Jump to content

Undervolting RTX 3080

OrangeLibra

Hey everyone!

 

I just recently got an MSI 10 GB GDDR6X Gaming Z Trio Card and after using it for a couple of days I am thinking of undervolting it, however im not sure. I still have a couple of questions regarding all of that

 

1. GPU-Z says the Boost clock is 1830 Mhz however ingame I noticed it goes up to 1970 Mhz. Does that mean that the card is factory overclocked and will I need to put it back to its Base Clock before undervolting?

2. I have never undervolted a GPU before, so I dont really know what to do and what to look out for. Is there any video or guide that you guys would recommend?

3. The TDP when playing games like RDR2 and Cyberpunk 2077 is usually around 350W sometimes goes all the way to 360W and the temperatures go to 71c. If the card is undervolted, will this decrease the Temperature AND TDP or will it just decrease the power consumption?

4. Is there anything really bad that can happen while doing this aka can the card somehow break while undervolting?

 

Sorry if these are stupid questions, but id much rather get informed before doing anything and potentialy ruining my build. Looking forward to your replies!

Link to comment
Share on other sites

Link to post
Share on other sites

Undervolting is pretty much pointless if you arent having specific issues. Generally there are 3 main reasons to undervolt. 

 

Thermals, this tends to be something done in the Tiny case builds and such where you have to undervolt to meet thermal thresholds or noise levels since higher fan RPM. Your at 71C, you do not need to undervolt.

 

Instability: Sometimes undervolting/Underclocking memory to fix it can help stabilize a card, this is usually an issue on the 1st few batches of GPUS.  At this point unless you bought a used card, this shouldnt be an issue.

 

Crashing/Power trip: Some 3080s/3090s/ ti variants have Transient spikes that will trigger one of the power trips on the PSU, undervolting a bit can certainly help alleviate this a bit but dont expect major results. 

 

Other cases are people that just like to tinker and play with their hardware to see where they can get a balance of Card performance to thermals, but overall honestly its pretty much a huge waste of time for the vast majority of people, since they have to make sure settings stick often and most people are lazy.

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

42 minutes ago, OrangeLibra said:

Hey everyone!

 

I just recently got an MSI 10 GB GDDR6X Gaming Z Trio Card and after using it for a couple of days I am thinking of undervolting it, however im not sure. I still have a couple of questions regarding all of that

 

1. GPU-Z says the Boost clock is 1830 Mhz however ingame I noticed it goes up to 1970 Mhz. Does that mean that the card is factory overclocked and will I need to put it back to its Base Clock before undervolting?

2. I have never undervolted a GPU before, so I dont really know what to do and what to look out for. Is there any video or guide that you guys would recommend?

3. The TDP when playing games like RDR2 and Cyberpunk 2077 is usually around 350W sometimes goes all the way to 360W and the temperatures go to 71c. If the card is undervolted, will this decrease the Temperature AND TDP or will it just decrease the power consumption?

4. Is there anything really bad that can happen while doing this aka can the card somehow break while undervolting?

 

Sorry if these are stupid questions, but id much rather get informed before doing anything and potentialy ruining my build. Looking forward to your replies!

Like stated above, don’t undervolt unless you have a very specific reason to. Your temps are not an issue at all, so I doubt you have a good reason to do this…

 

Just let the card behave as it wants to, no reason to try and alter it. It is boosting because it has temperature and power headroom. 

Rig: i7 13700k - - Asus Z790-P Wifi - - RTX 4080 - - 4x16GB 6000MHz - - Samsung 990 Pro 2TB NVMe Boot + Main Programs - - Assorted SATA SSD's for Photo Work - - Corsair RM850x - - Sound BlasterX EA-5 - - Corsair XC8 JTC Edition - - Corsair GPU Full Cover GPU Block - - XT45 X-Flow 420 + UT60 280 rads - - EK XRES RGB PWM - - Fractal Define S2 - - Acer Predator X34 -- Logitech G502 - - Logitech G710+ - - Logitech Z5500 - - LTT Deskpad

 

Headphones/amp/dac: Schiit Lyr 3 - - Fostex TR-X00 - - Sennheiser HD 6xx

 

Homelab/ Media Server: Proxmox VE host - - 512 NVMe Samsung 980 RAID Z1 for VM's/Proxmox boot - - Xeon e5 2660 V4- - Supermicro X10SRF-i - - 128 GB ECC 2133 - - 10x4 TB WD Red RAID Z2 - - Corsair 750D - - Corsair RM650i - - Dell H310 6Gbps SAS HBA - - Intel RES2SC240 SAS Expander - - TreuNAS + many other VM’s

 

iPhone 14 Pro - 2018 MacBook Air

Link to comment
Share on other sites

Link to post
Share on other sites

ibid
Cards are pretty good at keeping themselves healthy these days. 

If you're curious you can look at HWMonitor and see if it's being limited by temp or power while you game

5950X/3080Ti primary rig  |  1920X/1070Ti Unraid for dockers  |  200TB TrueNAS w/ 1:1 backup

Link to comment
Share on other sites

Link to post
Share on other sites

Out of the box most cards are WAY overvolted.  Aside from borderline chips.  Its unnecessary power usage, heat and fan noise.  I'm quite sensitive to noise.  The 50-80W I knocked off my 3080 made the fan way more tolerable and the performance is the same.  Actually, eve n if I really push the card to like 450W it gives me less than 5% performance increase.

AMD 7950x / Asus Strix B650E / 64GB @ 6000c30 / 2TB Samsung 980 Pro Heatsink 4.0x4 / 7.68TB Samsung PM9A3 / 3.84TB Samsung PM983 / 44TB Synology 1522+ / MSI Gaming Trio 4090 / EVGA G6 1000w /Thermaltake View71 / LG C1 48in OLED

Custom water loop EK Vector AM4, D5 pump, Coolstream 420 radiator

Link to comment
Share on other sites

Link to post
Share on other sites

15 hours ago, LIGISTX said:

Like stated above, don’t undervolt unless you have a very specific reason to. Your temps are not an issue at all, so I doubt you have a good reason to do this…

 

Just let the card behave as it wants to, no reason to try and alter it. It is boosting because it has temperature and power headroom. 

Nope

Manufacturers tend to overvolt/underclock everything so that it meets base specs and they won't have a lot of RMA and bad press if it doesn't

So usually you can save some watts or gain some clocks or timings on all GPU CPU and RAM

System : AMD R9 5900X / Gigabyte X570 AORUS PRO/ 2x16GB Corsair Vengeance 3600CL18 ASUS TUF Gaming AMD Radeon RX 7900 XTX OC Edition GPU/ Phanteks P600S case /  Eisbaer 280mm AIO (with 2xArctic P14 fans) / 2TB Crucial T500  NVme + 2TB WD SN850 NVme + 4TB Toshiba X300 HDD drives/ Corsair RM850x PSU/  Alienware AW3420DW 34" 120Hz 3440x1440p monitor / Logitech G915TKL keyboard (wireless) / Logitech G PRO X Superlight mouse / Audeze Maxwell headphones

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, PDifolco said:

Nope

Manufacturers tend to overvolt/underclock everything so that it meets base specs and they won't have a lot of RMA and bad press if it doesn't

So usually you can save some watts or gain some clocks or timings on all GPU CPU and RAM

I mean... sure. Most would typically go the route of upping clocks vs reducing voltage, but in a desktop where heat isn't really a concern, I would typically say its not worth introducing potential instability just because someone may have seen a youtube video about undervolting and go ahead and try it.

 

Its similar to RAM overclocking...... So many posts on this forum have started to pop up with people who have instability issues and can't figure out why and have OCed their RAM without actually understanding how difficult a task that is and how hard it is to properly test. Obviously RAM OCing is not quite the same as GPU uindervolting, but the point remains; for most people, just leaving things alone is likely the best approach unless they want to actually invest the time and energy into learning what they are doing, understanding how to properly test, and be ready for when things may become an issue/unstable. This is what drives most of my answers asking about RAM OCing and GPU underclocking (especially for desktop, laptop there is a bit more of an argument), to just let the GPU do its thing.

Rig: i7 13700k - - Asus Z790-P Wifi - - RTX 4080 - - 4x16GB 6000MHz - - Samsung 990 Pro 2TB NVMe Boot + Main Programs - - Assorted SATA SSD's for Photo Work - - Corsair RM850x - - Sound BlasterX EA-5 - - Corsair XC8 JTC Edition - - Corsair GPU Full Cover GPU Block - - XT45 X-Flow 420 + UT60 280 rads - - EK XRES RGB PWM - - Fractal Define S2 - - Acer Predator X34 -- Logitech G502 - - Logitech G710+ - - Logitech Z5500 - - LTT Deskpad

 

Headphones/amp/dac: Schiit Lyr 3 - - Fostex TR-X00 - - Sennheiser HD 6xx

 

Homelab/ Media Server: Proxmox VE host - - 512 NVMe Samsung 980 RAID Z1 for VM's/Proxmox boot - - Xeon e5 2660 V4- - Supermicro X10SRF-i - - 128 GB ECC 2133 - - 10x4 TB WD Red RAID Z2 - - Corsair 750D - - Corsair RM650i - - Dell H310 6Gbps SAS HBA - - Intel RES2SC240 SAS Expander - - TreuNAS + many other VM’s

 

iPhone 14 Pro - 2018 MacBook Air

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, LIGISTX said:

I mean... sure. Most would typically go the route of upping clocks vs reducing voltage, but in a desktop where heat isn't really a concern, I would typically say its not worth introducing potential instability just because someone may have seen a youtube video about undervolting and go ahead and try it.

 

Its similar to RAM overclocking...... So many posts on this forum have started to pop up with people who have instability issues and can't figure out why and have OCed their RAM without actually understanding how difficult a task that is and how hard it is to properly test. Obviously RAM OCing is not quite the same as GPU uindervolting, but the point remains; for most people, just leaving things alone is likely the best approach unless they want to actually invest the time and energy into learning what they are doing, understanding how to properly test, and be ready for when things may become an issue/unstable. This is what drives most of my answers asking about RAM OCing and GPU underclocking (especially for desktop, laptop there is a bit more of an argument), to just let the GPU do its thing.

Ok, that's a position good for standard users that just want the thing to work and not bother

And indeed if you tinker with anything you've got to test it and possibly roll back indeed if system crash or is unstable. 

But my point is that stock components operate always under their max capacity because they're sampled and they're given safe settings

System : AMD R9 5900X / Gigabyte X570 AORUS PRO/ 2x16GB Corsair Vengeance 3600CL18 ASUS TUF Gaming AMD Radeon RX 7900 XTX OC Edition GPU/ Phanteks P600S case /  Eisbaer 280mm AIO (with 2xArctic P14 fans) / 2TB Crucial T500  NVme + 2TB WD SN850 NVme + 4TB Toshiba X300 HDD drives/ Corsair RM850x PSU/  Alienware AW3420DW 34" 120Hz 3440x1440p monitor / Logitech G915TKL keyboard (wireless) / Logitech G PRO X Superlight mouse / Audeze Maxwell headphones

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, PDifolco said:

max capacity because they're sampled

Ish. They are all binned to some extent, but "max" is a bit of a stretch. Some chips run at higher voltages than others, some run less, its somewhat dependent on the silicon quality and somewhat on the cooling the chip is currently seeing, all mixed in with the voltage table that was determined for the chip from the manufacturer/board partner. They can run at higher voltages safely, they just become "not worth the power or heat" trade off for a slight bump in performance.

 

But, yes, as a result of this you usually can reduce the voltage a bit since they are given voltage tables that will make statistically all of the produced chips happy, which means on average yours will require less. 

Rig: i7 13700k - - Asus Z790-P Wifi - - RTX 4080 - - 4x16GB 6000MHz - - Samsung 990 Pro 2TB NVMe Boot + Main Programs - - Assorted SATA SSD's for Photo Work - - Corsair RM850x - - Sound BlasterX EA-5 - - Corsair XC8 JTC Edition - - Corsair GPU Full Cover GPU Block - - XT45 X-Flow 420 + UT60 280 rads - - EK XRES RGB PWM - - Fractal Define S2 - - Acer Predator X34 -- Logitech G502 - - Logitech G710+ - - Logitech Z5500 - - LTT Deskpad

 

Headphones/amp/dac: Schiit Lyr 3 - - Fostex TR-X00 - - Sennheiser HD 6xx

 

Homelab/ Media Server: Proxmox VE host - - 512 NVMe Samsung 980 RAID Z1 for VM's/Proxmox boot - - Xeon e5 2660 V4- - Supermicro X10SRF-i - - 128 GB ECC 2133 - - 10x4 TB WD Red RAID Z2 - - Corsair 750D - - Corsair RM650i - - Dell H310 6Gbps SAS HBA - - Intel RES2SC240 SAS Expander - - TreuNAS + many other VM’s

 

iPhone 14 Pro - 2018 MacBook Air

Link to comment
Share on other sites

Link to post
Share on other sites

On 7/3/2022 at 8:20 AM, Shimejii said:

Undervolting is pretty much pointless if you arent having specific issues. Generally there are 3 main reasons to undervolt. 

 

Thermals, this tends to be something done in the Tiny case builds and such where you have to undervolt to meet thermal thresholds or noise levels since higher fan RPM. Your at 71C, you do not need to undervolt.

 

Instability: Sometimes undervolting/Underclocking memory to fix it can help stabilize a card, this is usually an issue on the 1st few batches of GPUS.  At this point unless you bought a used card, this shouldnt be an issue.

 

Crashing/Power trip: Some 3080s/3090s/ ti variants have Transient spikes that will trigger one of the power trips on the PSU, undervolting a bit can certainly help alleviate this a bit but dont expect major results. 

 

Other cases are people that just like to tinker and play with their hardware to see where they can get a balance of Card performance to thermals, but overall honestly its pretty much a huge waste of time for the vast majority of people, since they have to make sure settings stick often and most people are lazy.

 

 

This is not correct - undervolting can lead to lower power usage and lower temperatures. I have undervolted my 3080 to use about 70 watts less for almost the same performance. This also results in less heat as well. So it is definitely worth a try. 

CPU i7 14700K | CPU Cooler Noctua NH-U12A | Motherboard MSI Pro Z690-A | GPU Zotac Airo RTX 4080 | RAM 32 GB GSkill Ripjaws V 4400
Mhz |
 Monitor Alienware AW2721D / Gigabyte M28U | PSU ASUS ROG Strix 850G

Link to comment
Share on other sites

Link to post
Share on other sites

7 hours ago, Bobbysixjp said:

This is not correct - undervolting can lead to lower power usage and lower temperatures. I have undervolted my 3080 to use about 70 watts less for almost the same performance. This also results in less heat as well. So it is definitely worth a try. 

it is 100% correct. Unless you are having issues with thermals, noise levels, or crashing,  most really wont care to do it unless that is something you enjoy doing. You said ALMOST the same performance, but not the same. Yes is it 98% of it? Sure, but the vast majority of people simply do not care if their card runs a bit hotter to get every last bit out of it. 

 

I did clearly state the three main reasons why you would undervolt, and some people genuinely enjoy tinkering and such and that is fine, its your hardware! Enjoy it to the full extent you can, but most simply wont give a shit. Its the same reason why most people dont Overclock, or try memory tuning, because in the end its marginal gains for a ton of stability testing wheras if it plugs in and just works without any tinkering and they are getting what they paid for, they are happy with it. Is it fun to OC, tune your card and cpu and such? Sure it can be, but it can also be a headache if things crash and other issues arise. Some people simply dont have the time nor will power to do that. 

Link to comment
Share on other sites

Link to post
Share on other sites

On 7/2/2022 at 4:12 PM, OrangeLibra said:

.

if you know what you are doing, undervolting is more efficient, and if the pc is close to you acting as a space heater, for ampere most start at 0.8v 1900mhz and start bumping the clocks from there.

 

If you have never done it before, go look at a guide, leave it stock, or just use the optimized oc in afterburner.

5950x 1.33v 5.05 4.5 88C 195w ll R20 12k ll drp4 ll x570 dark hero ll gskill 4x8gb 3666 14-14-14-32-320-24-2T (zen trfc)  1.45v 45C 1.15v soc ll 6950xt gaming x trio 325w 60C ll samsung 970 500gb nvme os ll sandisk 4tb ssd ll 6x nf12/14 ippc fans ll tt gt10 case ll evga g2 1300w ll w10 pro ll 34GN850B ll AW3423DW

 

9900k 1.36v 5.1avx 4.9ring 85C 195w (daily) 1.02v 4.3ghz 80w 50C R20 temps score=5500 ll D15 ll Z390 taichi ult 1.60 bios ll gskill 4x8gb 14-14-14-30-280-20 ddr3666bdie 1.45v 45C 1.22sa/1.18 io  ll EVGA 30 non90 tie ftw3 1920//10000 0.85v 300w 71C ll  6x nf14 ippc 2000rpm ll 500gb nvme 970 evo ll l sandisk 4tb sata ssd +4tb exssd backup ll 2x 500gb samsung 970 evo raid 0 llCorsair graphite 780T ll EVGA P2 1200w ll w10p ll NEC PA241w ll pa32ucg-k

 

prebuilt 5800 stock ll 2x8gb ddr4 cl17 3466 ll oem 3080 0.85v 1890//10000 290w 74C ll 27gl850b ll pa272w ll w11

 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×