KingEngineRevUp
FTW Member
- Total Posts : 1030
- Reward points : 0
- Joined: 2019/03/28 16:38:54
- Status: offline
- Ribbons : 9
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 10:53:10
(permalink)
gsrcrxsi I've used clamp meters. and yes i did read that post, well before you linked it, last time it was suggested earlier in the thread. clamp meter readings fluctuate a good bit, was he dumping the telemetry and averaging the readings? doubtful since cheap clamp meters dont really have that ability. which means he was eyeballing the readings. you can't expect great precision from a $10 Home Depot clamp meter. he also claims to have "verified with his UPS" but probably didnt take into account PSU efficiency, which will show higher power draw from the wall than what the component is using on the DC side.
"The good news is that most benchmarks/games will hit the stock 1068mv voltage limit before 550w. So there's some safety there, although there are some that might exceed that. 600w is probably right at 20amps on #2 (nope didn't test that). Balance seems to remains the same except PCIE appears to pick up 20w or so. I would think that reading was wrong except my UPS readings are just too close to make me think it is. UPS estimate was done by recording the idle power reading, subracting out the GPU idle power, a nd then subtracting that value from the reading and factoring it by 0.88 for efficiency of my PSU.UPS power = [(Total power) - (idle power) + (gpu idle power)]* 0.88"
|
gsrcrxsi
SSC Member
- Total Posts : 985
- Reward points : 0
- Joined: 2010/01/24 19:20:59
- Status: offline
- Ribbons : 5
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:00:58
(permalink)
re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe slot and not the power plugs. I detailed my observations about that earlier in the thread. EVGA power percentage is a calculation of pcie power. if you take observed pcie slot power / 66W you get the same % reading that PX1 reports. which is why GPU says the card is being power limited even though board power is only 85-90% PL
post edited by gsrcrxsi - 2021/07/22 11:03:40
Rig1: EPYC 7V12 | [4] RTX A4000Rig2: EPYC 7B12 | [5] 3080Ti + [2] 2080TiRig3: EPYC 7B12 | [6] 3070Ti + [2] 3060 Rig4: [2] EPYC 7742 | RTX A2000 Rig5: [2] EPYC 7642 Rig6: EPYC 7551 | [4] Titan V
|
KingEngineRevUp
FTW Member
- Total Posts : 1030
- Reward points : 0
- Joined: 2019/03/28 16:38:54
- Status: offline
- Ribbons : 9
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:04:02
(permalink)
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
|
gsrcrxsi
SSC Member
- Total Posts : 985
- Reward points : 0
- Joined: 2010/01/24 19:20:59
- Status: offline
- Ribbons : 5
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:05:31
(permalink)
talk about confirmation bias lol
Rig1: EPYC 7V12 | [4] RTX A4000Rig2: EPYC 7B12 | [5] 3080Ti + [2] 2080TiRig3: EPYC 7B12 | [6] 3070Ti + [2] 3060 Rig4: [2] EPYC 7742 | RTX A2000 Rig5: [2] EPYC 7642 Rig6: EPYC 7551 | [4] Titan V
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:07:50
(permalink)
KingEngineRevUp
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
It's not the "only reasonable explanation" for the issues we are having with the XC3 cards. These cards are power capped for some reason. It's not because of temp, that's for sure. Neither my 3080 Ti XC3 Hybrid or 3080 Ti XC3 Hydro Copper cards can match a 3080 TI FE card, that's terrible. My Hybrid beats my Hydro Copper.
post edited by kram36 - 2021/07/22 11:37:54
|
KingEngineRevUp
FTW Member
- Total Posts : 1030
- Reward points : 0
- Joined: 2019/03/28 16:38:54
- Status: offline
- Ribbons : 9
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:36:17
(permalink)
kram36
KingEngineRevUp
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
It's not the "only reasonable explanation" for the issues we are having with the XC3 cards. These cards are power capped for some reason. It's not because of temp, that's for sure. Neither my 3080 Ti XC3 Hybrid or 3080 Ti XC3 Hydro Copper cards can beat a 3080 TI FE card, that's terrible. My Hybrid beats my Hydro Copper.
Understood. I'm sharing with you that other cards visually seem to be power capped in GPU-Z. They, like you, are seeing similar results. The problem seems to be almost across their board for all 3080 Tis. The 3080 Ti FTW3 and Strix are supposed to be able to do 450W if you max the slider up, but they're not averaging that kind of draw, not even close. They might average 400W, the stock TDP they're rated for. That's far from 450W. Is that not a similar behavior to your XC3? You're hovering around 300W which is the stock TDP, even though you slid the slider to max out and you should be getting 350-363W? Is it hardware? Did NVIDIA have the 3080 Ti designed a certain way to be capped? Or is it software, can we change and manipulate these caps? If it is the latter, it would give hope that a simple update can fix this. If it's hardware, we're screwed. Or is it that we are hitting these power targets, and the sensors are not correctly reporting them? There's a lot to be investigated here with the 3080 Tis.
|
lobstar
Superclocked Member
- Total Posts : 208
- Reward points : 0
- Joined: 2007/03/05 13:08:02
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:41:58
(permalink)
KingEngineRevUp
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
Welcome to the ****ty card design from EVGA. Seems you guys have the same issue we have with the 3090. Check out the XOC thread for the 3090 to see if your issues are similar.
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:43:02
(permalink)
KingEngineRevUp
kram36
KingEngineRevUp
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
It's not the "only reasonable explanation" for the issues we are having with the XC3 cards. These cards are power capped for some reason. It's not because of temp, that's for sure. Neither my 3080 Ti XC3 Hybrid or 3080 Ti XC3 Hydro Copper cards can match a 3080 TI FE card, that's terrible. My Hybrid beats my Hydro Copper.
Understood. I'm sharing with you that other cards visually seem to be power capped in GPU-Z. They, like you, are seeing similar results. The problem seems to be almost across their board for all 3080 Tis. The 3080 Ti FTW3 and Strix are supposed to be able to do 450W if you max the slider up, but they're not averaging that kind of draw, not even close. They might average 400W, the stock TDP they're rated for. That's far from 450W. Is that not a similar behavior to your XC3? You're hovering around 300W which is the stock TDP, even though you slid the slider to max out and you should be getting 350-363W? Is it hardware? Did NVIDIA have the 3080 Ti designed a certain way to be capped? Or is it software, can we change and manipulate these caps? If it is the latter, it would give hope that a simple update can fix this. If it's hardware, we're screwed. Or is it that we are hitting these power targets, and the sensors are not correctly reporting them? There's a lot to be investigated here with the 3080 Tis.
The 3080 Ti XC3 cards have a total power draw of 350W, not 300W. The 3080 Ti FE card also has the same 350W.
|
KingEngineRevUp
FTW Member
- Total Posts : 1030
- Reward points : 0
- Joined: 2019/03/28 16:38:54
- Status: offline
- Ribbons : 9
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 11:54:32
(permalink)
gsrcrxsi talk about confirmation bias lol
Confirmation bias is having an argument with someone and then doing research selectively to find evidence that supports your own and ignoring any counter evidence. Not sure how I've participated in that. I showed some findings from another user about the FTW3 and asked if anyone here is capable of doing something similar or better to the XC3 to try and measure its power draw to compare to software sensors. That's it. kram36
KingEngineRevUp
kram36
KingEngineRevUp
gsrcrxsi re-read my edits. but especially suspect is that he only measured one clamp at a time, and cant dump telemetry. these are big sources of error in my opinion. still no evidence that GPUz readings are wrong IMO. and it's not applicable to the XC3 card anyway, since the limit is clearly the PCIe and not the power plugs.
Well it doesn't explain why several if not all 3080 Tis I have seen DO NOT average, hit and float around their rated maximum TDP like other cards are able to do in the 30 series. The only explanation I have seen is his. So you're free to do your own experiments. He used two different things to do his own readings and that's all we have for now. It's the only reasonable explanation.
It's not the "only reasonable explanation" for the issues we are having with the XC3 cards. These cards are power capped for some reason. It's not because of temp, that's for sure. Neither my 3080 Ti XC3 Hybrid or 3080 Ti XC3 Hydro Copper cards can match a 3080 TI FE card, that's terrible. My Hybrid beats my Hydro Copper.
Okay well Understood. I'm sharing with you that other cards visually seem to be power capped in GPU-Z. They, like you, are seeing similar results. The problem seems to be almost across their board for all 3080 Tis. The 3080 Ti FTW3 and Strix are supposed to be able to do 450W if you max the slider up, but they're not averaging that kind of draw, not even close. They might average 400W, the stock TDP they're rated for. That's far from 450W. Is that not a similar behavior to your XC3? You're hovering around 300W which is the stock TDP, even though you slid the slider to max out and you should be getting 350-363W? Is it hardware? Did NVIDIA have the 3080 Ti designed a certain way to be capped? Or is it software, can we change and manipulate these caps? If it is the latter, it would give hope that a simple update can fix this. If it's hardware, we're screwed. Or is it that we are hitting these power targets, and the sensors are not correctly reporting them? There's a lot to be investigated here with the 3080 Tis.
The 3080 Ti XC3 cards have a total power draw of 350W, not 300W. The 3080 Ti FE card also has the same 350W.
Okay, well gsrcrxsi here sounds like they know the best method to verify what the card is actually drawing so maybe they can do some testing.
post edited by KingEngineRevUp - 2021/07/22 11:58:37
|
gsrcrxsi
SSC Member
- Total Posts : 985
- Reward points : 0
- Joined: 2010/01/24 19:20:59
- Status: offline
- Ribbons : 5
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 12:29:03
(permalink)
I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
Rig1: EPYC 7V12 | [4] RTX A4000Rig2: EPYC 7B12 | [5] 3080Ti + [2] 2080TiRig3: EPYC 7B12 | [6] 3070Ti + [2] 3060 Rig4: [2] EPYC 7742 | RTX A2000 Rig5: [2] EPYC 7642 Rig6: EPYC 7551 | [4] Titan V
|
Aruzedragon
New Member
- Total Posts : 82
- Reward points : 0
- Joined: 2014/12/28 02:52:15
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 12:38:23
(permalink)
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 12:47:05
(permalink)
Aruzedragon
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Go ahead and put it in Step-Up so you are not as far back in the Step-Up queue. It's not like you have to do the Step-Up, but have your options open as soon as possible.
|
Aruzedragon
New Member
- Total Posts : 82
- Reward points : 0
- Joined: 2014/12/28 02:52:15
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 12:48:06
(permalink)
kram36
Aruzedragon
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Go ahead and put it in Step-Up so you are not as far back in the Step-Up queue. It's not like you have to do the Step-Up, but have your options open as soon as possible.
Yeah the only thing that is stopping me is the fact I have to pay 10% sales tax 🤣 again.
|
gsrcrxsi
SSC Member
- Total Posts : 985
- Reward points : 0
- Joined: 2010/01/24 19:20:59
- Status: offline
- Ribbons : 5
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 12:48:46
(permalink)
Personally. My card is “good enough” and the low clock speeds while running Einstein dont seem to impact performance too much since Einstein tends to benefit from memory performance. It’s still the fastest single GPU on the whole project. Beating out RX6900s and Radeon VIIs (the previous king). If I had a case that could fit the FTW3, I might consider swapping out my XC3, but performance is OK and it fits in my case now and came with a waterblock.
Unfortunately I use my card primarily in Linux and the tools to play with voltages just don’t exist. I can only adjust power limit and clock speeds.
Rig1: EPYC 7V12 | [4] RTX A4000Rig2: EPYC 7B12 | [5] 3080Ti + [2] 2080TiRig3: EPYC 7B12 | [6] 3070Ti + [2] 3060 Rig4: [2] EPYC 7742 | RTX A2000 Rig5: [2] EPYC 7642 Rig6: EPYC 7551 | [4] Titan V
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:01:13
(permalink)
Aruzedragon
kram36
Aruzedragon
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Go ahead and put it in Step-Up so you are not as far back in the Step-Up queue. It's not like you have to do the Step-Up, but have your options open as soon as possible.
Yeah the only thing that is stopping me is the fact I have to pay 10% sales tax 🤣 again.
If you're talking about paying the full tax again, that's a bug in EVGA's system. I called them about the tax deal and they told me when my queue time hits I will only get charged tax on the difference in the card prices.
|
Aruzedragon
New Member
- Total Posts : 82
- Reward points : 0
- Joined: 2014/12/28 02:52:15
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:03:16
(permalink)
kram36
Aruzedragon
kram36
Aruzedragon
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Go ahead and put it in Step-Up so you are not as far back in the Step-Up queue. It's not like you have to do the Step-Up, but have your options open as soon as possible.
Yeah the only thing that is stopping me is the fact I have to pay 10% sales tax 🤣 again.
If you're talking about paying the full tax again, that's a bug in EVGA's system. I called them about the tax deal and they told me when my queue time hits I will only get charged tax on the difference in the card prices.
Oh really? So it's not the full sales tax on the card? If that's the case, then here I come FTW3 🤣
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:06:16
(permalink)
Aruzedragon
kram36
Aruzedragon
kram36
Aruzedragon
gsrcrxsi I’ve done lots of testing and reported all my findings. It’s in the thread. You only need to read it. Long story short, memory controller load plays a big part in the behavior outcome on the 3080Ti XC3 for most cards. Sometimes someone gets a good sample and it clocks up higher (and reports a higher power draw across the board) with accompanying increase in clock speeds and benchmark scores. But the vast majority of 3080Ti XC3 users experience similar power and clock limiting behavior with lower than normal power caps.
But I’ve more or less accepted the situation. Until EVGA addresses it, I don’t think anything else can be gained from more testing other than verifying existing results.
EVGA doesn’t seem interested in addressing it though. Unless it gets the same kind of exposure that the Amazon game issue got, EVGA will likely continue to ignore it because they can. XC3 cards are made in far fewer quantities, and will always be a small % of the cards in the wild as a result. The Amazon game issue effected the FTW3 cards in huge numbers so there was enough exposure that they couldn’t ignore it.
This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Go ahead and put it in Step-Up so you are not as far back in the Step-Up queue. It's not like you have to do the Step-Up, but have your options open as soon as possible.
Yeah the only thing that is stopping me is the fact I have to pay 10% sales tax 🤣 again.
If you're talking about paying the full tax again, that's a bug in EVGA's system. I called them about the tax deal and they told me when my queue time hits I will only get charged tax on the difference in the card prices.
Oh really? So it's not the full sales tax on the card? If that's the case, then here I come FTW3 🤣
Nope, you do not pay the full card price tax again. I was on the phone with EVGA about it yesterday and they told me when the queue hits for me the tax will be applied to the difference in the card prices, not the full card price. Here is when I did a Step-Up from a 2080 Super Hydro Copper to a 3080.
post edited by kram36 - 2021/07/22 13:10:09
|
illestrolla
New Member
- Total Posts : 13
- Reward points : 0
- Joined: 2017/05/23 05:34:52
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:07:20
(permalink)
Got all excited when I saw: 12G-P5-3958-KR 6/3/2021 7:47:54 AM PT Yes But now reading the posts here about the XC3 issues has me thinking otherwise.
post edited by illestrolla - 2021/07/22 13:12:52
|
Aruzedragon
New Member
- Total Posts : 82
- Reward points : 0
- Joined: 2014/12/28 02:52:15
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:11:47
(permalink)
illestrolla Got all excited when I saw: 6/3/2021 7:47:54 AM PTYes But now reading the posts here about the XC3 issues has me thinking otherwise.
Tbh, go for it. It's still a decent buy and if your card is power limited, you can always undervolt it
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:33:30
(permalink)
Aruzedragon
illestrolla Got all excited when I saw: 6/3/2021 7:47:54 AM PTYes But now reading the posts here about the XC3 issues has me thinking otherwise.
Tbh, go for it. It's still a decent buy and if your card is power limited, you can always undervolt it
Plus he could always put it in Step-Up and use the card until his turn hits in the queue. That's the only bright side to buying this card TBH. Cost more in the long run, but he would have a card in the meantime.
|
sentient12
New Member
- Total Posts : 19
- Reward points : 0
- Joined: 2021/05/06 12:27:22
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 13:58:00
(permalink)
illestrolla Got all excited when I saw: 12G-P5-3958-KR 6/3/2021 7:47:54 AM PT Yes But now reading the posts here about the XC3 issues has me thinking otherwise.
There has been much hyperbole in this post. The reality is, all of the rtx 3080 and up cards I've seen being toyed with are power limited, the question is just at what power level. Some are closer to spec, but even the 3 pin cards are reporting power draw levels ten percent under their max spec. The self reported power draw on this card is less than the spec for most work loads. It appears to be range from a few percent off of an average performance for the card to pretty typical. This card won't compete with a card with a more careful power design and more available power unless you hit the lottery. The increased cost of the cooler seems to have little effect. If you don't yet have a 3080 level card, they're pretty fun to toy with regardless of power limitations.
|
kram36
The Destroyer
- Total Posts : 20362
- Reward points : 0
- Joined: 2009/10/27 19:00:58
- Location: United States
- Status: offline
- Ribbons : 72
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 14:38:00
(permalink)
sentient12
illestrolla Got all excited when I saw: 12G-P5-3958-KR 6/3/2021 7:47:54 AM PT Yes But now reading the posts here about the XC3 issues has me thinking otherwise.
There has been much hyperbole in this post. The reality is, all of the rtx 3080 and up cards I've seen being toyed with are power limited, the question is just at what power level. Some are closer to spec, but even the 3 pin cards are reporting power draw levels ten percent under their max spec. The self reported power draw on this card is less than the spec for most work loads. It appears to be range from a few percent off of an average performance for the card to pretty typical. This card won't compete with a card with a more careful power design and more available power unless you hit the lottery. The increased cost of the cooler seems to have little effect. If you don't yet have a 3080 level card, they're pretty fun to toy with regardless of power limitations.
The XC3 cards have the same power limit of the FE card, but we are not getting anywhere near what EVGA is advertising the power spec on this card. We are averaging around 300W, some even lower. Even overclocking the XC3, it can't match a Stock FE card in performance. This card is trash, it's the worst as far as 3080 Ti cards go. Just glad there is Step-Up to get me out of EVGA's mess of a card.
|
jboud47
Superclocked Member
- Total Posts : 118
- Reward points : 0
- Joined: 2021/05/10 14:49:25
- Location: Silver Spire
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 15:22:21
(permalink)
This thread is frustrating to read. This card is the closest in my queue and one I was likely to grab.
At this point I'll be passing on it if I get my email. Unacceptable to pay >$1,000 for a card and it not perform as expected. And without a response from EVGA either.
I'll be looking at other manufacturers and FE cards instead this go around.
Pitiful.
"It was I who helped the Prime Evils mastermind their own exile to your world. The plan we set in motion so long ago cannot be stopped by any mortal agency. Hell, itself, is poised to spill forth into your world like a tidal wave of blood and nightmares. You and all your kind... are doomed."
|
neteng101
Superclocked Member
- Total Posts : 153
- Reward points : 0
- Joined: 2018/02/03 06:56:57
- Status: offline
- Ribbons : 1
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 15:40:38
(permalink)
jboud47 I'll be looking at other manufacturers and FE cards instead this go around.
The FE is not better - just different flaws - hot memory being the number one issue, also has a non transferrable warranty. Zotac/Gigabyte/etc all have their issues too. MSI is overpriced. Your only choice if you want about the best you can get right now for Ampere is Asus. And good luck finding them in stock. If you want more performance buy a 3-connector card period, all 3080+ cards are held back if they aren't a 3-connector power design PCB. 3080 Ti's are also overpriced out the gate and not great value - that's Nvidia pricing them to start at $1199 instead of $999. 3080s are much better value if you can find them at manufacturer's prices.
post edited by neteng101 - 2021/07/22 15:46:04
|
RangerRobAZ
New Member
- Total Posts : 15
- Reward points : 0
- Joined: 2020/11/20 10:32:25
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 16:52:35
(permalink)
Aruzedragon This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Would you mind giving some details on the settings that worked for you? Did you undervolted to 816mV (that seems really low)? At what clock speed? Any other adjustments you made that worked? Just trying to get a starting point for when I fiddle with mine. I was already planning to undervolt this card before I read this thread. I am hopeful based on previous undervolting experience that I can shave at least 50-75W off the power draw to make a 300W limit acceptable. I just got the queue pop today and ordered despite the issues. Obviously in a normal market it would be a pass.
|
jboud47
Superclocked Member
- Total Posts : 118
- Reward points : 0
- Joined: 2021/05/10 14:49:25
- Location: Silver Spire
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 16:56:41
(permalink)
neteng101
jboud47 I'll be looking at other manufacturers and FE cards instead this go around.
The FE is not better - just different flaws - hot memory being the number one issue, also has a non transferrable warranty. Zotac/Gigabyte/etc all have their issues too. MSI is overpriced. Your only choice if you want about the best you can get right now for Ampere is Asus. And good luck finding them in stock. If you want more performance buy a 3-connector card period, all 3080+ cards are held back if they aren't a 3-connector power design PCB. 3080 Ti's are also overpriced out the gate and not great value - that's Nvidia pricing them to start at $1199 instead of $999. 3080s are much better value if you can find them at manufacturer's prices.
Sorry should have clarified I meant non-Ti version. Yes the value for any 3080ti's are meh and the ONLY reason I was considering it was because of my position in the queue and my member discount. But I don't NEED a card and don't mind waiting for something of better value for my money.
"It was I who helped the Prime Evils mastermind their own exile to your world. The plan we set in motion so long ago cannot be stopped by any mortal agency. Hell, itself, is poised to spill forth into your world like a tidal wave of blood and nightmares. You and all your kind... are doomed."
|
sentient12
New Member
- Total Posts : 19
- Reward points : 0
- Joined: 2021/05/06 12:27:22
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 17:03:01
(permalink)
RangerRobAZ
Aruzedragon This, I accepted the situation too. It sucks that my card is so limited that I have to overclock at .816 due the my card's limits but at least it's somewhat matching a 3080ti FE by doing that. Still debating if I should do a step up or not but would just like an real reason from EVGA on why the cards are falling into different piles of power draws.
Would you mind giving some details on the settings that worked for you? Did you undervolted to 816mV (that seems really low)? At what clock speed? Any other adjustments you made that worked? Just trying to get a starting point for when I fiddle with mine. I was already planning to undervolt this card before I read this thread. I am hopeful based on previous undervolting experience that I can shave at least 50-75W off the power draw to make a 300W limit acceptable. I just got the queue pop today and ordered despite the issues. Obviously in a normal market it would be a pass.
Beware that doing this isn't an end all fix. It can seriously limit performance in workloads that have a different power vs voltage scaling. In other words, his settings may not be helpful. And synthetic testing is dramatically different than games. I can go +265 core clock in synthetic benchmarks and get about 1,850 mhz stably. When playing some games, this +265 is unstable as the power load is lower and the card boosts well into the 2k range and crashes. But with a +120 in those games it will still boost to just over 2k and be stable. If I had just set an undervolt that took whatever synthetic benchmark I was tuning against off the power limit, I would have given up the top end of the frequency range. Sadly if you want to undervolt, Ie limit the max voltage and specify a core clock, it needs to be application specific or you can give up much performance.
|
sentient12
New Member
- Total Posts : 19
- Reward points : 0
- Joined: 2021/05/06 12:27:22
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 17:06:49
(permalink)
neteng101
jboud47 I'll be looking at other manufacturers and FE cards instead this go around.
The FE is not better - just different flaws - hot memory being the number one issue, also has a non transferrable warranty. Zotac/Gigabyte/etc all have their issues too. MSI is overpriced. Your only choice if you want about the best you can get right now for Ampere is Asus. And good luck finding them in stock. If you want more performance buy a 3-connector card period, all 3080+ cards are held back if they aren't a 3-connector power design PCB. 3080 Ti's are also overpriced out the gate and not great value - that's Nvidia pricing them to start at $1199 instead of $999. 3080s are much better value if you can find them at manufacturer's prices.
I mostly agree with you if you are talking FE card pricing. But the 3080 I had is retailing for 1099 and had a similar power limiting behavior as the card I have now. Moving to the XC3 for $1199 didn't seem like too bad of a hit for a pretty decent performance gain.
|
gsrcrxsi
SSC Member
- Total Posts : 985
- Reward points : 0
- Joined: 2010/01/24 19:20:59
- Status: offline
- Ribbons : 5
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 17:24:03
(permalink)
neteng101
jboud47 I'll be looking at other manufacturers and FE cards instead this go around.
The FE is not better - just different flaws - hot memory being the number one issue, also has a non transferrable warranty. Zotac/Gigabyte/etc all have their issues too. MSI is overpriced. Your only choice if you want about the best you can get right now for Ampere is Asus. And good luck finding them in stock. If you want more performance buy a 3-connector card period, all 3080+ cards are held back if they aren't a 3-connector power design PCB. 3080 Ti's are also overpriced out the gate and not great value - that's Nvidia pricing them to start at $1199 instead of $999. 3080s are much better value if you can find them at manufacturer's prices.
I bought 2 GPUs around the same time. 1 EVGA RTX 3080 Ti XC3 Ultra HC 1 EVGA RTX 3080 XC3 Ultra HC Under the same exact loading (Einstein@home Gamma Ray tasks, which are all very homogenous): the 3080Ti runs limited right around 285W (power limit set to 350W), with clock speeds ~1700MHz the 3080 runs limited at its 320W default power limit, with clocks around 1950MHz Tell me how that makes sense.
Rig1: EPYC 7V12 | [4] RTX A4000Rig2: EPYC 7B12 | [5] 3080Ti + [2] 2080TiRig3: EPYC 7B12 | [6] 3070Ti + [2] 3060 Rig4: [2] EPYC 7742 | RTX A2000 Rig5: [2] EPYC 7642 Rig6: EPYC 7551 | [4] Titan V
|
sentient12
New Member
- Total Posts : 19
- Reward points : 0
- Joined: 2021/05/06 12:27:22
- Status: offline
- Ribbons : 0
Re: 3080 Ti XC3 Ultra Hydro Copper, low power consumption, hidden power cap?
2021/07/22 17:33:15
(permalink)
Considering gsrcrxsi I bought 2 GPUs around the same time. 1 EVGA RTX 3080 Ti XC3 Ultra HC 1 EVGA RTX 3080 XC3 Ultra HC
Under the same exact loading (Einstein@home Gamma Ray tasks, which are all very homogenous): the 3080Ti runs limited right around 285W (power limit set to 350W), with clock speeds ~1700MHz the 3080 runs limited at its 320W default power limit, with clocks around 1950MHz
Tell me how that makes sense.
Along with gsrcrxsi It’s still the fastest single GPU on the whole project
I'm a little confused. Is the 3080 also being used on the same project? If so, how different is the performance between the 3080 and 3080 ti during the same load?
|