon I would go for the 800$ upgrade...since you have only to gain...I mean if you already had a quad core, then I would say no....but theres no point going for a Q9650 for 400$ when for only an extra 400$ you can get a whole new system with much higher performance gains.

Though you do play at 1920 resolution, your not that cpu dependant as you are to GPU...soooo I dunno how much cpu would help in your case for gaming:p ...and I saw the phenom 2s beat the i7s at high resolution:S


February 17, 2009 8:06:36 PM

Quote:
Wow, I sense so much jealousy in that statement.


I sense so much ignorance in your statement.

I praised him on his system and then commented on the numbers. Leave it alone or dont post.
February 17, 2009 8:19:21 PM

Annisman said:
Anyone think I need to go to i7 for improvements ? Or would a Q9650 overclocked to say 3.8-4.0Ghz get me bid improvements as well ? It's either a 400$ upgrade, or an 800$ upgrade.


As l1qu1d said, there are quite a few instances where Phenom 2 beats i7 at higher resolutions. You could take that $400 and get a 790FX board and P2 940. Or wait a few more weeks and get the P2 945 (AM3 model). Anand did a review with the AM3 X3's, and their conclusion was that DDR3 is totally and completelly worthless right now, and for a future ground up build. It had zero performance impact over DDR2 with the AM3 chips. So you wont have to worry about spending extra cash on DDR3 and a AM3 model board.
February 17, 2009 8:22:51 PM

Three 4870 strapped together to a dually ? Heck crisis is and old game and recommends quad core but nice numbers btw.
February 22, 2009 7:50:47 AM

Hey spathotan, do you know if the 945 is going to be a BE or not? If not wouldnt you be just as well off (if not better) getting the 940? (assuming you didnt get an AM3 mobo and DDR3)

Also, unrelated but I love your avatar. Cracks me up every time.
February 22, 2009 4:56:07 PM

bonanzaguy said:
Hey spathotan, do you know if the 945 is going to be a BE or not? If not wouldnt you be just as well off (if not better) getting the 940? (assuming you didnt get an AM3 mobo and DDR3)

Also, unrelated but I love your avatar. Cracks me up every time.


Yes the 945 is BE. The 910 and what not are the regular models.
a b U Graphics card
February 28, 2009 3:36:32 PM

There's just something about going AMD that seems so mainstream lol
February 28, 2009 3:43:34 PM

yup AMD rocks and you will make them happy since you have radeon cards too.
February 28, 2009 4:20:59 PM

Amd is owned when it comes to apps most of the time, but in games its been showig much better fps at higher resolutions.

I would love to go AMD, but I'm always afraid of going AMD, ever since I left the XP Series and the 5000+ BE (amazing CPu for the price). After that it was just not good enough for my standards:) 
a b U Graphics card
March 1, 2009 4:25:58 AM

yea, my first ever dual core, the 4800+ was a huge upgrade over my single core 3700, I used to rock AMD but Intel + overclock = the winning factor.
a b U Graphics card
April 3, 2009 10:55:42 PM

Update, just added a Quad Core 9650 @ 3.8Ghz, many of the benchmarks that were CPU limited have been raised significantly. Such as World in Conflict and Far Cry 2, however Crysis saw no gains, seems like it is not Quadcore optimized afterall.
April 4, 2009 3:17:46 AM

Annisman said:
Update, just added a Quad Core 9650 @ 3.8Ghz, many of the benchmarks that were CPU limited have been raised significantly. Such as World in Conflict and Far Cry 2, however Crysis saw no gains, seems like it is not Quadcore optimized afterall.

It would be great if you updated the benchie with the numbers you got on the OC'ed 9650. I'm interested to see the gains you got with it :) 
May 23, 2009 11:33:11 PM

Thanks for your report on the 4870x2 to 4870x3. Already having a 4870x2, I'm wanting to add support for a third monitor so another 4870 1GB definitely seems the way to go as it'll likely be boosting my FPS significantly whenever I'm not using SoftTH. I'm on a 3.8GHZ Q6600 so this seems the perfect CPU for the bundle.

Monitor setup is 2x20" 1600x1200 Samsungs (204B), 1x24" 1920x1200 Dell (2407WFP-HC) in the middle. Target of SoftTH is iRacing.
May 24, 2009 6:35:23 AM

yeah... what g3force said:
could u please link some more benchmarks, if u have the time :D 

a b U Graphics card
May 24, 2009 4:12:31 PM

alright when I get a second I will post updated benches with the quad core.
a b U Graphics card
May 26, 2009 2:41:43 AM

Alright, some of you want to know how adding two more cpu cores helps squeeze out some more frames from these three 48701GB gpus. Here are some updated results:


Lost Planet benchmark: Dual = 86fps snow. 66fps cave
Quad= 84 fps snow 93fps cave

Apparently the quad really comes in handy during the cave benchmark, which is seemingly cpu intensive.


World In Conflict Benchmark: Dual= Min.22 avg.48 max. 118
Quad= Min.31 avg.55 max. 117

Oh yea baby! WIC loves my new quadcore, as we can see the minimum fps go up by almost 50%!

Far Cry 2 Benchmark Ranch Small, 3 loops: Dual= Min. 43 avg.62 max. 98
Quad= Min.44 avg.65 max. 104

Here we can see a slight increase in scores, but nothing worth writing home to mom about.


Counter Strike Source Video Stress Test: Dual=292fps
Quad= 297fps

Hey, it's a gain, right ?


World At War Fraps recording: Dual= Min.28 avg.82 max.93
Quad= Min.51 avg.79 max. 94

Again, the minimum fps gets a significant boost from the quad core.


Crysis Warhead Benchmark: Althought I have heard people say that Crysis benefits from a Quad Core, my fps count stays exactly the same when I run this benchmark, and so there is nothing to show.

Stalker ClearSky Benchmark: Stalker also shows no use of a quadcore, though this is not a surprise because A. The game is VERY GPU limited, and B. the Stalker engine only takes full use of ONE cpu core, no surpises here.


And, an updated pic of my setup.


May 26, 2009 9:55:01 AM

ooh... nice increase for 'playability' at waw... the miinimum is what u want improved... nice idea... so quads actually do increase gaming.... haha

a c 86 U Graphics card
May 26, 2009 10:45:53 AM

Are you still using those 9.1 drivers? or have you updated them to something more recent? :D 

and those memory heatsinks look so grazy! 0_o
May 26, 2009 1:47:23 PM

Nice benchmark
Thanks for sharing

Cheers!
a b U Graphics card
May 26, 2009 1:50:14 PM

Nice. This means that the E8400 is still a top of the line GPU good for any gaming situation.
a b U Graphics card
May 26, 2009 6:42:17 PM

Kari said:
Are you still using those 9.1 drivers? or have you updated them to something more recent? :D 

and those memory heatsinks look so grazy! 0_o



No, I'm using 9.4 right now, as the 9.5's don't look like anything great, so I haven't switched yet.
a b U Graphics card
May 26, 2009 6:44:29 PM

micky_lund said:
ooh... nice increase for 'playability' at waw... the miinimum is what u want improved... nice idea... so quads actually do increase gaming.... haha


Yea, I would rather have the minimum go up rather than anything else, it makes games that much more playable like you said. However, you should note that a few games, like crysis and stalker do not benefit at all from a quad core, at least not for me. So my 340$ upgrade is worthless in that respect, I guess it depends on what games you play to see if it is worth getting a quad.
May 26, 2009 10:02:42 PM

true... or if u have another life... like doing 'work', such as video editing
a b U Graphics card
May 26, 2009 10:18:47 PM

micky_lund said:
true... or if u have another life... like doing 'work', such as video editing


another life ? What is this 'work' word that you speak of ?
May 27, 2009 5:45:40 AM

lols .... work is one thing over 3% of the world's population do... its kinda a hard thing to constantly do.... and very pointless for the other 97% the world do...
!