Resident Evil 5 video benchmark

[SSChevy2001]

@evolucion8
Here's screenshot showing the difference between HDR low and high.
http://i31.photobucket.com/alb...1/RE5HDRComparison.jpg

DX10 is supposed to use less vram yet takes a 35% performance hit over DX9 with nothing to show for it.

 

[Udgnim]

have a 4870 512 using Cat 9.7

in the Fixed benchmark, anyone noticing hardcore aliasing primarily in certain areas of the wooden platform that the loudspeaker man & executioner are standing on?

I've run the benchmark up to 8x AA and aliasing is still very noticeable

CCC settings are app controlled for AA with the "Box" filter setting

 

[kkk60091]

Originally posted by: AdamK47

Before running the test I hit "Restore" in the nVidia control panel to put everything in there as default.

tried, no change.

I guess i will load a new windows on a ssd and uplug everything other than video cards.


loaded up 7 rtm and only video card driver.

yeah...works, generally got the same score as yours.

something must messed up the system.

 

[evolucion8]

Originally posted by: SSChevy2001
@evolucion8
Here's screenshot showing the difference between HDR low and high.
http://i31.photobucket.com/alb...1/RE5HDRComparison.jpg

DX10 is supposed to use less vram yet takes a 35% performance hit over DX9 with nothing to show for it.

The difference is more subtle than that, the texture looks sharper, the HDR looks better, those are things that cannot be captured in pictures. In DX9 mode the HDR looks milky and blurs the screen very often, in DX10, the HDR adjust only the areas close to it's light exposure, making the colors more cleaner and realistic, also the jaggies are a little bit less noticeable, but the performance impact is just too great for such little gains in image quality, if it was made for DX10.1, the performance would be improved, specially when Anti Aliased was used.

Also I found that changing the HDR mode to High in the config file found on the My Document's Capcom folder, cause stuttering in DX9 mode, proving that DX9 in such scenarios uses more VRAM, in DX10 such stuttering never happened, but the issue is the low performance found in the Level 3 using DX10.

 

[SSChevy2001]

Originally posted by: evolucion8
The difference is more subtle than that, the texture looks sharper, the HDR looks better, those are things that cannot be captured in pictures. In DX9 mode the HDR looks milky and blurs the screen very often, in DX10, the HDR adjust only the areas close to it's light exposure, making the colors more cleaner and realistic, also the jaggies are a little bit less noticeable, but the performance impact is just too great for such little gains in image quality, if it was made for DX10.1, the performance would be improved, specially when Anti Aliased was used.

I'm not here to argue that DX10 has a very slightly better image quality, but it's performance on ATi cards is crap. What good is DX10.1 if even without AA this title runs like crap. ( DX9 80.0fps vs DX10 63.9fps )

Also I found that changing the HDR mode to High in the config file found on the My Document's Capcom folder, cause stuttering in DX9 mode, proving that DX9 in such scenarios uses more VRAM, in DX10 such stuttering never happened,

Well I ran the benchmark yet again, but I don't have any stuttering issue with DX9 HDR high with 8xAA. It might have to do with me changing shaderlog.slg or 8.634 catalyst drivers, but I haven't tested.

If you want to see stuttering though just run a FarCry2 benchmark with DX10.1, yet DX9 will run just fine. Does that mean that FarCry2 DX10.1 is running out of vram also?

but the issue is the low performance found in the Level 3 using DX10.

What about dropping about 50fps scene 2 and 4 that doesn't count? Heck I've lost 36% of my fps even in scene 1!

1080P DX9 HDR high 8xAA no motion blur
68.2 / 73.6 / 62.0 / 69.9 = 68.5fps

1080P DX10 HDR high 8xAA no motion blur
43.4 / 21.0 / 8.0 / 20.8 = 21.5fps

 

[dflynchimp]

Originally posted by: SSChevy2001
@AmberClad
Your score looks a little low for a OC GTX285. Did you edit the config.ini to use both threads of that E8400?

waaaaait what's this about a config.ini???

 

[SSChevy2001]

Originally posted by: dflynchimp

Originally posted by: SSChevy2001
@AmberClad
Your score looks a little low for a OC GTX285. Did you edit the config.ini to use both threads of that E8400?

waaaaait what's this about a config.ini???

The config.ini is in your "Documents\CAPCOM\RESIDENT EVIL 5 Benchmark Version" folder.

there's two things you might want to change.

HDR = low or high
JobThread = 2 or 4 or 8 depending on your CPU

Edit
You can also create a blank shaderlog.slg if you don't want to wait for the 1st load time on DX9. That's in "C:\Program Files (x86)\CAPCOM\RESIDENT EVIL 5 Benchmark Version\nativePC" folder.

 

[evolucion8]

Originally posted by: SSChevy2001
I'm not here to argue that DX10 has a very slightly better image quality, but it's performance on ATi cards is crap. What good is DX10.1 if even without AA this title runs like crap. ( DX9 80.0fps vs DX10 63.9fps )

That's your opinion which everybody here respects, but is just far too fetched and subjective, you have no proof that all ATi cards are crap running DX10 code. The performance impact with my card using DX10 mode in RE5 is significant, but is still very enjoyable, probably the performance in your system is hampered by your CPU's little cache.

Well I ran the benchmark yet again, but I don't have any stuttering issue with DX9 HDR high with 8xAA. It might have to do with me changing shaderlog.slg or 8.634 catalyst drivers, but I haven't tested.

If you want to see stuttering though just run a FarCry2 benchmark with DX10.1, yet DX9 will run just fine. Does that mean that FarCry2 DX10.1 is running out of vram also?

DX10 simply uses less VRAM whenever your like it or not, it's documented. I also had the same issue with Far Cry 2, and what I did to fix it was simply pressing CTRL+ALT+DEL to switch screen, then going back to the game and it worked great, even after new level loads. It also worked for Cryostasis which had the same stuttering issue in DX10 mode.

What about dropping about 50fps scene 2 and 4 that doesn't count? Heck I've lost 36% of my fps even in scene 1!

I only lost less than 3fps in level 1, probably is again your CPU small cache, the Core architecture needs larger caches to perform great in memory intensive applications, like games, a measly Pentium Dual Core 2.20GHz overclocked to 2.80GHz is much slower than a Core 2 Duo E6600 at stock speeds because of its tiny cache. I use the latest drivers for my videocard, chipset along with the latest Windows updates, otherwise my videocard would have the same performance hit as yours, but at my LCD resolution of 1280x1024 I should be CPU limited and the game is heavily multi threaded (I monitored the CPU usage with Rivatuner to confirm this)

 

[SSChevy2001]

Originally posted by: evolucion8
That's your opinion which everybody here respects, but is just far too fetched and subjective, you have no proof that all ATi cards are crap running DX10 code. Stop right here

I'm only talking about DX10 and THIS title, which the benchmarks are proof enough. IMO is this problem has to do with Capcom.

DX10 simply uses less VRAM whenever your like it or not, it's documented.

Thanks, now tell me something I don't know.

I also had the same issue with Far Cry 2, and what I did to fix it was simply pressing CTRL+ALT+DEL to switch screen, then going back to the game and it worked great, even after new level loads.

Any proof? Give me one FarCry2 DX10 benchmark with your 4870 that doesn't stutters. I've run the benchmarks with the lastest driver recently and it still stutters.

I only lost less than 3fps in level 1. Again stop right there

Here's proof your wrong again! We're both GPU limited right now, so stop this whole cache BS. With a lower resolution scene 1 runs slightly faster.

DX9 1280x1024 max settings 8xAA
110.1 / 104.8 / 88.8 / 100.7 Total: 100.3
& yours
107.3 / 102.9 / 85.2 / 99.4 / Total: 97.9

DX10 1280x1024 max settings 8xAA
111.9 / 54.6 / 23.0 / 53.8 / Total: 56.2
& yours
104.8 / 52.0 / 22.4 / 51.0 / Total: 53.2

 

[evolucion8]

Originally posted by: SSChevy2001
I'm only talking about DX10 and THIS title, which the benchmarks are proof enough. IMO is this problem has to do with Capcom.

You are right.

Any proof? Give me one FarCry2 DX10 benchmark with your 4870 that doesn't stutters. I've run the benchmarks with the lastest driver recently and it still stutters.

I don't have to proof you anything, I played the game 4 months ago and then uninstalled it and the trick that I mentioned before worked fine for me, believe what you want to believe, try it yourself! I will not install such boring game just to satisfy your incredulity.

Here's proof your wrong again! We're both GPU limited right now, so stop this whole cache BS. With a lower resolution scene 1 runs slightly faster.

I just said that probably was your smaller cache, never said that definitively WAS your smaller cache, and please if you don't stop your bad boy attitude using words like "BS", then I'm done talking with you, keep it civilized, thanks.

DX9 1280x1024 max settings 8xAA
110.1 / 104.8 / 88.8 / 100.7 Total: 100.3
& yours
107.3 / 102.9 / 85.2 / 99.4 / Total: 97.9

DX10 1280x1024 max settings 8xAA
111.9 / 54.6 / 23.0 / 53.8 / Total: 56.2
& yours
104.8 / 52.0 / 22.4 / 51.0 / Total: 53.2

Even though the game is heavily multi threaded, shows that the game is GPU bound instead of CPU bound.

That data is interesting, specially when you claimed that you lost 36% of your FPS in the level 1 and now you posted something different, showing a small performance boost.

What about dropping about 50fps scene 2 and 4 that doesn't count? Heck I've lost 36% of my fps even in scene 1!

 

[vj8usa]

Originally posted by: evolucion8
That data is interesting, specially when you claimed that you lost 36% of your FPS in the level 1 and now you posted something different, showing a small performance boost.

What about dropping about 50fps scene 2 and 4 that doesn't count? Heck I've lost 36% of my fps even in scene 1!

The 36% FPS drop in scene 1 he was talking about was at 1080p. Odds are you're too CPU limited at 1280x1024 to see the same drop there.

 

[SSChevy2001]

Originally posted by: evolucion8
I don't have to proof you anything, I played the game 4 months ago and then uninstalled it and the trick that I mentioned before worked fine for me, believe what you want to believe, try it yourself! I will not install such boring game just to satisfy your incredulity.

Well I hate to break it to you, but your trick doesn't work.

I just said that probably was your smaller cache, never said that definitively WAS your smaller cache, and please if you don't stop your bad boy attitude using words like "BS", then I'm done talking with you, keep it civilized, thanks.

You claimed that you've monitored the CPU usage in this title, but can make such a false statement. From my point of view your just busting my chops, and I personally don't care for it.

That data is interesting, specially when you claimed that you lost 36% of your FPS in the level 1 and now you posted something different, showing a small performance boost.

Yeah I would say it's very interesting also, and clearly shows there's a problem on Capcom side. Either way the problem is beyond just scene 3 and having AA on.

 

[evolucion8]

Originally posted by: SSChevy2001
You claimed that you've monitored the CPU usage in this title, but can make such a false statement. From my point of view your just busting my chops, and I personally don't care for it.

Monitoring CPU usage with Rivatuner is a piece of cake, and RE5 put an average load of 65% in each core, so I'm not making false claims, I'm not busting your chops, just keeping it civilized, I'm done talking with you.

 

[SSChevy2001]

Originally posted by: evolucion8
Monitoring CPU usage with Rivatuner is a piece of cake, and RE5 put an average load of 65% in each core, so I'm not making false claims, I'm not busting your chops, just keeping it civilized, I'm done talking with you.

I have no problem using rivatuner also, only using RE5 I'm get average load under 60% in each core using the same 12x10 settings.

http://i31.photobucket.com/alb...2001/RE5_CPU_12x10.jpg

Average CPU Usage
Core 0 / Core 1 / Core 2 / Core 3 | Total Avg

DX9
55.77 / 46.97 / 48.97 / 59.86 | Total Avg 52.89

DX10
25.54 / 33.23 / 25.98 / 25.08 | Total Avg 27.46

Still think your not making false claims?

 

[dflynchimp]

Originally posted by: SSChevy2001

Originally posted by: evolucion8
Monitoring CPU usage with Rivatuner is a piece of cake, and RE5 put an average load of 65% in each core, so I'm not making false claims, I'm not busting your chops, just keeping it civilized, I'm done talking with you.

I have no problem using rivatuner also, only using RE5 I'm get average load under 60% in each core using the same 12x10 settings.

http://i31.photobucket.com/alb...2001/RE5_CPU_12x10.jpg

Average CPU Usage
Core 0 / Core 1 / Core 2 / Core 3 | Total Avg

DX9
55.77 / 46.97 / 48.97 / 59.86 | Total Avg 52.89

DX10
25.54 / 33.23 / 25.98 / 25.08 | Total Avg 27.46

Still think your not making false claims?

so how many background tasks are the two of you having running while these benchies are going on? Maybe one of you just have too much fat you need to trim off the process tree XD

interesting numbers in any case!

 

[SSChevy2001]

@dflynchimp
Thanks! 28 processes
I keep my machine very light on background tasks.

 

[dflynchimp]

yeah, I have on average 70 tasks from random drivers (printer, free antivirus progs, backgroudn services) running that really eat up services. Normally with a quadcore you never notice the difference but once I fire up a multithreaded game it becomes more evident.

 

[dflynchimp]

bump!

Just ran the benchmark again with the Cat 9.9 drivers. MASSIVE IMPROVEMENT.

Here's my original post, IIRC I did not have Vsync enabled, but otherwise the settings are pretty much the same.

Originally posted by: dflynchimp
119-40-15-40 respectively in FPS for my rig (specs listed below) for the variable bench with 45FPS overall

Settings at 60Hz 1920X1200 8X AA and all settings to max.

I'm noticing alot of micro stutter and flickering in certain parts of the benchmark on my 4870X2. Is anyone with a similar rig getting the same issue?

First off, the ATI flickering and microstuttering issue with DX10 is completely gone. That alone would've sold me, but the numbers themselves have also went up. It's also worth noting that my CPU for the previous bench was running at 2.66 GHz, but when I ran the 9.9 benchmarks I actually had it clocked lower (2.16GHz) for power savings. This really goes to show the improvement in the graphics drivers.

Settings: 1920X1200, 8XAA, DX10 with all options on high with motion blur

no Vsync: 127.3/79.5/54.0/76.7 with 80.5fps overall
With Vsync: 59.9/59.9/53.7/59.9 with 54fps overall

Strange is the third part of the variable bench, which ran like shit on previous versions of catalyst, namely 9.2, 9.6 and 9.7 (I skipped 9.8). Now it never dipped below 50fps with or without Vsync and with 8xAA on. Whatever ATI's driver team did, they came through for us end users. Kudos to them!

It's notable that while without vsync the game could run higher than 60fps I could notice alot more screen tearing, which Vsync does a good job of eliminating.

 

[AdamK47]

Decided to run the benchmark again with my new hardware. Specs in sig. Old screenshot and specs posted above.