Why does my HD4850 suck!?

uninstall the drivers, clean them with driver cleaner pro and driver sweeper, run crap cleaner , install Catalyst 8.7 beta , then report back .

also check if the image quality settings (AA/AF and others ) in ur CCC are set to application preference .
 
get a visiontek 4850 they offer lifetime warranty for a reason
rma the gigbyte 4850 and sell it
 
Not sure if it matters but I'm getting 201FPS on the Lost Coast benchmark @ 1360 x 768 (4xAA / 16xAF).
 
CamaroZ28 said:
The other thing is that I'm worried about RMA'ing the card because I already cut the UPC off the box for the MIR. Will this be a problem, or should they just be able to replace it?

**Bought from Newegg on 7/3/2008, Manf. is Gigabyte**
Click to expand...

Should be no problem. The UPC is not necessary for the RMA, just the bar-code on the device itself and a copy of your receipt I think.
 
I have the ASUS and love it. You should consider due to the motherboard you're buying Comaro.

Though I hope the driver set will work for you. Get with me before 5PM Alaska time and I will correlate how to get those drivers to you. Thinks he knows what the problem is.
 
im getting 164fps@1900x1200 6xmsaa/16xaf/full adaptive aa/ingame settings full with a 4870/e6600/2gigs ddr2
 
Install 8.6 Hotfix by themselves, as said before, these are a standalone set of drivers, not a patch for the 8.6. Failing that, RMA the card. Seeing these issues persist over several installs and a whole new motherboard makes you having a retarded card all the more likely.
 
If you sent in the MIR that only means you can't RMA it to Newegg if that's what you're wondering. The original manufacturer can handle it for you.

Make sure you never get a card from Sapphire... when I got my Sapphire HD 3650, a day after I sent in my rebate the fan started to buzz. I later learned that the guys at Sapphire wanted me to pay them a $15 administrative fee for sending the card back to Hong Kong... yeah right. What a ripoff...
 
Maybe you need to reinstall your OS. Driver problems can cause serious performance issue. Your hardware should be performing better than it is. If the hardware is not defective, then it's a configuration, driver, OS, or software issue. When you got your new motherboard, did you use the same hard drive with the same old install of Vista? Format that hard drive and reinstall everything.
 
A lot of people seemed to not have read my first post because I stated that this is most defeinetly a fresh install of Vista x64 and that I have JUST the hotfix drivers installed.

Also as for my latest findings, I "borrowed" another 4850 from Fry's, this one being a Diamond, and I seem to be having the exact same problem with it. Not to mention that I can't even install drivers for having them in xfire.
 
That's odd. Didn't you already have xfire drivers installed before you put the 2nd one in? If you did, once you put the second one in, you just have to enable cf in the CCC panel. P.s. can you post a pic of your system, maybe we can figure out an issue if we can see what's going on in there.

You really should of installed the 4850 in your buddy's system like you said you were gonna do. If his system ran the card well, then we can start targeting your cpu, ram, or motherboard as the bottleneck.

I'd blame your board at this point...or cpu. 2160 might still be bottlenecking...a good OC would be 3.4.
 
BigCactus said:
I'd blame your board at this point...or cpu. 2160 might still be bottlenecking...a good OC would be 3.4.
Click to expand...
Why would overclocking it from 3.1 to 3.4 make a world of difference in something that's not even CPU-bound?
 
I don't know. I'm pulling rabbits out of my you know what at this point. If I was this guy I would of returned the cards and bought a totally different card already.

I think it has more to do with his CPU perhaps than with his clock speed....What's the difference between a E2160 vs. an E8400 clock for clock? I don't think you would get the same performance clock for clock....or would you? :confused:
 
Dude - bring the system back to earth. Stock speed and voltage - everything. Once you figure out the deal then you can go crazy. Help us help you.
 
BigCactus said:
That's odd. Didn't you already have xfire drivers installed before you put the 2nd one in? If you did, once you put the second one in, you just have to enable cf in the CCC panel. P.s. can you post a pic of your system, maybe we can figure out an issue if we can see what's going on in there.

You really should of installed the 4850 in your buddy's system like you said you were gonna do. If his system ran the card well, then we can start targeting your cpu, ram, or motherboard as the bottleneck.

I'd blame your board at this point...or cpu. 2160 might still be bottlenecking...a good OC would be 3.4.
Click to expand...

Sorry, forgot to mention that I ran the system in my friends board with the specs I listed before. Got 135.4 on the same stress test. Which is way better then I've been able to get on my rig. Not to mention that his was an Nvidia chipset running at PCI Express 1.1.

Also I will post a pic as soon as I can, and as for the crossfire, I completely wiped drivers and installed the 2nd card then installed new drivers. Right now I'm playing Crysis to figure out whats going on.
 
Mega666 said:
Dude - bring the system back to earth. Stock speed and voltage - everything. Once you figure out the deal then you can go crazy. Help us help you.
Click to expand...

Already did that on my previous system, decided it was cpu limited... so overclocked. I think you might need to refer to my previous post if you need more information, mind you I was running a Foxxconn G33M board during that whole post.
 
What cpu did you friend's pc have? At least at this point we can eliminate the 4850 as the problem. It's strange that you were getting only slightly better performance on your other board...so I'd eliminate the motherboard as the issue as well. You did try other ram? I know it's a stretch but maybe you can borrow your buddy's system, put his cpu, and ram on your motherboard...and put his psu in your rig one by one. I'd start with the easy stuff like psu and ram first.

Oh yah, you could also try installing windows xp instead...
 
I think it's a driver issue. I think your drivers bung, but you haven't pm'd me to see when you want to get together and do that.
 
BigCactus said:
What cpu did you friend's pc have? At least at this point we can eliminate the 4850 as the problem. It's strange that you were getting only slightly better performance on your other board...so I'd eliminate the motherboard as the issue as well. You did try other ram? I know it's a stretch but maybe you can borrow your buddy's system, put his cpu, and ram on your motherboard...and put his psu in your rig one by one. I'd start with the easy stuff like psu and ram first.

Oh yah, you could also try installing windows xp instead...
Click to expand...

Ya he has a E6600 and his ram is almost the same as mine except better timings, and I don't see how it could be my ram because I've run plenty of memtest's on it for stability testing.

Also I refuse to go to XP because its obvious that some people are getting it to work on Vista so I should be able to as well.

AKeZtArGeT said:
I think it's a driver issue. I think your drivers bung, but you haven't pm'd me to see when you want to get together and do that.
Click to expand...

Sent PM.
 
However if you load XP and it works fine, you can rule it as a software/driver issue. If you have an XP licence, consider partitioning your drive and loading XP on the new partition.
 
Cheffy said:
However if you load XP and it works fine, you can rule it as a software/driver issue. If you have an XP licence, consider partitioning your drive and loading XP on the new partition.
Click to expand...

Ok, I agree, that is good advice. Would you recommend 64bit or 32bit as I have both flavors of Professional.
 
Just to toss it out there, I got 179.54 FPS average on a 4850 at stock speeds, a Q6600 at 3.0GHz, 2GB RAM at 800MHz, running in XP SP3 on the Hotfix 8.6. This was at 1600x1200, the most comparable 4:3 resolution to the one you specified at all of the settings you mentioned in the OP.
 
krameriffic said:
Just to toss it out there, I got 179.54 FPS average on a 4850 at stock speeds, a Q6600 at 3.0GHz, 2GB RAM at 800MHz, running in XP SP3 on the Hotfix 8.6. This was at 1600x1200, the most comparable 4:3 resolution to the one you specified at all of the settings you mentioned in the OP.
Click to expand...

Thanks for testing that out. So far my conclusions seem to lead me to believing I either have a driver issue or the lack of cache in my cpu is making a difference (1Mb vs. 4Mb or 6Mb in most C2D's)

Also tried playing a little crysis, all high settings and no AA and at first it was doing really good but when I got into physics intesive fights with explosions my screen would literally stay on 1 frame for about 20 seconds before moving to the next frame. Then I switched everything down to all low settings and lowest res my screen would support and I got the exact same results. So it's either drivers or cpu.
 
This one is easy.. Your comparing numbers on your system to people with Full blown Quads and newer e8400s and e8500s all of which will beat your overclocked processor. I dont see anything wrong with the numbers for your setup so far. I bet you if you stole your budies quadcore for a day youd get the same results or even better on your board. Notice how you got 85 fps at 1680x1050 and 90 at 1280x1024. This is a pretty big indicator your cpu bound.
 
FlyinBrian said:
This one is easy.. Your comparing numbers on your system to people with Full blown Quads and newer e8400s and e8500s all of which will beat your overclocked processor. I dont see anything wrong with the numbers for your setup so far. I bet you if you stole your budies quadcore for a day youd get the same results or even better on your board. Notice how you got 85 fps at 1680x1050 and 90 at 1280x1024. This is a pretty big indicator your cpu bound.
Click to expand...

I agree with you somewhat.. However right now i'm comparing with other processors thats are WAY underclocked compared to mine. My brothers E6750 @ 2.66 and my friends E6600 @ 2.4 and both min. 40fps+ better then me all around. So if you can confidently say that is all to be blamed on my lack of cache, then so be it. However, I'm not fully convinced its just cache. We'll have to see though, should be borrowing my brothers cpu today because he's getting a new heatsink and has to take his proc out anyways.
 
YeuEmMaiMai said:
uhm intel 3 series chipsets support pcie 2.0. This includes the G33 chipset as well
Click to expand...

Actually the P35 and G33 do not have PCI-E 2.0, only PCI-E 1.1. I think what you meant to say is that PCI-E 2.0 is backwards compatible with PCI-E 1.1.

Anyway, here are a few links detailing the performance increase from a larger L2 cache:
Does cache size really boost performance? @ Tom's
Influence of L2 Cache Size on Conroe Processors Performance @ XBit
UT3 Review: Cache Scaling: 1MB, 2MB, 4MB @ AnandTech
2MB or 4MB Cache? @ AnandTech
Influence of L2 Cache Size on Wolfdale Performance @ Xbit
 
that's funny because both my 3870 and my 4870 show the interface as PCIE 2.0 on my foxconn G33-M board on GPU-Z
 
YeuEmMaiMai said:
that's funny because both my 3870 and my 4870 show the interface as PCIE 2.0 on my foxconn G33-M board on GPU-Z
Click to expand...

It's showing the interface of the cards themselves, not the motherboard. Nowhere does it say on the Foxconn siie for the G33M about it having PCI-E 2.0. Same for the Intel website: No mention of the G33 chipset with PCI-E 2.0.

EDIT: Just checked GPU-Z on my own rig. Yup I was right: GPU-Z is showing the interface of the card themselves. Otherwise, I must have been lucky in getting a P965 motherboard with PCI-E 2.0 :eek:
 
sure its not just the card? im looking on newegg and none of the g33 foxconn's have 2.0 but as far as i know all 3-series chipsets are 1.0 or 1.1
 
CamaroZ28 said:
Ok, I agree, that is good advice. Would you recommend 64bit or 32bit as I have both flavors of Professional.
Click to expand...

I suspect support would be best for 32 bit, but I don't have the experience to really comment on that.

I must say, I'd be surprised if the CPU made that much of a difference, but maybe that cache is dragging you down.
 
I wouldn't put too much faith in the new driver release, although I wish you the best of luck. Just so you know you aren't alone out there:

I've been having similar problems on my Gigabyte 4870. FPS in Half-Life 2/TF2 range from 20 to 60, in game... at 1280 by 1024 (!!) .. AA settings don't noticeably change the in game experience.

Lost Coast stress tests avg fps:
1280 by 1024, 6x MSAA (first run): 59.40 fps
1280 by 1024, no AA (first run): 40.20 fps
1280 by 1024, 6x MSAA (second run): 27.04 fps
1280 by 1024, no AA (second run): 46.14 fps
1280 by 1024, 6x MSAA (third run): 57.35 fps
Don't think it ever topped 60 fps..

I'm a little surprised? Both at the low numbers and the spread, and no difference between AA/non-AA

I did this with the new drivers on ATI's site after driver cleaning... In-game results were similar with 8.7 beta and 8.6 hotfix although i didn't run the lost coast test.

Specs: (brand new system)
Vista Home Premium 64 bit SP1
Intel Q9450 @ 2.66, stock cooler
Gigabyte GA-X48-DS4 rev 01
Gigabyte Radeon 4870
G.Skill 2 x 2GB DDR2 1066... 5-5-5-15 @ 2.1V
Western Digital Caviar SE16 WD6400AAKS 640GB
Corsair 750TX

I've tried both the hotfix as well as the beta drivers (after driver cleaning). I'll report back after I try XP, swapping the card into another computer, doing a crysis benchmark, and probably running prime95 again... for what it's worth, memtest runs fine.

Any ideas are appreciated of course, but try to stick with what CamaroZ28 hasn't tried, since it's his thread and all (although of course I might have screwed up on something totally unrelated)
 
It's not the CPU cache, there have been several CPU reviews that show the higher cache usually nets you 5%-10% more performance. On my e2180 @ 3.2Ghz I was getting 160+ fps in Lost Coast stress test.
 
Latest drivers did absolutely crap... I'm just totally perplexed, I checked all those links Danny Bui posted and it does show that less cache does have a signifigant effect on gaming, but not near as signifigant as what I'm experiencing.

Although the Anandtech article did mention how as clock speeds increase the difference gets greater due to being more data hungry, and considering all of those sites were testing at speeds much lower then mine, it could be possible that I see that much of a deficiency at this speed.

More to be determined tomorrow in relation to the cpu since my brother will be loaning his E6750 to me. Also I'm going to try installing 32bit XP on another partition to see how that fans out.
 
safekeeping said:
I wouldn't put too much faith in the new driver release, although I wish you the best of luck. Just so you know you aren't alone out there:

I've been having similar problems on my Gigabyte 4870. FPS in Half-Life 2/TF2 range from 20 to 60, in game... at 1280 by 1024 (!!) .. AA settings don't noticeably change the in game experience.

Lost Coast stress tests avg fps:
1280 by 1024, 6x MSAA (first run): 59.40 fps
1280 by 1024, no AA (first run): 40.20 fps
1280 by 1024, 6x MSAA (second run): 27.04 fps
1280 by 1024, no AA (second run): 46.14 fps
1280 by 1024, 6x MSAA (third run): 57.35 fps
Don't think it ever topped 60 fps..

I'm a little surprised? Both at the low numbers and the spread, and no difference between AA/non-AA

I did this with the new drivers on ATI's site after driver cleaning... In-game results were similar with 8.7 beta and 8.6 hotfix although i didn't run the lost coast test.

Specs: (brand new system)
Vista Home Premium 64 bit SP1
Intel Q9450 @ 2.66, stock cooler
Gigabyte GA-X48-DS4 rev 01
Gigabyte Radeon 4870
G.Skill 2 x 2GB DDR2 1066... 5-5-5-15 @ 2.1V
Western Digital Caviar SE16 WD6400AAKS 640GB
Corsair 750TX

I've tried both the hotfix as well as the beta drivers (after driver cleaning). I'll report back after I try XP, swapping the card into another computer, doing a crysis benchmark, and probably running prime95 again... for what it's worth, memtest runs fine.

Any ideas are appreciated of course, but try to stick with what CamaroZ28 hasn't tried, since it's his thread and all (although of course I might have screwed up on something totally unrelated)
Click to expand...

My setup is the exact same, resolution or AA/AF don't chage the fps one bit, and I got the same results with 8.6-Hotfix, 8.7 Beta, and the official 8.7 drivers.

Things I would definetly recommend for you to try if you have the oppertunity. Try a different graphics card such as 8800gt or the like (I bought a random card at Fry's because they have a 15 day "no questions asked" return policy) on your board to see if you are getting handicapped fps. Also if you have a friend/family member with a pci express comp, try it in their computer and see what you get. This could help you at least determine if the card is at fault here.
 
Yea, I would recommend that too. Use family/friend computer to see if the video card is at fault otherwise go to Frys and get yourself a Diamond 4850(70) and maybe a case fan. Also check out the car audio section, lcd monitors, and 32" lcd tv's if you're there too. This way you can eliminate if it is the graphics card. Maybe get yourself an Nvidia chipset too, they seem to get better FPS than those damn P45 boards. Asus is a must. Maybe consider overclocking your cpu and determine if you need to get a aftermarket HSF. Noctua are nice but pricey. Looks like the Artic Cooler Freezer 7 Pro seems like the best bang for you buck. Other than that, I would pour water all over my system.
 
You must log in or register to reply here.
Back
Top