GTX 3070 Gaming Performance

[CLEAR RTC]

I just got a 3070 yesterday and have terrible performance.  I sold my GTX1080 (To help finance the new card purchase) and was running a Vega 56.  The 3070 is not even performing on par with the Vega 56 in COD Warzone let alone the GTX1080 with the same settings.  In fact it doesn't matter what graphics settings I chose in the game it ranges from 70-90 fps.  I was getting 90-110 with the Vega.

 

When I launched Warzone it detected the new hardware and asked if I wanted it to change settings. I click NO I wanted to compare the performance directly from my previous settings on the Vega 56 and to my surprise, the frame rates were the same or WORSE than the Vega 56. OK Settings must have changed right. I open up the game settings and Sure enough Ray Tracing is enabled, OK Turn that junk off. Frame rates remain in the mid 70 to high 80s Back into settings to do a more thorough review. Render resolution was set to 170 not 100, ok drop that back down, and check all other details settings, most are the same other than shadows had gone to max, adjusted all those to previous settings. The frame rate remains the same maybe 5fps better. How is it possible this thing is performing worse than a GTX1080? I am playing at 1080p on an Acer KS271 144hz display and I hoped to at least be able to utilize it. G-Sync on and off the results are the same.

 

In PUBG I see wild swings in performance 200+fps to below 90 averaging about 110, this again was a game I was getting 90-100fps in with the Vega 56.

 

When I removed the Vega and installed the 3070 I rebooted into safe mode.  I ran DDU and removed all drivers Nvidia/AMD/Intel.  I Disconnected my patch cable and booted windows.  I installed driver 461.09 as a clean install.  In 3DMark TimeSpy Extreme I am see about 11500 points a good jump from the Vega and leads me to believe its not a hardware issue. 

 

I have seen lots of threads now with people complaining about 3070 performance and they go on for pages and then just end with no real resolution.  I can't believe this is actually that common though, I mean every review I have watched/read was positive about the card but the stated performance is no where close to what I am seeing.

 

I am running at 1080P with the following specs

 

GPU: EVGA XC3 Ultra 3070

CPU: AMD R5 2600

Mobo: MSI B450 Tomahawk MAX
Mem: Thermaltake ToughRAM 3600Mhz @3200 (bought the ram with expectation of upgrading to Zen2/3 down the road)

PS: Corsair AX850M

Cooler: Gamdias Chione 360 AIO

OS: Win10 Pro 20H2 latest feature release

Monitor: Acer KS271 144hz TN

[CLEAR RTC]

I did just see Nvidia released a driver on Jan 26th 461.40, So I will give that a try tonight.

[SFFDesigns]

Try DDU'ing your drivers, then install them again. Your 2600 will bottleneck the 3070 though

[CLEAR RTC]

1 minute ago, Hymenopus_Coronatus said:

Try DDU'ing your drivers, then install them again. Your 2600 will bottleneck the 3070 though

Thanks for the reply.  Yeah I will try DDU again and 461.40 driver when I get home.  I thought I may get bottlenecked a little but CPU usage is not maxed and I can't get close to performance I had with the GTX1080 in Warzone at least.  The PubG performance is just weird, maybe it is bottlenecking there.

 

I'd like to move to a 5600x or 5800x, however while we joke about graphics cards being unobtanium I haven't even been able to backorder a Zen 3 part.

[Mark Kaine]

the new ryzens are mostly made of snake oil anyway, I would recommend to get a "last gen" ryzen that's almost on par and *much* cheaper. And maybe even available lol. 

 

 

As for you gpu - I also think this is a software / user issue probably - any game I've tested my performance like tripled or quadrupled... 

 

But I'm using Studio drivers, I think that may make a difference, I'd suggest to DDU again and install Studio drivers therfore. 

 

 

if that doesn't help I've seen a post here where a "fresh" windows install but keeping your data fixed any 30xx "performance issues". I recommend to make a backup first, should you want to try that solution tho. Better safe than sorry! 

[CLEAR RTC]

So I did the fresh install and performance remained the same.  First thought was well crap maybe it is a CPU bottleneck.  Anyway I had to put my main boot drive back in to do some work with my machine.

 

When I finished I fired up Afterburner and ran some Warzone.  The results were interesting to say the least.  

 

Soo I was getting about 60% GPU usage and about the same for CPU usage.  first 6 CPU are the cores and the 7th is average usage.  I saw the primary core hit 80% once while playing.

 

I had asked Afterburner to measure other things but it said it could start all the hooks and to restart the app, which I did but still didn't get all the measurements I wanted average fps, 1% .1%, min and max.  However you can see the Warzone presented FPS from the screen shot in between the green GPU and orange percentage.

 

 

[Mister Woof]

Online games are hard to measure. There's a lot of variables and sometimes they will bog turn even in amazing hardware because of network, server issues, etc.

 

Try your tests and single player games.

[CLEAR RTC]

2 minutes ago, Mister Woof said:

Online games are hard to measure. There's a lot of variables and sometimes they will bog turn even in amazing hardware because of network, server issues, etc.

 

Try your tests and single player games.

I see what you are saying, Problem is I don't have a reference for single player games I really only play PUBG and Warzone, online a few times a week with friends etc.  I will find something to compare between the V56 and 3070.  In the meantime I was just catching up in the other thread and see I totally missed built in tools in Warzone so I'll give those a quick test.

 

Thanks for the assistance thus far

[CLEAR RTC]

Well crap on a cracker.  CPU time is high.  Same as the 1700x in the other thread.  Process reduction and clock increase yielded no improvement?   I am starting to feel like I drank the cool-aid.  More testing still required and I am going to get a Zen 2 part to test with.

 

Good lord .. I thought to myself this Vega 56 is pretty fast almost as fast as my GTX 1080 ...  F-bombs was I bottle necking my last GPU, would it have had the performance I wanted all along? 

 

 

[Purgent]

I just upgraded from a 1080ti to a 3070, but saw little fps increase in Warzone.  Reason is it is very CPU dependent.

 

Upgraded from a 2700x to 5600x and saw massive differences in every game I play.

[CLEAR RTC]

9 hours ago, Purgent said:

I just upgraded from a 1080ti to a 3070, but saw little fps increase in Warzone.  Reason is it is very CPU dependent.

 

Upgraded from a 2700x to 5600x and saw massive differences in every game I play.

Looks like that is the root I will eventually take.  How did you come by the unobtanium that is Zen3? 

[Purgent]

On 2/3/2021 at 9:29 AM, CLEAR RTC said:

Looks like that is the root I will eventually take.  How did you come by the unobtanium that is Zen3? 

Microcenter

[CLEAR RTC]

I ordered a 5600X today and Expect it to be in tomorrow or Friday, will update this thread with results.

[CLEAR RTC]

So just to close the thread off as everyone said, it was a CPU bottle neck. I see 160-200 FPS with lows in the 130's @ 1080P in Warzone with everything on high except RT which is disabled.  GOU Time is 5-7ms and CPU is 4-6.  I haven't done any CPU or GPU tuning for these results.

 

 

[CLEAR RTC]

Resurecting my own thread just to add a note.  Myself and others running older Ryzen CPU's that saw performance regressions moving from AMD Cards Vega 56 and 5600XT in my case on a R5 2600 to a RTX 3070 Performance dropped from 100-110 FPS to 80-90.

 

Everyone did correctly identify a CPU bottleneck, but as Hardware Unboxed has demonstrated in a Video released earlier today the Nvidia Driver is adding a lot of overhead further adding to the CPU load.