Just Cause 2 demo benchmark with steam update. Post your scores.

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Apocalypse23

Golden Member
Jul 14, 2003
1,467
1
0
1920x1200, same settings as everyone else (4xAA, 16xAF, etc.)

5850 @ stock (725/1000): 53.41
5850 @ XXX speed (755/1125): 56.02 (4.9% faster than stock)
5850 @ 5870 speed (850/1200): 61.58 (15.3% faster than stock)

And while I don't game at 1920x1080, I benched it to make a direct comparison to Apocalypse's 5870:

5850 @ 5870 speed (850/1200): 64.02 (5.4% slower than Apocalypse's 5870)

This confirms my observations in several other games I've benchmarked: 5850 at 5870 speed is 15% faster than stock 5850, 5% slower than stock 5870, so you get a 5% bonus for the 160 extra shaders of the 5870. And overclocking the 5850 makes a great card even better...

By the way, I also played the demo (it gives you 30 minutes of playtime), and while it was amusing, it didn't convince me to buy it. The grappling hook has been done better in Batman, and the third-person gunplay has been done better in many other games (particularly Uncharted 2 on the PS3). Just Cause presents an interesting combo of the two, but it's not a must-have.

Here are my Asus 5850 scores at stock and OC:

5850@stock (725/1000):56.95
justcause22010042402264.png


5850@oc(1000/1175):72.15
justcause22010042402223.png

 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
Going to run quite a few scenarios and post back:

Had to break this up into two posts. 10 image limit. So this post is single GTX480 stock and o/c'd to 800/1600/2000 (stock 701/1402/1848)

GTX480 stock clocks (single) 4xAA all else maxxed but CUDA effects off
JustCause2single_noCUDA.jpg


GTX480 stock clocks (single) 4xAA all else maxxed with CUDA effects on
JustCause2single_CUDA.jpg


GTX480 o/c (single) 4xAA all else maxxed but CUDA effects off
JustCause2singleOC_4x_noCUDA.jpg


GTX480 o/c (single) 4xAA all else maxxed with CUDA effects on (COMING)
JustCause2singleOC_4x_CUDA.jpg



I'll post a screenie for each. All stock clock benches done, now have to o/c.
Post back in about 30min.

I notice most are running 1920x1080, so don't forget I'm at 1920x1200. Pushing a few more pixels.

Keys
 
Last edited:

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
SLI stock and o/c'd. o/c'd to 800/1600/2000 (stock 701/1402/1848)

GTX480SLI stock clocks 4xAA all else maxxed but CUDA effects off
JustCause2SLI_4x_noCUDA.jpg


GTX480SLI stock clocks 4xAA all else maxxed with CUDA effects on
JustCause2SLI_4x_CUDA.jpg


GTX480SLI stock clocks 32xCSAA all else maxxed but CUDA effects off
JustCause2SLI_32x_noCUDA.jpg


GTX480SLI stock clocks 32xCSAA all else maxxed with CUDA effects on
JustCause2SLI_32x_CUDA.jpg


GTX480SLI o/c 4xAA all else maxxed but CUDA effects off
JustCause2SLIoc_4x_noCUDA.jpg


GTX480SLI o/c 4xAA all else maxxed with CUDA effects on
JustCause2SLIoc_4x_CUDA.jpg


GTX480SLI o/c 32xCSAA all else maxxed but CUDA effects off
JustCause2SLIoc_32x_noCUDA.jpg


GTX480SLI o/c 32xCSAA all else maxxed with CUDA effects on (COMING)
JustCause2SLIoc_32x_CUDA.jpg
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
So a 5870 overclocked to 1000 core is 6fps behind a stock gtx 480?
Is this a Nvidea sponsored game also? Better yet, which good games are not TWIMTBP?

Good stuff guys.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
So a 5870 overclocked to 1000 core is 6fps behind a stock gtx 480?
Is this a Nvidea sponsored game also? Better yet, which good games are not TWIMTBP?

Good stuff guys.

Probably be behind a few frames more if I benched at 1080 instead of 1200.

And it is a TWIMTBP game.
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
Wow the gtx480 scores are crazy high and at a slightly higher resolution too. The overclocked 5850 score is really impressive too.

The cuda effects are kinda like going from a "high" setting to a "very high" setting. It looks better, but only a little bit.
 

Apocalypse23

Golden Member
Jul 14, 2003
1,467
1
0
So a 5870 overclocked to 1000 core is 6fps behind a stock gtx 480?
Is this a Nvidea sponsored game also? Better yet, which good games are not TWIMTBP?

Good stuff guys.

Also note that my benchmarks for the 5870 and 5850 stock and OC were conducted on a 3.9Ghz E8400 platform, I will reconduct each test on my new i7 930@4.0Ghz platform soon to compare results, the fps should be marginally higher.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Also note that my benchmarks for the 5870 and 5850 stock and OC were conducted on a 3.9Ghz E8400 platform, I will reconduct each test on my new i7 930@4.0Ghz platform soon to compare results, the fps should be marginally higher.
I dont think the average framerate will really change going with a 4.0 i7. your 3.9 E8400 probably gives right at 100% of what 5870 can do in this game.
 

Apocalypse23

Golden Member
Jul 14, 2003
1,467
1
0
An additional thing I did notice was that even with an Ocd 5850, the BFBC2 performance was worse than my ocd 5870, overall the 5850 felt sloppy in that game.
 

Termie

Diamond Member
Aug 17, 2005
7,949
48
91
www.techbuyersguru.com
Here are my Asus 5850 scores at stock and OC:

5850@stock (725/1000):56.95
justcause22010042402264.png


5850@oc(1000/1175):72.15
justcause22010042402223.png


Interesting - in comparison to your 5870 scores at 1000/1300, you're only 5% behind, which suggests that the much higher memory speed on the 5870 isn't doing anything at all, since you're only getting the 5% that the 160 extra shaders would deliver anyway. Thus these cards aren't memory limited at stock 5870 memory speeds.

Curious - what voltage do you need to get to 1000 on the 5850, and why happens with a 1200 memory clock? I've seen a lot of people leave there 5850s below 1200 on the memory.
 

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
I re-ran the Desert Sunrise benchmark at 1920x1200, 4xAA, 16xAF, everything maxed and got 100.70FPS average. This is at my everyday clocks of 900/1250.
 

Apocalypse23

Golden Member
Jul 14, 2003
1,467
1
0
Interesting - in comparison to your 5870 scores at 1000/1300, you're only 5% behind, which suggests that the much higher memory speed on the 5870 isn't doing anything at all, since you're only getting the 5% that the 160 extra shaders would deliver anyway. Thus these cards aren't memory limited at stock 5870 memory speeds.

Curious - what voltage do you need to get to 1000 on the 5850, and why happens with a 1200 memory clock? I've seen a lot of people leave there 5850s below 1200 on the memory.

My OC settings for the Asus 5850 were: 1000/1175 @ 1.287V.

5850100011751287vstable.png


I'm not really impressed by the 5850, it runs choppier in BFBC2 when compared to an Ocd 5870, so I believe the 5870's extra shaders and the nitty gritty do matter, but perhaps not so much in this game.

why happens with a 1200 memory clock? I've seen a lot of people leave there 5850s below 1200 on the memory.

Results are posted here: http://forums.anandtech.com/showpost.php?p=29732311&postcount=6
 
Last edited:

Apocalypse23

Golden Member
Jul 14, 2003
1,467
1
0
I re-ran the Desert Sunrise benchmark at 1920x1200, 4xAA, 16xAF, everything maxed and got 100.70FPS average. This is at my everyday clocks of 900/1250.

Wow, nice, have you tried overclocking your core up to 1000 and leaving your memory at stock?
 

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
Wow, nice, have you tried overclocking your core up to 1000 and leaving your memory at stock?
Thanks. I can try that later when I get home. What would be interesting numbers to see? 1000/1000, 1000/1250, and maximum on both?
 

Termie

Diamond Member
Aug 17, 2005
7,949
48
91
www.techbuyersguru.com
My OC settings for the Asus 5850 were: 1000/1175 @ 1.287V.

I'm not really impressed by the 5850, it runs choppier in BFBC2 when compared to an Ocd 5870, so I believe the 5870's extra shaders and the nitty gritty do matter, but perhaps not so much in this game.



Results are posted here: http://forums.anandtech.com/showpost.php?p=29732311&postcount=6

Interesting, thanks. I don't believe I've had instability at 1200 on the memory, but I might bump it down just to be safe, and try to get a bit more out of the core. Right now I'm at 850 core at stock voltage, though, which I like for lower heat/power. I might experiment later on. I use the card mainly for BFBC2, and I think I get a distinct advantage with it overclocked versus stock (7fps or so). I can see why you'd find that an OCd 5870 would be even better. That game is just at the margin of high-res/AA/AF playability on the new cards.
 

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
So I ran some different clocks and here are the outcomes. Note that in each run, all options are maxed.

1920x1200, 4xAA, 16xAF - 1000/1000 - 96.24
1920x1200, 4xAA, 16xAF - 1000/1250 - 103.27
2560x1600, 4xAA, 16xAF - 1000/1250 - 70.09

Pretty interesting I'd say. First, lowering the RAM frequency by 20% (1250->1000) but increasing the core frequency by 11% (900->1000) shows a net loss in FPS. So, there's some memory bandwidth limitations for sure. Raising the memory clock again still only nets a few more FPS. I thought I'd test for a CPU bottleneck, but even raising the resolution to 2560x1600 (almost double the pixels) doesn't show much gain. To summarize, an 11% core overclock got me ~3% gain at both resolutions. There's clearly some bottleneck here which clock speed really isn't addressing. It might be software (drivers, etc.) or hardware (architecture), any thoughts? Interesting nevertheless.
 

Termie

Diamond Member
Aug 17, 2005
7,949
48
91
www.techbuyersguru.com
So I ran some different clocks and here are the outcomes. Note that in each run, all options are maxed.

1920x1200, 4xAA, 16xAF - 1000/1000 - 96.24
1920x1200, 4xAA, 16xAF - 1000/1250 - 103.27
2560x1600, 4xAA, 16xAF - 1000/1250 - 70.09

Pretty interesting I'd say. First, lowering the RAM frequency by 20% (1250->1000) but increasing the core frequency by 11% (900->1000) shows a net loss in FPS. So, there's some memory bandwidth limitations for sure. Raising the memory clock again still only nets a few more FPS. I thought I'd test for a CPU bottleneck, but even raising the resolution to 2560x1600 (almost double the pixels) doesn't show much gain. To summarize, an 11% core overclock got me ~3% gain at both resolutions. There's clearly some bottleneck here which clock speed really isn't addressing. It might be software (drivers, etc.) or hardware (architecture), any thoughts? Interesting nevertheless.

I'm no expert on crossfire, but I bet your results differ from singlefire systems because with two cards, you no longer need clock speed due to the dual GPUs, which are now sufficient to max out your memory bandwidth. Thus you are more memory speed starved, while a single card is clockspeed starved. Just a guess.