• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The Ampere RTX 3080 Owners Thread

doug_1986 said:
That's really helpful, downloaded. Thanks!



Pretty sure this is to be expected.
Click to expand...


So when I paste his DL folder in my MSI Profiles folder it does not read any of it. Also it looks like you can only undervolt these cards with reducing the core clocks as well as vcore? When I did undervolt/underclock, the temps were very low but did notice the 1-3fps drops as the card could no longer reach 2040Mhz on the core (MSi Card). But the card used like 100W less somewhat.

OP, undervolting a cpu usually reduces heat,temps and it still runs at normal speed, but seems for the gpu you loose some speed.
 
Last edited:
Sarge2255 said:
So when I paste the his DL in the Profiles folder it does not read any of it. Also it looks like you can only underclock with reducing the core clocks with vcore? When I did the temps were very low but did notice the 1-3fps drops as the card could no longer reach 2040Mhz on the core (MSi Card). But the card used like 100W less somewhat.
Click to expand...

You may need to copy and paste the contents in your own profile, as I think afterburner may not use if it detects its from a different GPU.

My profiles were made with 0 set in the main UI for clock. As the clocks are configured in the curve. Try not to be obsessed with what clock speed you see, the performance is what matters, the % drop in performance is tiny even if the gap seems bigger on the clock speed, if you are the type of person who likes to see their card hitting 2ghz then my profiles are not for you.

They about power efficiency. The profile I think is the best is the one that reaches 1830mhz, but there is also a 1905mhz profile (same clocks as chroniclard). Profile 3 and 4 respectively. 1800-1900 seems the sweetspot.
 
chrcoluk said:
You may need to copy and paste the contents in your own profile, as I think afterburner may not use if it detects its from a different GPU.

My profiles were made with 0 set in the main UI for clock. As the clocks are configured in the curve. Try not to be obsessed with what clock speed you see, the performance is what matters, the % drop in performance is tiny even if the gap seems bigger on the clock speed, if you are the type of person who likes to see their card hitting 2ghz then my profiles are not for you, they about power efficiency. The profile I think is the best is the one that reaches 1830mhz, but there is also a 1905mhz profile (same clocks as chroniclord). Profile 3 and 4 respectively.
Click to expand...


Thanks, is there a way to undervolt the GPU without reducing the core clocks?
 
Sarge2255 said:
Thanks, is there a way to undervolt the GPU without reducing the core clocks?
Click to expand...

You would need to just set the top part of your voltage curve to stop at your desired clock speed, but of course realise it will need more voltage than the lower speeds, as you go for those last couple of % of performance power efficiency goes out the window, but you should still be able to beat the stock voltage. My profile 5 is closer to 2ghz but isnt stable on my card. That could be used as your starting point.

Also there is an option in afterburner to keep profiles for "similar cards" that "might" allow it to work without copy and paste of contents.
 
To be fair, It's a pretty big ask to drop 100w and still maintain full performance. When I was running benchmarks I was finding that sometimes it was outperforming stock, sometimes just off. I think it's basically the same for me.
 
doug_1986 said:
To be fair, It's a pretty big ask to drop 100w and still maintain full performance. When I was running benchmarks I was finding that sometimes it was outperforming stock, sometimes just off. I think it's basically the same for me.
Click to expand...

Yeah thats what my profile 4 does, in every test I ran it maintains the same performance or even has more performance (because lower temps and power prevents throttling). But my card doesnt reach 2ghz at stock, so its compared to what my card reaches at stock. But profile 3 is much more power efficient than 4, albeit 4 is still noticeably better than stock.

Profile 3 uses over 30% less peak power than profile 1, to show how impressive it is.

Port royale scores

profile 1 score 11331 52.46fps
profile 3 score 11118 51.48fps
profile 4 score 11454 53.03 fps

--

I just checked the profiles guys and its this file that has the curve data inside it.

VEN_10DE&DEV_2206&SUBSYS_146710DE&REV_A1&BUS_1&DEV_0&FN_0.cfg

Note the filename matches the device name in device manager, if it doesnt match your existing filename then contents would have to be copied and pasted.
 
Last edited:
chrcoluk said:
Yeah thats what my profile 4 does, in every test I ran it maintains the same performance or even has more performance (because lower temps and power prevents throttling). But my card doesnt reach 2ghz at stock, so its compared to what my card reaches at stock. But profile 3 is much more power efficient than 4, albeit 4 is still noticeably better than stock.

Port royale scores

profile 1 score 11331 52.46fps
profile 3 score 11118 51.48fps
profile 4 score 11454 53.03 fps

--

I just checked the profiles guys and its this file that has the curve data inside it.

VEN_10DE&DEV_2206&SUBSYS_146710DE&REV_A1&BUS_1&DEV_0&FN_0.cfg

Note the filename matches the device name in device manager, if it doesnt match your existing filename then contents would have to be copied and pasted.
Click to expand...

I'll try that.
This is your Profile 4 Voltage/Frequency curve.
chrcoluk said:
Yeah thats what my profile 4 does, in every test I ran it maintains the same performance or even has more performance (because lower temps and power prevents throttling). But my card doesnt reach 2ghz at stock, so its compared to what my card reaches at stock. But profile 3 is much more power efficient than 4, albeit 4 is still noticeably better than stock.

Profile 3 uses over 30% less peak power than profile 1, to show how impressive it is.

Port royale scores

profile 1 score 11331 52.46fps
profile 3 score 11118 51.48fps
profile 4 score 11454 53.03 fps

--

I just checked the profiles guys and its this file that has the curve data inside it.

VEN_10DE&DEV_2206&SUBSYS_146710DE&REV_A1&BUS_1&DEV_0&FN_0.cfg

Note the filename matches the device name in device manager, if it doesnt match your existing filename then contents would have to be copied and pasted.
Click to expand...


Your profile 4 crashes during TimeSpy :( Your profile 1 does work with my card but in Time Spy I loose almost 1k points and it runs just as hot a my stock temps at 2040mhz core. My undervolt runs at 63c but loose 1-3fps/500 points in Time Spy

I think load profiles of different cards do not work well.
 
Last edited:
doug_1986 said:
In Timespy mine scores 15148

That's dropping the Core to -250, then taking 850mV to 1890 MHz and running the curve from there. This is on a FE.

https://www.3dmark.com/spy/19080919
Click to expand...

My card crashes with Core to -250, then taking 850mV to 1890 MHz and running the curve from there.

Also I can't see RT in Timspy options.

I'm just going to leave my card on stock tbh , reducing the Core by 200 and setting the 850mV to 1815Mhz works fine for me with 63c but I loose 1-3fps and it us not worth it to me at 4k.

My stock card comes overclocked at 2040Mhz and hits 73c max so I would only undervolt if I get the same performance which I just can't seem to get.
 
Sarge2255 said:
I'll try that.
This is your Profile 4 Voltage/Frequency curve.



Your profile 4 crashes during TimeSpy :( Your profile 1 does work with my card but in Time Spy I loose almost 1k points and it runs just as hot a my stock temps at 2040mhz core. My undervolt runs at 63c but loose 1-3fps/500 points in Time Spy

I think load profiles of different cards do not work well.
Click to expand...

Yes each card has different quality silicon, so this may not be a good profitable experiment sadly.

My profile 1 I didnt change anything, I just saved it, so I had an easy way to get back to my card's stock. So either your card has a different voltage curve to mine at stock or maybe you previously had a higher power limit or clock boost dialled in? as it would reset those also.
 
doug_1986 said:
In Timespy mine scores 15148

That's dropping the Core to -250, then taking 850mV to 1890 MHz and running the curve from there. This is on a FE.

https://www.3dmark.com/spy/19080919
Click to expand...

I forgot to mention my scores are on a custom test so wont match up if you just run stock timespy.

"port royale custom 1440p, async on, reflection filtering medium (non demo), reflection mode RT, all off next 5 options, max reflection samples 1, bloom on, filtering mode AF, max AF 16, max TAA 16, TAA on for tests, vsync and fixed fps off"

I can run a stock test later to compare to your score.
 
Sarge2255 said:
My card crashes with Core to -250, then taking 850mV to 1890 MHz and running the curve from there.

Also I can't see RT in Timspy options.

I'm just going to leave my card on stock tbh , reducing the Core by 200 and setting the 850mV to 1815Mhz works fine for me with 63c but I loose 1-3fps and it us not worth it to me at 4k.

My stock card comes overclocked at 2040Mhz and hits 73c max so I would only undervolt if I get the same performance which I just can't seem to get.
Click to expand...

You have an AIB factory o/c card? if so I wouldnt use any of my curves anyway. My card is a founders edition.
 
You must log in or register to reply here.
Back
Top Bottom