Tested by Christopher “Revenant” Moine
Senior Editor
 
Update 1-20-2024: More detailed fan tests were conducted in the thermals sections. Those test charts were added, and fan-speed normalized chart updated. These changes do not reflect a change in the overall recommendation of the card.
Update 1-21-2024: Added acoustic spectrum analysis. Added pictures confirming GPU width, fan diameter, and fan hub diameter.
 
Let’s not mince words: The GeForce RTX 4070 SUPER is what the GeForce RTX 4070 should have been a year ago. It performs like a RTX 3090, uses only 220 watts, and doesn’t feel like a giant step backward in pricing compared to the generation before it. The reference model is $599, which is $100 less expensive than the $699 GeForce RTX 3080, while performing on par with the $1,499 GeForce...

Continue reading...
 
Last edited:

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
I've also ordered a faster Arctic P12 Max fan to see if that helps with the cooling. There's definitely a significant difference when I leave the case open.
 

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
hoping you could also take a look at the Palit Dual and Inno3D 4070 Supers one day if possible - curious to see how they perform in an SFF build versus Zotac
 

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827
hoping you could also take a look at the Palit Dual and Inno3D 4070 Supers one day if possible - curious to see how they perform in an SFF build versus Zotac

I'd love to but SFFN doesn't have a media contact with those companies. Maybe if we can get to Computex we could get a media representative for them.
 

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827
Ugh, I missed that. Seems only Palit and Inno3D have true dual slots other than Zotac

It's very easy to miss. Their marketing team was deceptive with that picture. They made it look like they were comparing it to another card and that theirs was 2-slots by putting dual slot in massive fonts and 2.3 slot in tiny numbers. The truth is that it's a 3-slot card. It takes up 3 slots. It removes functionality of a third slot. It doesn't measure the same as 2 slots. Even if your card is 2.05 slots, than it's three slots.

That's the wonderful thing about objective measurements: They are what they are, and they don't care about your marketing or feelings.
 
  • Like
Reactions: Donmigs85

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
UPDATE: Installed a new P12 Max fan for exhaust earlier and set it to 70%. Much better temps now (GPU 75-78C max instead of 80-84C, hotspot 88-92C instead of 95C and memory junction 84C compared to 90+C before) and the GPU fans don't exceed 72% whereas before they would hit 80-90%. This is with +10% power limit and the mild OC too. Room temperature is 26C. So yeah, this card demands either an open-air case or very strong airflow especially if you don't undervolt.
 
  • Like
Reactions: rfarmer

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827
UPDATE: Installed a new P12 Max fan for exhaust earlier and set it to 70%. Much better temps now (GPU 75-78C max instead of 80-84C, hotspot 88-92C instead of 95C and memory junction 84C compared to 90+C before) and the GPU fans don't exceed 72% whereas before they would hit 80-90%. This is with +10% power limit and the mild OC too. Room temperature is 26C. So yeah, this card demands either an open-air case or very strong airflow especially if you don't undervolt.

Cut the +10 and leave the mild OC. I'm assuming +150 range for the core. See if it's stable.
 
  • Like
Reactions: Donmigs85

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
Discovered another weird quirk with this card - if I run the Time Spy Extreme stress test after a cold boot, the fans will spin up as high as 82-83%. But, if I run the test again right after the fan will only peak at 73% instead for the duration of the 2nd run. Temps are higher on the first run (GPU reaches 83C, hotspot 95C, VRAM 90C). On the 2nd run clocks are the same but the GPU peaks at 78-80C, hotspot 90-92C and VRAM 86 to 88C. It's very odd and consistently reproducible.
 
  • Like
Reactions: Phuncz

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827
Discovered another weird quirk with this card - if I run the Time Spy Extreme stress test after a cold boot, the fans will spin up as high as 82-83%. But, if I run the test again right after the fan will only peak at 73% instead for the duration of the 2nd run. Temps are higher on the first run (GPU reaches 83C, hotspot 95C, VRAM 90C). On the 2nd run clocks are the same but the GPU peaks at 78-80C, hotspot 90-92C and VRAM 86 to 88C. It's very odd and consistently reproducible.

Is this stock or with the overclock?
 

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
What's funny is the Japanese reviewer and a couple others from Hong Kong and Taiwan seem to be getting much better temps and lower noise than we are.
 

Revenant

Christopher Moine - Senior Editor SFF.N
Original poster
Revenant Tech
SFFn Staff
Apr 21, 2017
1,740
2,827
It actually does it both stock and with the +10% power limit and +100/1000 core/VRAM. Not sure if MSI Afterburner is influencing this or maybe it's a VBIOS bug.

I'll try to replicate it on the one we have.
 
  • Like
Reactions: Donmigs85

Donmigs85

Caliper Novice
Feb 6, 2024
24
7
Any idea also if the caps and chokes they're using are decent quality based on the pics on the Japanese site? I noticed they use fewer than some of the other cheap variants.