GTX 970 G1 Gaming SLI Performance issues.

[rtwjunkie]

The 3770K does have about 10% better performance than a 2600K at the same frequency, so that would have to account for some of the GPU variance as well.

 

[mull54]

almost the same? the ones with the 3770k show 2000 more pts on the physx test and 800-1000 pts more on the combined.

That's a variance of 15-20%

the 4770k physx and combined scores are even higher, more like the 23-28% variance range.

On top of that some of those are invalid as I said before because they are playing around with textures in the NV control panel. This makes it so that their scores are higher simply because the detail level is lower than the benchmark design.



Then you have the old disclaimer "results are not guaranteed even if you have the exact same hardware"

3dmark these day does a piss poor job at actually evaluating gaming performance. That's why W1zzard stopped using it in his reviews. What about actual gaming performance? Any stuttering, slowness? Are you able to set details to the highest level at 1080p? or if you have a bigger monitor does performance meet expextations at 1440p/1600p, 4k etc?

I got 53fps in metro last light with 4xssaa.

But i am not talking about the total score. Ofcoruse the Newer cpus have higher physics and combined score. Am still talking about the gpu score. Even if some of them get higher score becuase of cheating. There is clearly a difference between 17k and 20-22k. And two 780ti get 3600(oc maybe) in heaven and i get 2600. With Everything at ultra. Is that normal?

 

[rtwjunkie]

The cpu and how powerful it is will still affect your GPU score. That's what they've been telling you. Images first go to CPU, and processed, then passed to GPU for the rest of the work. So, even aside from the combined score being different, the GPU score will reflect in some part the CPU performance.

In any case, benchmarks aside, what is your real world gaming performance? Are you satisfied? Because that's all that really matters.

 

[mull54]

The cpu and how powerful it is will still affect your GPU score. That's what they've been telling you. Images first go to CPU, and processed, then passed to GPU for the rest of the work. So, even aside from the combined score being different, the GPU score will reflect in some part the CPU performance.

In any case, benchmarks aside, what is your real world gaming performance? Are you satisfied? Because that's all that really matters.

Yeah it does. But what i have trying to say it that the cpu doesnt affect the gpu score with 5-6k. I Think it is a kombination of Everything. But should i keep the 4 pin molex(cf/sli connector) between the motherboard and the psu or should i unplug it?

 

[Lopez0101]

I take 3DMark's results with a grain of salt now too. I can clock my 290 around 60Mhz higher on core in 3DMark than I can in any game. Clocks that work fine in 3DMark cause blue screen, CTDs, hard crashes and artifacting when actually playing a game.

 

[Aquinus]

wait a min. My mobo got a PLX chip:

1. "Integrated PLX PEX8608 chip onboard to offer sufficient PCI-E lanes"

http://www.asrock.com/mb/Intel/P67 Extreme4/

Is that the problem?

Not the PLX chip I was thinking about, I was thinking about one of those 16 to 48 (or 60?) bridge chips, but that doesn't matter as it's only switching lanes off the PCH. Your CPU does matter. You want to improve your score, overclock it bit, but clock for clock, a 2600k isn't going to run as fast as an SB-E, IVB, IVB-E, Haswell, or Haswell-E CPU. What you're probably seeing is difference in IPC between CPU generations as well as improvements to different instructions.

Make sure the GPUs are in the top two most PCI-E slots. The bottom one is limited to 4x (or slower because of the PLX chip) because it's running off the chipset and not off the PCI-E root complex on the CPU like the top two slots are.

Edit: You said you have 8GB of DDR3. How fast is the memory running? If it's DDR3-1066 or 1333, running 1600 could result in a noticable increase in performance. Also in the off chance that memory is running in single-channel mode, that could have an impact as well.

 

[mull54]

Not the PLX chip I was thinking about, I was thinking about one of those 16 to 48 (or 60?) bridge chips, but that doesn't matter as it's only switching lanes off the PCH. Your CPU does matter. You want to improve your score, overclock it bit, but clock for clock, a 2600k isn't going to run as fast as an SB-E, IVB, IVB-E, Haswell, or Haswell-E CPU. What you're probably seeing is difference in IPC between CPU generations as well as improvements to different instructions.

Make sure the GPUs are in the top two most PCI-E slots. The bottom one is limited to 4x (or slower because of the PLX chip) because it's running off the chipset and not off the PCI-E root complex on the CPU like the top two slots are.

Edit: You said you have 8GB of DDR3. How fast is the memory running? If it's DDR3-1066 or 1333, running 1600 could result in a noticable increase in performance. Also in the off chance that memory is running in single-channel mode, that could have an impact as well.

They are running at 1333mhz. I belive i clocked it down becuase of some voltage problems with Sandy bridge/ram.

And singel Channel i Will look Into that.

Btw guys. To those of you that don't belive me. Look at this thread.

http://forums.guru3d.com/showthread.php?t=393069&page=18

People there got the exact same problem as me. Most likely driver problem they say.

 

[Frag_Maniac]

People there got the exact same problem as me. Most likely driver problem they say.

Well I hope this isn't Nvidia trying to level the playing field for their exclusive partner EVGA.

 

[mull54]

Well I hope this isn't Nvidia trying to level the playing field for their exclusive partner EVGA.

Hope not.

I saw that my ASIC Quality was 66% for one card and 74.8 for the other. Isn't this very low. Should I return them for a better pair?

 

[z1tu]

I got 53fps in metro last light with 4xssaa.

But i am not talking about the total score. Ofcoruse the Newer cpus have higher physics and combined score. Am still talking about the gpu score. Even if some of them get higher score becuase of cheating. There is clearly a difference between 17k and 20-22k. And two 780ti get 3600(oc maybe) in heaven and i get 2600. With Everything at ultra. Is that normal?

The new 970 cards score lower in Unigine Heaven than the 780, my friend, who recently purchased a 970, scores lower than me and we have the same cpu, he even has it clocked higher than me with 100 mhz.

 

[eidairaman1]

Pioneer of technology gets arrows in back.

 

[mull54]

Here is one more with SB who got 22k http://www.3dmark.com/3dm/4185861

 

[rtwjunkie]

What I see is it's a 2700k vs your 2600k...not a big difference, but some. I also see his overclock is 4.6, whereas you stated 4.2. That WILL affect GPU performance more. To truly compare, you have to make exact comparisons of all major components. BTW, you said you were going to look into if you are running dual-channel or not. I looked through and didn't see a follow up, unless I missed it.

 

[mull54]

What I see is it's a 2700k vs your 2600k...not a big difference, but some. I also see his overclock is 4.6, whereas you stated 4.2. That WILL affect GPU performance more. To truly compare, you have to make exact comparisons of all major components. BTW, you said you were going to look into if you are running dual-channel or not. I looked through and didn't see a follow up, unless I missed it.

It's a 100MHz difference between 2600 kand 2700k(stock). He also said it was almost no difference between 4.2GHz and 4.6GHz when it comes to GPU score. Yes they where running in Dual-channel. I even clocked them to 1600MHz. With no luck. i have returned the cards and i will probably buy 2 new, when they are in stock just to be sure it was no the cards. Should i still go for the G1 Gaming or should i buy another brand or parhaps 1 GTX 980?

 

[eidairaman1]

I wouldnt have jumped the gun. Stuck with your existing card(s) for 6 months longer then made a move

 

[Lopez0101]

I'd go with the single highest card you're willing to spend money on. I wouldn't bother with SLI of either of those cards if you're not going higher than 1440p.

 

[mull54]

I'd go with the single highest card you're willing to spend money on. I wouldn't bother with SLI of either of those cards if you're not going higher than 1440p.

Yeah just my thinking. But when gtx 970 sli only cost 100$ more than 1 GTX 980, it's really hard to even consider that card.

 

[D007]

Hey guys.

I reviced my two GTX 970 G1 Gaming(Really good looking cards, and damn they run quiet att 80% even) and i have a problem with performance.

First of, i ran a custom fan curve and got 67C on the top card on 80% fan speed. And 60C on the bottom card. Is that normal temps?

But to my main question. I ran Firestrike(free with demo) and only got 17000 in score. I know this is not normal. Do you guys have any idea what the problem is? Also i got a bit over 2600 on Heven bench on 1080P Ultra 8xaa and Extreme tess.

The drivers and performance of this 9 series, is absolute shit so far from what I am seeing.
I get worse performance in arma 3 than i did with my gtx 680's.

I scored well under what 780's scored on Valley benchmark by 1k points. That is ridiculous.
Nvidia needs to get their head out of their ass and start making good drivers again.

I made a forum post on nvidia about the crap performance.

Yeah just my thinking. But when gtx 970 sli only cost 100$ more than 1 GTX 980, it's really hard to even consider that card.

Unless you are doing 4k gaming.
In which case those 970's won't cut it and you want/need the 980's.

 

[rtwjunkie]

With brand new cards, it will be a bit before SLI profiles are working perfectly for them I imagine.

Since you asked though, and already sent them back I suggest the fastest single card you can get, which would currently be the 980.

Personally tho, I would have stuck it out and either figured it out or waited for good drivers for tje SLI setup.

Let us know what you do!

 

[mull54]

With brand new cards, it will be a bit before SLI profiles are working perfectly for them I imagine.

Since you asked though, and already sent them back I suggest the fastest single card you can get, which would currently be the 980.

Personally tho, I would have stuck it out and either figured it out or waited for good drivers for tje SLI setup.

Let us know what you do!

I really cant justify the Price of 1 gtx 980 against two 970. So i Will Most likely order two as fast as possible.

 

[BooStFeD]

I too am experiencing something very similar on a x58 platform Rampage III Formula and a Xeon X5660 at 4.6 GHz. GPU score is insanely low while physics and combined score are fine. GPU score of only 18K while boosting to 1518 core on both cards no throttling. Also my Rampage III Formula supports PCIE 2.0 x16 in slots 1 and 3 for full x16 / x16 configuration. GPU-Z reports card in slot 1 running at PCIE 2.0 x16 but card in slot 3 running at PCIE 2.0 x8 during full screen 3D applications. I have not been able to resolve this, been browsing google like a mad man

Not sure if its a driver or MB issue. Maybe PCIE 2.0 @ x8 is causing a decent bottleneck with this beast. GPU utilization is not always pegged in benchmarks.

 

[Aquinus]

I too am experiencing something very similar on a x58 platform Rampage III Formula and a Xeon X5660 at 4.6 GHz. GPU score is insanely low while physics and combined score are fine. GPU score of only 18K while boosting to 1518 core on both cards no throttling. Also my Rampage III Formula supports PCIE 2.0 x16 in slots 1 and 3 for full x16 / x16 configuration. GPU-Z reports card in slot 1 running at PCIE 2.0 x16 but card in slot 3 running at PCIE 2.0 x8 during full screen 3D applications. I have not been able to resolve this, been browsing google like a mad man

Not sure if its a driver or MB issue. Maybe PCIE 2.0 @ x8 is causing a decent bottleneck with this beast. GPU utilization is not always pegged in benchmarks.

More cores doesn't mean faster. If you search 3DMark scores by GPU and only have dual 970 scores, you'll see that all the 2600k scores are about the same (the the OPs score being almost the highest.) Additionally, you may have your Xeon clocked at 4.6Ghz, but it's also several generation of CPU behind which is going to make it run slower than CPUs that have a faster turnaround time when it comes to the pre-processing the CPU has to do before the GPU gets to work.

But come on. Look at the GPU and physics scores. The cpu score are almost the same as mine and he gets 24k on GPU.

http://www.3dmark.com/fs/2832646
http://www.3dmark.com/fs/2807735
http://www.3dmark.com/fs/2806068
http://www.3dmark.com/fs/2817211
http://www.3dmark.com/fs/2808450
http://www.3dmark.com/fs/2816970
http://www.3dmark.com/fs/2784943

The flaw with your logic is all of these CPUs are newer than yours, there's a 3770k with a similar score but many are 4770ks and a couple 5820Ks. You're comparing apples and oranges. If the OP actually listened in the first place, he would have learned that it was his CPU that was slowing things down. I suspect you (@BooStFeD ) are having a similar issue.

 

[mull54]

More cores doesn't mean faster. If you search 3DMark scores by GPU and only have dual 970 scores, you'll see that all the 2600k scores are about the same (the the OPs score being almost the highest.) Additionally, you may have your Xeon clocked at 4.6Ghz, but it's also several generation of CPU behind which is going to make it run slower than CPUs that have a faster turnaround time when it comes to the pre-processing the CPU has to do before the GPU gets to work.



The flaw with your logic is all of these CPUs are newer than yours, there's a 3770k with a similar score but many are 4770ks and a couple 5820Ks. You're comparing apples and oranges. If the OP actually listened in the first place, he would have learned that it was his CPU that was slowing things down. I suspect you (@BooStFeD ) are having a similar issue.

I have one theory though. The slow memory speed is compensated with Third Generation Delta Color Compression and the PCI-E 2.0 lanes can't fully handle this. So Parhaps PCI-E 20 8X is the bottleneck here . Because everyone here with older MB and CPU doesn't have the faster PCI-E lanes and this is why we are getting worse score.

 

[z1tu]

I have one theory though. The slow memory speed is compensated with Third Generation Delta Color Compression and the PCI-E 2.0 lanes can't fully handle this. So Parhaps PCI-E 20 8X is the bottleneck here . Because everyone here with older MB and CPU doesn't have the faster PCI-E lanes and this is why we are getting worse score.

For me, going from FX-6350 and PCIE 2.0 to i5 4690K and PCIE 3.0 meant about 1K difference in overall score in 3dmark. With the same video card of course. Gaming wise however, huge difference in FPS.

 

[mull54]

For me, going from FX-6350 and PCIE 2.0 to i5 4690K and PCIE 3.0 meant about 1K difference in overall score in 3dmark. With the same video card of course. Gaming wise however, huge difference in FPS.

Yeah overall score will be improved ofcourse. The interesting thing would be seeing if the GPU score itself increased.