4850 Benchmarks - Something Wrong Here?

[Stabgotham]

First, the specs:

Lian Li PC61
MSI K9A2 Platinum 790FX
AMD Athlon 64 X2 5000+ Black Edition @ 3.2GHz
Xigmatek S1283 Rifle CPU Cooler
4GB OCZ SLI-Edition DDR2 800
Visiontek HD4850
WD Caviar 250GB
X-Fi XtremeMusic
Samsung 17" 740n

Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

My GPU-Z also caught me by surprise. Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

 

[Marty502]

Fan speed on zero? Is that a bug?

If it isn't, the only thing I can think of is that your GPU is throttling back.

Also, did you try different drivers? Does the 4850 have a PSU connection, and if it does, is it well plugged?

 

[Quiksilver]

Originally posted by: Stabgotham
Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

For starters, 3DMark doesn't mean a damn really, it hardly ever translates into real world performance numbers and it can vary(eg. run multiple tests, chances are you'll have a different score each time).

Crysis is a terrible benchmark, it's a poorly optimized game engine that only has really nice graphics going for it, I would not go by these numbers to prove performance.

This of course is all assuming you installed the Hotfix on AMD's site.

Also a C2D or C2Q will out benchmark your system even if you have it overclocked so the low number just could be because of that. As far as the prime stability, it could be because the card is putting a greater power draw on the PSU and your PSU efficiency dropped a little or could be for the extra heat being poured into the case by the card.

My GPU-Z also caught me by surprise.
Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

I don't know about this, it could be a GPU-Z reading the card wrong but I don't know. I did notice though your card is still using the old card BIOS that hasn't had the card idle issue fixed yet though.

 

[Stabgotham]

Yes, I have the most up to date drivers with the hotfix. I have confirmed that the PCI-E power plug is firmly connected to the videocard. I show the fan speed constantly running at 5%. To my knowledge there is not yet a way to increase the fan speed on the 4850.

My temps idle at 71C, but under load reach 86C.

 

[Stabgotham]

Originally posted by: Quiksilver

Originally posted by: Stabgotham
Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

For starters, 3DMark doesn't mean a damn really, it hardly ever translates into real world performance numbers and it can vary(eg. run multiple tests, chances are you'll have a different score each time).

Crysis is a terrible benchmark, it's a poorly optimized game engine that only has really nice graphics going for it, I would not go by these numbers to prove performance.

This of course is all assuming you installed the Hotfix on AMD's site.

My GPU-Z also caught me by surprise.
Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

I don't know about this, it could be a GPU-Z reading the card wrong but I don't know. I did notice though your card is still using the old card BIOS that hasn't had the card idle issue fixed yet though.

Updated BIOS? Where is this located? I can't find anything on Visiontek's or AMD's website with a BIOS update.

 

[Quiksilver]

I don't know where to get it from, It was announced though, so it might not be availible yet. It fixed the idle speed of the card, to it drops to like 100Mhz for less power consumption, also you quoted me before my edit... so

Also a C2D or C2Q will out benchmark your system even if you have it overclocked so the low number just could be because of that. As far as the prime stability, it could be because the card is putting a greater power draw on the PSU and your PSU efficiency dropped a little or could be for the extra heat being poured into the case by the card.

 

[Stabgotham]

If 3dMark06 and Crysis are not good benchmarks, then what is?

 

[rgallant]

Originally posted by: Stabgotham
First, the specs:

Lian Li PC61
MSI K9A2 Platinum 790FX
AMD Athlon 64 X2 5000+ Black Edition @ 3.2GHz
Xigmatek S1283 Rifle CPU Cooler
4GB OCZ SLI-Edition DDR2 800
Visiontek HD4850
WD Caviar 250GB
X-Fi XtremeMusic
Samsung 17" 740n

Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

My GPU-Z also caught me by surprise. Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

your sig didn't get in this quote but it shows you had a Nvidia card ,is it possible you did not run driver sweeper or Driver Cleaner Pro ,and cab cleaner,(part of DCP) then run CCleaner to clear the reg.,and still have parts of the Nvidia drivers still in your machine,
-do a search or check the system32 files for nv. also reload chip set drivers and POWER off the system so the bios can reconfig.then reload the video drivers and maybe reload the OS/repair if need be . not sure if 3d06 and games might have to be reloaded ?
-and\or re seat the card and check the power connections
-there's a new gpu-z out , some were borked

-Iam not a expert by the standards of this forum , but that's what I would look at.

-I was going to say to get Rivatuner ,for the fan , but it look's like unless someone sends Unwinder a 48xx card to test ,the end of July for it.

-http://forums.guru3d.com/showthread.php?t=264811

 

[daveybrat]

Something is definitely wrong. I get over 9500 3dMarks in 06 with my lowly 8800GTS 320MB (stock) and my 5600+ (stock).

 

[nitromullet]

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

 

[Stabgotham]

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

 

[Stabgotham]

Originally posted by: rgallant

Originally posted by: Stabgotham
First, the specs:

Lian Li PC61
MSI K9A2 Platinum 790FX
AMD Athlon 64 X2 5000+ Black Edition @ 3.2GHz
Xigmatek S1283 Rifle CPU Cooler
4GB OCZ SLI-Edition DDR2 800
Visiontek HD4850
WD Caviar 250GB
X-Fi XtremeMusic
Samsung 17" 740n

Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

My GPU-Z also caught me by surprise. Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

your sig didn't get in this quote but it shows you had a Nvidia card ,is it possible you did not run driver sweeper or Driver Cleaner Pro ,and cab cleaner,(part of DCP) then run CCleaner to clear the reg.,and still have parts of the Nvidia drivers still in your machine,
-do a search or check the system32 files for nv. also reload chip set drivers and POWER off the system so the bios can reconfig.then reload the video drivers and maybe reload the OS/repair if need be . not sure if 3d06 and games might have to be reloaded ?
-and\or re seat the card and check the power connections
-there's a new gpu-z out , some were borked

-Iam not a expert by the standards of this forum , but that's what I would look at.

-I was going to say to get Rivatuner ,for the fan , but it look's like unless someone sends Unwinder a 48xx card to test ,the end of July for it.

-http://forums.guru3d.com/showthread.php?t=264811

When I installed the card, I completely wiped my hard drive and reinstalled Vista 64. No concerns over any Nvidia drivers still lurking.

 

[Stabgotham]

Also, it looks like I am using the latest version of GPU-z. Not sure where you believe I am using an older version. I have 0.2.4 loaded.

 

[rgallant]

Originally posted by: Stabgotham

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

whats it like at Dx9 all high as ,shadows very high are tough

edit : wasn't sure on the software and GPU-Z the last one crashed my system when ever I was changing cards in the sensor tab

 

[nitromullet]

Originally posted by: Stabgotham

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

It's possible that your cpu is holding you back a little from matching AT's numbers. They used a QX9770 @ 3.20GHz as opposed to your A64 @ 3.2GHz. Even in situations where the card is not cpu limited per se, the cpu does still have an effect on the overall performance by a few fps.

 

[AzN]

Originally posted by: nitromullet

Originally posted by: Stabgotham

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

It's possible that your cpu is holding you back a little from matching AT's numbers. They used a QX9770 @ 3.20GHz as opposed to your A64 @ 3.2GHz. Even in situations where the card is not cpu limited per se, the cpu does still have an effect on the overall performance by a few fps.

Perhaps his 3dmark but not crysis. Maybe 1 to 2 fps difference, nothing major.

 

[AzN]

Originally posted by: Stabgotham
First, the specs:

Lian Li PC61
MSI K9A2 Platinum 790FX
AMD Athlon 64 X2 5000+ Black Edition @ 3.2GHz
Xigmatek S1283 Rifle CPU Cooler
4GB OCZ SLI-Edition DDR2 800
Visiontek HD4850
WD Caviar 250GB
X-Fi XtremeMusic
Samsung 17" 740n

Okay, so I picked up a 4850 a few days ago (Visiontek). I've been trying to figure out whats up with my low benchmarks.

3DMark06 in Vista = 7416

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

Additionally, I have my processor overclocked to 3.2GHz. This thing is stable (Prime ran for 24 hours). Since having popped in my 4850, I cannot get Prime run Stable at these speeds. Even 3DMark06 will cause my system to reboot on occasion.

My GPU-Z also caught me by surprise. Most people running a PCI-e 2.0 card on a 2.0 board will show in their Bus Interface:

PCI-E 2.0 x16 @ x16 2.0

I don't. I show:

PCI-E x16 @ x16

I do have a PCI-E 2.0 board. So that's my story thus far. Any ideas??

Your pci-e bus @ x16 is normal considering you are running PCI-E 1.0 board.

Your benchmark seems a little low but that could just be your system or something you did to the system.

I would wipe the drivers and reinstall again.

ARe you using 4 sticks of ram or 2? What about your power supply?

 

[rgallant]

you know after reading the post again about the system not being stable after the 4850 could mean a mother board bios problem ,needs updating , resetting or a power issues, real noob question but are all the extra power plugs to the mother board plugged in? not sure on that board but some have a extra 12v ,or your power supply rails are off balanced ?if multi rail supply.
and no point looking at bench marks unless the system is stable

-if power is not the issues after adding a more powerful card what about the voltage settings in the bios ,could explain the same OC not working

 

[Piuc2020]

Originally posted by: Stabgotham
If 3dMark06 and Crysis are not good benchmarks, then what is?

3DMark is a horrible benchmark, Futuremark should go under and the 3DMark application and any trace of it should be eliminated from the surface of earth forever. Just don't use 3DMark ever again.

Crysis on the other hand, IS NOT A POORLY OPTIMIZED GAME, it's just very demanding on the higher settings, games that are supposedly "well optimized" don't look anywhere near as good as Crysis on High and Very High more like Crysis on Medium and on these settings Crysis runs blistering fast (as fast as these "well coded" games", so I don't see any "bad coding" here, just a HUGE perceptual error. Crysis is just about the best benchmark you'll get.

 

[nitromullet]

Originally posted by: Azn

Originally posted by: nitromullet

Originally posted by: Stabgotham

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

It's possible that your cpu is holding you back a little from matching AT's numbers. They used a QX9770 @ 3.20GHz as opposed to your A64 @ 3.2GHz. Even in situations where the card is not cpu limited per se, the cpu does still have an effect on the overall performance by a few fps.

Perhaps his 3dmark but not crysis. Maybe 1 to 2 fps difference, nothing major.

AT HD4850 Crysis 1680x1050 0xAA = 29.8fps

OP HD 4850 Crysis 1024x768 0xAA = 31.4
OP HD 4850 Crysis 1280x1024 0xAA = 29.9

We're not talking about much more than just a few fps.

Something is holding him back other than the gpu... He only loses 2fps by applying 4xAA. I think the 0xAA benchmarks are getting limited by his cpu, and the 4xAA runs shift over to being gpu limited, which is why they are so close.

If it was me, I'd lower the OC on the cpu a tad to get the system Prime stable, play Crysis with 4xAA at 1280x1024, and not worry about it.

 

[Tempered81]

Originally posted by: Stabgotham
Lian Li PC61
MSI K9A2 Platinum 790FX
AMD Athlon 64 X2 5000+ Black Edition @ 3.2GHz
Xigmatek S1283 Rifle CPU Cooler
4GB OCZ SLI-Edition DDR2 800
Visiontek HD4850
WD Caviar 250GB
X-Fi XtremeMusic
Samsung 17" 740n

3DMark06 in Vista = 7416



Any ideas??

go play some games other than crysis, and try not to worry about your 06 scores.
IMO this rig won't perform as well as a fresh copy of XP pro, and with your games installed like so...

defrag - install - defrag

Also you want minimal HD clutter & regular defrags anyway (which Vista does, yet vista sucks, imo).

Also don't expect your 3.2ghz athlon x2 to perform like a conroe/penryn of speeds between 3.0 to 4.0ghz.

shot in the dark here - but perhaps the 790FX isn't a great performer @ 3D? Make sure, as others have said, that your bioses are up to date on all hardware. I see you already have the beta cat's & the hotfix.

IMO, Vista & AMD aren't going to perform as well as a 4GHZ penryn & XP PRO. If that 3d06 Score of 7xxx is wayyyyyy off the norm, then maybe it could be something else. Good luck!

 

[Stabgotham]

Your pci-e bus @ x16 is normal considering you are running PCI-E 1.0 board.

Your benchmark seems a little low but that could just be your system or something you did to the system.

I would wipe the drivers and reinstall again.

ARe you using 4 sticks of ram or 2? What about your power supply?

What are you talking about? I have a PCI-E 2.0 board. I've mentioned that already as well. I'm using 2x2GB of RAM. What power supply? I'm using a Corsair HX520W. It's has three 12v rails @ 18amps each.

 

[Stabgotham]

Originally posted by: rgallant
you know after reading the post again about the system not being stable after the 4850 could mean a mother board bios problem ,needs updating , resetting or a power issues, real noob question but are all the extra power plugs to the mother board plugged in? not sure on that board but some have a extra 12v ,or your power supply rails are off balanced ?if multi rail supply.
and no point looking at bench marks unless the system is stable

-if power is not the issues after adding a more powerful card what about the voltage settings in the bios ,could explain the same OC not working

It is a multi rail PSU. Three 12v rails @18amps each. I'm have the most up to date BIOS from MSI, version 1.4, but it was released some months back before the 4000 series was released.

My voltage settings are relatively low for the CPU, 1.44v. RAM is at 2.1v for which it is rated. No other voltage changes currently in the BIOS.

 

[AzN]

Originally posted by: nitromullet

Originally posted by: Azn

Originally posted by: nitromullet

Originally posted by: Stabgotham

Originally posted by: nitromullet

Crysis Benchmarks:
Run #1- DX10 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 31.365
Run #2- DX10 1024x768 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 30.105
Run #3- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 29.915
Run #4- DX10 1280x1024 AA=4x, 32 bit test, Quality: Custom ~~ Overall Average FPS: 27.265

It's really hard to say if your numbers are good or not, as you are running at a custom setting. You should definitely play the game at whatever settings you prefer, but if you want to put up any numbers and have them mean anything to anyone else you need to pick something we all know (like everything on high). I believe that if you use the Crysis benchmark tool http://downloads.guru3d.com/download.php?det=1791, you can benchmark at whatever settings you want without affecting your custom settings in the game.

Those numbers are using the Crysis Benchmarking Tool. All settings on High, Shaders set to Very High. These are the settings Anand used in its tests.

It's possible that your cpu is holding you back a little from matching AT's numbers. They used a QX9770 @ 3.20GHz as opposed to your A64 @ 3.2GHz. Even in situations where the card is not cpu limited per se, the cpu does still have an effect on the overall performance by a few fps.

Perhaps his 3dmark but not crysis. Maybe 1 to 2 fps difference, nothing major.

AT HD4850 Crysis 1680x1050 0xAA = 29.8fps

OP HD 4850 Crysis 1024x768 0xAA = 31.4
OP HD 4850 Crysis 1280x1024 0xAA = 29.9

We're not talking about much more than just a few fps.

Something is holding him back other than the gpu... He only loses 2fps by applying 4xAA. I think the 0xAA benchmarks are getting limited by his cpu, and the 4xAA runs shift over to being gpu limited, which is why they are so close.

If it was me, I'd lower the OC on the cpu a tad to get the system Prime stable, play Crysis with 4xAA at 1280x1024, and not worry about it.

We are talking about crysis here that is GPU limited not cpu limited. Perhaps a powerful gpu or running low settings/resolutions, cpu might be the culprit but I doubt it in this case.

When you start pushing a little more than 40fps than the CPU starts to see subtle changes. I've done benchmarks with crysis on high settings running lower cpu frequency and fps doesn't change for the most part. But that was with a g92. AMD 4850 might be different. Couldn't tell you unless I get a 4850 to play with.

 

[Stabgotham]

To avoid the confusion over whether or not the board is PCI-E 1.1 or 2.0...

http://www.newegg.com/Product/...x?Item=N82E16813130136
http://global.msi.com.tw/index...no=1&cat2_no=&cat3_no=

Hopefully that will answer the questions. I have a PCI-E 2.0 card and board.