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

NukeDukem

2[H]4U
Joined
Feb 15, 2011
Messages
2,447
I have a window AC unit (Frigidaire 8,000BTU on low/quiet mode) blowing into an open-air case. Air flow about 1.5' away; no ducting. So I'm not sure if that would qualify for 'extreme' cooling or air. The FE cooler does an excellent job by itself, and when you feed that big upper intake fan with some cool air it kicks ass.

edit: Yeah looking at OP's criteria it should go extreme. I can definitely push it harder so I'll do another run tomorrow when I get the faster memory.
 
Last edited:

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
I have a window AC unit (Frigidaire 8,000BTU on low/quiet mode) blowing into an open-air case. Air flow about 1.5' away; no ducting. So I'm not sure if that would qualify for 'extreme' cooling or air. The FE cooler does an excellent job by itself, and when you feed that big upper intake fan with some cool air it kicks ass.
I put you down in the extreme category since some effort was made it sounds like to cool the GPU beyond the norm. If you want to post a score without the AC aimed at the computer, I can put you in the air category, just so the testing scenarios remain comparable to one another. You're the first to try anything beyond regular air so far.
 

kenpomasta

Gawd
Joined
Jun 21, 2004
Messages
903
Damn really nice :) Im on the order list for the Optimus Waterblock for my FTW3. Cant wait for this thing, to really push the GPU. The preprod block that GN had looked amazing, cant wait to see the full prod block in action.
 

sk3tch

[H]ard|Gawd
Joined
Sep 5, 2008
Messages
1,844
19,188
19,216
19,330

I have some room to grow...best of night 1 with my 4266MHz DDR4. :)

9900KS on AIO water and 3090 FE on air.

4133_XMP_32GB.png

EDIT: got a little higher - looking like 4266MHz will be tough to hit - it's a big OC - still tweaking.
EDIT2: 3090 not hitting what it did previously, so I will be up more soon enough!
 
Last edited:

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
19,188
19,216
19,330

I have some room to grow...best of night 1 with my 4266MHz DDR4. :)

9900KS on AIO water and 3090 FE on air.

View attachment 301762
EDIT: got a little higher - looking like 4266MHz will be tough to hit - it's a big OC - still tweaking.
EDIT2: 3090 not hitting what it did previously, so I will be up more soon enough!
hoping I can beat you today. ;)
 

talon95

Limp Gawd
Joined
Aug 13, 2006
Messages
159
Here's my Zotac 3090. Stock cooling. Gigabyte bios at 390w. Custom undervolt curve (although stock curve scores nearly identically for some reason)

Fans were running at 80 to 85% in auto with a somewhat more aggressive curve than stock. At 100% fans the gpu score will go up 100 or 150 but crazy loud.

https://www.3dmark.com/spy/15552003
 

kenpomasta

Gawd
Joined
Jun 21, 2004
Messages
903
Had to push it further :)

20377

https://www.3dmark.com/3dm/53448394

EVGA 3090 FTW Ultra XOC bios +95C, +750M

Only got up to 116% power limit, it can go higher with the bios, but it doesnt seem to want to go much more on Air. Funny thing, is I got a worst CPU score this time around lol, vs before, With the same settings, love 3dmark lol.
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
FYI, anyone who wants to buy the full version if 3dmark, it is on sale for $4.49 through the 3dmark site or through steam, and steam version includes port royale as well.

edit: my new high score, 19677. Don't think I can do any better with the 5800x/3090 on air.

https://www.3dmark.com/3dm/53591462
 
Last edited:
  • Like
Reactions: DFenz
like this

kenpomasta

Gawd
Joined
Jun 21, 2004
Messages
903
ok I can push just a little more also.

Air cooled 3090 FTW3 Ultra (can't wait to get a hybrid kit for this bad boy!):

https://www.3dmark.com/spy/15580054

Overall: 20413
GPU: 22401
CPU: 13583


Edit: Was able to eek out a teeny bit more before I topped out with the GPU:

https://www.3dmark.com/spy/15580398
Overall: 20468
GPU: 22484
CPU: 13573


Ok you keep pushing me lol. I think Im pushed just about as far as it will go. Looks like you have slightly better GPU silicon then I do :) But 10900k pushes me just over the top. Pushed it up to 5.3, and upped GPU a little more


20496

+115 C, +750 M

10900k @ 5.3

https://www.3dmark.com/3dm/53593547

 

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,672
ok I can push just a little more also.




Ok you keep pushing me lol. I think Im pushed just about as far as it will go. Looks like you have slightly better GPU silicon then I do :) But 10900k pushes me just over the top. Pushed it up to 5.3, and upped GPU a little more


20496

+115 C, +750 M

10900k @ 5.3

https://www.3dmark.com/3dm/53593547

Force me off my daily driver 4.7Ghz all core will you? So be it!

https://www.3dmark.com/spy/15583849
Overall: 20517
GPU: 22455
CPU: 13781
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
Is this test reporting the wrong CPU? I thought you were running a 5800X.

Also, what are you doing to average 2,002 MHz and keep the GPU at 46C?
I sold the 5800x and got a 3950x instead, the extra cores are helpful in unreal engine more than the IPC gains. I sold it to a member here for cost. When zen 3 is in stock, maybe I will get a 5950x in a year or something.

For the GPU, running 114% power limit, +180MHz on the core +600 on the memory, and fan at 100%. Ambient is 68f/20c. Just stable enough to get through time spy, not much else.
 
  • Like
Reactions: erek
like this

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
Here is my latest:
19,494
https://www.3dmark.com/spy/16019033

GPU Info:
Power Limit 114%, Core Clock +150 MHz Offset Curve, +500 MHz Memory Overclock, All fans at 100%. Ambient temp inside the house is around 71-72f/~22C.

I haven't tried upping the core clock much more yet but I am clearly running into a power limitation that's keeping me from going much further.

The more I up the memory speeds, the lower my average GPU clock speeds become, so it becomes a trade off of what does the application benefit more from. At no memory overclock but all other settings left the same (fans at 100%), I can average a clock speed of 1,993 MHz. If I bump the memory by +250 MHz, my average clock speed drops to 1,980 MHz. When I go to +500 MHz on the memory, average clock speed drops even further to 1,976 MHz. Peak clock speed with no memory overclock was hitting 2,070 MHz while it drops to 2,055 MHz once I add any memory overclock to it. the 250 MHz memory overclock makes little to no difference than without the overclock, thus not offsetting the lost of GPU clock speed. But when I go to a 500 MHz memory overclock, there is a performance gain to offset the loss GPU clock speed, but it is only about a .5% gain across the board. For the added heat it creates, not sure it is worth it over having faster GPU clock speeds.

The fan speeds play a factor in power consumption too and overall power availability to the GPU, but the power it consumes to keep the GPU cool doesn't seem to overtake the throttling that would occur when temperatures increase. So for us air cooled peeps, the power consumption of the fans isn't much of a hindrance to performance. Now, if we go to water cooling and don't need any fans, that extra power freed up will become a benefit and allow the GPU to squeak out a few more MHz on top of the added benefit of running cooler when on water.
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
Here is my latest:
19,494
https://www.3dmark.com/spy/16019033

GPU Info:
Power Limit 114%, Core Clock +150 MHz Offset Curve, +500 MHz Memory Overclock, All fans at 100%. Ambient temp inside the house is around 71-72f/~22C.

I haven't tried upping the core clock much more yet but I am clearly running into a power limitation that's keeping me from going much further.

The more I up the memory speeds, the lower my average GPU clock speeds become, so it becomes a trade off of what does the application benefit more from. At no memory overclock but all other settings left the same (fans at 100%), I can average a clock speed of 1,993 MHz. If I bump the memory by +250 MHz, my average clock speed drops to 1,980 MHz. When I go to +500 MHz on the memory, average clock speed drops even further to 1,976 MHz. Peak clock speed with no memory overclock was hitting 2,070 MHz while it drops to 2,055 MHz once I add any memory overclock to it. the 250 MHz memory overclock makes little to no difference than without the overclock, thus not offsetting the lost of GPU clock speed. But when I go to a 500 MHz memory overclock, there is a performance gain to offset the loss GPU clock speed, but it is only about a .5% gain across the board. For the added heat it creates, not sure it is worth it over having faster GPU clock speeds.

The fan speeds play a factor in power consumption too and overall power availability to the GPU, but the power it consumes to keep the GPU cool doesn't seem to overtake the throttling that would occur when temperatures increase. So for us air cooled peeps, the power consumption of the fans isn't much of a hindrance to performance. Now, if we go to water cooling and don't need any fans, that extra power freed up will become a benefit and allow the GPU to squeak out a few more MHz on top of the added benefit of running cooler when on water.
Good stuff. Remember that any change in afterburner to core or memory frequency locks in your curve based on where it sits currently, so you always want to make these changes 'cold', not right after finishing a run. I played around with fan speeds a bunch thinking they might be pulling some power from the core, but like you found, it is negligible and the lower temp is more benificial to aid in boost clocks than the power saved by turning them down.

CPU speed makes a pretty big difference as well, but between the GPU dumping 400 watts of heat and my cpu being on a 120mm aio, I can't do much about it.

As a side note, for regular use outside of benchmarks, I set my founders card to run the "exhaust" fan faster than the pass through fan in afterburner, so it pulls more heat out of the case before dumping it inside the case.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
Good stuff. Remember that any change in afterburner to core or memory frequency locks in your curve based on where it sits currently, so you always want to make these changes 'cold', not right after finishing a run. I played around with fan speeds a bunch thinking they might be pulling some power from the core, but like you found, it is negligible and the lower temp is more benificial to aid in boost clocks than the power saved by turning them down.

CPU speed makes a pretty big difference as well, but between the GPU dumping 400 watts of heat and my cpu being on a 120mm aio, I can't do much about it.

As a side note, for regular use outside of benchmarks, I set my founders card to run the "exhaust" fan faster than the pass through fan in afterburner, so it pulls more heat out of the case before dumping it inside the case.
When modifying the curve, I was always modifying the existing curve that I first edited, but would always wait for my temps to cool down and level out. All but one time when I edited the curve the peak clock speed was the same, it would be at 2,085 MHz. The one time it wasn’t it was at 2,100 MHz.

When tinkering with the curve, I would increase it by 15 MHz over the previous then do multiple tests; at 100% power curve, then at 114%, then increase fans, then bump memory to 250, then bump memory to 500, and then run it again at default fans. Each time I would cool it down between runs. Then I would try all that again with a clock speed bump of another 15 MHz. By doing it this way, the curve is not being changed when trying to figure out how other factors impact the results. During these tests it also made me notice that setting the voltage slider to 100% makes no difference on the FE model.

Increasing the memory speeds starts adding extra heat quickly and I am not so sure if the added speed and heat versus the performance bump it gives is worth it when on air...and being power constrained. So while fun for benchmarking, I am most likely going to leave the memory speeds alone for day to day gaming. At most, maybe just a 250 bump.
I have my CPU on a fixed voltage and frequency, so it’s results don’t change too much between runs. This most recent run I am running a newer clock speed and voltage on my CPU that bumps it up around 75-100 MHz (and an added 250 mV) over my previously submitted runs. It’s on a custom loop going to a single 240mm radiator, but that radiator is in a separate chamber in the case fed by its own fans, so heat from the video card has little to no impact to the CPU temps. At most the GPU is heating up the coolant tubes going to the cpu water block and increasing overall ambient temp in the motherboard chamber. I am toying with the idea of adding a second 240mm radiator to the case later, especially if I get a waterblock for the GPU.
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
When modifying the curve, I was always modifying the existing curve that I first edited, but would always wait for my temps to cool down and level out. All but one time when I edited the curve the peak clock speed was the same, it would be at 2,085 MHz. The one time it wasn’t it was at 2,100 MHz.

When tinkering with the curve, I would increase it by 15 MHz over the previous then do multiple tests; at 100% power curve, then at 114%, then increase fans, then bump memory to 250, then bump memory to 500, and then run it again at default fans. Each time I would cool it down between runs. Then I would try all that again with a clock speed bump of another 15 MHz. By doing it this way, the curve is not being changed when trying to figure out how other factors impact the results. During these tests it also made me notice that setting the voltage slider to 100% makes no difference on the FE model.

Increasing the memory speeds starts adding extra heat quickly and I am not so sure if the added speed and heat versus the performance bump it gives is worth it when on air...and being power constrained. So while fun for benchmarking, I am most likely going to leave the memory speeds alone for day to day gaming. At most, maybe just a 250 bump.
I have my CPU on a fixed voltage and frequency, so it’s results don’t change too much between runs. This most recent run I am running a newer clock speed and voltage on my CPU that bumps it up around 75-100 MHz (and an added 250 mV) over my previously submitted runs. It’s on a custom loop going to a single 240mm radiator, but that radiator is in a separate chamber in the case fed by its own fans, so heat from the video card has little to no impact to the CPU temps. At most the GPU is heating up the coolant tubes going to the cpu water block and increasing overall ambient temp in the motherboard chamber. I am toying with the idea of adding a second 240mm radiator to the case later, especially if I get a waterblock for the GPU.
I am in an itx case, but I kind of want to try custom loop CPU and GPU... I don't have the extra cash right now thankfully otherwise I'd be diving into a whole new can of worms!
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
I am in an itx case, but I kind of want to try custom loop CPU and GPU... I don't have the extra cash right now thankfully otherwise I'd be diving into a whole new can of worms!
Have you looked at growing into a slightly larger case, like the Fractal Design Node 804?

It is a mATX case which I have a mDTX board in. It is more of a cube in shape, will take mITX boards, have room for water cooling if you don’t use 3.5” drives, albeit it does take some clever effort to get a custom loop to work.

my previous case was a mITX case, a Corsair 250d to be exact. That was a great little case too that I had a 240mm AIO squeezed into along with a GTX 1080 and 5 drives. I liked how the GPU was positioned to suck cool air from outside the case. But it was too small to fit these newer RTX cards.
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
Have you looked at growing into a slightly larger case, like the Fractal Design Node 804?

It is a mATX case which I have a mDTX board in. It is more of a cube in shape, will take mITX boards, have room for water cooling if you don’t use 3.5” drives, albeit it does take some clever effort to get a custom loop to work.

my previous case was a mITX case, a Corsair 250d to be exact. That was a great little case too that I had a 240mm AIO squeezed into along with a GTX 1080 and 5 drives. I liked how the GPU was positioned to suck cool air from outside the case. But it was too small to fit these newer RTX cards.
I chose this itx case for its size and built in handle, for occasional transporting. Lian Li TU150.

407356_20201204_151650.jpg
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
I chose this itx case for its size and built in handle, for occasional transporting. Lian Li TU150.
Yeah, mine I don't need to transport. Instead of a handle on top, mine has 4 fans. Aside from width, the outer dimensions between your case and mine are not too far off. Mine is about .6" deeper but yours is .2" taller. Width, there is a 5.5" difference. This allows me to put my power supply, cabling, radiator, and pump/reservoir behind the motherboard. Both are limited to 320mm GPUs or smaller. But with how much I had to finagle to get the RTX 3090 to fit, I can't imagine how one could squeeze in a full 320mm card.

Node804.jpg
 

mnewxcv

[H]F Junkie
Joined
Mar 4, 2007
Messages
8,606
Yeah, mine I don't need to transport. Instead of a handle on top, mine has 4 fans. Aside from width, the outer dimensions between your case and mine are not too far off. Mine is about .6" deeper but yours is .2" taller. Width, there is a 5.5" difference. This allows me to put my power supply, cabling, radiator, and pump/reservoir behind the motherboard. Both are limited to 320mm GPUs or smaller. But with how much I had to finagle to get the RTX 3090 to fit, I can't imagine how one could squeeze in a full 320mm card.

View attachment 306475
I only had to remove cpu cooler, ram, and both case fans to fit the GPU. Just the normal stuff :D
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
I only had to remove cpu cooler, ram, and both case fans to fit the GPU. Just the normal stuff :D
I had to remove the front fan and the front filter, I can't remember if the rear fan had to be removed too. Then from there it had to go in tilted at around a 45 degree angle on both its x and y axis, inserting it front side first through the hole the filter was covering, then swing the backside in as I rotated it and slowly pulled out the front side to get it to line up with the motherboard. There was some acrobatics involved. :ROFLMAO:
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
Just got my 3090 XC3 ULTRA HYBRID GAMING yesterday.

My score kinda sucks since I only have a 5600X CPU, but I guess it is good enough for now to be the first entry on the water cooled list. :)

https://www.3dmark.com/3dm/54596660?

View attachment 307555

CPU is clocked at 4.7 and on water.
Your graphics score is around what I was seeing with my 3090 FE running at stock clocks. At stock on all settings, my graphics score is 19,720 and an average temp of 56c. With just the power limit increased to 114% I can get a graphics score of 20,336 and an average temp of 59c.

Pull back your memory clocks some or entirely and see if your score stays about the same or if it improves. Because of power limitations, I find that average core clocks are lower with higher memory clocks since they both share from the same power limit. For me, I noticed core clocks start to dip by 13-17MHz when adding +250 to +500. But the +500 memory did help increase my score by a whopping .5% despite the core clock dip, the +250 made little to no difference. Also, sometimes going too high on the memory clocks can have an adverse effect on performance but the memory ECC keeps it from crashing. So, instead, your performance drops. Those memory chips create a lot of heat when cranked up and could be having an adverse effect as well.

Also, take a look at the undervolting thread on here and look at applying a curve to your overclock instead of a full on offset. It might net you more performance and less heat. I am currently running a curve offset of +150 MHz. With that along with a +500 MHz on the memory and the fans set to 100%, I can get a graphics score of 21,169.

Is 104% the highest power limit your card will allow? I am kind of surprised by that considering it is water cooled. Even the FE allows a power limit of 114%. Of course, what would be a more accurate measurement is peak wattage. I hit around 410w or so at peak when my power limit is set to 114%.

I am also a little surprised by your temps considering you are on a water cooling hybrid and have the fans at 100%. I would have expected it to be lower. What's your ambient room temperature? Are the radiator fans pulling in cool air from outside the case or pulling air from within the case?

With fans at 100% I can keep the 3090 at around 50c when running this benchmark and around 56c-62c if fans are left at default settings. 69-72c is about the highest I see my GPU temps get when fans are at default when playing something like RDR2 for extended periods. My idle temps at night are around 24-26c and 28-29c during the day when in use at default fans.

Your card has more potential. You just need to get your settings dialed in.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
470
I'm starting to think, which I believe was mentioned earlier, that we should probably be comparing graphics scores instead of overall scores to get a better idea as to what each of us are able to squeeze out of the graphics card. I am seeing some people's scores get really skewed because of the CPU they are using.

Also, and this maybe too much of a request, can the OP attach the URL next to the score in the 1st post for easier reference. As this thread gets bigger, it becomes more and more difficult to chase down where the link to the score is so that we can do side by side comparisons.
 
Top