GeForce RTX 2080 Overclocking Preview with Scanner @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
56,437
GeForce RTX 2080 Overclocking Preview with Scanner

We have had very little time to play with our new GeForce RTX 2080 Founders Edition video cards, but we did want to see just how well this new GPU overclocked out of the box. We put NVIDIA's OC Scanner technology to work and give you our thoughts about just how well it works and what results we got.

If you like our content, please support HardOCP on Patreon.
 
My current system is housed inside a small NZXT S340 Elite case. I've decided that with the huge RTX 2080 Ti that I want additional air-flow so going to re-house my system inside a new Thermaltake View 71 RGB.

Hoping to get good results with this scanner tool after doing so.
 
That's way better results from a beta level program than I thought was going to happen.

Thanks for digging into this, [H]
 
re: the large difference between the scanner and manual overclock

Its possible the scanner takes account loading the RT and Tensor cores at the same time as the GPU core, a max overclock must take these into account to be fully stable?
I'd like to know what it really does.
 
Last edited:
Nenu said:
Its only 4% with the scanner tool and 7% manual.
Not amazing.
Click to expand...

Have to concur here. Really hoping AIB cards do this a lot better, but this is looking a lot more like Fury's OC headroom than Pascal/Maxwell.
 
Way to go with the 2010mhz. I think that might be the first I've seen so far that's gotten that far over 2000.
 
lostin3d said:
Can't wait to see what a custom loop does for one of these.
Click to expand...
From what I have read they are already power limited unfortunately.
With how Pascal reacts to lower temps with the newer power control mechanisms (no doubt similar on Turing) there will be a fair MHz gain dropping 30C.
But it might not be worth the cost.
I felt this on my 1080ti. Despite already having a custom water cooler on my CPU I went for an Accelero Xtreme III cooler @ £50 instead and got within 10 to 15C of water temps.
 
1. Bust out that ice chest chiller loop!

2. is there BIOS access to get rid of those pesky hard limits?

3. Your NVLink show up?
 
Good to see you guys have these cards in. Looking forward to the full review. (without any of the NDA restrictions).

jbc029 said:
Have to concur here. Really hoping AIB cards do this a lot better, but this is looking a lot more like Fury's OC headroom than Pascal/Maxwell.
Click to expand...

I am agreeing with this. With the 2080 coming in so close to the 1080ti Nvidia had to get their clocks as high as possible. As is it, the 2080 is barely beating 1080ti. Last gen, the 1080 smoked the 980ti. I just picked up a used 1080ti to tide me through the next year or so. Either the 2080s need to come down in price or one of the tech improvements (DLSS perhaps) needs to come thru to make the 2080 worthwhile.

Observation: Someone was excited to get these cards! (3am post :coffee:)
 
Last edited:
DNMock said:
1. Bust out that ice chest chiller loop!

2. is there BIOS access to get rid of those pesky hard limits?

3. Your NVLink show up?
Click to expand...
2. Doubt it. A vBIOS editor never happened for Pascal as far as I know since they are using signage that couldn't be duped or disabled. NVIDIA seems to want the power of their new cards locked down.
 
Comparing the cuda cores and boost speeds of the 2080 vs the 1080ti and comparing that to the review totals....

Pretty obvious that NV literally took Pascal, updated to GDDR6 and slapped on some Tensor cores (with two of them ghetto rigged for Ray Tracing).



I'm highly confident now that with the huge lead Nvidia had going for them they decided to dump all their R&D into Deep Learning and commercial application stuff. Even though there was no attempt to advance the base Maxwell architecture and no real node shrink to leech on (pascal) they needed to put out a new consumer line of GPU's, so enter turing, Pascal re-release with their commercial Tensor cores slapped on.
 
Armenius said:
2. Doubt it. A vBIOS editor never happened for Pascal as far as I know since they are using signage that couldn't be duped or disabled. NVIDIA seems to want the power of their new cards locked down.
Click to expand...

I know it didn't for the Titan's, but I was under the impression that one eventually came out for the 1080ti. At least I know you could flash the bios of an AMP! 1080ti onto, lets say, an Asus Strix 1080ti.
 
DNMock said:
I know it didn't for the Titan's, but I was under the impression that one eventually came out for the 1080ti. At least I know you could flash the bios of an AMP! 1080ti onto, lets say, an Asus Strix 1080ti.
Click to expand...
Yes, flashing an already existing one to another card was possible because the digital signature was intact, but actually editing was not as far as I'm aware.
 
You could definitely edit the bios of Pascal for TDP. You just couldn’t modify max voltage since that was programmed into the voltage controller itself (tiny chip that is on the pcb).

There were ways to hard mod around it if you had the balls.
 
Nenu said:
Its only 4% with the scanner tool and 7% manual.
Not amazing.
Click to expand...

I was about to say lol.
What was Vega's average overclock %? I'm pretty sure it's higher than this, like 12%.
I remember when I overclocked my Vostro's 8400m GS from 400Mhz to 650Mhz. Talk about percentages!!! 62% OC!!!
 
Dayaks said:
You could definitely edit the bios of Pascal for TDP. You just couldn’t modify max voltage since that was programmed into the voltage controller itself (tiny chip that is on the pcb).

There were ways to hard mod around it if you had the balls.
Click to expand...

Shunt mod, right? Can you imagine trying that on a $1200 card?
 
Dayaks said:
You could definitely edit the bios of Pascal for TDP. You just couldn’t modify max voltage since that was programmed into the voltage controller itself (tiny chip that is on the pcb).

There were ways to hard mod around it if you had the balls.
Click to expand...
I see. Shorting the shunt resistor is always an option regardless of BIOS accessibility.
kamikazi said:
Shunt mod, right? Can you imagine trying that on a $1200 card?
Click to expand...
I've seen plenty of people do it to their Titan X or Xp. Don't see why someone wouldn't try it on a 2080 Ti.
 
I'm curious if the Scanner Technology produces different results based on card temps; how well would the same card OC in a well ventilated case vs one that's struggling to keep cool.
 
Armenius said:
I see. Shorting the shunt resistor is always an option regardless of BIOS accessibility.

I've seen plenty of people do it to their Titan X or Xp. Don't see why someone wouldn't try it on a 2080 Ti.
Click to expand...
der8auer(sp?) is probably already working on it.
 
tsuehpsyde said:
I'm curious if the Scanner Technology produces different results based on card temps; how well would the same card OC in a well ventilated case vs one that's struggling to keep cool.
Click to expand...
I would suggest heat is a big factor, as it is with pretty much every other overclocking scenario on the planet.
 
Chris_B said:
Whats your opinion so far on the cooler? Seen 2 conflicting opinions of it so far one (jay) saying it got upto near 80c on an open air test bench and the shroud was very enclosed making air movement hard (it could only escape the shroud at the top and not the bottom), the other (oc3d) saying it done a good job and remained at around 70c and was very quiet. o_O

Also, any vids in the works of these?
Click to expand...
The cooler did what it was supposed to do. Temps, sound, and clocks noted.
 
Well, this Precision 1 beta app is going downhill pretty fast. I have not gotten it to complete another OC scan all day. Meh. Needs a lot of work.
 
pendragon1 said:
have you tried afterburner?
Click to expand...
Yeah, that is what I have been using today. Much easier to use than Precision 1. Precision 1 seems to be looking for problems to fix that do not exist. That said, this is my first time to use P1. Afterburner has been the standard here for years. On another note on GPU OC hardware, we did meet with ASUS engineers that build GPU Tweak not too long ago and went over our full list of gripes about that program, and we should see a new version with "fixes" early next year.
 
I wouldn't be surprised if this is already on the radar, but a 2080Ti vs 1080Ti SLI comparison would be interesting, especially with the price so close.
 
jfreund said:
I wouldn't be surprised if this is already on the radar, but a 2080Ti vs 1080Ti SLI comparison would be interesting, especially with the price so close.
Click to expand...

Can't recall the last time [H] did any SLI testing. It's always interesting to see the views on SLI/CF/mGPU. From it's totally dead to hey let's compare an SLI to the latest and greatest single GPU.

SLI is great when it works well, which is sometimes. But there's no way it is a substitute for the fastest single cards across the board.
 
I'm pissed... my Zotac RTX 2080 from Amazon is delayed till Monday. Was supposed to be here today. I can't wait to see how it compares.
 
heatlesssun said:
Can't recall the last time [H] did any SLI testing. It's always interesting to see the views on SLI/CF/mGPU. From it's totally dead to hey let's compare an SLI to the latest and greatest single GPU.

SLI is great when it works well, which is sometimes. But there's no way it is a substitute for the fastest single cards across the board.
Click to expand...

You're right. With the 2080Ti costing about twice as much as a 1080Ti, I think the comparison becomes more relevant. It does suck that developers have slacked off when it comes to SLI/XFire support.

1080Ti owners wanting to upgrade should be looking at the games they play, and deciding if a 2nd 1080Ti for about half the price of a 2080Ti is the better choice.
 
Is the shy OC just a matter of Nvidia clocking the FE so high to begin with compared to the after market cards?
 
You must log in or register to reply here.
Back
Top