Sign in with
Sign up | Sign in
Your question
Closed

So frustrating!!!!!! GTX 590 performance way too low in games..

Last response: in Graphics & Displays
Share
August 1, 2011 3:27:09 AM

I bought and installed Mafia 2, which seems to have recognized the dual - gpus in SLI (profile automatically set in nvidia control panel), and set everything to the nvidia-recommended settings (maxed w/o vsync) and ran the benchmark. The Nvidia website said that the average should be 57 fps, and reviewing websites gave even higher scores such as overclockersclub which says i should be getting 88fps, and hardwareheaven.com saying that i should be getting 109 fps, all at my resolution, with the same settings enabled (physx on, in-game settings to high)

so why the hell is it, when i run the benchmark, my average frames per second at the end of it is 47??? I wouldn't expect anything else to be a bottleneck, as i have an intel core i7 2600k overclocked to 3.8ghz, 8GB of G.Skill Ripjaws RAM, and 1000W Power Supply powering it all.

This is really frustrating, because all of my benchmark scores have been lower than they are in reviews for this video card!!!!!!!!!!!! for this game, Just cause 2, Crysis, and other games!! I would really appreciate help sorting this out ):

GPU-Z screenshot for one GPU:
GPU-Z screenshot for the other GPU:


EDIT: Mafia 2 does have a benchmark tool = that is where i got my 47 fps from.

I dedicated one GPU to Physx - this improved my Avg FPS to about 53 fps. This is closer to the official Nvidia score (that i assume they also got from the benchmark tool), 57 fps.

But it still does not explain the 80-110 fps that other reviewing websites got... !!

Also, I opened up Nvidia inspector while running it - The GPUs are not even at 100% usage!!! What could be going on with my $2600 computer... :(  plz help..
August 1, 2011 3:48:19 AM

I am a little confused. Do you have one 580 or two? You stated that it seemed to recognize the dual gpus.
a c 216 U Graphics card
August 1, 2011 4:17:13 AM

Given all the range of FPS given from different websites, I'd guess that there isn't a becnhmark program for the game, so people are just taking FPS from random locations. Depending on what zone/level you are in, you may see higher FPS.
Related resources
a b U Graphics card
August 1, 2011 6:07:57 AM

bystander said:
Given all the range of FPS given from different websites, I'd guess that there isn't a becnhmark program for the game, so people are just taking FPS from random locations. Depending on what zone/level you are in, you may see higher FPS.

Mafia 2 does have a benchmark
a c 216 U Graphics card
August 1, 2011 6:13:59 AM

FlintIronStagg said:
Mafia 2 does have a benchmark


Interesting. How did they all get so radically different results?
a b U Graphics card
August 1, 2011 6:22:06 AM

bystander said:
Interesting. How did they all get so radically different results?

That I couldn't say; maybe they aren't using the benchmark tool? In game I get a steady 55-60 fps on my 570
August 1, 2011 6:28:09 AM

I'm not sure guys... Mafia 2 does have a benchmark tool = that is where i got my 47 fps from.

I dedicated one GPU to Physx - this improved my Avg FPS to about 53 fps. This is closer to the official Nvidia score (that i assume they also got from the benchmark tool), 57 fps.

But it still does not explain the 80-110 fps that other reviewing websites got... !!!

xerxces, Nvidia GTX 590 is a dual-gpu card, essentially i am running two cards in SLI with this card, and Mafia 2 seems to recognize it.

amd655, I am running the 275.33 Drivers.

Also, I opened up Nvidia inspector while running it - The GPUs are not even at 100% usage!!! What could be going on with my $2600 computer... :( 
a c 216 U Graphics card
August 1, 2011 6:31:33 AM

It's possible that the one benchmark didn't include PhysX. However, I'm suspicious about those results. It's usually better performance when you don't dedicate one card to physX.

Are you running that benchmark from a window? Or is it fullscreen? If it's not fullscreen, it will not run in SLI, which is why you gained performance assigning one to PhysX.
August 1, 2011 6:36:09 AM

I run everything under fullscreen, monitor at 60hz. the benchmark too. The benchmark is also designed in order to display massive amounts of physx..
a c 216 U Graphics card
August 1, 2011 6:41:01 AM

Well, with your adjustment, you were within 3 FPS of the one benchmark, which is pretty good. That other benchmark may not have included physX.

How are other benchmarks and games running?
a b U Graphics card
August 1, 2011 4:20:07 PM

Ok, I just ran the benchmark, and completely maxed out at 1920x1080 i got 56.3 fps average. This was without physx on with a phenom iix4 925 @ 2.8 ghz and a GTX570.
August 1, 2011 7:49:14 PM

FlintIronStagg said:
Ok, I just ran the benchmark, and completely maxed out at 1920x1080 i got 56.3 fps average. This was without physx on with a phenom iix4 925 @ 2.8 ghz and a GTX570.



Yeah, without physx - the whole point of the benchmark is to showcase a ridiculous amount of physx to see hwo your video card can handle it.

bystander, other benchmarks are most of the time also lower than they should be. Just cause 2 for example, my benchmark scores are lower than other benchmark scores, also Crysis 2 (DX11, with the high rez textures and 1.9 patch) drops to about 30 fps when there are a lot of particle effects and stuff going on, which seems way too low for my video card (which could get it up to 100+ fps on other versions of crysis 2...)

amd655, I paid a computer store to set my video card up for me - I really have no idea how to check if it's connected properly, i mean in terms of pcie 8 pin and 6 pin and etc... I played around in NCP and didnt get far, and yeah it is all stock settings, only the processor is OCed to 3.8.

Any other ideas or help would be appreciated? :) 
a c 216 U Graphics card
August 1, 2011 7:53:45 PM

Could you list the actual scores? On the Crysis 2 benchmark, in Dx11, it will drop to 30-40 at the mins. That is normal. How much lower are you talking about on the benchmarks? You might want to try running 3dmark11.
a b U Graphics card
August 1, 2011 8:20:15 PM

By using one 580 for physx, you are only getting half the graphics performance. Perhaps if you had a 560 or something for physx you might get near 100.
a c 504 U Graphics card
August 1, 2011 8:25:14 PM

When you run CPU-Z what is the DRAM Frequency reported on the Memory tab page? Is it 665 MHz or something close to it?
a c 216 U Graphics card
August 1, 2011 8:25:28 PM

Example of benchmark results for Crysis 2. Now this is 470's in SLI, which will be slightly slower than a 590, but this is more for reference.

Avg = 48.065, min = 34.423, max = 58.685

I would not expect more than +3-5 FPS with a 590.
August 1, 2011 10:25:14 PM

Bystander - Well, for example.. In just cause 2, when i'm in a plane flying over the city and all the billboards and decals start drawing in, or when I use the cluster grenade thing that causes a lot of explosions the FPS can drop to as low as 40. Now considering i run it with shadows and objects detail set to medium with 16xAF and 8xAA And it runs at a constant 60FPS at all other times with Vsync, I don't know if that kind of a drop is even supposed to happen at all... I guess I'll run 3dMark 11 and see my score in that.

and really??? You mean Dx11 really requires that much power? sometimes though the game can drop to as low as 18 fps.. Thats what happens when i stand still and shoot at water which causes it to splash up in the character's face, or when i shoot at the floor and stare right at it - I'm guessing its the particle effects which cause this drop in performance?

Also, arent 470s really outdated? I would expect one of the GPUs in the 590 to outclass one 470 by quite a bit (no offense meant of course) If Crysis 2 has a benchmarking tool i would use it and tell you my scores..

ko888 said:
When you run CPU-Z what is the DRAM Frequency reported on the Memory tab page? Is it 665 MHz or something close to it?


ko888, the DRAM frequency it says is 665.3 MHz. This is normal?
a c 216 U Graphics card
August 1, 2011 10:31:07 PM

DTsoy said:
Bystander - Well, for example.. In just cause 2, when i'm in a plane flying over the city and all the billboards and decals start drawing in, or when I use the cluster grenade thing that causes a lot of explosions the FPS can drop to as low as 40. Now considering i run it with shadows and objects detail set to medium with 16xAF and 8xAA And it runs at a constant 60FPS at all other times with Vsync, I don't know if that kind of a drop is even supposed to happen at all... I guess I'll run 3dMark 11 and see my score in that.

and really??? You mean Dx11 really requires that much power? sometimes though the game can drop to as low as 18 fps.. Thats what happens when i stand still and shoot at water which causes it to splash up in the character's face, or when i shoot at the floor and stare right at it - I'm guessing its the particle effects which cause this drop in performance?


Dx11 Crysis 2 is very demanding with all the details up. With everything maxed, I drop into the 30's on occasion.


DTsoy said:

Also, arent 470s really outdated? I would expect one of the GPUs in the 590 to outclass one 470 by quite a bit (no offense meant of course) If Crysis 2 has a benchmarking tool i would use it and tell you my scores..


I am using two 470's in SLI, which is almost as fast as two 560 ti's in SLI, which is only a little behind a 590. The 590 will be faster, but it won't be a lot faster.

With a single gtx 470, it drops down to 19 fps min on the benchmark, with 30-34 FPS average.
a c 504 U Graphics card
August 1, 2011 10:41:39 PM

DTsoy said:
ko888, the DRAM frequency it says is 665.3 MHz. This is normal?

Yes it's normal. That's what I was expecting to see.

When you initially said "The GPUs are not even at 100% usage!!!" would seem to indicate a CPU bottleneck so I was wondering if your 8GB of G.Skill Ripjaws RAM was running at the right speed.
a c 216 U Graphics card
August 1, 2011 10:57:31 PM

I think I see the problem. It's a vary common problem. He bought a vary fast GPU, and expected it to always put out insane FPS. What he hasn't taken into account is that the CPU also needs to keep up with it and can hold it back at times. This is why you aren't seeing 100% usage all the time and also why FPS will drop into the 30's. Sometimes the games require a lot of CPU power.

The 590 will be more power than you need most the time, but when playing demanding games like Crysis 2 and Metro 2033, it will be put to the test and give you a good experience.

However, that 19 FPS dip is more than I'd expect, although these things do happen in rare circumstances, usually when files are loaded. I am assuming this rarely ever happens.

What was the benchmark results for the 590, min, ave and max, while playing Crysis 2 DX11?
August 2, 2011 12:45:44 AM

bystander said:
I think I see the problem. It's a vary common problem. He bought a vary fast GPU, and expected it to always put out insane FPS. What he hasn't taken into account is that the CPU also needs to keep up with it and can hold it back at times. This is why you aren't seeing 100% usage all the time and also why FPS will drop into the 30's. Sometimes the games require a lot of CPU power.

The 590 will be more power than you need most the time, but when playing demanding games like Crysis 2 and Metro 2033, it will be put to the test and give you a good experience.

However, that 19 FPS dip is more than I'd expect, although these things do happen in rare circumstances, usually when files are loaded. I am assuming this rarely ever happens.

What was the benchmark results for the 590, min, ave and max, while playing Crysis 2 DX11?


Well bystander.. it was the assumption that i was making when i bought my video card ): i overclocked my processor to 3.8ghz and got it a liquid cooler for the reason to not bottleneck it!

I used FRAPS to make a user made benchmark on the level Semper Fi or Die, and I got:
Minimum: 44
Max: 68
Avg: 56

While playing though, like normally (with fraps active to see fps) I get like 30 fps at times, dropping even lower on rare occasions...
Once i get back in about an hour, i am going to use FRAPS to make another user benchmark for myself.
a c 216 U Graphics card
August 2, 2011 1:55:04 AM

As you can see, I have had 6950's in my machine. I had 6950's unlocked and overclocked past 6970 stock settings. With that setup, I had nearly the same results as yours. I had a min about 4x, a max around 65, and an average of about 52, but it was at 1920x1200, which is slightly more demanding.

Those scores are what you should expect on average, the above setup would be faster than yours although we didn't run the same exact benchmark or at the same exact resolution, but your score looks about right on for a 590.
August 2, 2011 2:52:50 AM

bystander said:
As you can see, I have had 6950's in my machine. I had 6950's unlocked and overclocked past 6970 stock settings. With that setup, I had nearly the same results as yours. I had a min about 4x, a max around 65, and an average of about 52, but it was at 1920x1200, which is slightly more demanding.

Those scores are what you should expect on average, the above setup would be faster than yours although we didn't run the same exact benchmark or at the same exact resolution, but your score looks about right on for a 590.


Thanks bystander (: a few last issues though.. My resolution was 1920x1200, so are the scores still about right?

and also, I cool my i7 2600k 3.4ghz stock with a liquid cooler (kuhler H2O 920) and at 100% usage for all 4 cores, each core averages at around 50 Degrees celcius. I am thinking of overclocking it further to 4.0 ghz to prevent future (or current) bottlenecks, do you think this is a safe bet for me? or should i stick with 3.8 ghz?

thanks again very much for your help :) 
a c 216 U Graphics card
August 2, 2011 3:31:38 AM

DTsoy said:
Thanks bystander (: a few last issues though.. My resolution was 1920x1200, so are the scores still about right?

and also, I cool my i7 2600k 3.4ghz stock with a liquid cooler (kuhler H2O 920) and at 100% usage for all 4 cores, each core averages at around 50 Degrees celcius. I am thinking of overclocking it further to 4.0 ghz to prevent future (or current) bottlenecks, do you think this is a safe bet for me? or should i stick with 3.8 ghz?

thanks again very much for your help :) 


Ya, those scores look good for 1920x1200. Slightly more than my ATI setup, which is somewhat expected, as it is an Nvidia backed title.

A lot of people with the 2600k's are getting their CPU's up to 4.5ghz. It might not be needed to go that high, but you should be safe to go higher than you have it. Find a good guide to follow. 50C at load is also quite good.
August 2, 2011 6:20:27 AM

bystander said:
Ya, those scores look good for 1920x1200. Slightly more than my ATI setup, which is somewhat expected, as it is an Nvidia backed title.

A lot of people with the 2600k's are getting their CPU's up to 4.5ghz. It might not be needed to go that high, but you should be safe to go higher than you have it. Find a good guide to follow. 50C at load is also quite good.



Thanks bystander, you're a great help.. sorry to bother you again but one last request, I wouldn't be able to distinguish between a "good" or a "bad" guide if my life depended on it :??:  The store overclocked it for me for extra charge. Would you be able to reccomend me one?

Best solution

a c 216 U Graphics card
August 2, 2011 6:24:10 AM
Share

No, I don't have one for your CPU, I'd recommend going to the Overclocking forum on this site: http://www.tomshardware.com/forum/forum-29.html

They could probably point you in the right direction.
a b U Graphics card
August 2, 2011 10:07:00 AM

DTsoy said:
Yeah, without physx - the whole point of the benchmark is to showcase a ridiculous amount of physx to see hwo your video card can handle it.

bystander, other benchmarks are most of the time also lower than they should be. Just cause 2 for example, my benchmark scores are lower than other benchmark scores, also Crysis 2 (DX11, with the high rez textures and 1.9 patch) drops to about 30 fps when there are a lot of particle effects and stuff going on, which seems way too low for my video card (which could get it up to 100+ fps on other versions of crysis 2...)

amd655, I paid a computer store to set my video card up for me - I really have no idea how to check if it's connected properly, i mean in terms of pcie 8 pin and 6 pin and etc... I played around in NCP and didnt get far, and yeah it is all stock settings, only the processor is OCed to 3.8.

Any other ideas or help would be appreciated? :) 

My point in stating my results was to show you how taxing physx is and how well your gpu is running with it enabled if you are getting close to the same results with physx on. Try turning off physx and see what it can do and if you see a Jump in performance then there's your issue. Oh, and benchmarking is to see where your computer runs optimally for its power and not just to test physx.
August 3, 2011 10:22:06 PM

Best answer selected by DTsoy.
August 3, 2011 10:22:32 PM

Thank you everybody for helping me out :) 
a c 271 U Graphics card
August 4, 2011 12:40:40 AM

This topic has been closed by Mousemonkey
!