gtx 1070, low performance

[Termie]

Do you have benchmarks with your previous card?

Planetside 2 is CPU bottlenecked, The Witcher 3 and Unigine are completely GPU bound, so they should see improvement.

Use MSI Afterburner to check GPU load levels.

 

[SinOfLiberty]

cpu is not a botteneck. period.

I monitored the settings and the fact is gpu usage never goes above 40% unless you downsample to 4K or framerate drops too low(what must not happen). i have 100> fps when I tested csgo, dota bf4 etc...

NCP no result. Clean windows install(most likely wont help although i am testing it now). Motherboard BIOS updated to the latest version, did not help. Previos gpu works perfectly. It is either the cable(hdmi1.3b) which is a cause of the issue or gpu, however, there is a QC sticker on the bottom of the pcb, suggesting that it passed quality control.

Memory test passed.

 

[HeXen]

This thread is a fail. Where are the benchmarks that show the previous card was on par? You should always bench the old right before swapping cards and test the comparison with comparisons of nominal data. Now you'll have a baseline first to confirm everything is balanced to begin with.
How about posting some freaking temps? Maybe some high temps are causing some throttling? How the redacteddo we know if your PC case is a shoebox or not? Maybe not enough PSU wattage, what's the full load wattage output...not the rated but the actual output?
What drivers have been tried? What are full specs? There is more to gaming performance than just the CPU and GPU alone...a lot more actually.

Or just send the card back, drop the attitude and learn to post the data you have already analyzed in the first post including some more benchmarks outside of just Unigine. OP is creating his own headache imo.




No profanity in the tech forums.


esquared
Anandtech Forum Director

 

[Termie]

Is the monitor directly connected to the video card? Try another input and cable. HDMI 1.3 could be issue.

What was previous card? And motherboard? Could be PCIe issue.

 

[SinOfLiberty]

My previous card is gtx 570. P8p67 deluxe. Pcie is not an issue. I tested the card in both PCIe slots, 1(closest to the CPU) and the 2(in between 1 and 3). Monitor is connected to the card directly through hdmi 1.3b cable. The same cable i use to connect gtx 570.
Monitor is 2012 dell with only 1 hdmi slot and vga(wtf). I gotta test with another hdmi cable which i dont have.

After upgrading bios version, I got a big fps boost in BF4 on gtx 570, yes 570. Prior, i could only have such fps with it running at 860mhz and 4.5Ghz cpu. This time everything was at stock.

Gtx 1070 Valley benchmark has dips in usage, gpu is at 95-97% for the most time but 70,80% do occur.

 

[SinOfLiberty]

@HeXen

Temps are fine, 60C unless you want lower?...

There is only 1 driver for 1070!!!

No point in running valley on gtx 570.

 

[Ketchup]

Wow, never seen someone with such an attitude. Expecting help? Do you think people here are getting paid to help you. Don't take out your frustrations on people who are trying to help you.

Easy test would be to overclock the CPU and prove that's not your issue.

 

[mjdupuis]

Did you use DDU to uninstall the driver? In safe mode?

 

[Termie]

The 570 is five years old. Your motherboard is five years old. Your hdmi cable is nine years old. Because your system worked fine before doesn't mean you won't run into compatibility issues now that you add a GTX 1070.

You probably do not have a cpu bottleneck holding back all your benchmarks, but you certainly have SOMETHING holding them back, at least you think so. You need to test methodically to find out what it is.

Your unigine cpu usage numbers tell me the system is working pretty well, but that's such an old benchmark that it really isn't relevant anymore. Try 3Dmark Fire Strike and we'll know if something is truly the matter.

 

[VirtualLarry]

Is this the same OP that started a thread that claimed it was massively slowing down his FPS, because his video card or monitor refresh rate was set to 59.97 rather than 60, or something like that?

 

[SinOfLiberty]

I have something better. Here is the real world game benchmark I have recorded. BF4 short play session. Cpu is at 4.5ghz Prime95 48hours stable. This must not be like this. Low fps and something is blocking GPU throughput. In fact, when gpu fails to deliver the desired framerate, CPU kicks in taking more load than it should!

BF4 ULTRA 1080P

Sfc/scannow showed no integrity violations

I will try 3dmark benchmark as it says cant install cause i am running as admin wtf? is this even a legit thing... However, they optimize it for the newer CPUs, thus, the result is not the most truth telling.

 

[mnewsham]

However, they optimize it for the newer CPUs, thus, the result is not the most truth telling

Do you think games are optimized for older CPUs??

They optimize for the newest stuff first generally. Not for a 2600k which came out in 2011. That being said, it shouldnt make a huge difference as the architecture is mostly the same. But regardless, a benchmark isn't going to be biased because your CPU is old, it's just going to demonstrate how old it is by comparing it to newer CPUs.

 

[SinOfLiberty]

gtx 570 valley:

fps:39.9
score:1585
21.7 min fps
max.67.8
ULTRA quality 4x AA. Extreme would be harder to run

 

[Iron Woode]

Is this the same OP that started a thread that claimed it was massively slowing down his FPS, because his video card or monitor refresh rate was set to 59.97 rather than 60, or something like that?

no that was wingman. the fps was 59.94 Hz.

what a train wreck that thread was.

 

[Termie]

OP - you should try 3DMark. It's a legitimate benchmark, and no, it's not biased against your Core i7-2600K.

Based on your BF4 numbers, I would say you're getting great performance, far better than a GTX 570. I think your initial posts suggested performance went down. That's clearly wrong.

But there is something keeping your GPU and CPU from 99%, and yes, in BF4, one or the other should be at 90-99% most of the time. Are you running DDR3-1333 by any chance? That could be a bottleneck. I'm also still a bit concerned about your motherboard's PCIe 2.0 x16 slot, which could also have this result.

Now, it would be easy to say "hey, you're getting great performance, just play your games," but I can understand your desire to get the most performance possible out of your new card. If it's in fact a case of a platform limitation, at least you know you're pretty close to maxing out your 1070, and using multi-sampling/DSR to push it harder is an option, as there's not much use for 100fps on a 60Hz screen anyway. And your next upgrade could be a CPU/motherboard, rather than a GPU.

 

[SinOfLiberty]

according to the graph, there is definitely a problem. I should not getting these fps with gpu usage. I have seen others with the same set up as mine. PCIe 2.0 is not a bottleneck. I have an unhealthy behavior. Again, not only with one game, with all. I run gddr 3 1666mhz. BIOS says it is 1333, cause it is at stock settings. I stil have to manually adjust it.

The only reason for such drop in gpu usage would be if you had 300-400 fps.

I tried another hdmi cable, high speed rated. Worked perfectly, no issues on 570. After installing gtx 1070, got distorted image quality when I booted in standard vga mode). While installing the driver, image got more distorted, occupying more space of the screen. After the install has completed, i got no image when booting into windows. System started up but gpu did not put the signal. Restarting(2 times) did not help. The only way to fix it was to change back to 570. Afterwards, took 570out, put gtx 1070 back in and the problem was gone(image quality). Will do more testing.

 

[SinOfLiberty]

I have ran multiple gpu stability tests and none of the showed an issue.

What exacly 3d mark going to do? Say that my cpu score is low unlike the optimized hexa cores and i7 6700K and for everyone to bla bla bla bla about my cpu?
5 gigas wow

Valley is a very reliable benchmark what only stress tests the gpu. Little cpu usage.

 

[Termie]

Heaven is fine, but there's no database of scores you can compare to. I have no idea what a 1070 should score in it.

Run 3DMark to get a better understanding of how your entire system is working. I realize you think everything is fine and that your video card is broken, but the fact that you're running your memory at DDR3-1333 suggests you are leaving performance on the table.

 

[SinOfLiberty]

I have run with ddr 3 1666mhz. Before upgrading the bios version. Problem is there.

Valley scores are higher than alot of people, from what i looked on the onternet. Various forums. overlock.net, anandtech ect..

My 3d mark score, well, it is gonna be low as i aforementioned. So if i run with cpu+570 and get the score and compare it to cpu+ 1070 all stock, then it is all gonna be cause of cpu. Is this what you are telling?

 

[SinOfLiberty]

The issue has been re-created in the combined score test.

Idiot Mgur

See the tidbit where gpu usage is all over the place in the end of a graph, that is the combined score test. CPU was at 58-61% load at that time. All stock.

 

[StinkyPinky]

I'd reinstall the OS, start from a clean slate. Have you thought about Windows 10? I imagine all the benchmarks would have been done on that OS.

 

[SinOfLiberty]

I'd reinstall the OS, start from a clean slate. Have you thought about Windows 10? I imagine all the benchmarks would have been done on that OS.

No way in hell.

I tried on both win 7 drives. Each separately. No result. 2nd drive win is only 5 month old.

 

[Termie]

OP, you posted 3dMark screenshots, but covered up all the scores. The overall score is not the one we need to see.

But based on the GPU usage, it seems there's nothing wrong with your system, and that it's performing exactly as it should.

 

[SinOfLiberty]

 

[Ketchup]

17557 is what videocardz is showing for the graphics score of your card, so it looks like your card is performing exactly as it should.