How bottlenecked will this be?

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

toyota

Lifer
Apr 15, 2001
12,957
1
0
In your case you had your CPU @ 1.8ghz turn out faster results next and slower on another day. Seems to me that is inaccurate when you don't get the same consistent results as you are testing different scenarios on different occasions.

While I got consistent results and eliminated the anomaly. You spent hours on a demo and benchmark? Perhaps you should play the game and stop concentrating on benchmark numbers as I have no problems playing the game with my cpu @ 2.2ghz or 3.045ghz. Then again I have GTX260 216SP @ faster clocks and pulling faster frame rates than you with your E8500 and GTX-260 192SP @ lower clocks. :D

If you don't use the built in benchmark that's fine but you used built in benchmark in RE5 only to test CPU portion of this test where it's not testing actual game play. And you have NO evidence of Batman benchmark being flawed. Then again reviewers like legionhardware used the exact same benchmark in their reviews. All benches have some margin of error BUT I've eliminated these anomalies to give you consistent results on mine. :)
theres plenty of evidence that the benchmark is off for me and others. if someone is getting in the 20s with a gtx285 and dedicated 9600gt for physx and I am getting the same or higher with my plain gtx260 then yes the benchmark is flawed. also like I said I can get wildly different max framerate number just doing the test back to back. this is true on both computers I have ran the built in benchmark on. its just a flythrough and can be very inaccurate.

I only used the RE5 benchmark because it seemed to be very consistent when doing repeated runs. you are right though that it wasnt reflective of how good the actual gameplay is. in the actual game I rarely go below 40-45 fps with my cpu at 1.8 and only during lots of enemies do I dip below 40. I think what little sluggishness I feel is just the game itself. with my cpu at 3.16 I rarely come off below 55-60fps even when using vsync though. either way the game does play well and that fixed benchmark is a worst case scenario.
 
Last edited:

AzN

Banned
Nov 26, 2001
4,112
2
0
theres plenty of evidence that the benchmark is off for me and others. if someone is getting in the 20s with a gtx285 and dedicated 9600gt for physx and I am getting the same or higher with my plain gtx260 then yes the benchmark is flawed. also like I said I can get wildly different max framerate number just doing the test back to back. this is true on both computers I have ran the built in benchmark on. its just a flythrough and can be very inaccurate.

Whatever it is. Your benches aren't consistent. Now if it was consistent and made some sense I would agree with some of those numbers but it isn't. You can't create the same scenario twice. That's where the problem is I think. Anyways Batman was more than playable with PhysX off even for you. Now if you had faster GPU with physX support it would have been more playable with PhysX enabled.


I only used the RE5 benchmark because it seemed to be very consistent when doing repeated runs. you are right though that it wasnt reflective of how good the actual gameplay is. in the actual game I rarely go below 40-50 fps with my cpu at 1.8 and only during lots of enemies do I dip below 40. I think what little sluggishness I feel is just the game itself. with my cpu at 3.16 I rarely come off below 60fps even when using vsync though. either way the game does play well and that fixed benchmark is a worst case scenario.

Then again it's not actual game play but countless zombies on screen to showcase worst case scenario in the game and not reflecting real case scenario. The variable benchmark is the actual game play where the frame rates double if not triple. But you quickly point to the fixed benchmark. How convenient.. Anyways this game is 'more' than playable with 5600 x2 with decent enough graphics. Now if this game was a proper PC game and not console port that's been coded with 7900gt performance level in mind it would be GPU limited and slow to a craw no matter how fast your CPU was.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Whatever it is. Your benches aren't consistent. Now if it was consistent and made some sense I would agree with some of those numbers but it isn't. You can't create the same scenario twice. That's where the problem is I think. Anyways Batman was more than playable with PhysX off even for you. Now if you had faster GPU with physX support it would have been more playable with PhysX enabled.




Then again it's not actual game play but countless zombies on screen to showcase worst case scenario in the game and not reflecting real case scenario. The variable benchmark is the actual game play where the frame rates double if not triple. But you quickly point to the fixed benchmark. How convenient.. Anyways this game is 'more' than playable with 5600 x2 with decent enough graphics. Now if this game was a proper PC game and not console port that's been coded with 7900gt performance level in mind it would be GPU limited and slow to a craw no matter how fast your CPU was.

I wasnt using the RE5 benchmark to deceive. I was using it because it was consistent and I wasnt even playing the actual game at that time. as soon as I fired up the actual game I told you what the results were for it which were of course much better then that benchmark would indicate.

actually having physx on in Batman, a faster gpu wouldnt help much with with my cpu at 1.8 and even 2.0 for the most part. overclocking the gtx260 had basically little to no effect with the cpu at that speed with physx on. only when I bumped the cpu speed back up could I take advantage of overclocking the card. I have showed that many times and not just in this thread.

in fact here are some runs posted in this thread. to be clear these runs are from the entire first fight scene in the demo. this was a somewhat demanding part but being in one fairly small space with just a few enemies made the runs very easy to replicate.

1920x1080 all very high settings, high physx no AA

E8500 at 2.0 GTX260 at 576/1188/1990
Frames, Time (ms), Min, Max, Avg
2092, 63045, 21, 42, 33.183


E8500 at 2.0 GTX260 at 666/1392/2200
Frames, Time (ms), Min, Max, Avg
2086, 61778, 21, 40, 33.766


then somebody said I need to use AA to show the difference...

1920x1080 all very high settings, 4x AA and high physx

E8500 at 2.0 GTX260 at 576/1188/1990
Frames, Time (ms), Min, Max, Avg
1786, 63457, 18, 34, 29.035


E8500 at 2.0 GTX260 at 666/1392/2200

Frames, Time (ms), Min, Max, Avg
2086, 67780, 18, 36, 30.476


E8500 at 3.16 GTX260 at 666/1392/2200

Frames, Time (ms), Min, Max, Avg
2286, 65699, 27, 43, 34.795.
 
Last edited:

AzN

Banned
Nov 26, 2001
4,112
2
0
A "C2D @ 3Ghz or better" is one holy hell of a lot faster than an X2 Classic @ 2.8.

Take the numbers out and you pretty much left with this.

"video card is more important MOST of the time than the CPU"
 

AzN

Banned
Nov 26, 2001
4,112
2
0
I wasnt using the RE5 benchmark to deceive. I was using it because it was consistent and I wasnt even playing the actual game at that time. as soon as I fired up the actual game I told you what the results were for it which were of course much better then that benchmark would indicate.

But you hand picked benches to show 5600x2 not being playable in this game. Anyone looking at the 25fps average will think the game isn't playable which isn't true.

actually having physx on in Batman, a faster gpu wouldnt help much with with my cpu at 1.8 and even 2.0 for the most part. overclocking the gtx260 had basically little to no effect with the cpu at that speed with physx on. only when I bumped the cpu speed back up could I take advantage of overclocking the card. I have showed that many times and not just in this thread.

By what you are proposing you would be getting same FPS whether you enabled physX or not. But in your case your frame rate improved 80% when you disabled PhysX with the same CPU and settings. Another 8800gt along with your GTX260 would massively improve your PhysX benchmark as shown by many review sites already. Overclocking can do so much especially on the exact same GPU where it's being stressed by the game and PhysX.



1920x1080 all very high settings, high physx no AA

E8500 at 2.0 GTX260 at 576/1188/1990
Frames, Time (ms), Min, Max, Avg
2092, 63045, 21, 42, 33.183


E8500 at 2.0 GTX260 at 666/1392/2200
Frames, Time (ms), Min, Max, Avg
2086, 61778, 21, 40, 33.766


then somebody said I need to use AA to show the difference...

1920x1080 all very high settings, 4x AA and high physx

E8500 at 2.0 GTX260 at 576/1188/1990
Frames, Time (ms), Min, Max, Avg
1786, 63457, 18, 34, 29.035


E8500 at 2.0 GTX260 at 666/1392/2200

Frames, Time (ms), Min, Max, Avg
2086, 67780, 18, 36, 30.476


E8500 at 3.16 GTX260 at 666/1392/2200

Frames, Time (ms), Min, Max, Avg
2286, 65699, 27, 43, 34.795.

Again my benchmarks which doesn't reflect your findings.

Batman highest settings @ 1920x1080 4xAA physx Normal
315x7=2205mhz
3rd run
Min = 26
Max = 64
Avg = 45

435x7=3045mhz
3rd run
Min = 27
Max = 63
avg = 47

480x7=3360mhz
Min = 27
Max = 61
avg = 47
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
But you hand picked benches to show 5600x2 not being playable in this game. Anyone looking at the 25fps average will think the game isn't playable which isn't true.



By what you are proposing you would be getting same FPS whether you enabled physX or not. But in your case your frame rate improved 80% when you disabled PhysX with the same CPU and settings. Another 8800gt along with your GTX260 would massively improve your PhysX benchmark as shown by many review sites already. Overclocking can do so much especially on the exact same GPU where it's being stressed by the game and PhysX.





Again my benchmarks which doesn't reflect your findings.

Batman highest settings @ 1920x1080 4xAA physx Normal
315x7=2205mhz
3rd run
Min = 26
Max = 64
Avg = 45

435x7=3045mhz
3rd run
Min = 27
Max = 63
avg = 47

480x7=3360mhz
Min = 27
Max = 61
avg = 47


well on the RE 5 benchmark thats all I had at the time and I already explained that. as for Batman test the actual game not the benchmark and you will see. I have ran that same fraps run dozens and dozens of times with multiple settings and that is the real game results. anybody that doubts me can get the free demo and test it for themselves at the exact place I did. I start the bench 5 seconds after the cutscene, fight the guys and then right before the last blow to the final guy I stop the benchmark before that cutscene. I have taken screenshots and posted them before as well as just looking at the fraps numbers even when not benching and my results are as accurate as possible and reflect real gameplay not a flythrough.

EDIT: I can only find the screenshots for when I tested at 1.6 but here they are.

1920x1080 all very high settings high physx no AA

1.6 27 fps standing still
http://img42.imageshack.us/img42/7277/shippingpcbmgame2009102.png

3.16 39 fps same spot and even getting hit
http://img203.imageshack.us/img203/7277/shippingpcbmgame2009102.png

dont blame fraps or anything else because thats real gameplay and shows that the built in bench is waste of time. I already told you that an Eidos mod even said the flythrough doesnt reflect gameplay.
 
Last edited:

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
It's quite interesting the fact that AZN with a lower end CPU and a slighly faster card is getting better frame rate than the toyota's better CPU with a slighly slower GPU in Batman AA.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
It's quite interesting the fact that AZN with a lower end CPU and a slighly faster card is getting better frame rate than the toyota's better CPU with a slighly slower GPU in Batman AA.
he is using the built in benchmark which is a flythrough and not reflective of gameplay. they even tell you that on the official forums.

I am running a fraps benchmark at the exact part of the demo that I have tested so many times that it is ridiculous. my tests are representative of actual gameplay and have always been consistent on those fraps runs. the screenshots show exactly what I am talking about.
 
Last edited:

AzN

Banned
Nov 26, 2001
4,112
2
0
well on the RE 5 benchmark thats all I had at the time and I already explained that. as for Batman test the actual game not the benchmark and you will see. I have ran that same fraps run dozens and dozens of times with multiple settings and that is the real game results. anybody that doubts me can get the free demo and test it for themselves at the exact place I did. I start the bench 5 seconds after the cutscene, fight the guys and then right before the last blow to the final guy I stop the benchmark before that cutscene. I have taken screenshots and posted them before as well as just looking at the fraps numbers even when not benching and my results are as accurate as possible and reflect real gameplay not a flythrough.

EDIT: I can only find the screenshots for when I tested at 1.6 but here they are.

1920x1080 all very high settings high physx no AA

1.6 27 fps standing still
http://img42.imageshack.us/img42/7277/shippingpcbmgame2009102.png

3.16 39 fps same spot and even getting hit
http://img203.imageshack.us/img203/7277/shippingpcbmgame2009102.png

dont blame fraps or anything else because thats real gameplay and shows that the built in bench is waste of time. I already told you that an Eidos mod even said the flythrough doesnt reflect gameplay.

I don't know what to tell you. There might be something with your system. I get very little difference with PhysX enabled. PhysX is done by your GPU not your CPU anyway and shouldn't reflect by your CPU.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
I don't know what to tell you. There might be something with your system. I get very little difference with PhysX enabled. PhysX is done by your GPU not your CPU anyway and shouldn't reflect by your CPU.
wait so earlier when I said I was using physx to be very gpu limited you told me to turn physx off as it might affect my performance even on the cpu since I was complaining about minimum framerates. well I did turn it off and you were right because min framerate went way up with physx off. you cant now change the argument back to physx is only having an impact on gpu. that makes no sense and you know it.

there is nothing wrong with my system. the only thing wrong is you trying to compare the flythrough with actual gameplay. we have both thrown out the RE 5 fixed bench as being not reflective of gameplay so you need to man up and admit that is true for the Batman flythrough too. even on the official Eidos forums, jaycw2309 that actually worked on the game says its NOT representative of real gameplay.

again I have tested the shit out that part of the demo and my numbers are consistent. even if you dont bench with fraps it doesnt take a genius to see the lower numbers on the screen and feel the game being slightly more sluggish. anybody can test that exact part for themselves with fraps if they doubt me. and yes i have tested other parts of the game too. I just mainly used that part of the demo for most testing and to get consistent results for a small bit of action.
 
Last edited:

toyota

Lifer
Apr 15, 2001
12,957
1
0
I did what you did Toyota by taking screen shot with fraps. Absolutely makes no difference.

2205mhz

3360mhz
so do you think the one place I tested at the most was some fluke? I would find that hard to believe wouldnt you? you can see even when not really fighting the framerate difference from 1.6-3.16 was very large at that part of the game. again I would never post that if I hadnt tested it so much. I dont want to call you a liar without testing that exact part of the game so where is that in the game? it would probably be easier if we tested at the same spots but I am only very early into the full game.

EDIT: the part I was at had enemies moving and papers flying around so maybe that was having an impact. the part you are at has absolutely nothing going on.
 
Last edited:

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
Take the numbers out and you pretty much left with this.

"video card is more important MOST of the time than the CPU"

Of which I obviously agree :)

There is a point though, when using one of the CPU's that is getting very advanced in age, where even the very best video card won't help gaming performance as far as raising minimums beyond a point of severely diminished returns. The last couple of years saw the sun set on the Pentium 4, Pentium D, and Athlon 64 processors for being logically paired with a new high-end video card, and now it's the early C2D and Athlon X2 / Opteron's time to be in that sunset position.

I'm not saying that a decently balanced older C2D (non-overclocked E6300/E6400) or Athlon X2 can't provide a decent gaming experience, but that something around the 4850 or so was about as fast a GPU as made sense to pair with such a setup. Even with Crossfired 5870's, such a setup is bogged down by a lot of current games by the lack of CPU power, and it's only going to get more pronounced going into the 2010 release schedule.

Luckily, Athlon X4's, E5xxx's, even budget PhII builds aren't that $$$, and make a world of difference.

Given the choice of gaming on a 2.8ghz X2 Classic w/5850 vs. gaming on an 3ghz+ X4 w/4870, I'd take the more balanced 2nd option every time, and twice on Sunday.
 

AzN

Banned
Nov 26, 2001
4,112
2
0
wait so earlier when I said I was using physx to be very gpu limited you told me to turn physx off as it might affect my performance even on the cpu since I was complaining about minimum framerates. well I did turn it off and you were right because min framerate went way up with physx off. you cant now change the argument back to physx is only having an impact on gpu. that makes no sense and you know it.

I didn't tell you to do anything. I gave you my results and then you soon followed and posted yours without PhysX. This is what I said.

It's so devious of you to point to PhysX benchmark of Batman where you cripple the performance into more GPU limited situation and then point to the CPU which I benchmarked 3 times only to have 2fps difference in a GPU limited situation.

I'm pointing at your results and saying error obviously.



there is nothing wrong with my system. the only thing wrong is you trying to compare the flythrough with actual gameplay. we have both thrown out the RE 5 fixed bench as being not reflective of gameplay so you need to man up and admit that is true for the Batman flythrough too. even jaycw2309 that actually worked on the game sasy its NOT representative of real gameplay.

It seems like there is something wrong with your system. When you bench a GPU limited situation and your CPU is capable of putting out 30fps minimum frame rates without PhysX than bench with PhysX in more GPU limited situation your minimum frame rate shouldn't be dropping because PhysX is done by your GPU not your CPU!

again I have tested the shit out that part of the demo and my numbers are consistent. even if you dont bench with fraps it doesnt take a genius to see the lower numbers on the screen and feel the game being slightly more sluggish. anybody can test that exact part for themselves with fraps if they doubt me. and yes i have tested other parts of the game too. I just mainly used that part of the demo for most testing.

Again something wrong with your system. Drivers? Did you enable GPU physX. You probably need to flush your drivers.
 

AzN

Banned
Nov 26, 2001
4,112
2
0
so do you think the one place I tested at the most was some fluke? I would find that hard to believe wouldnt you? you can see even when not really fighting the framerate difference from 1.6-3.16 was very large at that part of the game. again I would never post that if I hadnt tested it so much. I dont want to call you a liar without testing that exact part of the game so where is that in the game? it would probably be easier if we tested at the same spots but I am only very early into the full game.

EDIT: the part I was at had enemies moving and papers flying around so maybe that was having an impact. the part you are at has absolutely nothing going on.

Again you might want to check your system as my system doesn't reflect your results.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
I didn't tell you to do anything. I gave you my results and then you soon followed and posted yours without PhysX. This is what I said.

It's so devious of you to point to PhysX benchmark of Batman where you cripple the performance into more GPU limited situation and then point to the CPU which I benchmarked 3 times only to have 2fps difference in a GPU limited situation.

I'm pointing at your results and saying error obviously.





It seems like there is something wrong with your system. When you bench a GPU limited situation and your CPU is capable of putting out 30fps minimum frame rates without PhysX than bench with PhysX in more GPU limited situation your minimum frame rate shouldn't be dropping because PhysX is done by your GPU not your CPU!



Again something wrong with your system. Drivers? Did you enable GPU physX. You probably need to flush your drivers.

I promise nothing is wrong with my system. same EXACT results on Vista as in 7 too. only difference is lowering the cpu and thats the results. if you are really being honest with me then download the demo and go to that exact spot. I know that the area I am in the demo had most of the effects removed in the full game like breakable walls and such. that still wouldnt have an impact on exactly whats going on in my screenshots but my benchmark also included my breaking those walls.
 

AzN

Banned
Nov 26, 2001
4,112
2
0
I promise nothing is wrong with my system. same EXACT results on Vista as in 7 too. only difference is lowering the cpu and thats the results. if you are really being honest with me then download the demo and go to that exact spot. I know that the area I am in the demo had most of the effects removed in the full game like breakable walls and such. that still wouldnt have an impact on exactly whats going on in my screenshots but my benchmark also included my breaking those walls.

How can you be so sure? you talk as if you don't make any mistakes.

I was running around at the settings I said prior @ 2205mhz in batman I don't even drop to teens like you. If anything I'm hovering 40-60fps which only points to your doings.

It's like that time you benched Crysis @ medium settings with your 8600gt then swore by your CPU wasn't up to task to run high settings. CPU matters little in a GPU limited situation like Batman with PhysX or Crysis.

I guess history repeats itself.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
How can you be so sure? you talk as if you don't make any mistakes.

I was running around at the settings I said prior @ 2205mhz in batman I don't even drop to teens like you. If anything I'm hovering 40-60fps which only points to your doings.

It's like that time you benched Crysis @ medium settings with your 8600gt then swore by your CPU wasn't up to task to run high settings. CPU matters little in a GPU limited situation like Batman with PhysX or Crysis.

I guess history repeats itself.
again with that? I like how you take that completely out of context and overblow it. some guy made a claim about getting really good framerates in Crysis with an 8600gt and what I thought was a very low end cpu. all I was trying to say back to him was that what he was claiming might be possible on the 8600gt at a low res if he a had a better cpu. in other words yes I know an 8600gt is almost completely gpu bound but I figured at least having a modern dual core cpu would allow him to turn up a setting or two at a very low res. please start your bs with someone else.

when testing earlier today I took a screenshot after the cutscene finished. with my cpu at 1.8 and then 3.16 there was clearly a difference there too.

3.16 56 fps
http://img28.imageshack.us/img28/7064/shippingpcbmgame2009113.png

1.8 44 fps
http://img28.imageshack.us/img28/294/shippingpcbmgame2009113d.png

now either you are a liar are you just happened to be in place that had little to no effect on framerate. its odd that I can get these differences and you cant.
 
Last edited:

AzN

Banned
Nov 26, 2001
4,112
2
0
Toyota just check your system and reinstall drivers and stop calling me a liar. I'm about as honest and blunt as they come.

I've already mentioned that I was running around @ 2205mhz in batman only to hover 40-60fps. If you don't believe it. Tough shit your loss. If my system isn't behaving like your system and your system is the one dropping frames like crazy it's obvious there's something wrong with your system.
 
Last edited:

AzN

Banned
Nov 26, 2001
4,112
2
0
You know what I think is happening? Your GPU accelerated PhysX is fucked up. You are using CPU PhysX when you enable physX. This happened to me once. I corrected by updating PhysX package.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
You know what I think is happening? Your GPU accelerated PhysX is fucked up. You are using CPU PhysX when you enable physX. This happened to me once. I corrected by updating PhysX package.
those last shots that showed a 12fps difference between 1.8 and 3.16 were with no physx at all. also my performance is exactly where it should be when running physx and my cpu at 3.16. I dont know what else to tell you but dropping cpu down to 2.0 then 1.8 and 1.6 each take the performance and especially minimum framerate down with it. the game is mostly playable but certainly more sluggish during action the further I drop the cpu speed. at 1.8 and especially 1.6 I would not be happy if this was my real cpu because fights are a little too poky feeling at times. oh and I got the same results in Vista as in Windows 7. I have played around a bit more and I can get within 2 to 3 fps of each other with my cpu at 2.0 and 3.16 when nothing at all is around me. sometimes walking in an empty area will show very similar framerates but as soon some action starts the difference is very noticeable.
 
Last edited: