• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
The Return Of GPU Cryptocurrency Mining
#41
(06-11-2017, 02:31 PM)SickBeast Wrote: I'm keeping the cards undervolted all the time.  0.825v for a constant boost speed of 1700mhz.  It's keeping my tempertures much lower and I hear this cuts the power consumption in half.  I'm still getting great performance in games.  So far Ghost Recon Wildlands and Crysis 3 are the most impressive.  I have a feeling I'm getting more performance than a single GTX 1080 Ti in those games.  Wildlands in particular looks absolutely stunning.

I'm getting 8X+8X PCIe for each card.  I also have the memory overclocked to +500mhz for both cards.
I am certain you are getting more performance from SLI's GTX 1070s than from a single GTX 1080 Ti - when SLI scales.
:cookie:

Here is a TITAN XP (2016 - the equal of GTX 1080 Ti) vs. GTX 1070 SLI
http://www.babeltechreviews.com/titan-x-.../view-all/
[Image: Untitled-1-3.jpg]
[Image: Untitled-2-3.jpg][Image: Untitled-3-3.jpg][Image: Untitled-4-3.jpg?w=910][Image: Untitled-5-3.jpg]


And here is GTX 1080 Ti SLI performance - just to compare

http://www.babeltechreviews.com/gtx-1080...-25-games/
  Reply
#42
Thanks for that. I have been referring to both of those articles quite a bit just to see if my performance looks decent.

I'm now up to 1734mhz at 0.825v. I'm slowly pushing my speed a bit higher. I'm hoping for a steady boost of 1800mhz at low voltage.

One strange problem I'm having is a black screen when I quit games. I'm not sure what's happening. The only way to fix it is to unplug and then replug the HDMI cable. That's not ideal though. I'm wondering if it's a driver bug. So far it's happened in the Witcher 3, Crysis 3, and Wildlands.
  Reply
#43
I've always had issues with HDMI and just use DP now. I always thought the bug was in Windows 10.

Since your SSD is also on a M2 slot and uses PCIe bandwidth, does it run at 4x or at 8x?

How does your performance compare with my 1070 SLI evaluation?
  Reply
#44
(06-11-2017, 04:00 PM)apoppin Wrote: I've always had issues with HDMI and just use DP now.  I always thought the bug was in Windows 10.

Since your SSD is also on a M2 slot and uses PCIe bandwidth, does it run at 4x or at 8x?

How does your performance compare with my 1070 SLI evaluation?

Almost identical, except I'm getting 5-10fps more than you in Crysis 3, I'm not sure why. Mind you, you are probably benchmarking a different more demanding part of the game.

I will run some benchmarks later. For now I'm just enjoying gaming with the cards and getting the mining up and running properly.

I'm not sure about the M2 SSD. I kind of regret buying it to be honest.

I'm up to 1800mhz sustained boost speed, undervolted to 0.850v, it was stable for a half hour gaming session in Wildlands. I seem to have solved the "black screen" issue also. I set the drivers to "single display performance mode" and I'm also using YCbCr444 instead of RGB. The top card hit 74C and the bottom card was ice cool at only 62C.
  Reply
#45
Not bad
Cool

You can always return the SSD.

I benchmark the very first scene in Crysis 3. As soon as you are able to move, hit the benchmark key and follow Psycho up the tower in the rain until he opens the door. It's totally repeatable and it is the most demanding scene in the game.
  Reply
#46
Thanks. The undervolting is working for both cards in gaming. It's just not working for mining, it's some kind of bug. It's too bad because it would cut my power consumption in half for the second card. Right now I'm undervolting the top card for mining because it runs hot. I don't think I'm going to push my speeds any further. 1800mhz boost speed is giving me very nice performance and reasonable temperatures.

I will have to run GTA5 to test my stability. I have a feeling I'll need a bit more voltage or else I'll have to dial back the clockspeed a bit.
  Reply
#47
Everything seems stable in GTA5. Good performance too. What was strange is that the cards didn't even break a sweat in this game. I ran it at 4k completely maxed out with 4xMSAA on top. It was generally pegged at 60fps in the benchmark. My cards stayed cool, that was the strange part. They were both around 60C, showing 100% usage.
  Reply
#48
(06-11-2017, 05:54 PM)SickBeast Wrote: Everything seems stable in GTA5.  Good performance too.  What was strange is that the cards didn't even break a sweat in this game.  I ran it at 4k completely maxed out with 4xMSAA on top.  It was generally pegged at 60fps in the benchmark.  My cards stayed cool, that was the strange part.  They were both around 60C, showing 100% usage.

Sounds great!

You can run Fraps while the benchmark in GTA 5 is running to get exact details of the run.
Thinking

I'm still working on my VR evaluation BtW. I am just not satisfied with the way it looks and I am doing a lot of re-writing. It got way too long and technical, and now I am trying to make it more concise and readable. And I am pretty sure I will get complaints for covering Nvidia's VR works in more depth than AMD's LiquidVR.
:emo:
  Reply
#49
The AMD guys are never happy. Good luck wrapping it up! Long and technical sounds good to me. I hope you saved the revision, maybe you could do a more technical follow up article. It seems a shame to waste all that work.

I will do some benchmarking later. GTA5 is a PITA; it always makes you restart the game when you change the graphics settings and also every time you run the benchmark.

Ok I ran my first benchmarks in Wildlands.

4k ultra w/ temporal AA 37.53 (these are the settings BTR uses)
4k ultra w/ FXAA 42.81fps
very high w/ ultra textures, SMAA, and 16x anisotropic filtering 62.74fps (49.80fps min) - these are the settings I use in game. It looks almost as good as "ultra" to me but the FPS is way higher and way more playable.

My score was virtually indentical to BTR's benchmarks of the GTX 1080 Ti. Pretty good IMO.
  Reply
#50
(06-11-2017, 06:37 PM)SickBeast Wrote: The AMD guys are never happy.  Good luck wrapping it up!  Long and technical sounds good to me.  I hope you saved the revision, maybe you could do a more technical follow up article.  It seems a shame to waste all that work.

I will do some benchmarking later.  GTA5 is a PITA; it always makes you restart the game when you change the graphics settings and also every time you run the benchmark.

Ok I ran my first benchmarks in Wildlands.

4k ultra w/ temporal AA 37.53 (these are the settings BTR uses)
4k ultra w/ FXAA 42.81fps
very high w/ ultra textures, SMAA, and 16x anisotropic filtering 62.74fps (49.80fps min) - these are the settings I use in game.  It looks almost as good as "ultra" to me but the FPS is way higher and way more playable.

My score was virtually indentical to BTR's benchmarks of the GTX 1080 Ti.  Pretty good IMO.
I hate GTA V benchmark since it takes longer to set-up and run than any other benchmark. It's a little quicker if you run the benchmark from "story mode". It looks like you got GTX 1070 SLI running great for a real 4K experience.

Maybe I didn't explain myself very well. My VR evaluation IS long and technical. However, I am totally unsatisfied with my explanations and am re-writing them for conciseness and for clarity. I don't (really) mind taking an extra day to get it right. Also, I didn't like my videos - they were all "run together" but I see now that I need to separate them and run ONE video for EACH benchmark with an explanation for each one - It's also a lot of work re-doing the videos (i.e. 12 videos instead of 2 - one for each of 6 cards times 2 video benchmarks). Also, my FCAT VR graphs need some work. It's hard to put 6 cards on each graph and not confuse the viewer.
Blush
  Reply


Forum Jump:


Users browsing this thread: 12 Guest(s)