* OcUK Crysis DX9 TopList *

zytok · 18 Jan 2008 at 12:42

gurusan said:
pedros that cant be right

no it can't.... Should be getting around 40ish. Could he have the config files mixed up? Sure I read somewhere that you can change the values in say the high config to very high values. Not saying Pedros has done anything on purpose but that benchmark is way off.

Oh and re Vista, I can personally vouch for 20%+ losses in Vista. Get about 10fps less in Vista at same settings, even more so if I use DX10. For a game that was supposedly designed for Vista......... :rolleyes:

Scougar · 18 Jan 2008 at 12:42

Well I'm getting 32-33fps average with my 3870 clocked at 805/2600 and E4300 @ 2.66Ghz, on a 4X pci-e slot on XP Pro.

I gotta say I'm feeling seriously shortchanged right now (ONLY with Crysis) considering how fast some peeps systems are running with ONLY a single 8800GT. Double the speed, that is just flippin' crazy.

Oh bizarrely, If I use the 'very high' SYSTEM.CFG file in the crysis directory, I actually gain 0.75 - 1fps in the test! Something very strange is going on here.

Matthew

zytok · 18 Jan 2008 at 12:46

Seriously, you need to discount Pedros' result. (No offence Pedros). Realistically, 45-47 is about the max on a single GT regardless of CPU.

Pedros · 18 Jan 2008 at 12:48

Maybe those that have similar specs are running at 16x PCI-e not 4x

I don't know how i'm getting these results to be honest.

zytok · 18 Jan 2008 at 12:55

Pedros said:
Maybe those that have similar specs are running at 16x PCI-e not 4x

I don't know how i'm getting these results to be honest.

any chance of you posting a screengrab not sourced from two separate images?? Judging from your screenie, the benchie and the gpu-z are from different source images

Be nice to see a cpu-z alongside as well just for reference. Maybe some bread wallpaper might be handy.

Scougar · 18 Jan 2008 at 13:02

Are you using customised (altered) cvar files (config files) ?

If I use the custom 'very high' system.cfg file, it tells me and asks me if I want to continue or not.

[There is also a result in here with a 2900pro getting much higher even though its only clocked at 600mhz.]

Matthew

zytok · 18 Jan 2008 at 13:08

Scougar said:
Are you using customised (altered) cvar files (config files) ?

If I use the custom 'very high' system.cfg file, it tells me and asks me if I want to continue or not.

[There is also a result in here with a 2900pro getting much higher even though its only clocked at 600mhz.]

Matthew

Sorry Matthew, are you asking me? :confused:

if so, are you talking about the benchmark or in game? Shouldn't be using custom cfg's for the benchie otherwise it's pointless. My 45 ish was based on the benchie and people's results on here. 60+ is pretty much impossible without either error or cheating.

Scougar · 18 Jan 2008 at 13:16

No not you Zytok

I was referring to Pedros, that he may inadvertantly tried out some custom configs (like willhub uses and claims it is better than 'very high'), and forgotten to remove the config change before he ran the tests.

You know, the "use this config/cvar hack and crysis will run soo much better" style sites/links/comments.

Matthew

melbourne720 · 18 Jan 2008 at 13:18

gurusan said:
^^ ehh? From the scores it looks like CF gives 50-65% boost

I'm not sure tbh. The difference between single and crossfire setups is hard to gauge, given the low number of ATi single card setups but it seems there are bigger differences between crossfire setups than there are between crossfire and single cards.

e.g.

VortexA1, 3870 crossfire, 43.825 fps
gurusan, 3850 Pro single card, 40.835

gurusan has a 500MHz greater clock on his/her CPU, but otherwise they have similar rigs (both quad, both Vista 64 bit). However, if we add another user to the equation:-

marscay, 3870 crossfire, 58.75 fps (Quad at 3800MHz)

So it could be CPU bound. This is potentially reinforced by the simonmaltby result:-

simonmaltby, 3870 crossfire, 63.705 fps (Quad at 4500MHz)

But who knows what the effect of the higher clockspeed could be on a single card rig. If one of the higher crossfire people could disable crossfire and then run the bench again it would be interesting.

zytok said:
wouldn't you have gotten more performance by just oc'ing the GT a bit more? I got a better score than you with a slower cpu. sure it's more GPU limited than CPU.

You're right, plus I dropped the PCIE to 115 to avoid data corruption.

gurusan · 18 Jan 2008 at 13:53

melbourne720 said:
I'm not sure tbh. The difference between single and crossfire setups is hard to gauge, given the low number of ATi single card setups but it seems there are bigger differences between crossfire setups than there are between crossfire and single cards.

e.g.

VortexA1, 3870 crossfire, 43.825 fps
gurusan, 3850 Pro single card, 40.835

gurusan has a 500MHz greater clock on his/her CPU, but otherwise they have similar rigs (both quad, both Vista 64 bit).

Ahh but you missed the part where my 3850pro was clocked to 1ghz core and 2200mem

Also I'm thinking that VortexA1's score should be higher....not sure what is holding it back

Scougar · 18 Jan 2008 at 14:14

Gurusan > Why haven't you benched you put up your results with 3850 crossfire ?

Matthew

melbourne720 · 18 Jan 2008 at 14:17

gurusan said:
Ahh but you missed the part where my 3850pro was clocked to 1ghz core and 2200mem

Good point, I missed that. It is hard to find a comparison of single card vs crossfire, as no 2 rigs are a like.

gurusan said:
Also I'm thinking that VortexA1's score should be higher....not sure what is holding it back

Motherboard or CPU maybe? Excellent thread btw

VortexA1's image says crossfire disabled...?

Pedros · 18 Jan 2008 at 15:00

Ok, A complete screen grab rather than Alt+Print Screen which captures just the front window. I know it sounds WAY too good to be true and I'm puzzled myself. The only thing I've changed is updated the drivers to xxx.xx.28 Beta's and added a 80mm, 120mm fans.

I currently don't have an altered config, did have at one point but took it off straight after as I'd rather have real Directx 10 Graphics from Vista.

There's no other way I can explain these weird happenings, I know I should DEFINATLY NOT be beating most of the PC's posted but somehow I am.

gurusan · 18 Jan 2008 at 15:07

Scougar said:
Gurusan > Why haven't you benched you put up your results with 3850 crossfire ?

Matthew

Ehh, sorry what? Crossfire?

Pedros said:
**snip**

Only thing I can think of is that for some reason the app isn't applying the settings properly. Try going in-game and changing them to global high manually...and set the res to 1280x1024 then run the bench

Cyanide · 18 Jan 2008 at 15:21

melbourne720 said:
So it could be CPU bound.

It is, my system gets a pitifully low score with Crossfire enabled, though it's still higher than a single card. I've got a feeling that it's down to the 1mb cache on my cpu which'll be getting upgraded when the new quads come out and drop in price

Scougar · 18 Jan 2008 at 15:29

gurusan said:
Ehh, sorry what? Crossfire?

Slightly confused, I was sure you had 2 3850's? I was sure i'd seen you post up 60FPS in Crysis with a suicide run (as you called it)???

Cyanide said:
It is, my system gets a pitifully low score with Crossfire enabled, though it's still higher than a single card. I've got a feeling that it's down to the 1mb cache on my cpu which'll be getting upgraded when the new quads come out and drop in price

A lot of peeps seem to have issues with setting crossfire up correctly. Also, the best way to run crossfire (AT THE MOMENT!) is to use vista 64 with dx9 I believe. That gives the best multicard performance. Also, please make sure you have all the multicard updates for vista

If a single card is scoring higher than 2, then something is DEFINATELY not setup correctly, and it is not your 1mb cache.

Matthew

Cyanide · 18 Jan 2008 at 16:33

Scougar said:
A lot of peeps seem to have issues with setting crossfire up correctly. Also, the best way to run crossfire (AT THE MOMENT!) is to use vista 64 with dx9 I believe. That gives the best multicard performance. Also, please make sure you have all the multicard updates for vista If a single card is scoring higher than 2, then something is DEFINATELY not setup correctly, and it is not your 1mb cache.

Matthew

Crossfire is scoring more than single card, but not by much. I'm running vista 64 and benchmarking with dx9. Got the 1.1 patch, all the hotfixes that I know of and the latest Catalyst drivers with AI set to standard. Crossfire's enabled and everythings connected as it should be. In 3dmark if I overclock the cpu more the overall score increases by 2-3 times the cpu increase which to me suggests that the cpu is limiting the performance of my crossfire setup.

gurusan · 18 Jan 2008 at 17:37

Scougar said:
Slightly confused, I was sure you had 2 3850's? I was sure i'd seen you post up 60FPS in Crysis with a suicide run (as you called it)???

Nope you must have me confused with someone else...I've always run just a single card.

dr. sarf · 18 Jan 2008 at 19:11

Pedros said:
Ok, A complete screen grab rather than Alt+Print Screen which captures just the front window. I know it sounds WAY too good to be true and I'm puzzled myself. The only thing I've changed is updated the drivers to xxx.xx.28 Beta's and added a 80mm, 120mm fans.

I currently don't have an altered config, did have at one point but took it off straight after as I'd rather have real Directx 10 Graphics from Vista.

There's no other way I can explain these weird happenings, I know I should DEFINATLY NOT be beating most of the PC's posted but somehow I am.

I have just about the same system as oyu, here's my results, gonna try the Beta drivers to see if it makes any difference. Those results are very hard to believe, and I'm running a stripped down version of XP.

dr. sarf · 18 Jan 2008 at 19:40

My imageshack seems to be down atm, but just ran the same benchmark with 169.28 BETA - 45.505 FPS.

Competitor rules

*** OcUK Crysis DX9 TopList ***

* OcUK Crysis DX9 TopList *