The Official Graphics Card and PC gaming Thread

Nope, there is a 1GB and a 2GB HD4850X2, so I wanted to make sure you'd get the right one. Sapphire are the only company to produce the 4850X2, it's their product, not ATI's.

A Infinity Ward game developed by EA sums it all up - a good game (infinity ward) that has bugs (EA).

 

Hahahaha. Well put Sam.

I DID NOT KNOW that sapphire made another flavor of the 4850x2 for a few bucks less with not as much memory. Thanks for covering my back. That would be a DEADLY mistake to make. No question about it - the 2 gig version for sure - and max VRAM from now on I should add. LOL.

By the way, since we're on the subject of sapphire, correct me if I am wrong, but I think you own two Sapphire 4870x2 boards, is that right? But even if one or both is a different make, each of your boards is basically the reference ATI single turbine fan design, isn't it? I recall THAT fan is supposed to get rather loud fan under high-rev conditions, yet you said that you could barely hear the fans four feet away from the case. Did you mean in idle or under load?

Are you satisfied with the brand of 4870x2 that you own, from the noise perspective, or would you still rather you had Palits, if two of those would fit on a motherboard (did we ever decide if two Palits would fit on an x48 board with at least one slot separation for breathing?) I'm asking in the remote chance this house sale doesn't die - it's hanging on by a thread lol.

 

Nope, I own one Sapphire 4870X2 and one ATI engineering sample card. Both are almost identical in design though, minor changes to the board layout, and the Sapphire does not have the green OK leds at the front.
I meant under idle, both cards are very quiet. Audible alright, but not behind the idle noise of the Corsair PSU.
Under load they get a bit raucous, but that's to be expected.
From what I hear, the Palits aren't really much different in this regard, though they probably produce a less irritating sort of noise. That said, even if you could fit two into a motherboard, the lack of angled venting means the top card could be starved for airflow and overheat.
I'm eagerly awaiting a review of Arctic Cooling's official product for the 4870X2.
http://www.hardwaresphere.com/2009/02/18/accelero-xtreme-4870x2-cooler-by-arctic-cooling/

Not a cheap cooler though, buying a pair of them for a quad CF config will cost as much as an HD4850.

 

Yeah that's me.

I studied really damn hard. I didn't want to be held back by my weakest subject.

So would I, LOL. I didn't study at all and I choked in some parts. I was too sure of myself. This time around I took some practice tests to acquaint myself with the subject matter. It helped keep my memory fresh.

Palit tries a lot of cool ideas and releases some great cards. Their 8800GT 1GB was rocking. They've done both ATi and Nvidia cards now both with impressive OC results. The coolers are usually really good too.

 

No question about it, practice tests work really well. Very smart thing to have done, Jeff!

LOL I am so used to Estuansis - but I'll slip Jeff in there from time to time - good ol boozer!

Well, we all sort of love the Palit,



but the 3-slot width is a little problematic for considering using two of the beasts.

Sam, that 4870x2 arctic cooler you referenced looks freakin awesome



especially when you flip it over and look at the back - 8 heat pipes - look at the fins.



That brings me back to my Asus 3 fan 4870x2 design that I had read a lot of good reviews about on newegg - despite having zero heatpipes and aluminum fins - basically the 3 fans seemed to be at the sound level of one 120mm fan many reviewers said. Here it is:



You don't get all the heat pipes and you may not get as good of cooling results therefore, but the cost is buried in the competitive board price. You said cost of 4850 - single card - for two of them, so price of about $75 each??

By the way Sam, I had just started to get used to db ratings - I know that 20 db is pretty silent, 30 is somewhat silent but noticeable, 40 is not loud but beginning to make itself heard, etc.

What the hell does 0.5 sones translate into with all three artic cooling fans going at max 2000 rpm? Well I googled it, and wikipedia says 20-30 db, a "very calm room", translates to 0.15 to 0.4 sones, therefore 0.5 sones is slightly above that. "Normal talking within 3 feet" is 40-60 db, or 1 to 4 sones. I guess 0.5 sones is pretty quiet, maybe 35 db.

The arctic is a nice looking product!

Rich

 

How loud is a 120mm fan? How long is a piece of string?
Loud 120mm fans certainly do exist, so that doesn't really tlel us much.
The Sone is a useless measurement, it can't be compared to the decibel scale, and that's how AC get away with not having to post any noise levels of their products.
In general, 20 is ultra-quiet, 25 is quiet, 30 is about normal, 35 a bit on the noisy side, 40 and beyond would be considered loud. The Freezer 7 Pro cooler is about 38dB at full speed, Zalman coolers are about 40dB. The fans in the Antec 900 at max are abut 37dB. The HD4870X2 cooler at full speed is about 48dB.

 

Hmmmm. I had thought that 120mm fans - compared to 80 and 92 - were generally considered quiet. But I do have two of them going at 1600 rpm, and yes, I can hear them. The real noise though, comes from the hsf when the p4 gets hot.

Well, you'd have to glance through the newegg reviews - the overall tone of most of the reviews of the 3-fan asus is that it is fairly quiet - compared to the reference turbine design which you say is at 48 db. I guess some of them had friends with the standard sapphire or ati reference design board.

But for an additional $75, 8 heat pipes looks worth it to me, over the Asus with zero heat pipes, if the arctic performs as well as they say. One could pick up the bottom price design - take off the stock cooler - and pop the arctic on.

But I can see why you're waiting for an actual review of the product if the sone rating is useless.

By the way Sam, two boards producing 48 db each, equates to how many db total (I know it's not 96!) Does that get into the 55-60 db range?

Rich

 

Two boards at 48dB are 51dB, as the decibel scale doubles in noise for a 3dB increase.

 

Too noisy for you, I know that much.

I am having my catalyst nightmares again - everything crashes xp now. So 9.2 is not compatible with my 3850 card - I used driver clean and everything.

I'll take it off and put on an earlier catalyst.

 

Unlucky. 9.2 has proven far more stable for me than 9.1.

 

I think I had 8. something - drivers only - worked great. This 3850 is nothing fancy - I'll just go back to when they had that working smoothly, and try to get catalyst to work this time. The only problem I had before was 3dmark6 - now even just running a quicktime movie to test the osd placement crashes the system. And when I went further into ATT - it couldn't find the hardware for the fan profile - that tells me something right there I suppose.

 

Hmm, I can't really explain that.

 

I was running the drivers only from 8.10. With Catalyst installed I couldn't get 3dmark to run at the time - in November. But now with driver clean, maybe it will work.

I'll try 9.2 one more time, before I give up on it.

Oh, catalyst oh catalyst - you bitch! LOL

 

Heh, don't be quite so hard, amazingly, I haven't had quite as many issues, and I'm using the vista version!

 

Hah, hah. Actually it is appearing to be a problem between Ati tray tools and catalyst. I might need to get over to ati tray tools and get the latest version. This time with 9.2 I left ati tray tools installed. Each time I go into it to exit - system crashes. I haven't tested for fan profiles yet. But that makes me think there is a basic incompatibility there. I use ati tray tools to overclock the 3850 gpu and memory.

 

I had big issues with ATT and Vista64 due to insigned drivers, so I don't use it. I see no reason to use it really, GPU-Z gives me more stats, Catalyst is fine for overclocking and fan control.

 

You might have a point. I never used catalyst - I liked ATT when I learned to use it a few years ago, and I like the on-screen display. But I have to say I haven't given the more modern catalysts a try.

So I am going to remove Ati Tray Tools for now - I'll leave the registry settings just in case. I did find the newest ATT - I'm several revisions behind - and Ray updated it just a few days ago - so it's new as of the CCC 9.2. In uninstalling ATT just now, again the system crashed. So this isn't a fair test of 9.2 - there is some incompatibility with the version of ATT that I had been running.

I'll get ATT completely off, driver clean, put 9.2 on, boot up a couple of times until it stabilizes - takes at least twice to get the font sizes correct - and then see if 3dmark6 runs. Then I'll know catalyst is okay, and then I'll try the newest version of ATT - 1382 (I'm running 1368.)

OK - I have a bad feeling about 9.2. I'm booting up again to make sure it's cleanly installed - but I looked at catalyst just now and the fan slider control is missing - that's the same thing I experienced with ATT - it found no fan under CCC 9.2.

Well, the computer booted, and I tried dxdiag - and it DID find hardware acceleration - the quicktime movie didn't crash - and 3dmark6 is running right now. (There is a little revolving thing over the door that opens later in the attack which I never noticed before. You don't miss anything on the BIG SCREEN lol.) I'll let 3dmark6 finish and see if I get my 4800+ score from November. Well, no, I won't, because that was with graphics hardware acceleration. Maybe I'll get around 4300.

By the way, the CCC talks about a fan slider to set the fan speed - and leave it set at that speed. My CCC doesn't show a fan slider, so I need ATT for that function. I like better how ATT allows not just a manual fan speed setting like that, but you can also set variable fan settings depending on the temperature in 7 discrete steps - as temps rise the fan speed goes up - obviously IMO that's a more clever way to do it. If 3dmark6 works, I'll install the latest version of ATT - if it finds the fan I'm in business - if not I'm dropping back to CCC 8.10.

I need some kind of fan control, the default Ati on-board setting allows the card to run really hot in the 90s, all the time.

Holy smokes, no ATT, no acceleration: 3dmark6 at 5116! (Don't laugh all you 15,000 - 22,000 types!) I guess 9.2 is a better set of drivers.

I just loaded the newest ATT - it found the fan and I have full fan control - it pulled all my settings where I had left them in the registry - one of the uninstall options. I'll implement my auto overclocking and run 3dmark6 again I guess.

I did, it dropped to 4910, with overclocking!! Hmmmm. What the hell?

Not sure what that means but the number is back to the 4850 or so level where it used to be. Maybe the first score was just an anomaly. I think I'll exit ATT, which will stop overclocking, and also stop the on-screen display and background task, and see what the third run produces. Third run with no ATT, stock settings, is 5066.

So acceleration reduces 3dmark6 scores!!! (Or running att reduces 3dmark6 scores.) Makes no sense. Sam or Jeff, you have a theory about that? Shaff doesn't put much stock in 3dmark6 as an indicator of game playing power - maybe all this means is that it gives you a rough indication, but 200 points over 5,000 is 4% - so 3dmark6 is a good system power indicator, plus or minus 5%. Maybe that's what it all means.

Finally, I am turning off att, then using catalyst to increase the gpu clock from stock 668 to max 720 (vs 769-att at my most conservative and very stable overclock profile setting) and increase memory from stock 840 to max 950 (vs 945-att on my stable overclocking profile). This will give me some CCC acceleration - not as much as with att.

THE RESULTS ARE IN: The new CCC-only accelarated score is: 5086, and another run with CCC stock settings, no acceleration, produces 5066. (So the 5166 of the very first run is gone. Maybe some background tasks have kicked in. I haven't used End-it-all which punkbuster told me about, to kill all the killable threads.) Trying once more with ATT turned on and normal acceleration, drops me to 4858.

So, running ATT, accelerated or not, costs me 200 points. What is going on? I thought Ati Tray Tool didn't use virtually any system resources. Well, maybe with my weak p4, I am affected more than normal core 2 duo or quad cpus.

If I take the scores literally - if 3dmark6 is spot-on indicative of real-world, then I am better off not accelerating the gpu. But obviously that can't be correct! I ran 3dmark6 several more times, same thing. Consistently about 200 points less when ATT is loaded and running, and with ATT accelerated versus non-accelerated, very slight difference in score - less than 1% difference.

My conclusion is that 3dmark6 is plus or minus 5% a good indicator of game-playing performance - it's a rough tool - you can't worry about 5%. Anybody have some differing opinions?

DESPITE THE WEIRD 3DMARK6 SCORES, AND NO FAN SLIDER, I LIKE CATALYST!
I'm glad I installed 9.2 - After all that trouble and multiple 3dmark6 runs, I decided to play some more Medal of Honor, with proper aspect ratio.

I scaled up Pacific Assault for a couple hours at 4:3 full screen height, with black borders on each side. As I thought, totally immersive still and you don't even notice the borders. Everybody got back to full height, lol.

Sam, thanks for the tip about the aspect ratio. CCC works very simply and cleanly - you can alt Tab to CCC, change it - alt Tab back - and see the difference immediately. I also tried the letterbox at true 1600x1200 1:1 pixel matching - not bad at all - very crisp of course - but full height looked excellent. Everything got slightly taller and looked really good.

For the time being I have disabled overclocking the gpu - the 3D profile is the same as the 2D profile - but I can change that back instantly in ATT. I have 3 overclocking profiles, but the most conservative is very stable - and overclocks 8% more than CCC allows, 769 on clocks whereas CCC maxes at 720.

For games that tax the gpu at near 100% (per the Ati Tray Tool onscreen display which shows me fps, cpu %load, gpu %load, as well as gpu temp and fan speed) I'll use the 3d accelerated profile, despite what 3dmark6 is telling me, and see what that does to gpu %load utilization.

Rich

 

I can't explain why the fan control slider goes missing, but you don't need to use ATT for fan control, RivaTuner will do it, and as far as I'm concerned, will do a better job of it.
I use Rivatuner primarily because it is compatible with Vista64, ATI Tray Tools is not.
Also don't forget Auto fan speed is designed to run graphics cards up to about 90ºC. That's perfectly normal, and under normal circumstances will not cause crashes.
As far as 3dmark dropping with an OC goes, that's very common, 3dmark scores do slowly degrade if you run them several times in succession, and there is also significat variance between the tests. You should also be closing every background application before runing the test.
With a P4, a program as simple as TrayTools will significantly hurt your CPU score.

 

I just got my evga gtx 260 and so far it is awesome. I have it overclocked to 680/1500/2120 and it gets 16802 in 3dmark 06. But I have not been able to get ati tool to work so i can get the temp. I am running windows server 2008 64bit and i am having trouble getting the driver it needs to load. So what program do you guys use to see gpu temp in 64bit windows?

 

GPU-Z.