[SOLVED] New AMD Ryzen Build Seeking Help - Very Low 3DMark Graphics Scores

[thessdreview]

Hoping someone can assist I am in a real bind and my specialty is not graphics cards whatsoever. I just built a new system and my scores on 3D Mark are horriffying. See attached.

http://thessdreview.com/wp-content/uploads/2019/11/PCMark-8-Timespy.png

My system is here and I have tried switching between this, a Quadro M5000 and PNY 1080 I have. No change with graphics one and two in Time Spy very low.

The system can be found here: http://www.thessdreview.com/hardwar...70-pcie-4-test-bench-pcie-4-0-goes-mainstream

It is a Ryzen 3700 in a new ASRock Creator. I have a few memory options but right now am using Corsair Vengeance RGB Pro 2666Mhz.

I get the same results overclocked to 4300 with Vengeance LPX at 3600Mhz as I get stock.

I have also tried several drivers and redownloaded 3DMark which is getting darned expensive as I live remotely and pay for data...

Would love ANY ADVICE and will do any test you would like. Going to DL GPUz now.

Thank you ahead.

 

[Roland Of Gilead]

Hey there,

Firstly, I love that review Very well put together, but also a very interesting read! Love your new office BTW I get it totally!

So, with the results of your 3dmark/PCmark test, what exactly is underwhelming? The scores are roughly where they should be for that GPU. I've a GTX1060 6gb, which would be considered a more powerful GPU, and don't get that much a higher score in Timespy.

Although Ryzen does love fast ram, going above 3600 yields practically zero benefit. For the most part the biggest jump with the expected performance boost for faster ram comes, between 2400mhz and 3200mhz, which can be huge. Often between 15-25% task dependent. After 3200mhz the performance boost tapers of very quickly. To the point where the difference between 3200mhz and 3600mhz can be as little as just a few % increase. Actually the best solution for optimum Ryzen mem, is 4 x8gbs of 3600mhz, after that it's not worth the OC, or more expensive DIMMS. The point being, that the 3d mark result (having ruled out the fast ram component) will not improve with faster ram.

What results are you getting with the 1080. I'm presuming when you say PNY 1080, you mean a GTX1080? In your review, you say you have a 1080ti, I think. Can you clarify?

 

[thessdreview]

Thank you very much for the comps. Maybe it is because I am using Time Spy and nothing less but, with this system OC'd to 4300 and at 3600Mhz, I never figured that GPU result would be in the low end of other testers.. 32% zone actually.

As I said, I am not a GPU guy whatsoever. I just thought I would get much higher that 20fps average for Time Spy. Maybe I am reaching too high for that GPU and should look at another test.

I guess I got blurred vision from this result previously...

http://www.thessdreview.com/featured/building-asrock-x299-fatal1ty-test-bench-i9-7900x-cpu/3/

Is there any way of pushing that CPU result into say the top 60% of results bu OC'ing? Always time to learn another hobby...

 

[thessdreview]

So I just tried Firestrike Ultra and the combined test was only getting like 5fps...horrible. This is the result.

https://www.3dmark.com/3dm/40684463

I find it hard to believe this GPU is getting results that are under 20% of all other tests...

Furmark

http://www.ozone3d.net/gpudb/score.php?which=817818

 

[Roland Of Gilead]

Thank you very much for the comps. Maybe it is because I am using Time Spy and nothing less but, with this system OC'd to 4300 and at 3600Mhz, I never figured that GPU result would be in the low end of other testers.. 32% zone actually.

As I said, I am not a GPU guy whatsoever. I just thought I would get much higher that 20fps average for Time Spy. Maybe I am reaching too high for that GPU and should look at another test.

I guess I got blurred vision from this result previously...

http://www.thessdreview.com/featured/building-asrock-x299-fatal1ty-test-bench-i9-7900x-cpu/3/

Is there any way of pushing that CPU result into say the top 60% of results bu OC'ing? Always time to learn another hobby...

Hey

So, i don't mean to be disrespectful when i say this. But the idea behind 3dmark etc is to task your system, hardware based on what it could look like. I get similar or slightly more FPS during those very same tests. I'm happy to post result for you if it helps. Those 3dmark tests are not meant to hit 60fps smooth whilst benching. They are meant to push your GPU capabilities to the max. However, they are synthetic benchmarks, and don't necessarily show real world gaming performance. If I might say, you are getting hung up on these benches. If you are getting 20fps on Timespy, I'm getting 25fps. If using a GTX 1080 you night hit 30 or 35fps. This is still below the considered threshold of 'smooth' gameplay (considered 60fps). Do you see what I mean? It's a benchmark. That all. It doesn't indicate anything other than a result in a benchmark. Your GPU is mainstream 1080p card. If you bumped in a GTX 1080 you would still not get much more in Timespy, It's the benchmark suite, rather than the hardware that is the limiting factor here.

 

[Roland Of Gilead]

Also, even though I've a more powerful GPU, i also get the same 32% percentile result. 3dmark, is a bench. Don't put too much reliance on it to determine the performance of a system, let alone that of a GPU. There are other benches too, that can help evaluate performance.

 

[thessdreview]

Yes I am well aware that I am a bit picky with these things and thank you very much...

 

[dstln]

GPU score appears to be in line with your 570. Those tests basically go off of your GPU power and nothing else really. A GTX 1080 is a much more powerful card, thus the higher score...

If you want to boost your GPU score "percentage" then try overclocking your GPU and/or get a more capable 570, if it matters to you.