I'm all out of ideas for helping a pal with 4870x2 problem, anyone?

[cocainewhite]

Hi guys.

My friend is having some seriously choppy performance in Crysis and showing lower than expected scores on 3D Mark Vantage. Was wondering if you guys might be able to help

Specs

Q9400 @ 2.6Ghz (with Zalman Heatsink)
Radeon 4870x2 - 9.4 drivers
4GB Corsair Twin3xDHx
Antec Quattro 1000w psu
Western Digital 1Tb Greentech HD
Gigabyte GA-X48T-DQ6 - bios v.F5 21/07/08
Vista Ultimate 64
Antec 1200 case

Crysis performance @1680x1050 with all settings at 'Gamer' and 2x AA is choppy at all times (DX10)

3D Mark Vantage score is approx 12000

We have replaced PSU, vid card and RAM, and it made no difference

We also tried swapping out the 4870x2 with a mates less powerful Geforce - after rerunning Vantage the CPU score went from 9000 to close to 20000 - which is very strange.

Basically, I'm all out of ideas for the poor guy!

If anyone needs more info then let me know and I'll try and get it up here.

 

[apoppin]

try overclocking the CPU; you need to aim for ~3.4GHz as a minimum
.. then try other games besides Crysis: Warhead

Here are my Crysis results {all very high with 4xAA/16xAF} with rig in sig [q9550s at 4.0 GHz]
4870-X2 = 27 min. - 33.98 av. - 46 max.


15,000 in Vantage
21,000 in 3DMark 06

ignore the GeForce's inflated CPU score
- his CPU is soimply too slow at 2.4GHz to take advantage of a 4870-X2

overclock it and his performance will [probably] increase dramatically

 

[Keysplayr]

The CPU speed is definitely not the problem. If my Q6600 at stock speeds with 2GB of RAM and a GTX280 can play Crysis at the OP's settings without choppiness, the OP's rig should do it easily. Maybe it's an issue with Vista64 and drivers.

OP: Is this a fresh Vista 64 install? Or was there another vid card in there before the 4870X2?

Anyway, make sure all motherboard chipset drivers are up to date/installed, completely uninistall any Cat drivers. Reinstall Cats.
An optimum test to see if it's just your current software configuration is to take another hard drive and do a fresh install of Vista64 and all the appropriate drivers for your
mobo and vid card.

 

[apoppin]

The CPU speed is definitely not the problem.

How do you know for sure that it is not at least part of his problem ?
- it is simple for him to test without you dismissing it completely - out of hand
. . . GTX280 and X2 are quite different

if i underclock my CPU to his speeds, i get choppiness with my 4870-X2, playing Crysis at 16x10 and i am also using Vista64

 

[Keysplayr]

Originally posted by: apoppin

The CPU speed is definitely not the problem.

How do you know for sure that it is not at least part off the problem ?
- it is simple for him to test without you dismissing it completely - out of hand
. . . GTX280 and X2 are quite different

if i underclock my CPU to his speeds, i get choppiness with my X-2, playing Crysis at 16x10 and i am using Vista64

I just explained why.
It does not make sense that a slower card (my GTX280) is not choppy on a slower CPU (Q6600 at 2.4 Ghz) with 1/2 of the OP's RAM. You're suggesting that if I place this faster video card, (4870X2) in my PC, it will be choppy when the GTX280 isn't? What if I put a 4890 in my system? A GTX295? Still choppy?

You must have the same issue as the OP. I've never heard of somebody getting "worse" performance upgrading to a faster video card and all other things in the system being the same (CPU speed, RAM, OS, etc.) At worst, it will be CPU bottlenecked BUT perform NO WORSE than it did before the vid card upgrade.

unless........ There are OS/driver/Xfire/SLI issues.

Which leads me to believe it may have something to do with Vista64 and/or the Drivers "for" Vista 64.

Or are there CPU overhead issues with X2's? I don't think there is. Nor have I seen any evidence to support that.

 

[apoppin]

Originally posted by: Keysplayr

Originally posted by: apoppin

The CPU speed is definitely not the problem.

How do you know for sure that it is not at least part off the problem ?
- it is simple for him to test without you dismissing it completely - out of hand
. . . GTX280 and X2 are quite different

if i underclock my CPU to his speeds, i get choppiness with my X-2, playing Crysis at 16x10 and i am using Vista64

I just explained why.
It does not make sense that a slower card (my GTX280) is not choppy on a slower CPU (Q6600 at 2.4 Ghz) with 1/2 of the OP's RAM. You're suggesting that if I place this faster video card, (4870X2) in my PC, it will be choppy when the GTX280 isn't? What if I put a 4890 in my system? A GTX295? Still choppy?

You must have the same issue as the OP. I've never heard of somebody getting "worse" performance upgrading to a faster video card and all other things in the system being the same (CPU speed, RAM, OS, etc.) At worst, it will be CPU bottlenecked BUT perform NO WORSE than it did before the vid card upgrade.

unless........ There are OS/driver/Xfire/SLI issues.

Which leads me to believe it may have something to do with Vista64 and/or the Drivers "for" Vista 64.

Or are there CPU overhead issues with X2's? I don't think there is. Nor have I seen any evidence to support that.

You must have the same issue as the OP.

Sure .. when i underclock my CPU to his speeds
- i tried it just now at 2.60 GHz with a single 4870 - 19.95 average with 4.0 Ghz q9550s and 19.19 with 2.6 GHz underclocked q9550s


i am finishing up my benching with a single 4870 for my upcoming CrossFire-X review and i can stick in X-2 next

i don't care about your "explanations"; you can theorize all you want - i have the testbed .. i can post 4870-X2 results with an underclocked q9550s vs. q9550s at 4.0Ghz

 

[Zstream]

Apoppin just please.....

*facepalm*

 

[Keysplayr]

Originally posted by: apoppin

Originally posted by: Keysplayr

Originally posted by: apoppin

The CPU speed is definitely not the problem.

How do you know for sure that it is not at least part off the problem ?
- it is simple for him to test without you dismissing it completely - out of hand
. . . GTX280 and X2 are quite different

if i underclock my CPU to his speeds, i get choppiness with my X-2, playing Crysis at 16x10 and i am using Vista64

I just explained why.
It does not make sense that a slower card (my GTX280) is not choppy on a slower CPU (Q6600 at 2.4 Ghz) with 1/2 of the OP's RAM. You're suggesting that if I place this faster video card, (4870X2) in my PC, it will be choppy when the GTX280 isn't? What if I put a 4890 in my system? A GTX295? Still choppy?

You must have the same issue as the OP. I've never heard of somebody getting "worse" performance upgrading to a faster video card and all other things in the system being the same (CPU speed, RAM, OS, etc.) At worst, it will be CPU bottlenecked BUT perform NO WORSE than it did before the vid card upgrade.

unless........ There are OS/driver/Xfire/SLI issues.

Which leads me to believe it may have something to do with Vista64 and/or the Drivers "for" Vista 64.

Or are there CPU overhead issues with X2's? I don't think there is. Nor have I seen any evidence to support that.

You must have the same issue as the OP.

Sure .. when i underclock my CPU to his speeds
- i tried it just now at 2.60 GHz with a single 4870 - 19.95 average with 4.0 Ghz q9550s and 19.19 with 2.6 GHz underclocked q9550s


i am finishing up my benching with a single 4870 for my upcoming CrossFire-X review and i can stick in X-2 next

i don't care about your "explanations"; you can theorize all you want - i have the testbed .. i can post 4870-X2 results with an underclocked q9550s vs. q9550s at 4.0Ghz

Great! Knock yourself out. No, really...

 

[brblx]

Originally posted by: apoppin

i don't care about your "explanations"; you can theorize all you want - i have the testbed .. i can post 4870-X2 results with an underclocked q9550s vs. q9550s at 4.0Ghz

it's not a 'theory,' he's using something called 'logic.'

well, actually, yeah, it is a theory. he used the scientific process to eliminate the CPU as a variable- in science theories are essentially fact. ;P

there is practically never a situation where a more powerful graphics card, all other things the same, will cause worse 3d performance. the same is true of a cpu upgrade (thought it may not improve). of course a 4.0ghz processor is going to faster than a 2.6 of the same architechure, but that in no way proves that a 4870x2 and a 2.6ghz proc is going to be SLOWER than the same proc with a gtx280. one card doesn't see the CPU as a 'bigger' bottleneck and automatically crap itself.

 

[alyarb]

all of this is in reference to apoppin's *irrelevant* "testbed" data. The data in question is the OP's. Reproducing contrary results on a different machine neither explains nor solves his problem. Something is wrong with the OP's configuration. Can you verify both GPUs are working at the correct clock freqs and all multiGPU fixes are installed? Have you tried drivers other than the latest release? can you verify that this anomalous delta in performance between the geforce and x2 is consistent among every game?

 

[apoppin]

Originally posted by: brblx

Originally posted by: apoppin

i don't care about your "explanations"; you can theorize all you want - i have the testbed .. i can post 4870-X2 results with an underclocked q9550s vs. q9550s at 4.0Ghz

it's not a 'theory,' he's using something called 'logic.'

well, actually, yeah, it is a theory. he used the scientific process to eliminate the CPU as a variable- in science theories are essentially fact. ;P

there is practically never a situation where a more powerful graphics card, all other things the same, will cause worse 3d performance. the same is true of a cpu upgrade (thought it may not improve). of course a 4.0ghz processor is going to faster than a 2.6 of the same architechure, but that in no way proves that a 4870x2 and a 2.6ghz proc is going to be SLOWER than the same proc with a gtx280. one card doesn't see the CPU as a 'bigger' bottleneck and automatically crap itself.

Logic is something you are clearly short of in this case, at the least

REread the OP's comments in the first place and you will see Keys misunderstood

ALL the OP says is that his X2 is "choppy" in Crysis - that is all he says - so what? - it can be choppy in crysis
- THEN he says a SLOWER Nvidia GPU gets a higher Vantage CPU score
- again .. so what - UNrelated to Crysis or anything at all - and that is a known issue comparing Nvidia CPU scores to ATi CPU scores in Vantage

and Keys is coming off with HIS GTX280 - which is COMPLETELY *unrelated* to the OP's graphics; at least my testbed is similar to the OP's
:roll:

try logic yourself for a change, before criticizing others

A 4870-X2 is held back by a slow ass CPU

 

[brblx]

congrats, add another worthless post to your count.

 

[apoppin]

Originally posted by: brblx
congrats, add another worthless post to your count.

And another to yours; you have ZERO useful info to contribute here
:thumbsdown:

My suggestion to the OP remains the same - TRY overclocking your CPU and let us know what happens in Crysis
- better still try measuring the difference with FRAPS

 

[Tempered81]

the cpu could use a little more speed, thats for sure. But it's probably not the main issue plagueing his computer.

Get a cpu-z screenshot of the PCI-E speed... it's probably 1x or 4x

 

[Keysplayr]

Originally posted by: jaredpace
the cpu could use a little more speed, thats for sure. But it's probably not the main issue plagueing his computer.

Get a cpu-z screenshot of the PCI-E speed... it's probably 1x or 4x

Defo a place to start. :thumbsup:

 

[Zap]

Alright guys, take it easy with the personal stuff or there may be a visit from the video mods.

Originally posted by: brblx
but that in no way proves that a 4870x2 and a 2.6ghz proc is going to be SLOWER than the same proc with a gtx280.

Well, a single 4870 is (overall) slower than a single GTX 280, right? What if for some reason only one GPU was being used? I don't have experience with Radeon X2 cards, but I do have experience with dual GPU NVIDIA cards and multi-GPU needs to be enabled to use both of them. Also, what if the choppiness has to do with micro-stutter?

 

[Keysplayr]

Originally posted by: Zap
Alright guys, take it easy with the personal stuff or there may be a visit from the video mods.

Originally posted by: brblx
but that in no way proves that a 4870x2 and a 2.6ghz proc is going to be SLOWER than the same proc with a gtx280.

Well, a single 4870 is (overall) slower than a single GTX 280, right? What if for some reason only one GPU was being used? I don't have experience with Radeon X2 cards, but I do have experience with dual GPU NVIDIA cards and multi-GPU needs to be enabled to use both of them. Also, what if the choppiness has to do with micro-stutter?

I had thought about that also. What if Xfire is actually "enabled" in the driver, but isn't actually enabled?
Maybe we can get a frame grab. OP do you have FRAPS for Vista(64)?

 

[brblx]

to apoppin's credit, i went benchmark diving and it DOES appear that the 4870x2 is more CPU limited than the gtx280. so sorry for that, duder. but, it is still probably a software or driver-related occurence, and i don't think it's related to the OP.

but yeah, he can try kicking it up 4-500mhz and see what it does. the problem is that if it DOES improve, there's no gauge for how much it 'should' improve. even with another issue, he'll probably net a few fps more.

i'm sure the OP has tried this, but since i didn't see it mentioned- you did check all your driver-level settings, right?

 

[Tempered81]

Originally posted by: Keysplayr

Originally posted by: Zap
Alright guys, take it easy with the personal stuff or there may be a visit from the video mods.

Originally posted by: brblx
but that in no way proves that a 4870x2 and a 2.6ghz proc is going to be SLOWER than the same proc with a gtx280.

Well, a single 4870 is (overall) slower than a single GTX 280, right? What if for some reason only one GPU was being used? I don't have experience with Radeon X2 cards, but I do have experience with dual GPU NVIDIA cards and multi-GPU needs to be enabled to use both of them. Also, what if the choppiness has to do with micro-stutter?

I had thought about that also. What if Xfire is actually "enabled" in the driver, but isn't actually enabled?
Maybe we can get a frame grab. OP do you have FRAPS for Vista(64)?

Isn't there some kind of option in CCC to display a crossfire indicator on screen. You might try to enable that option to see if crossfire is working

 

[Elfear]

OP- Make sure Cat AI is turned on as that is what enables Crossfire to function.

Just to back up what Appopin said, Crossfire/SLI solutions in general need more cpu power to compensate for the increased overhead. Seems plausible to me that a dual card solution might be choppier than a single, less-powerful card with a weak cpu in certain situations.

I experimented with that a bit when I had dual 7800GTs. When I displayed the SLI utilization bar with a stock clocked cpu (I think it was a dual core Athlon at the time) it varied quite a bit and didn't get much over 50-60%. Overclocking the cpu resulted in a much higher utilization and much better performance.

 

[brblx]

...but he's not crossfiring?

i was under the impression he had a 4870x2, not two 4870's.

 

[Elfear]

Originally posted by: brblx
...but he's not crossfiring?

i was under the impression he had a 4870x2, not two 4870's.

The 4870x2 still has Crossfire overhead even if both chips are on one card.

 

[brblx]

ah. my mistake. i thought they were all hardware.

 

[Keysplayr]

Originally posted by: brblx
to apoppin's credit, i went benchmark diving and it DOES appear that the 4870x2 is more CPU limited than the gtx280. so sorry for that, duder. but, it is still probably a software or driver-related occurence, and i don't think it's related to the OP.

but yeah, he can try kicking it up 4-500mhz and see what it does. the problem is that if it DOES improve, there's no gauge for how much it 'should' improve. even with another issue, he'll probably net a few fps more.

i'm sure the OP has tried this, but since i didn't see it mentioned- you did check all your driver-level settings, right?

Yes, it is CPU limited. You put a GTX280, 4890 or 4870X2 is a 1GHz PIII system (as if that were possible, but for example purposes) The all would be accutely limited by the CPU. All three graphics cards would have the same exact ceiling of fps in a given game. None would do worse than the other, nor would they do better. You put these same 3 cards in a 4.0GHz i7 rig, and the 4870X2 would dust them both. So yes, the 4870X2 would benefit from a faster CPU. But unless Xfire isn't working as Zap mentioned, or there is a driver/OS/hardware problem, there isn't any reason a 4870X2 would ever perform slower than a GTX280/4890 in any system where the system remains the constant.

 

[apoppin]

Originally posted by: Keysplayr

Originally posted by: brblx
to apoppin's credit, i went benchmark diving and it DOES appear that the 4870x2 is more CPU limited than the gtx280. so sorry for that, duder. but, it is still probably a software or driver-related occurence, and i don't think it's related to the OP.

but yeah, he can try kicking it up 4-500mhz and see what it does. the problem is that if it DOES improve, there's no gauge for how much it 'should' improve. even with another issue, he'll probably net a few fps more.

i'm sure the OP has tried this, but since i didn't see it mentioned- you did check all your driver-level settings, right?

Yes, it is CPU limited. You put a GTX280, 4890 or 4870X2 is a 1GHz PIII system (as if that were possible, but for example purposes) The all would be accutely limited by the CPU. All three graphics cards would have the same exact ceiling of fps in a given game. None would do worse than the other, nor would they do better. You put these same 3 cards in a 4.0GHz i7 rig, and the 4870X2 would dust them both. So yes, the 4870X2 would benefit from a faster CPU. But unless Xfire isn't working as Zap mentioned, or there is a driver/OS/hardware problem, there isn't any reason a 4870X2 would ever perform slower than a GTX280/4890 in any system where the system remains the constant.

the OP has run away .. and what the hell does "choppy" mean anyway?
- he left us with NOTHING to go on .. but a discussion

ANYWAY -
q9550s at 4.0 GHz [Very high/4xAA/16xAF/16x10]:
4870-X2 = 25 min / 33.77 average/45 max
q9550s at 2.6 GHz [Very high/4xAA/16xAF/16x10]:
4870-X2 = 17 min / 27.92 average/42 max

i'd say it is a little choppy with the slower CPU down from 33.72 > 27.92 average

Also, by my testing at MAX detail, most of the work has been offloaded to the GPU
- i live this stuff now

there isn't any reason a 4870X2 would ever perform slower than a GTX280/4890 in any system where the system remains the constant.

So what? You have MISREAD the original post - what you are saying has nothing to do with anything he stated here

Crossfire is working fine .. there are no driver issues in my PC
-the OP should at least attempt an OC to check out the performance differences