Post your 3dmark06 HD 3K Series Results

[freaksavior]

woot im top of dual core crossfireX woot

and the only one

 

[cdawall]

woot im top of dual core crossfireX woot

like hell you are

http://service.futuremark.com/resultComparison.action?compareResultId=6687940&compareResultType=14

thats an AX2 5000BE @3.2ghz and xfired 3850 256mb @674/832

 

[erocker]

Well, this thread is specifically for 3dMark 06 scores. You are more than welcome to post them in a different thread though.

 

[MadClown]

HD 3870, 3dmark06 default settings

 

[Zax85]

Hmmm..

Think I can break 11k on air?

I'm sure it will be possible once we get some even better drivers..

14207 easy!...

 

[Wile E]

·
why ?

CPU speed.

 

[HELLSPAWNPR]

my score

 

[Dr. Spankenstein]

I could give some Riva screenshots later. What are you trying to determine?

 

[trt740]

not too bad nice bench

 

[Dr. Spankenstein]

whether or not crossfire/SLI sucks compared to one-card of same type in real games instead of benchmarks

CF/SLI sucking is totally subjective.

Either you WANT to pay for a second card or you DON'T.
I (personally) appreciate everything my CF setup has done for me. Sure it doesn't scale well in a FEW games and doesn't scale at all in fewer, but I'll take it when I can get it...

 

[Dr. Spankenstein]

Screenies as promised:

I have provided the graph that is created by the secondary card. Nothig lower that 53% useage on 2nd card, the rest is pretty self-explanitory.

Here is Call of Juarez @ 1920x1200, AF, all others "HIGH":

Sucks....?

Methinks not. Especially for DX10 games.

 

[sinister_steve]

3d Mark Score

Heres mine...I like the ofuture mark on the web site as you knoe no one can change settings and show fulse readings..Steve

http://service.futuremark.com/home.action?resultId=184115&resultType=22

 

[sinister_steve]

thanks, btw a 3000 mhz phenom with a hd4850 15k is good ? I heard the phenom's suck at anything that isn't multithreaded is that true ?

Hey man..No the phenoms are sweet as hell..they are fast for everything I can throw at it..and they OC damn goot to ..jUST TONIGHT i HAD IT AT 3.6GHZ AND STILL NEVER CRASHED..So just letting you know these 9850 black editions are very very fast...I read out there about them beeing slow a lot ,and I sortof thought so myself untill I oc a bit and then look at her rip it up now...Steve

 

[Dr. Spankenstein]

I'm still on stock coolers. During those tests I only had the fan set to 70%, always 1.6v and clocked @ 875 core for both. Ambient is more like 25C.

Do you have more than 1.6v going to your cards? I have a sneeking suspicion that your water cooling isn't up to the task. You've just got more heat going than can be transferred out of the loop.

The higher temps during games can be attributed to particular settings in BOTH CCC and in-game.

Try this as an example: Run a timedemo (or something that is repeatable) and record the results of temps while only changing MipMap detail and AF to MAX and MIN. The artifact scanner does not have any provision to change quality settings so if your settings in CCC are "Use game defaults" it will be on the lowest quality UNLESS you "force" these settings in CCC.

 

[budbik]

I thought I was getting low scores but this thread was a great tool for comparison
My temps are extremely hot for stock cpu speeds I need to upgrade the stock cpu cooler for sure

 

[Dr. Spankenstein]

do not have CCC, but I can modify the necessary settings in the registry, what are you sayin that the higher are the settings (AA/AF, mipmap) the higher is the temperature ? yes my registry parameters are at default

That is exactly what I'm saying. I'm guessing you are using "HIGH" settings in-game. That would account for the higher temps.

the voltage appears to be dropping over time, right now it has reached about 1.6 yhe, not much higher even if it would be any higher, do you think the temperature is something to be concerned about ?

I don't know where people have gotten the idea that these GPUs cam operate at 90C without issues. If anything (in my experience), the smaller fabrication seems to be less tolerant of temps. I'm not positive, but I think your leevated temps are what is keeping your card from reaching 1GHz stable @ 1.6v.

what you meen transferred out of the loop ? as in radiators or as in vgablock ?
idle temp is around 50, I do think that there may be some other parameter that doesn't show up, which support the possibility of a radiator/fan problem, the heat took a long time to reach 87 C, about 30 minutes, in the past I saw water cooling systems from other people and they had "heat buildups", regardless of the quality of the radiators, they just didn't have enough of them, usually those were people with powerful pumps with too little amount of radiators like only 120mm (heat increases faster because water moving faster), there is some dirt in the radiators, maybe I should clean them, didn't clean them up for like 6 months, how does it look when it is a block problem ? faster increase ? because i had a pipe problem a long time ago which caused temporary increase in temperature and then it had to throttle and then the water kicked in and reduced to 60 C and then back to 110 C, but this is constant, im gonna check the entire system make sure all is ok, thanks

it appears to be some kind of bug in sensors, when started something the game first and then riva the temp was no higher than 65 C after I quit the game and started it again, it shows 80 C and starts to increase slowly, as if there is some kind of bug in the sensors

appears to be in rivatuner or it's statistics server: switch to game, switch to riva, switch some process, swtich back to game and temperature changes in the same place

it appears to be also in gpu-z sensors, seems like a hardware sensors problem in the card
still, it looks like there was a lote of dust in the radiators, I cleaned it all up, and it seems the cpu temps with two prime95's reduced from 58-59 to 51-52 @ 12V, and there seems to be less noise from the radiator fans, usually at that speed they were loud

Kinda curious that both software would be having issues reporting temps, don't you think?
I was referring to your ENTIRE watercooling system. Each part is integral to the sytem's ability to effectively remove heat from the loop. The blocks must be of good design to transfer heat to the water, the pump has to move it through the tubing and radiator at a rate that works with the radiator's (and fan's) ability to remove the heat from the water. One weak link in that chain makes the whole setup poor. Your radiator that was clogged with dust was a big one. No need to "insulate" the fins!

With the kind of comp you are trying to run (and overclock), I still suggest you move to more "high end" watercooling components.

 

[Wile E]

yes, both software are reporting the same temperature indication problems, which means there is some kind of problem with the sensor itself or the data it sends, but there is no way to play a whole game for 30 minutes, switch to other process (2D clocks), switch back and see a temp that is 20-30 degrees higher than it was the entire game and then only to switch again and see the temps go back the way they were, did anyone experience any of these funny temperature behaviour ?

upgrading the watercooling system beyond what it currently is, is a waste of money, water has a limit and improving it is small difference at best, if im going to invest in cooling beyond what I already have, im gonna go xtreme, which I won't unless it was worth.

now that I have checked the entire system, it now operates at full performance, which leads me to believe that the temperature is 65 C maximum and no higher, because that is the first temperature that is reported before a process switch occurs, and it is also the only one that makes sense (at 85+ C there is gpu throttling, which does not occur, otherwise would be visible in-game) in whitch case you think the cooling is good ? I have tried to use the DMM thermometer on the waterblock but it wouldn't be accurate cause it's too far away from the actual chip.

I think that if I can't get reliable temperature information there is no way to know what the problem is, or whether or not there is a problem at all, except the gpu throttling, which so far I never seen. my only indication is your temp, if your temp is 78 C with stock cooling and no voltmod at 25 C room temp, and we know that water cooling limits the difference between highest and lowest temperature (water load-idle < air load-idle), assuming a 35 C room temp would put it about 88 C, using water cooling it reduces that max temp at load to about half of that (as far as I can remember since I had the stock cooling), assuming 44 C, add voltmod get +10 C which puts it about 54 C, but still if anyone could comment on temperatures of 3870 with waterblock and room temp could provide for a more accurate indication of the estimate.

about the 1000 mhz stable, it's possible that either room temp (although I do have 2x80 mm fans on the card at bottom) or I have the earlier design with the resistor problems they had at production (I bought it very early, the same day it was available actually) or like with CPU I just have a low overclockability problem as with any other chip (not all chips can clock the same) which so far leads me to believe that this is the case

Actually, the Bigwater 745 is not a very good water cooling system. Thermaltake makes poor water cooling systems. Sorry to put it so bluntly.

You would see a big difference by going with better water cooling parts. I bet just replacing your pump with a DDC-1 would make a noticeable difference.

 

[nINJAkECIL]

Crossfire HD3870, HIS HD3870 IceQ3 and Gecube HD3870 Xturbo III. all stock. using E4300@3,6ghz, and P965 chipset >.< Abit AB9-QuadGT:

 

[boogah]

wait y didn't you crossfire them

 

[nINJAkECIL]

wait y didn't you crossfire them

yeah, I Crossfire them. I don't know about the 'disabled but available' on GPU-Z, but indeed I check at the Catalyst Control Center and Crossfire already active. but then again, using AB9 QuadGT, I only got x16 and x4 PCI-E bandwidht

 

[Evo85]

I got 9105 last night with 800/1000 Win XP SP3

 

[Dr. Spankenstein]

yes, both software are reporting the same temperature indication problems, which means there is some kind of problem with the sensor itself or the data it sends, but there is no way to play a whole game for 30 minutes, switch to other process (2D clocks), switch back and see a temp that is 20-30 degrees higher than it was the entire game and then only to switch again and see the temps go back the way they were, did anyone experience any of these funny temperature behaviour ?

You're right. That does appear odd. If NOTHING changes between the differing temp readings (you don't kick down the fans on the rad) then there does appear to be some issue (that I've never heard of). I don't know how you have your overclock on you CPU, but sometimes enabling a "Loadline calibration" function can deliver significantly MORE volts to the CPU at Idle that it does under load (this stating beyond vDroop).

I, like Wile E, believe that your watercooling is being overwhelmed.

upgrading the watercooling system beyond what it currently is, is a waste of money, water has a limit and improving it is small difference at best, if im going to invest in cooling beyond what I already have, im gonna go xtreme, which I won't unless it was worth.

This is a misconception you need to let go. You have not reached the limits of watercooling, you have reached the limits of YOUR watercooling system. There are reasons that radiators are rated by the amount of heat (usually listed as BTU-British Thermal Units) that it can dissapate per hour. The effectivness of the radiator, it's fans, the flow within the loop, the efficiency of the water block AND ambient temperatures are the cornerstones of how effective your watercooling will be.

now that I have checked the entire system, it now operates at full performance, which leads me to believe that the temperature is 65 C maximum and no higher, because that is the first temperature that is reported before a process switch occurs, and it is also the only one that makes sense (at 85+ C there is gpu throttling, which does not occur, otherwise would be visible in-game) in whitch case you think the cooling is good ? I have tried to use the DMM thermometer on the waterblock but it wouldn't be accurate cause it's too far away from the actual chip.

This is a lot of conjecture. Where have you EVER read that these 55nm cards throttle @ 85C? And to be safe, don't confuse a DMM (Digital Multimeter) with a infrared thermometer. Taking temps with a DMM wil surely lead to a fried card when you bridge something with you probe. (He, he...I said "probe"!)

I think that if I can't get reliable temperature information there is no way to know what the problem is, or whether or not there is a problem at all, except the gpu throttling, which so far I never seen. my only indication is your temp, if your temp is 78 C with stock cooling and no voltmod at 25 C room temp, and we know that water cooling limits the difference between highest and lowest temperature (water load-idle < air load-idle), assuming a 35 C room temp would put it about 88 C, using water cooling it reduces that max temp at load to about half of that (as far as I can remember since I had the stock cooling), assuming 44 C, add voltmod get +10 C which puts it about 54 C, but still if anyone could comment on temperatures of 3870 with waterblock and room temp could provide for a more accurate indication of the estimate.

As I stated in my post previously, this was with the voltmod. Voltages were @ 1.6V.
You are pretty generous with your mathematical guestimations. Those MIGHT be valid IF you were on a primo watercooled rig. Search some posts from asb2106. He has an awesome setup much like what you are trying to achieve: CF 3870, voltmodded, watercooled with full coverage blocks. That would be a PERFECT resource.

about the 1000 mhz stable, it's possible that either room temp (although I do have 2x80 mm fans on the card at bottom) or I have the earlier design with the resistor problems they had at production (I bought it very early, the same day it was available actually) or like with CPU I just have a low overclockability problem as with any other chip (not all chips can clock the same) which so far leads me to believe that this is the case

I have the same type of cards. If you got them when they first came out, you have what are referred to as "reference" cards, since the design does not deviate from the original design in board layout or cooling.

I would really like to see some photos of your voltage readings and these screwy temp readings (if you would, include CPU-z, GPU-z and whatever probram you are using to overclock the cards.) From there we might be able to determine what exactly is happening with your system.

 

[Clutch442]

Well I'm new around here, Is this where you post your scores?
Also I hope you can read it, I don't know how to do that resize thing.

I see my System Specs didn't show up???

Well I changed that (obvious noobe)

 

[Wile E]

Personally, I would start with the pump. It's a known weak point of the Thermaltake systems. Low flow, and high failure rates. I would get a Laing DDC3.2 (Swiftech MCP-355).

Next I would change the blocks. I would go for a D-Tek Fuzion V2 for the cpu, and probably Dangerden Maze5 block for the gpu(s). I stick to universal gpu waterblocks because you can continue to use them if you upgrade your card. Just make sure you keep the vregs and mem of the card cool. The stock heatspreaders in combination with a couple well placed fans is usually enough to get the job done. Better would be to replace the factory heatspreader with some high quality ramsinks. If you get a full coverage block, not only does it add the heat of the vmem and vregs to the loop, potentially lowering your gpu overclock, you also have to buy a new block if you buy a new card and want to watercool it.

When that was done, I would move on to the rads. Rad choice would depend strictly on my budget. If I had $130 to spend, I would get a Thermochill 120.3 or a Dangerden 480. If I had a $50 or lower budget, I'd look at the Swiftech MCR offerings. Great bang for the buck.

The great thing about water cooling systems is their modularity. You can upgrade them piece by piece as money allows.

 

[boogah]

== Quadcore == == Single HD3870 ==

System Spec:

Q9450 @ 3.6 450x8 1800 FSB
P5E-Deluxe X48
OCZ 4 GB platinum DDR-1066 @ 900Mhz
MSI HD 3870 OC edition 850/1200 on stock fan @ 1.338 V
OCZ Pro X Stream 1000w PSU
Noctua NH-C12P CPU cooler