RTX 3090 [H] Owner's Official 3DMark Time Spy Leaderboard

bal3wolf said:
cool least its not just me for a while i was going crazy trying to figure out why i was getting lower scores on my 5950x vs my 3900x i had before.
Click to expand...
Odd, I have a 5950x and have rebar on and my CPU stays above 16k almost all the time. I wonder if its a mobo/bios thing for some people? I've not had any issues and score where I should for a 5950x.

https://www.3dmark.com/spy/21415635
 
Last edited:
III_Slyflyer_III said:
I leave rebar on at all times... unless nvifia enables it for a specific game, it should not make a difference, but c

Odd, I have a 5950x and have rebar on and my CPU stays above 16k almost all the time. I wonder if its a mobo/bios thing for some people? I've not had any issues and score where I should for a 5950x.

https://www.3dmark.com/spy/21415635
Click to expand...

Might be a asus board issue i see me and talon both have asus boards you have a msi. It worked for me on my 3900x but not the 5950x but i see talon has a 3900x and it dont work for him so its wierd.
 
Last edited:
lobstar said:
Oh yeah, finally got to the top of the air cooled list all of the lists in this thread :D (and 13th in the world for my hardware combo)

https://www.3dmark.com/3dm/60198840?

​

Graphics score 22 679​

Click to expand...
lobstar said:
Oh yeah, finally got to the top of the air cooled list all of the lists in this thread :D (and 13th in the world for my hardware combo)

https://www.3dmark.com/3dm/60198840?

​

Graphics score 22 679​

Click to expand...
How on earth did you get that sckre
lobstar said:
Oh yeah, finally got to the top of the air cooled list all of the lists in this thread :D (and 13th in the world for my hardware combo)

https://www.3dmark.com/3dm/60198840?

​

Graphics score 22 679​

Click to expand...
How on earth did you get that score and hold such high clocks on air. I have an rtx 3080ti ftw 3 hybrid card . My card is liquid cooled and I can't even lock 2100mhz clocks on average. Thanks
 
slyraptorx said:
How on earth did you get that score and hold such high clocks on air. I have an rtx 3080ti ftw 3 hybrid card . My card is liquid cooled and I can't even lock 2100mhz clocks on average. Thanks
Click to expand...
Room ambient was 62F. I spent literal days with those fucking broken cards dancing between not crashing and hitting the power limit while underclocking. On the FTW3U with the bad voltage controller if you ever hit power limit you would be clock limited the rest of the benchmark run.

I'm on water now. I'm not familiar with your card but to get high clocks now I just yeet some caveman numbers into the card via afterburner. This is my 'Lets game and youtube and not worry about shit' daily driver which benchmarks between 2145-2190mhz stable. I'm on the KPE 520w bios for this setup.
1627960299435.png


On the party bios for benching under water I'll start around +165 / +1848 at 'aint care' power limits.

It can really clock well but I must have hit some sort of unknown limit internally on this run to limit the score. I really don't know but it's my best clocks in a bench yet.
https://www.3dmark.com/pr/1019273
1627960684629.png

(This was before AE changed the mem slider to +2k so limited memory)
 
III_Slyflyer_III said:
Odd, I have a 5950x and have rebar on and my CPU stays above 16k almost all the time. I wonder if its a mobo/bios thing for some people? I've not had any issues and score where I should for a 5950x.

https://www.3dmark.com/spy/21415635
Click to expand...

Are you sure reBar was enabled for that run? I had some odd trouble where on my 3090, enabling reBar just for 3DMark worked, but for my 3080ti I had to enable it globally.

Also of interest is that I only see the slowdown in CPU score on TS, not TSE.
 
skypine27 said:
Kind of unimpressed with mine, graphics score of 20,228 and CPU 11,780 ! My memory is a lot slower that most of you guys, system doesnt seem stable with it clocked any faster.

https://www.3dmark.com/spy/22019147
Click to expand...

your clocks are quite a bit lower also most of us are pushing 2150+ when benchmarking you should be getting around 12k on cpu score with that cpu so your not to far off on that.
 
bal3wolf said:
your clocks are quite a bit lower also most of us are pushing 2150+ when benchmarking you should be getting around 12k on cpu score with that cpu so your not to far off on that.
Click to expand...
How do you get 2150? If I try to OC any more than +100, games seem to crash
 
skypine27 said:
How do you get 2150? If I try to OC any more than +100, games seem to crash
Click to expand...
Your card works in 15mhz chunks called bins. If you are just setting a flat OC across the curve use intervals of 15. So you would want 90 or 105.

Try this as a starting point.

(your power limits may be different; to lock your voltage press 'L' after you've selected your OC/voltage node)
Here is a comparison with your result and one of mine with a very similar setup. https://www.3dmark.com/compare/spy/22019147/spy/20245816#
1628521507144.png
 
Last edited:
lobstar said:
Your card works in 15mhz chunks called bins. If you are just setting a flat OC across the curve use intervals of 15. So you would want 90 or 105.

Try this as a starting point.

(your power limits may be different; to lock your voltage press 'L' after you've selected your OC/voltage node)
Here is a comparison with your result and one of mine with a very similar setup. https://www.3dmark.com/compare/spy/22019147/spy/20245816#
View attachment 383047
Click to expand...


So you get better perf locking it then doing a straight overclock wierd thing on my card if if use a curve and lock it i lose perf for some odd reason.
 
bal3wolf said:
So you get better perf locking it then doing a straight overclock wierd thing on my card if if use a curve and lock it i lose perf for some odd reason.
Click to expand...
I get more consistent results by locking it. I can get a little bit extra score when I go full curve but it seems more finicky on my FTW3U.
 
lobstar said:
I get more consistent results by locking it. I can get a little bit extra score when I go full curve but it seems more finicky on my FTW3U.
Click to expand...

Yea my cards really wierd i can hold higher mhz with a lock and curve but lose perf like i can hold 2190 no crashe but its slower then 2115 and funny uses less power also.
 
lobstar said:
Your card works in 15mhz chunks called bins. If you are just setting a flat OC across the curve use intervals of 15. So you would want 90 or 105.

Try this as a starting point.

(your power limits may be different; to lock your voltage press 'L' after you've selected your OC/voltage node)
Here is a comparison with your result and one of mine with a very similar setup. https://www.3dmark.com/compare/spy/22019147/spy/20245816#
View attachment 383047
Click to expand...
thx will mess with it again when I get home, am on vacation for a while. Thx again !
 
lobstar said:
Your card works in 15mhz chunks called bins. If you are just setting a flat OC across the curve use intervals of 15. So you would want 90 or 105.

Try this as a starting point.

(your power limits may be different; to lock your voltage press 'L' after you've selected your OC/voltage node)
Here is a comparison with your result and one of mine with a very similar setup. https://www.3dmark.com/compare/spy/22019147/spy/20245816#
View attachment 383047
Click to expand...
Ive never seen the "curve" option. I use MSI afterburner, a different skin than yours, but still where is the curve option usually?

Thx!
 
skypine27 said:
Ive never seen the "curve" option. I use MSI afterburner, a different skin than yours, but still where is the curve option usually?

Thx!
Click to expand...
It's the signal bars looking thing in my screen shot right above the mouse cursor.
1629047252357.png
 
lobstar said:
Your card works in 15mhz chunks called bins. If you are just setting a flat OC across the curve use intervals of 15. So you would want 90 or 105.
Click to expand...

Thanks bud!! Didnt know about the 15mhz intervals. Got it benching at +135 and score creeping up
https://www.3dmark.com/spy/22314089

(graphics 20,713)

Havent tried extended gaming to see if it crashes yet. I believe 135 will be the limit, because before I knew about the "bins", I was setting it at +150 (which would be the next higher 15mhz interval) and Cyberpunk would crash out often, and I'd lower it back down to +100 (again, not knowing about the bids) and the game was stable

Thx again
 
bal3wolf said:
You seem to know alot about curve overclocking you have any idea why when i use it and say lock voltage to 1.1 i get less perf btw im on a kingpin 3090 with the unlocked kpe bios.
Click to expand...

Probably need to increase MSVDD either with dip switches or classified tool. Most cards will perform better at 1093mv because of this. (My Zotac does)

Falkentyne talks about it here, although referring to increasing NVVDD also, but same idea.

https://www.overclock.net/threads/o...90-owners-club.1753930/page-877#post-28856599
 
Lamont, I'd seriously have a look at your card's cooling. Just about everyone else on the top 10 list that's running air cooling has an average card temp for a run in the 40-60°C range.

In contrast, the average temp for my best run was 20°C. (Granted that's on water, in the dead of winter, with an open window... a regular run for me is more like 40°C) But still...sheeesh! With an 81°C average, your card:

1630095529324.png
 
Last edited:
I'll take a look at the temps right now.
Thanks everyone :)

My ambient temp in this room is 33~35c. I have no AC.
 
Last edited:
Lamont said:
Ran the test on my other PC:

http://www.3dmark.com/spy/22463426

It's just hot af in this room due to no AC I am sure.
Click to expand...
Holy shit. It was 33°C here today outside. Fuck that. I have to ask, why not get a portable AC unit or something? We have like 80-95% humidity where I'm at so it is just unimaginable to not have AC.

Edit: I sound like a real jerk here. Not intended, I swear.
 
Last edited:
lobstar said:
Holy shit. It was 33°C here today outside. Fuck that. I have to ask, why not get a portable AC unit or something? We have like 80-95% humidity where I'm at so it is just unimaginable to not have AC.
Click to expand...
Hmm, I'd have agree. Lamont, you are running an i7 10700K and an RTX 3090, so power and $ don't seem big obstacles. Why not spring for a 10,000 BTU Window AC unit for ~$300?
 
I have a 18,000 BTU Window AC unit, 12 x 12 space :D. Wonderful to take the temperature down quickly to low 60F. No sweating in VR.
 
I have central AC, but my gaming room is in my upstairs office. In the summer, I always run my portable AC when gaming. Not only does it keep the PC cool, it keeps me comfortable as well! High end PCs are great for heat in the winter, not so much in the summer... lol
 
  • Like
Reactions: noko
like this
Thanks guys, I ran Afterbuner, changed my fan profiles, and kept the temps about 73c. This is about as good as it's gonna get till I finish my house, where there will be plenty of AC. I just need to get through one more summer in Kansai and I will be OK.

Right now I am in an apartment and don't wanna burn cash on a temp thing :D. I almost got a portable AC, but I changed my mind.

And yes, in winter time it's perfect. I don't run the heater in this room.

lobstar - No offense taken. It's hotter inside my place than it is outside, even with the windows open. I hydrate like I'm running a freakin' marathon. I didn't notice my temps until you guys brought it up, so thanks again :D.
 
1st SLI run.

Graphics score: 26,962

https://www.3dmark.com/spy/23010559

Air Cooling

Not sure why it says that hardware monitoring was disabled. It is enabled under options. Maybe that POS 1024x600 7" display I had connected for the run is the culprit?

I'll connect a "real" monitor for my next run. This was just a quick test to see if my new rig would even run the benchmark. Still dialing things in.

Also waiting on front and back blocks for both GPUs. They run hot on air in the current config:

IMG_2777.JPG


CPU (11900K) is the only thing on water right now.
 
Last edited:
But when I look at the top10 list, I see a lot of scores that are close to double what a single 3090 can do:

timespytop10.png


Maybe the fact that I'm on Win11 with the latest 472.12 drivers is working against me? I think I need to figure out what's going on with Hardware Monitoring being disabled during the runs first...
 
pclausen said:
But when I look at the top10 list, I see a lot of scores that are close to double what a single 3090 can do:

View attachment 396852

Maybe the fact that I'm on Win11 with the latest 472.12 drivers is working against me?
Click to expand...
Folks in that list are using liquid nitrogen to cool their hardware and pulling out all the stops (physical card mods involving voltage shunting to increase power delivery capability for example). You need to look far past even the top 100 range if you want to see somewhat more "normal" results to compare with.

For example, my best run was in the top 2% of all reported TimeSpy benches for a single 3090... but there are over 6000 entries listed that got a better GPU score than what I managed to get.

Timespy.jpg


The "average" is probably a better indicator of what to expect for a more normal "good" run.
 
Last edited:
You must log in or register to reply here.
Back
Top