• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

4850 Performance in HL2

C

CamaroZ28

Gawd
Joined
Jul 10, 2008
Messages
740
Hey guys, I just got my Gigabyte HD4850 a couple of days ago and i've been very disappointed in the performance I am getting in HL2 and mods. This is pretty much the primary game that I play so I was really hoping to play maxed out and smooth with this card.

Info: running Catalyst 8.7 Beta and all other hardware in sig desktop

First thing I did was run the stress test in Lost Coast on max everything including AA, AF @ 1024x768 (because i was using a crappy 15" lcd for the time being). I got about 85fps. I found this website where someone ran the exact same test, but at 1680x1050 and with Catalyst 8.6 and got 164.16fps. (His hardware is listed on the page under the HL2 Lost Coast picture).

Also my brother has a 8800gt on a 680i mobo with an E6750 and he was getting >200fps (forgot the exact number) on the same stress test at exact same settings/resolution.

Now from what my friend told me, HL2 is extremely CPU intensive compared to other games and he thinks my CPU might bottleneck it, which makes sense since the other cpus are much more powerful + more cache then mine. Also was able to watch gpuz during a stress test and it showed the load on my gpu to be 0% at times, never going above 25% load. Sometimes it would even down-clock to 500MHz core and 750MHz memory as if it was idling. So assuming gpuz is accurate, that means that the gpu wasn't even trying...

So basically I'm posting to see what you guys think because i know a good bit about computer hardware and how it works, but not very much on how specific games utilize that hardware and I'm trying to figure out if there is something wrong with my gpu or drivers (have already tried Catalyst 8.6 and the hotfix but got much worse performance) or whether its my other hardware. Bought this card because everywhere i read about it, it out performs the 8800gt but so far has got significantly less performance then my bro's 8800gt.

PS. This is my first post, but i've been stalking the forums for a while checking out deals and other hardware info/reviews here.

Thanks guys
 
dont run beta drivers run the hot fix drivers. First uninstall old drivers then Run a driver cleaner program and install hot fix drivers.
 
yes run the 8.6 hotfix driver.plus overclock your cpu more u should get to about 2.8-3.0ghz
 
8.7 beta work fine, running them myself and many people are saying the 8.7 run better then the 8.6+hotfix.
 
MrGuvernment said:
8.7 beta work fine, running them myself and many people are saying the 8.7 run better then the 8.6+hotfix.
Click to expand...

Yes, I said in my post that 8.6+hotfix gave me worse results then 8.7 beta.

stonewalker said:
yes run the 8.6 hotfix driver.plus overclock your cpu more u should get to about 2.8-3.0ghz
Click to expand...

And I would love to do that because i know the cpu could easily handle 3.0GHz but the limiter is my memory because my motherboard doesn't allow me to adjust the fsb-dram ratio, so overclocking my cpu force overclocks my memory, and its unstable past what it is currently at.
 
did you run a driver cleaner/sweeper??

Some people have had alot of problems with the beta drivers, some none at all. Am guessing its a mixed bag. To those who had no problems with the betas, did u run a game that uses the HL2 engine?

EDIT-

Its possible its your processor it is a little undersized (cache and core type) but it comes more in affect at higher rez, what rez do u game @.

Also note the architecture of a x2 is 2 years older then that of the Allendale core that you are running now.

EDIT-

One more thing how hot is the GPU running on idle and load
 
Spaxspore said:
did you run a driver cleaner/sweeper??

EDIT-

Its possible its your processor it is a little undersized (cache and core type) but it comes more in affect at higher rez, what rez do u game @.

Also note the architecture of a x2 is 2 years older then that of the Allendale core that you are running now.

EDIT-

One more thing how hot is the GPU running on idle and load
Click to expand...

The gpu at idle is at 45 degrees C and during game never goes above 55 degrees C but mostly stays around 50C
 
yes most definetly. I made two profiles one "Normal" and one "Gaming" the first profile sets the fan speed to 50% and with my desktop under my very thick wooded desk, i can't hear it all. The gaming profile is set to 75% fan speed, read about someone making their fan bearing buzz at 100% so i'm trying to stay away from that. Also once i get these performance issues sorted out i plan on overclocking a lot under my gaming profile.
 
Spaxspore said:
did you run a driver cleaner/sweeper??

Some people have had alot of problems with the beta drivers, some none at all. Am guessing its a mixed bag. To those who had no problems with the betas, did u run a game that uses the HL2 engine?

EDIT-

Its possible its your processor it is a little undersized (cache and core type) but it comes more in affect at higher rez, what rez do u game @.

Also note the architecture of a x2 is 2 years older then that of the Allendale core that you are running now.

EDIT-

One more thing how hot is the GPU running on idle and load
Click to expand...

Sorry, forgot to mention that I was running tests on a crappy 1024x768 monitor because was waiting for my new monitor to ship. But i plan to be gaming at 1680x1050.
 
Thanks, i'll use that tonight and run the 8.6 hotfix drivers and post as soon as i get some results.

EDIT:

However if you read in my first post it says that GPUZ is reporting that during the stress test my video card is hardly even taxed, running between 0-25% utilization and sometimes even down-clocking to 500MHz core, 750MHz Mem. What do you think this means, is GPUZ inaccurate, is it drivers, or could other hardware not be getting info to the card fast enough?
 
CamaroZ28 said:
Hey guys, I just got my Gigabyte HD4850 a couple of days ago and i've been very disappointed in the performance I am getting in HL2 and mods. This is pretty much the primary game that I play so I was really hoping to play maxed out and smooth with this card.

Info: running Catalyst 8.7 Beta and all other hardware in sig desktop

First thing I did was run the stress test in Lost Coast on max everything including AA, AF @ 1024x768 (because i was using a crappy 15" lcd for the time being). I got about 85fps. I found this website where someone ran the exact same test, but at 1680x1050 and with Catalyst 8.6 and got 164.16fps. (His hardware is listed on the page under the HL2 Lost Coast picture).

Also my brother has a 8800gt on a 680i mobo with an E6750 and he was getting >200fps (forgot the exact number) on the same stress test at exact same settings/resolution.

Now from what my friend told me, HL2 is extremely CPU intensive compared to other games and he thinks my CPU might bottleneck it, which makes sense since the other cpus are much more powerful + more cache then mine. Also was able to watch gpuz during a stress test and it showed the load on my gpu to be 0% at times, never going above 25% load. Sometimes it would even down-clock to 500MHz core and 750MHz memory as if it was idling. So assuming gpuz is accurate, that means that the gpu wasn't even trying...

So basically I'm posting to see what you guys think because i know a good bit about computer hardware and how it works, but not very much on how specific games utilize that hardware and I'm trying to figure out if there is something wrong with my gpu or drivers (have already tried Catalyst 8.6 and the hotfix but got much worse performance) or whether its my other hardware. Bought this card because everywhere i read about it, it out performs the 8800gt but so far has got significantly less performance then my bro's 8800gt.

PS. This is my first post, but i've been stalking the forums for a while checking out deals and other hardware info/reviews here.

Thanks guys
Click to expand...

An E2160 will definetly bottleneck you in Source games since they're so CPU dependent. The problem with that processor is that it has a very small 1mb cache compared to the normal 3-6mb found in most C2Ds, and that can have a negative effect on gaming performance. If you have a chance, try to upgrade it to something like the E7200 ($130 at newegg).
 
wolfdale core kicks ass. Thats what i would recommend.

But what am still puzzled about is this game was written in 04... 4 years ago. Even with only 1mb cache it should be fine.
 
orangeble said:
An E2160 will definetly bottleneck you in Source games since they're so CPU dependent. The problem with that processor is that it has a very small 1mb cache compared to the normal 3-6mb found in most C2Ds, and that can have a negative effect on gaming performance. If you have a chance, try to upgrade it to something like the E7200 ($130 at newegg).
Click to expand...

That's what i was suspecting myself, I would love to upgrade to an E8x00 but I was trying to hold out on my current processor until nehalem came out, expecting a huge price drop on C2D's.
 
For instance, look at this review.

http://www.techarp.com/showarticle.aspx?artno=339&pgno=5


They reviewed a 7900 GT w/ a 3200 AMD proc SINGLE CORE @ 2.2ghz-
512kb CACHE
Processor

AMD Athlon 64 3200+
- Socket 939 Venice core
- Overclocked to 2.7 GHz
Motherboard

DFI NF4 SLI-DR
- AMD NF4 6.70 driver
Memory

Two 512MB GSkill LE Memory Modules
- 2.5-4-4-8 timings
- 300MHz memory speed
Graphics Card

XFX GeForce 7900 GT Extreme Edition
NVIDIA GeForce 7800 GT
Sound Card

Creative SoundBlaster X-Fi Platinum
Hard Drives

Hitachi 160GB Deskstar T7K250 SATA 3Gb/s
Power Supply Unit

OCZ Powerstream 520W PSU

@1600x1200
hl2_video_stress.gif



Camaro should be running like a bat out of hell...


Still dont think its his proc...

DID you do a clean install of windows before u put the new card in. And what did u have before the 4850. Was it nvidia?

I
 
I just finished playing Half Life 2 episode 2 without a problem. I had everything set on high on 1920x1200 with the 8.6 hotfix driver.

amd opteron 165 (stock)
1gb mem
hd4850
 
At lower resolutions (below 1280x1024), many games are mostly CPU limited. Start cranking up the resolution and you'll become more and more GPU limited/dependent. Once you start playing above 1280x1024, the CPU doesn't matter too much. So basically it's a combination of your CPU and resolution. You're gaming at 1024x768 which means that you're CPU limited. So either get a bigger monitor or CPU if you don't want to be CPU limited.

Oh and 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
 
Idk I get about 223Fps avg with my 4850. I have everything maxed 16af 8aa @ 1680x1050. I just use the drivers supplied within the cd and that works well. Try ocing the card though since I got mine running at 675/1025.
 
WHOA WHOA WHOA!!! What is wrong with you? Bottleneck of Source games on a E2000 Series? Are you crazy? Dude I run my Socket A AMD 1800+ 1.8Ghz with my X800XT MAX settings on the Source Engine and get a solid 60FPS with 100Avg. FPS going in! My old Athlon is only running a 512k L2 cache. With my 4850 I get 200+ FPS, but I run an LCD so I get 60FPS constant. Which is double more then enough, since the eye only sees at 30FPS. So WTF do you need to spend 130$ on a E7300 for? Worth the money? Yes, but it looks like you need to spend 150$ on a newer MoBo so you can OC that processor you got. The problem with your card, is more then likely user error, or some funked up card. The ATi 4850 ROCKS/PWN/OWN/1337 PWNs nVidia 8 series cards.
 
Thank you! Finally some one with a memory. HALFLIFE 2 is an OLD GAME! You can run it well on a 4yrs old BOX. Thats what ive been trying to say for the PAST 6 posts.....
 
Spaxspore said:
Camaro should be running like a bat out of hell...


Still dont think its his proc...

DID you do a clean install of windows before u put the new card in. And what did u have before the 4850. Was it nvidia?

I
Click to expand...

Thats what I was thinking. I was able to max HL2 on my old system.
 
Spaxspore said:
Thank you! Finally some one with a memory. HALFLIFE 2 is an OLD GAME! You can run it well on a 4yrs old BOX. Thats what ive been trying to say for the PAST 6 posts.....
Click to expand...

Wow! QFT. I was thinking people were going crazy here. Considering I was able to run HL2 at 1280x1024 max settings with about 60-70FPS average on my old machine with a Pentium 4 3.0E and X850XT, just about any Core 2 Duo (or even Core Duo, lol) will be able to handle insane framerates with that game, no problem. There is clearly something wrong with the OP's computer if my old computer was getting those framerates. 1024x768 at 100+ FPS should be no problem for the 3850, never mind the 4850... hell, the X1800XT probably is able to do that, as my X850XT could almost get frames that high.

Nuby1Canuby said:
It sounds like vsync is turned on. Is there a maximum fps cap by default in Half Life 2?
Click to expand...

No, there isn't a maximum fps cap. But the OP would have to have a pretty weird LCD if it capped it at 90FPS with vsync...
 
Well really, Spaxspore that's pretty much the thing he should do. Just start from new, get rid of all the hassles and do it right. Especially if he moved from an nVidia driver set to an ATi catelyst. Dude Counter-Strike Source (my game) I run on my old GeForce3 Ti500. With yes, not great, but frames none the less! 20+ So to imagin your card running the game at the performance of a 9800XT sounds like a driver mess as Spaxspore was saying in a previous post. Yes, go home sweep the drivers, if that doesn't work, rebuild! MUHAHAHA!!! Okay, that's really not funny, actually that sucks! There's only a few upgrades that I even want to do. Another 4850 and an E8500.
 
Sorry guys, I have been out with family and not able to reply, let me do my best to respond to all the posts since my last post:

First off my system is very freshly reformatted, in fact I installed the HD4850 before I even installed Steam. So I know this is not a software issue in that sense because I bearly have any of my regular programs on here.

No there has been no other graphics card ever installed in this rig, was previously on G33 integrated graphics because I used the 7600 in my laptop for gaming.

I do not normally run 1024x768 resolution, I have a 22" 1680x1050 and that is the resolution I will be playing at, I just didn't have that monitor when I first got the card so I was testing on a crappy 15inch thing.

Right now I am going to move down to Catalyst 8.6 + Hotfix and see what happens there. I'll post back after I run some tests.
 
Ok no go here... I just attempted to Install the Catalyst 8.6 + Hotfix drivers and tried like 3 times and i got this exact error message every time "Driver Install: the INF file was not found" :mad:

So for now i'm going back to Catalyst 8.7 Beta with a driver sweep unless anyone has any idea what I am doing wrong... if anything.

EDIT:
Even while uninstalling what nonsense was on there, I just got some random error, I couldn't fully read it because it closed itself out quickly to keep uninstalling.

EDIT:
Here is a sample image of just standing still outside the first level in HL2. This can't be right for this card. Settings are max everything, AA, and AF @ 1680x1050. You can see the fps at thet top center of image.

hl2squarefpsxi3.jpg
 
Max everything at 49 FPS? Maybe its just me, but I don't see a problem. Why don't you trying settings that you see in reviews and see if they match up.
 
heatlesssun said:
Max everything at 49 FPS? Maybe its just me, but I don't see a problem. Why don't you trying settings that you see in reviews and see if they match up.
Click to expand...

Nah, its way too low.

I cant test his exact settings, but lets assume that a 4850 @ 1680x1050 is roughly equal to a 4870 @ 1920x1200.

I tried the exact same scene and got 200+ fps. And thats with everything maxed out + adaptive aa + edge detect.

So yeah. Something is definitely wrong with his setup.
 
lol I did some testing, forgot the exact numbers so I'll make them up.

Conroe Celron 420 1.8ghz = 56fps

Conroe Celeron 420 2.6ghz = 80fps

E4300 1.8ghz = 73fps

E4300 3.2ghz = 136

Q6600 3.6ghz = 179



You get the point, the source engine, especially playing online with a lot of people, LOVES high clocked cpu's.

Your low cache dual core + lack of overclocking is why the difference between the two pc is so drastic.
 
orangeble said:
An E2160 will definetly bottleneck you in Source games since they're so CPU dependent. The problem with that processor is that it has a very small 1mb cache compared to the normal 3-6mb found in most C2Ds, and that can have a negative effect on gaming performance. If you have a chance, try to upgrade it to something like the E7200 ($130 at newegg).
Click to expand...

There is no way that processor is bottlenecking that game. I play HL2 on my laptop that has a dothan 1.86 GHz processor and a neutered geforce 6600, and I play at 50-80 fps in the original HL2 with no aa and high settings. Lost coast can definitely tax the system, but by no means should it be causing gameplay issues with his setup. Something is clearly wrong here, just not sure what.
 
aZn_plyR said:
lol I did some testing, forgot the exact numbers so I'll make them up.

Conroe Celron 420 1.8ghz = 56fps

Conroe Celeron 420 2.6ghz = 80fps

E4300 1.8ghz = 73fps

E4300 3.2ghz = 136

Q6600 3.6ghz = 179



You get the point, the source engine, especially playing online with a lot of people, LOVES high clocked cpu's.

Your low cache dual core + lack of overclocking is why the difference between the two pc is so drastic.
Click to expand...

How badly were you making these numbers up or are they pretty close to legit?

I have been getting mixed answers between people saying "cpu is definetly bottleneck" and "cpu is definetly NOT bottleneck". If anyone else has some evidence such as this that may suggest one or the other I would like to see it. So far it really seems like my processor but I'm not quite convinced yet.

Another thing to consider: Is perhaps my chipset (G33) and lack of PCI Express 2.0 support draining some power out of the system?
 
Okay well I downclocked my e6850 to 2.4 to match your cpu (was at 3.6 before). And ran the same test as before. Fps went down to ~150 (and loading times got A LOT longer). I think this makes it pretty obvious that your cpu is NOT the bottleneck.

And neither is pci-e 2.0 since I dont have it either.
 
Here is a screen shot of HL2 Lost Coast stress test. I used exactly the same settings as seen on this page towards the bottom.

The only differences between our setups or settings are:

E8400 @ 4GHz vs. E2160 @ 2.4GHz
X38 vs. G33
4GB @ 1000MHz vs. 2GB @ 888MHz
Catalyst 8.6 vs. Catalyst 8.7 Beta

The website shows with that setup 164.16fps vs. my 76.97fps:
lostcoaststresstestui7.jpg


EDIT:
I'm going to bed now, so I'll be back on to check replies tomorrow morning around 9am EST.
 
I'm too lazy to read the rest of the thread but I was curious if you checked if your 4850 PCIe lane is running in 16x?

Also, that comparison you listed above isn't really a true depiction as the user had ran that benchmark in 1680x1050.
 
SolidBladez said:
I'm too lazy to read the rest of the thread but I was curious if you checked if your 4850 PCIe lane is running in 16x?

Also, that comparison you listed above isn't really a true depiction as the user had ran that benchmark in 1680x1050.
Click to expand...

First off i'm pretty darn positive that its in 16x because I only have 1 pci express x16 slot (being a micro board) and i'm pretty sure it defaults to 16x. I'll double check.

EDIT:
Yes, slot is definetly in 16x, here is screeny:
gpuztg5.jpg


BTW, for some reason GPUZ acts funny when I use the Catalyst 8.7 Beta drivers and reads the wrong clocks for core and memory. Under CCC it reads 500 core and 750 memory. So don't start posting thinking that my problem is that my card is downclocked to 1MHz memory clock :p
END EDIT

Second, I stressed that all my stuff was exactly the same, so yes my test was run at 1680x1050. Besides, if i was running a lower resolution my fps should only be much higher then that benchmark.
 
You must log in or register to reply here.
Back
Top