Nvidia rtx 3090 discussion

I just got the FE, is it just me or is this card ridiculously quiet for being FE? Also the temps hover between 68-71C playing RDR2, cyberpunk and crysis remastered, not sure how this is possible?!
The giant heat sink,
I just got the FE, is it just me or is this card ridiculously quiet for being FE? Also the temps hover between 68-71C playing RDR2, cyberpunk and crysis remastered, not sure how this is possible?!
How did you get your hands on one?
 
I just got the FE, is it just me or is this card ridiculously quiet for being FE? Also the temps hover between 68-71C playing RDR2, cyberpunk and crysis remastered, not sure how this is possible?!
Because it's a monument of overengineering.
 
Worth going from a 3080 to 3090 for gaming?
Worth it? No. If you have the patience just wait for the 3080 Ti.

I only went with the 3090 because I got tired waiting for the 3080 Ti which will likely also be extremely hard to get for months after release. And to be blunt, I have the disposable income to waste on impatience. :p
 
Worth going from a 3080 to 3090 for gaming?
I would say no as well. I have zero regrets going with a 3090, but the 3080 is the sweet spot for sure. The opportunity presented itself for a 3090 so I went with what was available at the time.
 
  • Like
Reactions: djoye
like this
I need more frames. I already run my card at 2040 average clocks. I can do higher but get instability after 2-3 hours of play time. Was thinking of nabbing a FTW3 hybrid 3090 that may run at 2100 sustained clocks and give about a 15% performance boost.
 
Worth going from a 3080 to 3090 for gaming?
Like others, I bought the 3090 because that's what was available (also direct from EVGA, not scalped). The 3090 probably does ~10FPS better than the 3080 (at least in the benchmarks I've seen), but I'm not sure the cost is worth it. I've also not seen anything use over 9GB VRAM yet (I saw that usage with 2K res scaled 200% in Mirror's Edge Catalyst). While you might get a few extra FPS, you might not even use half the RAM on the card yet. I'm hoping to have the card for ~4 years, so maybe it'll be worth it in the long run, but I agree with others that the 3080 is probably good enough for most gaming cases.

In games, my XC3 Hybrid seems to hit 2020MHz then settle around 1995MHz and max temp has been 53C, that's without any tweaking; it would be disappointing if these (especially the FTW) couldn't be easily pushed a bit more. The only thing I don't like about these EVGA AIOs is that their water block doesn't cover everything. Gigabyte's AIO cards cover everything (including the power components), but good luck getting one of those. I had a Gigabyte 1080 w/AIO and it was solid.
 
My EVGA 3090 XC3 Ultra came in today! Tossed it in the system and it runs great. It has very minor coil whine; at most 1/3rd the amount my 3090 TUF had. It’s hardly noticeable unless you listen for it.

The card clocks about the same as my 3090 TUF. Both boost to about 2050-2075 MHz before they begin running into stability issues. The XC3 definitely runs warmer and louder than the TUF, but that’s a non-issue for me since it’ll be under a waterblock in 2-3 weeks. Overall, very satisfied after initial impressions!
 
My EVGA 3090 XC3 Ultra came in today! Tossed it in the system and it runs great. It has very minor coil whine; at most 1/3rd the amount my 3090 TUF had. It’s hardly noticeable unless you listen for it.

The card clocks about the same as my 3090 TUF. Both boost to about 2050-2075 MHz before they begin running into stability issues. The XC3 definitely runs warmer and louder than the TUF, but that’s a non-issue for me since it’ll be under a waterblock in 2-3 weeks. Overall, very satisfied after initial impressions!
Some early testing to see what the card can do, haven't tried higher than this since its under air cooling:

+104% Power Limit
+150 Core
+500 Memory
100% Fan Speed (to get closer to what it'll be like once under the waterblock)

Has been stable running Heaven for the past 3 hours. Clock speeds consistently in the 1925-2050 range, max recorded boost of 2115. Stays around 65*C.

Might have gotten a decently good chip. Won't dial in the exact performance until its under a waterblock. 100% fan speed on this thing is obnoxiously loud.
 
  • Like
Reactions: JMCB
like this
Some early testing to see what the card can do, haven't tried higher than this since its under air cooling:

+104% Power Limit
+150 Core
+500 Memory
100% Fan Speed (to get closer to what it'll be like once under the waterblock)

Has been stable running Heaven for the past 3 hours. Clock speeds consistently in the 1925-2050 range, max recorded boost of 2115. Stays around 65*C.

Might have gotten a decently good chip. Won't dial in the exact performance until its under a waterblock. 100% fan speed on this thing is obnoxiously loud.
I got similar results without bumping the power limit up, max fan speed 60%. The card would throttle quickly with anything over power limit 100%.
 
I got similar results without bumping the power limit up, max fan speed 60%. The card would throttle quickly with anything over power limit 100%.
Just tried with 100% power limit and it’s stable after an hour of Heaven so far. Oddly, it now peak boosts to 2130 and is closer to 1950-2000 on average (occasionally 2000-2080) compared to the 104% power limit which would mostly stay around 1925-1950.

Not sure what is going on but it looks like I still have a lot of testing ahead of me. I’m wondering once it’s under water if I’ll be able to maintain 2000 minimum boost clock since it’ll be a few bins better in regards to temps.
 
I’m possibly kicking myself, but I passed on a MSI SUPRIM 3090 today at microcenter. I know nothing about that line and wasn’t sure it was worth ~350 over the founders edition.

I’m heading back in tomorrow morning - are there any 3090s I should absolutely get, or absolutely stay away from, if I’m planning to put a water block on them?
 
I’m possibly kicking myself, but I passed on a MSI SUPRIM 3090 today at microcenter. I know nothing about that line and wasn’t sure it was worth ~350 over the founders edition.

I’m heading back in tomorrow morning - are there any 3090s I should absolutely get, or absolutely stay away from, if I’m planning to put a water block on them?
for starters, make sure there is a waterblock available for them. IE, a founders card, reference card, or strix card. Someone can chime in if there are other cards with blocks available.
 
for starters, make sure there is a waterblock available for them. IE, a founders card, reference card, or strix card. Someone can chime in if there are other cards with blocks available.
Evga FTW has blocks available, so they are on my list with Strix and founders. Everything else, I’m not so sure
 
Evga FTW has blocks available, so they are on my list with Strix and founders. Everything else, I’m not so sure
just look up if a model is a 'reference' design. I think some gigabyte and zotac cards are reference (not founders), possibly msi as well.
 
Played around with my 3090 XC3 a bit more today. Got 20,949 in Time Spy. Tried to crack 21K but couldn’t do it without running into stability issues.

But this is running in my i7 6950X system which is about equivalent to a 6700K in gaming performance. Once it’s water cooled and paired with my Ryzen 9 5950X I’m hoping to be around 21.5K.

The GPU boosts to 1900-2000 under 80-100% load and up to 2100 when below 80% load.

Not bad for a GPU with a stock power limit of 366w. (y) Might flash it with a 390w BIOS in the future to see how much further it’ll go.
 
Played around with my 3090 XC3 a bit more today. Got 20,949 in Time Spy. Tried to crack 21K but couldn’t do it without running into stability issues.

But this is running in my i7 6950X system which is about equivalent to a 6700K in gaming performance. Once it’s water cooled and paired with my Ryzen 9 5950X I’m hoping to be around 21.5K.

The GPU boosts to 1900-2000 under 80-100% load and up to 2100 when below 80% load.

Not bad for a GPU with a stock power limit of 366w. (y) Might flash it with a 390w BIOS in the future to see how much further it’ll go.
My stock TimeSpy is a measly 13645 with the EVGA 3090 XC3 Ultra. That is a huge difference between yours and mine. What is your Time Spy Extreme if you can test at stock settings total and graphics score? Thanks
 
Played around with my 3090 XC3 a little more this morning, this time with undervolting. Currently running 2010 MHz (+185) @ 925mV and +400 MHz on the memory.

By undervolting I was able to increase my average clockspeed from 1905 MHz to 1937 MHz in Time Spy (despite a 400 lower score since it no longer peak boosts to 2100 but instead 1995).

With a bit more tweaking I think I should be able to hit 1950 sustained. I'm pretty confident that under a waterblock I'll be able to get it to maintain close to 1980-2000 MHz and if I flash with the 390w BIOS I think I'll be able to get it to 2025-2050 MHz sustained pretty easily with this chip as it looks like I'm unfortunately power limited with the current 367w BIOS. Not bad for a 2x 8-pin PCI-E connector 3090 power limited to 367w.
 
  • Like
Reactions: noko
like this
Anecdotal: I have an acquaintance that binned 5 different 3090's and kept the best performer. 3090 FE, 2 x STRIX, TrioX, FTW3. The TrioX with the KINGPIN 520W bios ended up beating everything else and he kept that. Noted the FE ran 17-20C cooler than all other cards in same test environment and tests (incredible) - 58C, vs 75C for the AIB cards.

Obviously ASIC RNG is a big factor - a different set of the same cards could yield completely different results.
 
Last edited:
Anecdotal: I have an acquaintance that binned 5 different 3090's and kept the best performer. 3090 FE, 2 x STRIX, TrioX, FTW3. The TrioX with the KINGPIN 520W bios flashed ended up beating everything else and he kept that. He noted the FE ran 17-20C cooler than all other cards in same test environment and tests (incredible) - 58C vs 75C for the AIB cards.

Obviously ASIC RNG is a big factor - a different set of the same cards could yield completely different results.
ASIC quality is the biggest factor. I’ve read of a couple 3090 FTW3s on OCN and on the EVGA forum that can’t even match my 3090 XC3 despite their 450w XOC BIOS. They couldn’t even boost to 2100 without crashing and their sustained core clocks were in the upper 1800s in TimeSpy despite VF curve optimizations. Unfortunately for them they just got complete duds. Probably have the same chance at having a golden chip as having a dud like they have.

At the same time, I’ve read of many 3090 FTW3s boosting as high as 2300 and maintaining clock speeds near 2100. There my 3090 XC3 can’t even come close due to being power limited. A lot of it is luck.

But for me, I don’t see value in pushing 400-500w+ through the GPU as a daily driver. The 5-6% max performance increase just isn’t worth the massive power draw and thermal increase. I only see value in them for benchmark leaderboards and that’s it.
 
Anecdotal: I have an acquaintance that binned 5 different 3090's and kept the best performer. 3090 FE, 2 x STRIX, TrioX, FTW3. The TrioX with the KINGPIN 520W bios ended up beating everything else and he kept that. Noted the FE ran 17-20C cooler than all other cards in same test environment and tests (incredible) - 58C, vs 75C for the AIB cards.

Obviously ASIC RNG is a big factor - a different set of the same cards could yield completely different results.
I just got a FE card and my Aorus Xtreme is faster, quieter and cooler but honestly it's expected with higher power limit with 3rd PCIE and a huge ass 4 slot cooler lol. Surprising thing is the FE has really bad coil whine while the Aorus one is silent. Not bothered trying to OC either of them. Honestly kind of considering dumping the FE card even though I really wanted one, I guess I should slap it in a case and see how bad it is when installed since open bench is deceiving.

Person asked above why I went 3090 over 3080? 3090 cards are like 10 times easier to buy because not everybody can afford them since the price is so absurd.
 
I just got a FE card and my Aorus Xtreme is faster, quieter and cooler but honestly it's expected with higher power limit with 3rd PCIE and a huge ass 4 slot cooler lol. Surprising thing is the FE has really bad coil whine while the Aorus one is silent. Not bothered trying to OC either of them. Honestly kind of considering dumping the FE card even though I really wanted one, I guess I should slap it in a case and see how bad it is when installed since open bench is deceiving.

Person asked above why I went 3090 over 3080? 3090 cards are like 10 times easier to buy because not everybody can afford them since the price is so absurd.

I got the FE 3090 for the same reason. I was able to snag a 3090 and 3060ti off bestbuy drop but not 3080. Looks like lot more buyers jump at 3080.

But totally different experience for me with the FE. 0 coil whine unless I just cant hear it in the case lol. But hands down cool and quiet. I have not seen the damn thing go above 60s. I am sure it will in some games but for me anything else more expensive would not have been worth the $$ over the founders edition. It is really top notch when it comes to power delivery and cooling.
 
  • Like
Reactions: noko
like this
I got the FE 3090 for the same reason. I was able to snag a 3090 and 3060ti off bestbuy drop but not 3080. Looks like lot more buyers jump at 3080.

But totally different experience for me with the FE. 0 coil whine unless I just cant hear it in the case lol. But hands down cool and quiet. I have not seen the damn thing go above 60s. I am sure it will in some games but for me anything else more expensive would not have been worth the $$ over the founders edition. It is really top notch when it comes to power delivery and cooling.
I noticed on mine the fins on the heatsink on the exposed part are also not secured that well and move and pop back in place if you push it. It is kind of odd and I did not expect it.

 
Anyone with a 3090 FTW3 have any issues? Small part of me is tempted to get a FTW3 (have an outstanding order on Amazon for one at the original MSRP) but am hesitant due to two reasons:

1) My 3090 XC3 clocks very well for its 367w PL. Essentially getting the same clocks as a higher leakage FTW3 and it has next to no coil whine. It’s basically as good as it gets for the XC3, so the FTW3 I get could show little to no advantage if I am not as lucky with the bin.

2) The EVGA forums have several threads about users experiencing a “red light of death” with their FTW3 cards. There is an audible pop and the GPU dies with a red light that shows above one of the 8 pin connectors. The cause hasn’t been identified and EVGA has remained silent about it. Replacement FTW3 cards are also experiencing the issue so it hasn’t been fixed [yet]. This is my bigger concern.
 
Last edited:
Interesting. My 3090 XC3 has a VBIOS power limit of 366w. Decided to try out GPU-Z today and noticed my card maxes out at 370.5w in each run. See it sit between 366-368w pretty regularly. Both above the power limit.

Really tempted to try out a 390w BIOS as my card is begging for more power based on testing I’ve done so far.
 
Interesting. My 3090 XC3 has a VBIOS power limit of 366w. Decided to try out GPU-Z today and noticed my card maxes out at 370.5w in each run. See it sit between 366-368w pretty regularly. Both above the power limit.

Really tempted to try out a 390w BIOS as my card is begging for more power based on testing I’ve done so far.
How are thermals?
 
How are thermals?
50°C. Have the fans maxed to simulate (or at least get close to) when I finally get a waterblock. It’s also with an overclock on an undervolt of 2070 @ 975mV (+195 I believe) and maxed power limit. Only stable for bench though. Otherwise I run 1995 @ 950mV (+125) which is 100% stable.
 
Anyone with a 3090 FTW3 have any issues? Small part of me is tempted to get a FTW3 (have an outstanding order on Amazon for one at the original MSRP) but am hesitant due to two reasons:

1) My 3090 XC3 clocks very well for its 367w PL. Essentially getting the same clocks as a higher leakage FTW3 and it has next to no coil whine. It’s basically as good as it gets for the XC3, so the FTW3 I get could show little to no advantage if I am not as lucky with the bin.

2) The EVGA forums have several threads about users experiencing a “red light of death” with their FTW3 cards. There is an audible pop and the GPU dies with a red light that shows above one of the 8 pin connectors. The cause hasn’t been identified and EVGA has remained silent about it. Replacement FTW3 cards are also experiencing the issue so it hasn’t been fixed [yet]. This is my bigger concern.
I have been running the 3090 FTW3 since mid November without issues on air. I have not bothered messing with the bios since I am still waiting for the EK water block(delayed until Feb now).
It has been stable at 1995mhz and even boosted up to 2025 on occasion with max temps around 75(usually around 65-70). All stock fan settings but I did add 125 to core with precision.
 
I have been running the 3090 FTW3 since mid November without issues on air. I have not bothered messing with the bios since I am still waiting for the EK water block(delayed until Feb now).
It has been stable at 1995mhz and even boosted up to 2025 on occasion with max temps around 75(usually around 65-70). All stock fan settings but I did add 125 to core with precision.
I’m guessing most are fine. Issues always are exacerbated on the internet as those with the issues complain while those that have no issues have no reason to post about it on the internet.
 
Yeah, my 3090 FTW3 has been rocking 450W power limit and never tops 70C under gaming load with average boost (with OC) around 2025Mhz ~ 2055Mhz. Also got the memory at +500Mhz.

I actually really like the HSF on this thing.
 
Averaging about 50-55C on my Kingpin on heavy gaming. Better cooling than i expected on it to be honest.
 
Interesting. My 3090 XC3 has a VBIOS power limit of 366w. Decided to try out GPU-Z today and noticed my card maxes out at 370.5w in each run. See it sit between 366-368w pretty regularly. Both above the power limit.

Really tempted to try out a 390w BIOS as my card is begging for more power based on testing I’ve done so far.
No shunt mod? Push that sucker to 450w :D.

Anyways with new 3960x, found out that turning off SMT makes a huge jump in performance in TimeSpy for the EVGA 3090 XC3 Ultra with Threadripper, this is with stock GPU settings. Also system memory speeds also has a big impact on performance. Anyways this is a comparison of before and after with no OC on the 3090, all stock settings and just changing system memory speeds from 3200mhz to 3733mhz and turning off SMT. Impact is huge!

https://www.3dmark.com/compare/spy/17329195/spy/17213622
 
No shunt mod? Push that sucker to 450w :D.

Anyways with new 3960x, found out that turning off SMT makes a huge jump in performance in TimeSpy for the EVGA 3090 XC3 Ultra with Threadripper, this is with stock GPU settings. Also system memory speeds also has a big impact on performance. Anyways this is a comparison of before and after with no OC on the 3090, all stock settings and just changing system memory speeds from 3200mhz to 3733mhz and turning off SMT. Impact is huge!

https://www.3dmark.com/compare/spy/17329195/spy/17213622
Wow, that's nuts how the CPU/SMT/Memory was absolutely killing your performance. Really enjoyed your comparison and somewhat unexpected.

Your AMD CPU should crush my overclocked 5960x (if you look at the CPU scores), but with your memory at a lower speed and with SMT, your graphics score was abysmal and held back horribly. For reference, my 3090 and stock was around 18K too before overclocking it, but that is also with "slower" 3000Mhz memory and my 5960x at 4.625Ghz. I get 21K+ in graphics score with a decent GPU OC.
 
Back
Top