2017 Hardware Thread

Dude said:

I have a Time Spy Anomaly, even with the new driver I didn't get any improvement

It's not a Time Spy Anomaly. It's a marketing job on the DX12 performance on that 378.78 driver. When I got the 12 point increase in the graphic score, my overall Time Spy score went down 30 points from my 7662 score. To get my 7702 score, I had to up the CPU to 4.8. In that score the graphic score is only 5 points higher, 8153 to 8158.

That 16% DX 12 improvement is comparing DX 12 games when they were released (and the driver at the time) to now, March 2017.

Who cares about game release date to now? Many thought 378.78 was going to give a significant DX 12 performance increase. The performance is flat compared to the previous hot fix driver.

essenbe said:

Another thing about the 1080 TI Time Spy scores could be that they are using Beta drivers. It may do better once the 1080 TI drivers mature a little.

Ya, maybe. When you look at the specific DX12 games, the 1080TI beats Pascal Titan in BF1, Hitman, and ROTTR (except ROTTR 4K where they are equal). GOW4, Sniper Elite 4 the numbers are pretty close. The Time Spy numbers themselves are pretty close but I thought looking at the specific DX12 games the 1080TI had a bit of an edge.

essenbe said:

@specialkone, you're not going to talk me out of it, so quit trying.

I noticed the Time Spy anomaly in another review. It sort of makes you wonder how the test is scored. I know when I went from my 780 Classified to the 980 Classified there was one test I just could not even come close to the score of the 780, even though the 980 was way faster. I finally researched and found out it was because the test heavily weighted cuda cores (I think). The 780 had more cuda cores than the 980. It would pay us benchmarkers to try to find out exactly what these tests measure I think.

I suspect @doorules knows though.

No, I'm trying to convince you to get a 1440P or better monitor with a 1080TI .

I agree we need to know more about what these tests measure. I talked to 3DMark people not long ago about my issue where Time Spy crashes if I have the XOC OSD enabled. I asked a lot of questions about Time Spy. They wouldn't say much, other than stating Time Spy tests more DX12 features than what are currently in any DX12 game.

Ya, I think there will be benchmark software and games coming out now "showing the benefit" of the latest hardware. Look at these Ghost Recon numbers. A GTX 1080 only achieves 61 FPS at 1080P and 47.7 FPS at 1440P. Are you kidding me? Guess I need a 1080 TI as well . Of course this is a Game Works title so we know who is behind this.



Ghost Recon Wildlands: Performance Analysis | techPowerUp