EVGA

Helpful ReplyEVGA GeForce RTX 3090 FTW3 XOC BIOS

Page: << < ..2122232425.. > >> Showing page 24 of 240
Author
Post
Essentials Only Full Version
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 17:53:29 (permalink)
raleighsmvp
wrt something repeatable - I just averaged 480W on the MSI kombustor suite's default test "(GL) MSI-01", run @ 3840x2160 and clicked the "X32" button (see red square in screenshot) . ftw3 ultra, gpu +111, 0 voltage, 0 mem, 119 power

edit: averaged == sustained, as confirmed by resetting gpuz counters after clicking "X32"

edit2: I'm running the stress test, not the benchmark



Very cool, thanks for posting the full log and average power.  I wonder if being at 83C and 1950 mhz @ 1.043v is keeping you from the last 20w.
post edited by kevinc313 - 2020/11/01 17:56:28
#691
kring
iCX Member
  • Total Posts : 274
  • Reward points : 0
  • Joined: 2007/01/19 09:06:42
  • Status: offline
  • Ribbons : 1
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 18:26:02 (permalink)
tubnotub1
A lot of us, including myself, have tested multiple video games at 4k, myself a sampling of 7 games using 4+ minute long sampling periods and recording Max/Avg board power and PerfLimit, in every case, I saw pwr usage <450 at max w/ PerfLimit Pwr being the limit, w/ the highest test being BL3 at something like 449.96 watts. I understand the last couple pages of posts have been pretty solidly about benchmarks, but prior to the last couple of days there was quite a bit of discussion and numbers recorded involving real games. At this point the only games that seem have been reported to hit >450 watts are Doom Eternal, Quake RTX and Civ 6, at least w/ what has been reported in this thread. 



I’m not disputing the low power a lot of people are seeing - I’m just saying the 3DMark benchies simply don’t push the GPU‘s power, so people shouldn’t be relying on them.  I still don’t get near 500W on my 3090 FTW3 Ultra, but I get more than original bios
post edited by kring - 2020/11/01 18:32:43
#692
tubnotub1
New Member
  • Total Posts : 27
  • Reward points : 0
  • Joined: 2006/01/10 23:29:49
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 18:59:23 (permalink)
Can confirm similarly promising results as to what raleighsmvp saw w/ Kombuster. GL MSI-01 test, 20 minute test run, max draw 493.9, avg board draw 458. Avg draw on the board was probably closer to 465 for the duration of the test. When I started I began the run w/ my standard gaming OC, +125/+900, 100% voltage, 119 power. Max power went up to but did not exceed 473, mostly bouncing between 445->465. At the 10 minute mark I took the card back to stock core/mem offsets, as I and others have seen higher power draw when the card is not running w/ an offset, upon setting +0/+0 I saw an immediate jump (literally as soon as I clicked the apply settings button) from 473 max watts to 486. Power draw from that point forward bounced between 455->475. Max temp was 72c, equilibrium was at 71c. Just another data point to add. 

https://imgur.com/a/8Qbrcc4
#693
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 19:26:08 (permalink)
tubnotub1
Can confirm similarly promising results as to what raleighsmvp saw w/ Kombuster. GL MSI-01 test, 20 minute test run, max draw 493.9, avg board draw 458. Avg draw on the board was probably closer to 465 for the duration of the test. When I started I began the run w/ my standard gaming OC, +125/+900, 100% voltage, 119 power. Max power went up to but did not exceed 473, mostly bouncing between 445->465. At the 10 minute mark I took the card back to stock core/mem offsets, as I and others have seen higher power draw when the card is not running w/ an offset, upon setting +0/+0 I saw an immediate jump (literally as soon as I clicked the apply settings button) from 473 max watts to 486. Power draw from that point forward bounced between 455->475. Max temp was 72c, equilibrium was at 71c. Just another data point to add. 

https://imgur.com/a/8Qbrcc4




Very interesting.  What happens if you run it at 119% power, then start stepping down power limit in small increments and resetting GPU-Z to see if the rolling average changes.  Does it have an immediate effect to average power or do you have to get down to 107% before anything really changes?
post edited by kevinc313 - 2020/11/01 19:28:42
#694
PadinnPlays
New Member
  • Total Posts : 53
  • Reward points : 0
  • Joined: 2020/10/07 23:56:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 19:33:42 (permalink)
I wonder if the increased power draw at lower offset is due to better sustained clocks. Gamers nexus discussed this in their 3080FE review, they found more stable fps and minimums with no offset.
#695
kring
iCX Member
  • Total Posts : 274
  • Reward points : 0
  • Joined: 2007/01/19 09:06:42
  • Status: offline
  • Ribbons : 1
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 19:36:14 (permalink)
You should lower you memory clocks - that speed is causing ECC errors which will actually pause calculations and cause the GPU to rerun them, which means you could see the power draw drop while its in that paused state. This memory doesn’t clock over ~275 without generating errors. Max power draw would happen under zero-error conditions. Same applies to Both clocks, so going over ~+120 / ~+240 can result in reduced power draw.
post edited by kring - 2020/11/01 19:41:17
#696
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 21:06:01 (permalink)
kring
tubnotub1
A lot of us, including myself, have tested multiple video games at 4k, myself a sampling of 7 games using 4+ minute long sampling periods and recording Max/Avg board power and PerfLimit, in every case, I saw pwr usage <450 at max w/ PerfLimit Pwr being the limit, w/ the highest test being BL3 at something like 449.96 watts. I understand the last couple pages of posts have been pretty solidly about benchmarks, but prior to the last couple of days there was quite a bit of discussion and numbers recorded involving real games. At this point the only games that seem have been reported to hit >450 watts are Doom Eternal, Quake RTX and Civ 6, at least w/ what has been reported in this thread. 



I’m not disputing the low power a lot of people are seeing - I’m just saying the 3DMark benchies simply don’t push the GPU‘s power, so people shouldn’t be relying on them.  I still don’t get near 500W on my 3090 FTW3 Ultra, but I get more than original bios


Just as a point of information: I've been testing these bios on my second card (X Trio) and get a steady 500W in Port Royale and Time Spy, peaking at between 506W-511W.
 
So I can confirm that 3DMark benches actually DO push the GPU. There is no reason why the FTW3 cards should be any different. I continue to go back and forth with whether I want to open my FTW3U box. Keeping an eye on this thread; no real rush to make a decision. I'm hoping for the best, as I'd really prefer to keep the FTW3. But I can't deny that a wee bit concerned about the situation and the radio silence regarding it.
 
Then again, maybe I shouldn't really care and the concern echo chamber is amplifying how much this matters more than is warranted. One thing I learned about 500W...is it's way too much heat on air. I greatly preferred to undervolt this card to about 925mV and lock it to about 2050MHz. It's like -85W of power draw and 10C off the core. Practical difference in gaming isn't even perceptible. And isn't actual gaming what matters here? If you're really competitive about this, you're probably under water anyway (or will soon be), and the EVGA cards will be even better there.
 
Just thinking out loud, I guess.
#697
tubnotub1
New Member
  • Total Posts : 27
  • Reward points : 0
  • Joined: 2006/01/10 23:29:49
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 21:12:22 (permalink)
kring
You should lower you memory clocks - that speed is causing ECC errors which will actually pause calculations and cause the GPU to rerun them, which means you could see the power draw drop while its in that paused state. This memory doesn’t clock over ~275 without generating errors. Max power draw would happen under zero-error conditions. Same applies to Both clocks, so going over ~+120 / ~+240 can result in reduced power draw.


I can't confirm this on my card through testing. Synthetics and gameplay show positive scaling (or within margin) up to +1050 at which point I see performance degradation until +1400 when the card artifacts/hard locks. Silicon quality is variable, I don't know that anyone can make a definitive statement that ampere (core clock) or the GDDR6X (mem clock) will only reach a certain pre-determined offset at which point performance degradation will occur, especially considering base clock changes from model to model, as can the VRAM IC (19 versus 21). Memory overclocking on GDDR6X is certainly more complicated w/ the advent of ECC on VRAM, but I am pretty comfortable w/ my offsets at the very least *not* being of detriment to performance. Of note, most (if not all) 3090s are equipped w/ the 21 Gb/s ICs, Nvidia is running them conservatively @ 9751 MHz which is ~19.5 Gb/s, whereas 21 Gb/s would be ~10501 MHz, or a +750 MHz offset. When taking that into consideration, seeing stability and performances increase up to a +950 offset doesn't seem so unlikely. You are only going ~200 MHz above spec, assuming your card did not receive lower binned ICs, which could have been a determining factor in Nvidia's choice to run them at 19.5 Gb/s, lower binned ICs would have a lower price tag.

@PadinnPlays I don't know that the clocks were any more stable, at least not that I could tell. Removing the offset certainly reduced the avg clock speed as can be seen when comparing the max clock and the avg clock in the image, but I don't know if the clocks were any more or less stable. 

@Kevin will do some more testing at a later time and post results. 

@ajreynolw I would leave it in the box until a week or two before big-big Navi drops then make a decision based on how this BIOS matures. We have seen repeated reports from other users of 3-pin cards from other AIBs that sustain 500w+ load on this BIOS in Timespy and other benchmarks. Honestly, that's where all the confusion is coming from and why a lot of us would like EVGA to just come forward and give us a definitive answer on the cause, even if we have no timeframe for a fix. For what it's worth, 2025 on .925 is *so* much better than my 3090 FTW3 Ultra can do, obviously, the lottery is the lottery, but if this FTW3 Ultra I have could do 2Ghz+ at that low of voltage I probably wouldn't be in this thread. My sample is 100% mediocre.  
 




 
 
post edited by tubnotub1 - 2020/11/01 21:26:54
#698
Kylearan
iCX Member
  • Total Posts : 288
  • Reward points : 0
  • Joined: 2013/12/26 04:04:40
  • Status: offline
  • Ribbons : 2
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 21:31:40 (permalink)
tubnotub1
Reflashed, let the benchmark run through all scenes to get all data points. Ended up running it twice because stupid centrix client disables PSC. Same(ish) results. 

https://imgur.com/a/U4AEDRu





I've seen similar things happen before.  I think I saw this on a Vega 64 where a softpowerplay mod performed differently when bypassing the power limit under certain conditions (e.g. after a reboot or normally, or after a driver crash+VPU recovery with clocks reset and reloading the same Afterburner profile).
 
When you got your 500W result, did you do it after running another game or application or benchmark?  Did you do it after a warm windows reboot or after a complete AC power off?  Try to see if you can replicate the condition that you managed to pull 500W, because there is always a way to reproduce a result like this.  Finding that way will be extremely helpful.
#699
drunknfoo
Superclocked Member
  • Total Posts : 117
  • Reward points : 0
  • Joined: 2018/10/02 22:12:55
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/01 23:20:11 (permalink)
xtra long fuzzy donut run
see current and avg

Attached Image(s)

#700
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 07:14:55 (permalink)
tubnotub1
@ajreynol I would leave it in the box until a week or two before big-big Navi drops then make a decision based on how this BIOS matures. We have seen repeated reports from other users of 3-pin cards from other AIBs that sustain 500w+ load on this BIOS in Timespy and other benchmarks. Honestly, that's where all the confusion is coming from and why a lot of us would like EVGA to just come forward and give us a definitive answer on the cause, even if we have no timeframe for a fix. For what it's worth, 2025 on .925 is *so* much better than my 3090 FTW3 Ultra can do, obviously, the lottery is the lottery, but if this FTW3 Ultra I have could do 2Ghz+ at that low of voltage I probably wouldn't be in this thread. My sample is 100% mediocre. 

Yea I'm one of those AIB owners.
 
I have a 3090 Trio X as well and 500W load is what it does in Port Royal and Time Spy, sustained, every time, no problem. Destiny 2 is also a good test if you all have it (it's F2P). Just crank up the render resolution to like 125%, put all the settings to max, and you'll see max power draw.


#701
Stickboy46
Superclocked Member
  • Total Posts : 158
  • Reward points : 0
  • Joined: 2019/07/17 08:16:22
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 07:21:20 (permalink)
It seems like we have some people that have both a FTW3 Ultra and another partner 3090.  Can one of you guys say if you see any ACTUAL benefit from the 500 watts?  Like an actual FPS difference in games ?  I'm curious how much of difference it makes.

i7-10700kf @ 5.1
Corsair H150i Elite Capellix
MSI Z490 ACE
8gbx4 G.Skill TridentZ DDR 4000CL15
EVGA 3090 Kingpin
Corsair AX1200i
Lian Li O11 XL Dynamic
#702
QueueCumber
Superclocked Member
  • Total Posts : 239
  • Reward points : 0
  • Joined: 2007/04/19 09:55:46
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 07:51:17 (permalink)
I have a two dongle MSI Ventus. That likely doesn't count lol....


#703
QueueCumber
Superclocked Member
  • Total Posts : 239
  • Reward points : 0
  • Joined: 2007/04/19 09:55:46
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 07:53:46 (permalink)
I haven't seen amy performance change in games from.the XOC bios. Nothing I notice at least. FPS doesn't seem different. I clock lower in OCs. I do get over 450W, but it seems to only work arbitrarily and with no performance difference. If anything my performance has been worse in benchmarks, even after trying afterburner and removing Precision X.


#704
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 07:56:21 (permalink)
Stickboy46
It seems like we have some people that have both a FTW3 Ultra and another partner 3090.  Can one of you guys say if you see any ACTUAL benefit from the 500 watts?  Like an actual FPS difference in games ?  I'm curious how much of difference it makes.


There really aren't many real benefits on air. People are getting a few hundred extra points in benchmarks. I can post a couple of benchmarks, though.  I have a few games with built-in benchmarks. No problem.
 
You'd really have to put these cards under water to avoid the thermal throttling the cards see when they eclipse 70C or so. The colder the card, the higher frequencies. Combined with the higher power available, you'd see a several percents worth of improvement over stock on air.
 
QueueCumber
I haven't seen amy performance change in games from.the XOC bios. Nothing I notice at least. FPS doesn't seem different. I clock lower in OCs. I do get over 450W, but it seems to only work arbitrarily and with no performance difference. If anything my performance has been worse in benchmarks, even after trying afterburner and removing Precision X.
Is PX1 recommended over Afterburner for these test? Afterburner seems to be doing the job well? Are there any advantages that PX1 offers other than RGB control?
post edited by ajreynol - 2020/11/02 07:58:23
#705
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:01:56 (permalink)
Stickboy46
It seems like we have some people that have both a FTW3 Ultra and another partner 3090.  Can one of you guys say if you see any ACTUAL benefit from the 500 watts?  Like an actual FPS difference in games ?  I'm curious how much of difference it makes.


As promised:
 
Basic Setup
FTW3: Stock bios. +90 / +800 / 107% 100% core voltage, 100% fan; capping at about 1950 in gaming, peak 2050. power draw 425W-450W, peak 453W.
XTrio: 500W bios. +160 / +1125 / 119% 100% core voltage, 100% fan; capping at about 2085 in gaming, peak 2145. power draw 450-495W, peak 505W.
 
Gears Tactics
FTW3: 113.9
XTrio: 110.1
 
Shadow of The Tomb Raider (min / max / average / 95%)
FTW3: 81 / 198 / 107 / 84
XTrio: 84 / 203 / 110 / 86
 
Unigine Heaven (Extreme preset) (min / max / average / score)
FTW3: 50.7 / 571.9 / 283.1 / 7132
XTrio: 48.5 / 588.6 / 288.8 / 7275
 
Port Royal (best run)
FTW3: 13,306
XTrio: 13,821
 
Time Spy (best run, Graphics Score)
FTW3: 19,642
XTrio: 20,127
 
Notes
CPU is not overclocked, IF @ 1900, RAM @3800 16-17-17-34-52. This is done on open air, with a fairly high ambient temp, and silicon RNG between these two cards should be considered. Performance would likely improve with a lower ambient temperature, but I'm in Florida and my AC is actually offline right now. If I were to guess, my ambient is probably around 25C. Afterburner used for frequency adjustments, and those were the max stable settings in all games tested and benchmarks.
 
500W bios was not flashed to the FTW3 because of the clear and widely reported issues. No reason to use that bios if the odds are good to great that I'd get lower performance from it.
 
Oh, and it's crazy how much quieter the XTrio is. I had the door off during testing and I easily forgot the XTrio was at max fan speed. XTrio @ 100% fan volume is roughly equivalent to FTW3 @ 70% fan speed. FTW3 is demonstrably louder. That said, its fans spin about 100rpm faster or thereabout, allowing the get about 1C cooler on max fan speed, with no load (41C vs 40C).
 
Game Observations
I found that Gears tactics isn't a good benchmark for comparing GPUs. Every run the game runs a different simulation than the run before. Only way to use it would be to average 10 runs and it's not that serious to me, so I just took the best result after a few runs. Everything else was pretty consistently placing the XTrio better, but only by a few frames.
 
I don't have any other games installed that have a benchmark built in and I wasn't about to download 50GB+ just to get another bench which will show the same thing.
 
Conclusions
So yea, at $150 more, the FTW3U feels like a tough sell. My understanding is that both companies are cool with swapping the cooler for a waterblock in the event of a problem. EVGA overall peace of mind has its own value of course (and why many of us have stuck with EVGA over the years). FWIW, I'd say the RGB on the FTW3 card is prettier from the side than the more conservative light bar that the MSI has. Whether makes the FTW3 worth the $150 premium over an XTrio (the lowest priced 3-pin card), however, is up to the individual and availability. And of course, if you're planning to dunk your card under water, how it looks doesn't matter at all. I'll continue to mull it, but I think the XTrio is going to stay. If another Bios update comes out in the next few days that allows the FTW3 to take a decisive lead, I'll keep the FTW3. Otherwise, I can't really justify keeping it.
 
Currently under-volting the XTrio - 2064MHz @ 950mV and saving about 80W.
 
Images of benches for those interested: https://imgur.com/a/b4chZaQ
#706
evgatorator
New Member
  • Total Posts : 38
  • Reward points : 0
  • Joined: 2016/10/15 10:07:01
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:08:38 (permalink)
ajreynol
Stickboy46
It seems like we have some people that have both a FTW3 Ultra and another partner 3090.  Can one of you guys say if you see any ACTUAL benefit from the 500 watts?  Like an actual FPS difference in games ?  I'm curious how much of difference it makes.


As promised:
 
Basic Setup
FTW3: Stock bios. +90 / +800 / 107% 100% core voltage, 100% fan; capping at about 1950 in gaming, peak 2050. power draw 425W-450W, peak 453W.
XTrio: 500W bios. +160 / +1125 / 119% 100% core voltage, 100% fan; capping at about 2085 in gaming, peak 2145. power draw 450-495W, peak 505W.
 
Gears Tactics
FTW3: 113.9
XTrio: 110.1
 
Shadow of The Tomb Raider (min / max / average / 95%)
FTW3: 81 / 198 / 107 / 84
XTrio: 84 / 203 / 110 / 86
 
Unigine Heaven (Extreme preset) (min / max / average / score)
FTW3: 50.7 / 571.9 / 283.1 / 7132
XTrio: 48.5 / 588.6 / 288.8 / 7275
 
Port Royal (best run)
FTW3: 13,306
XTrio: 13,821
 
Time Spy (best run, Graphics Score)
FTW3: 19,642
XTrio: 20,127
 
Notes
CPU is not overclocked, IF @ 1900, RAM @3800 16-17-17-34-52. This is done on open air, with a fairly high ambient temp, and silicon RNG between these two cards should be considered. Performance would likely improve with a lower ambient temperature, but I'm in Florida and my AC is actually offline right now. If I were to guess, my ambient is probably around 25C. Afterburner used for frequency adjustments, and those were the max stable settings in all games tested and benchmarks.
 
500W bios was not flashed to the FTW3 because of the clear and widely reported issues. No reason to use that bios if the odds are good to great that I'd get lower performance from it.
 
Oh, and it's crazy how much quieter the XTrio is. I had the door off during testing and I easily forgot the XTrio was at max fan speed. XTrio @ 100% fan volume is roughly equivalent to FTW3 @ 70% fan speed. FTW3 is demonstrably louder. That said, its fans spin about 100rpm faster or thereabout, allowing the get about 1C cooler on max fan speed, with no load (41C vs 40C).
 
Game Observations
I found that Gears tactics isn't a good benchmark for comparing GPUs. Every run the game runs a different simulation than the run before. Only way to use it would be to average 10 runs and it's not that serious to me, so I just took the best result after a few runs. Everything else was pretty consistently placing the XTrio better, but only by a few frames.
 
I don't have any other games installed that have a benchmark built in and I wasn't about to download 50GB+ just to get another bench which will show the same thing.
 
Conclusions
So yea, at $150 more, the FTW3U feels like a tough sell. My understanding is that both companies are cool with swapping the cooler for a waterblock in the event of a problem. EVGA overall peace of mind has its own value of course (and why many of us have stuck with EVGA over the years). FWIW, I'd say the RGB on the FTW3 card is prettier from the side than the more conservative light bar that the MSI has. Whether makes the FTW3 worth the $150 premium over an XTrio (the lowest priced 3-pin card), however, is up to the individual and availability. And of course, if you're planning to dunk your card under water, how it looks doesn't matter at all. I'll continue to mull it, but I think the XTrio is going to stay. If another Bios update comes out in the next few days that allows the FTW3 to take a decisive lead, I'll keep the FTW3. Otherwise, I can't really justify keeping it.
 
Currently under-volting the XTrio - 2064MHz @ 950mV and saving about 80W.
 
Images of benches for those interested: https://imgur.com/a/b4chZaQ


Thank you for taking the time to run these tests and record the results and your observations to post here. Very much appreciated.
#707
Stickboy46
Superclocked Member
  • Total Posts : 158
  • Reward points : 0
  • Joined: 2019/07/17 08:16:22
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:32:34 (permalink)
Interesting.  So the Gaming X Trio even with the 500Watts is only 2% or so better than stock bios FTW3.  That's definitely not worth giving up the EVGA warranty and CS for me .. especially if EVGA can release a fix.  

i7-10700kf @ 5.1
Corsair H150i Elite Capellix
MSI Z490 ACE
8gbx4 G.Skill TridentZ DDR 4000CL15
EVGA 3090 Kingpin
Corsair AX1200i
Lian Li O11 XL Dynamic
#708
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:34:04 (permalink)
evgatorator
Thank you for taking the time to run these tests and record the results and your observations to post here. Very much appreciated.

Sure thing. Not sure if it's helpful or not. I'm not sure if either card is a good chip or if they're both average and I'm not sure how much 50W will help this FTW3 in the first place. Just feels like lots of questions. 
#709
chester_306
New Member
  • Total Posts : 14
  • Reward points : 0
  • Joined: 2010/11/04 12:36:16
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:35:01 (permalink)
ajreynol
Stickboy46
It seems like we have some people that have both a FTW3 Ultra and another partner 3090.  Can one of you guys say if you see any ACTUAL benefit from the 500 watts?  Like an actual FPS difference in games ?  I'm curious how much of difference it makes.


As promised:
 
Basic Setup
FTW3: Stock bios. +90 / +800 / 107% 100% core voltage, 100% fan; capping at about 1950 in gaming, peak 2050. power draw 425W-450W, peak 453W.
XTrio: 500W bios. +160 / +1125 / 119% 100% core voltage, 100% fan; capping at about 2085 in gaming, peak 2145. power draw 450-495W, peak 505W.
 
Gears Tactics
FTW3: 113.9
XTrio: 110.1
 
Shadow of The Tomb Raider (min / max / average / 95%)
FTW3: 81 / 198 / 107 / 84
XTrio: 84 / 203 / 110 / 86
 
Unigine Heaven (Extreme preset) (min / max / average / score)
FTW3: 50.7 / 571.9 / 283.1 / 7132
XTrio: 48.5 / 588.6 / 288.8 / 7275
 
Port Royal (best run)
FTW3: 13,306
XTrio: 13,821
 
Time Spy (best run, Graphics Score)
FTW3: 19,642
XTrio: 20,127
 
Notes
CPU is not overclocked, IF @ 1900, RAM @3800 16-17-17-34-52. This is done on open air, with a fairly high ambient temp, and silicon RNG between these two cards should be considered. Performance would likely improve with a lower ambient temperature, but I'm in Florida and my AC is actually offline right now. If I were to guess, my ambient is probably around 25C. Afterburner used for frequency adjustments, and those were the max stable settings in all games tested and benchmarks.
 
500W bios was not flashed to the FTW3 because of the clear and widely reported issues. No reason to use that bios if the odds are good to great that I'd get lower performance from it.
 
Oh, and it's crazy how much quieter the XTrio is. I had the door off during testing and I easily forgot the XTrio was at max fan speed. XTrio @ 100% fan volume is roughly equivalent to FTW3 @ 70% fan speed. FTW3 is demonstrably louder. That said, its fans spin about 100rpm faster or thereabout, allowing the get about 1C cooler on max fan speed, with no load (41C vs 40C).
 
Game Observations
I found that Gears tactics isn't a good benchmark for comparing GPUs. Every run the game runs a different simulation than the run before. Only way to use it would be to average 10 runs and it's not that serious to me, so I just took the best result after a few runs. Everything else was pretty consistently placing the XTrio better, but only by a few frames.
 
I don't have any other games installed that have a benchmark built in and I wasn't about to download 50GB+ just to get another bench which will show the same thing.
 
Conclusions
So yea, at $150 more, the FTW3U feels like a tough sell. My understanding is that both companies are cool with swapping the cooler for a waterblock in the event of a problem. EVGA overall peace of mind has its own value of course (and why many of us have stuck with EVGA over the years). FWIW, I'd say the RGB on the FTW3 card is prettier from the side than the more conservative light bar that the MSI has. Whether makes the FTW3 worth the $150 premium over an XTrio (the lowest priced 3-pin card), however, is up to the individual and availability. And of course, if you're planning to dunk your card under water, how it looks doesn't matter at all. I'll continue to mull it, but I think the XTrio is going to stay. If another Bios update comes out in the next few days that allows the FTW3 to take a decisive lead, I'll keep the FTW3. Otherwise, I can't really justify keeping it.
 
Currently under-volting the XTrio - 2064MHz @ 950mV and saving about 80W.
 
Images of benches for those interested:




Very well detailed. much appreciated, Thank you! 
#710
Stickboy46
Superclocked Member
  • Total Posts : 158
  • Reward points : 0
  • Joined: 2019/07/17 08:16:22
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:36:49 (permalink)
ajreynol
evgatorator
Thank you for taking the time to run these tests and record the results and your observations to post here. Very much appreciated.

Sure thing. Not sure if it's helpful or not. I'm not sure if either card is a good chip or if they're both average and I'm not sure how much 50W will help this FTW3 in the first place. Just feels like lots of questions. 


At minimum it feels like a very minimal gain that isn't worth stressing over.  Definitely not worth going through the hassle of trying to get a different card at this point lol

i7-10700kf @ 5.1
Corsair H150i Elite Capellix
MSI Z490 ACE
8gbx4 G.Skill TridentZ DDR 4000CL15
EVGA 3090 Kingpin
Corsair AX1200i
Lian Li O11 XL Dynamic
#711
zogge
New Member
  • Total Posts : 68
  • Reward points : 0
  • Joined: 2020/10/12 02:59:38
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:43:37 (permalink)
I got an answer from EVGA on this topic through an informal channel, but in essence it is still being looked at and no conclusion is available yet. I will not give any further details as I do not want to spread any information they do not want to share as official statement.
#712
ajreynol
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2012/12/28 17:24:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 12:43:52 (permalink)
Stickboy46
Interesting.  So the Gaming X Trio even with the 500Watts is only 2% or so better than stock bios FTW3.  That's definitely not worth giving up the EVGA warranty and CS for me .. especially if EVGA can release a fix.  


Note: that's just the 2 samples of cards that I have. Someone else's cards may be better, I couldn't say. But I wouldn't expect more than a few percentage points of difference on air with no AC or something pushing temps down. Maybe 5% at most? On air, the 500W would give you more overhead to push power through without throttling due to heat. But I don't know how much better MY cards would fare.
 
Undervolting my XTrio, I get a PR score only 60 points lower. 2064MHz @ 0.950mV. That's 80W less power draw. Not sure why I can't get more out of 80W of additional power. Others are seeing ~500 points worth of uplift or more from it. But again, it may all be connected to ambient temps and AC. I'm not really sure.
post edited by ajreynol - 2020/11/02 12:46:00
#713
PadinnPlays
New Member
  • Total Posts : 53
  • Reward points : 0
  • Joined: 2020/10/07 23:56:48
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 16:54:21 (permalink)
kevinc313
raleighsmvp
wrt something repeatable - I just averaged 480W on the MSI kombustor suite's default test "(GL) MSI-01", run @ 3840x2160 and clicked the "X32" button (see red square in screenshot) . ftw3 ultra, gpu +111, 0 voltage, 0 mem, 119 power

edit: averaged == sustained, as confirmed by resetting gpuz counters after clicking "X32"

edit2: I'm running the stress test, not the benchmark



Very cool, thanks for posting the full log and average power.  I wonder if being at 83C and 1950 mhz @ 1.043v is keeping you from the last 20w.




Hmm, well, with no overclock or voltage increase, and just increasing power/temp limit, I am locked at 440w as my average. If I maxed out the voltage slider I see a small increase. What temperatures are you getting though, I am betting I am hitting a thermal limit because its around 76c (and thats with max fans)
#714
compuclinic
Superclocked Member
  • Total Posts : 121
  • Reward points : 0
  • Joined: 2013/12/27 12:33:37
  • Location: OHIO
  • Status: offline
  • Ribbons : 1
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 18:25:33 (permalink)
Still,   can't come anywhere near the 500 watts.  I'm barely able to hit over 450 :/     I have tried mimicking settings shown as well as my own testing.    100% fans,  and got the room down to like 40-50 F  for some testing.    Temps aren't limiting me I don't think unless this chip just likes to be colder than air can do.

My card crashes if I go anything over 120 GPU and memory feels unstable over 800.   Maybe I just lost out on the GPU lottery.  I really hope EVGA addresses this via RMA or something as we spent a lot of money for FTW performance and now I feel like I should have kept the ASUS TUF 3090 I had first.  

Now I wondering if I should just get the Strix and wash my hands clean of this thing and perhaps EVGA forever.   I'm OK with beta bios being the issue.   There are a TON of variables in play here.  I'm really glad to see people are getting great results. I'm ticked at the fact they aren't communicating anything with us.
 
 I bought an Optimus waterblock with the utmost faith in this EVGA card and now we see stuff like this coming out and we just have to not know.  Thankfully for EVGA most won't return because the resale value on these still supersedes MSRP so what do they care if we sell our card.   

This to me is just really frustrating,   I'd love to have someone just say hey we perhaps got you a bad card,  poo happens,  here is an RMA with a better card.  Maybe even bin it a bit for a brotha as its clear anyone returning for this,  cares.    Hell, anyone who paid the extra premium for this card to get the extra power connector cares.    

Also would be cool if they just said we screwed up,  the shunt resistor fixes it,   we won't deny warranty claims for the sheer fact you fixed our screw up.   Personally,  I'd be ok with that,  maybe my cards a gem that just needs moar powa,   sucks to RMA if 2 seconds of solder work demoed by countless pros,  like Gamers Nexus resolves the problem.   I'd never know.  

People should NOT rely on Youtube sources alone.     When something trends they make video as fast as possible before their competitors with clickbait.  The only ones I see handle this properly all the time is GamersNexus.  


post edited by compuclinic - 2020/11/02 18:27:34

2020 Pandemic build.
10900k @ 5.3 ghz All core  1.34v -(100 sp rating) 
Asus Maximus XII Formula -  EVGA RTX 3090 FTW   -  CORSAIR 1200i -  G.Skill Trident Royal z 4100 @cl15 -16-16-16 -  
2tb Corsair PCiE gen4,   1tb Samsung 960,   3x Samsung 860 Pro 1 TB 

Heatkiller Pro 4 -  Optimus XL Waterblock   EK 360 PE,  EK 360 XE, EK 480SE Radiators with a D5 pump.  

Samsung G9 Odyssey x2,  LG 34GK950F 144hz 1440p  and its 2560x1080 predecessor.    
Asus Chakram  -  Asus Strix Flare   -  
Sennheiser PC -350   Logitech Z906 5.1 
RGB to the max for the first time. 
#715
Den-ko
New Member
  • Total Posts : 64
  • Reward points : 0
  • Joined: 2017/06/20 16:17:13
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 19:40:43 (permalink)
zogge
I got an answer from EVGA on this topic through an informal channel, but in essence it is still being looked at and no conclusion is available yet. I will not give any further details as I do not want to spread any information they do not want to share as official statement.

Why was dr.disrespect banned?
#716
MK6gti12
New Member
  • Total Posts : 21
  • Reward points : 0
  • Joined: 2020/02/24 11:24:21
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 20:24:45 (permalink)
I'm not on the new bios but monitoring power limit with msi afterburner. Set it to +7 and I'm seeing 15-120% power limit in games with my in the 1900 range.
#717
Amped47
New Member
  • Total Posts : 6
  • Reward points : 0
  • Joined: 2012/05/04 06:44:56
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 20:54:14 (permalink)
I think some of these benchmarks just arent stressing the gpu enough or something. I was stressing over this trying to get my card to hit 500w on a bunch of 3dmark runs but the second i went to play 7 days to die.. at 1440p.. the card was sustaining 480-502w and hitting 120% power. Doesnt make sense to me but clearly the bios works.
#718
jdmwrxpower
Superclocked Member
  • Total Posts : 116
  • Reward points : 0
  • Joined: 2020/09/20 03:53:12
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 22:03:34 (permalink)
Damn well I just ordered the card because I received the notification email and I read back about 10 pages and now I am wondering if the ftw3 was worth the price...
#719
Amped47
New Member
  • Total Posts : 6
  • Reward points : 0
  • Joined: 2012/05/04 06:44:56
  • Status: offline
  • Ribbons : 0
Re: EVGA GeForce RTX 3090 FTW3 XOC BIOS BETA 2020/11/02 22:47:16 (permalink)
jdmwrxpower
Damn well I just ordered the card because I received the notification email and I read back about 10 pages and now I am wondering if the ftw3 was worth the price...

Its a amazing card. Most of these people are probably overreacting to a beta bios. Enjoy the card and dont stress over this.
#720
Page: << < ..2122232425.. > >> Showing page 24 of 240
Jump to:
  • Back to Mobile