6800/6900 Overclock Results

ZeroBarrier said:
I wouldn't say it's that close of a toss up at all. Nvidia has a slew of other value adds that people seem to forget. Shadowplay, Gamestream, Freestyle, Ansel, RTX Voice, NVENC just to name some.
Click to expand...
Lol people don't forget about it. It simply doesn't matter in most cases for what people are using a gaming PC for. Someone could start listing off the various bits of software that only Radeon cards come with, but we're talking raw performance which at the end of the day matters most in a GPU.
 
NightReaver said:
Lol people don't forget about it. It simply doesn't matter in most cases for what people are using a gaming PC for. Someone could start listing off the various bits of software that only Radeon cards come with, but we're talking raw performance which at the end of the day matters most in a GPU.
Click to expand...
Okay, it doesn't matter to you. Got it. 👍
 
ZeroBarrier said:
Okay, it doesn't matter to you. Got it. 👍
Click to expand...
Doesn't matter to me either and if you aren't a successful professional Streamer living off your income from Streams, then those are features that don't carry much weight. Sure, lots of people use those but are not Pro Streamers or make their whole income off streams. Besides the Streaming trendy craze..........

When you look at just performance and playing games, it does become a toss up and the hardware is overkill anyways for the games people play.
 
ZeroBarrier said:
Okay, it doesn't matter to you. Got it. 👍
Click to expand...
God these types of replies are so lazy. Why even talk about it then? What was even the point?

Those features are added frills that may or may not apply whereas the performance almost always matters. So why not just drop the technicalities and let's just talk performance. This is an overclocking thread afterall.
 
NightReaver said:
God these types of replies are so lazy. Why even talk about it then? What was even the point?

Those features are added frills that may or may not apply whereas the performance almost always matters. So why not just drop the technicalities and let's just talk performance. This is an overclocking thread afterall.
Click to expand...
We already did talk performance, and you disregarded it with your opinion. I'm just letting you know your opinion came across loud and clear. Sorry if it bothered you.
 
I improve the airflow in my case, and I add one Peltier (80w) on the backplate of the GPU. The result is more than 10 degrees low that I transfer for better overclock :D not bad for a ~10$ upgrade :)
The Peltier is connected with Molex outside of the case so I can unplug it when I do not need it.

Fire Strike Ultra result​

1658827585996.png

1658827743917.png
 
Below Ambient said:
24,598 https://www.3dmark.com/spy/30458734 Chilled Water 1300mv AMD reference 6900xt
Click to expand...
can i ask you, how the fuck did u almost get 14000 cpu score? static oc at heavy 4975 mhz??? and 4000mhz ram ? or was it holding boost that high with pbo? either way, is that running daily like that? what voltage and temps are you getting? that an insane cpu score for 5800x lol :D
im on 5800x aswell, and running 280mm aio nzxt full pump speed etc, but with pbo settings 139/160/200 and co with -21 2 best core, -27 2 second best, -30 rest, and +150 boost getting 13100 cpu score and boost avg in timespy at 4750, ram is 4x 16gb running 3600mhz with cl14-14-12-12-22 1T + subs but cant seem to make it much higher then that? tryed static but seems like im only able to hit 4750, which makes it suffer alot on the single core performance, atm r23 i get 1667/16080, any clue how i can make that even better?
 

Attachments

  • CPU-Z Rekord.jpg
    CPU-Z Rekord.jpg
    189.5 KB · Views: 0
  • timespy score 13100 cpu.jpg
    timespy score 13100 cpu.jpg
    283 KB · Views: 0
I don't run PBO..I run ram at 3933 single rank 2x8gb 1.65v it's on chilled water, for daily i just lower the multiplier and run 4.8

You're using too much ram and not enough fclk...at 4775 Im over 13k also the best thing I ever did was move over to a b550 strix from the ch8 I used to run...x570 are lousy clockers imo
 
Last edited:
Below Ambient said:
I don't run PBO..I run ram at 3933 single rank 2x8gb 1.65v def not daily it's on chilled water, although my CPU score daily is 13.6k at 4.8ghz

You're using too much ram and not enough fclk...at 4775 Im over 13k also the best thing I ever did was move over to a b550 strix from the ch8 I used to run...x570 are lousy clockers imo
Click to expand...
Im running 3600mhz 1800fclk. Higher and crash. Exept if i take out 2 sticks.

How much voltage for 4800mhz Daily ?? Daily with this pbo it hits 13100-13200 cpu score. But would be coool to go even higher But guess it takes a shit lot of voltage to go 4800+ allcore static ??
 
Nordskov said:
Im running 3600mhz 1800fclk. Higher and crash. Exept if i take out 2 sticks.

How much voltage for 4800mhz Daily ?? Daily with this pbo it hits 13100-13200 cpu score. But would be coool to go even higher But guess it takes a shit lot of voltage to go 4800+ allcore static ??
Click to expand...
I run 1.36 fixed, much lower than pbo
 
Below Ambient said:
I don't run PBO..I run ram at 3933 single rank 2x8gb 1.65v it's on chilled water, for daily i just lower the multiplier and run 4.8

You're using too much ram and not enough fclk...at 4775 Im over 13k also the best thing I ever did was move over to a b550 strix from the ch8 I used to run...x570 are lousy clockers imo
Click to expand...
X570 lousy clockers? You better take that back MOFO. My R23 was on a X570i and if I remember right I'm 4pts ahead of you lmao.
 
Hi all, it's been a while, and I had no successful attempts to report in the meantime ... but recently, the weather has gotten colder and I remembered this quote :
Catsonar said:
Ita a test bench setup and just stick it outside when it's cold enough.
Click to expand...
so I gave that a try :) and put my case outside in around -3 to -5°C:
24438: 1290mV, MPT 517W, 2855core, 2140mem
https://www.3dmark.com/3dm/85229256?

(CPU score is low, because I turned SMT off and no OC and also no additional RAM OC, for max stability / min heat)

Before I put my case outside, to try to understand why I was not successful so far in getting close to 24000 again, I logged some data with HWinfo.
The first graphics test is never an issue, but I always have frequency drops (or tests fail) in the first 30sec of the second test, so I focused on that part first. To my surprise, it seems that I was temperature limited: during that part of the test, the hotspot reaches 110° and caused throttling.
That was exactly the motivation needed to try out Catsonar's advice :D
After a while there was a clear shift for 'temperature limited' to 'power limited', so I finally had got some room to play again ...
(weather has changed again, next couple of days will no longer be freezing, but raining)

Here are some of the graphs that show the different scenario's:
  1. 22533, temperature limited > in first part of 2nd test, hot spot is reaching +110°C causing throttling
    22533.png


  2. 23050, power limited > in first part of 2nd test, power is reaching set limit of 400W causing throttling
    23050.png


  3. balanced > seems there is no limit reached yet, max 98°C and max 507W
    24422.png
 
Below Ambient said:
I took mine outside also and got a Timespy gpu score of 25695 using mpt at 1275 but with my mv at 1255 2940 highest clock
https://www.3dmark.com/3dm/85243423
Click to expand...
So you got a good +1000 points by taking it outside ? That's not too bad !
I only got +50 :D, but was happy already I could actually complete multiple +24000 runs again ... but needing to put this outside in below 0°C to get there, and the hotspot getting too easily to 110°C, feels like something is not right ... so I drained the loop, cleaned out the water blocks and re-pasted, and will soon do some tests again (on new AM5 platform)

Also, I just saw my RAM was running at default 2133 instead of 3200CL14 (probably because of a crash earlier, and I forgot to set it back to DOCP), I wonder how that would have influenced the score ?

Nordskov said:
this is my score with room temps, havent tryed to move pc outside lol, thats to fucked for my taste xD

heres my daily settings atm

https://www.3dmark.com/spy/33561351
Click to expand...
if +1000 is any indication, makes us wonder what score you could get when moving your PC outside :D
 
bl4C3y3 said:
So you got a good +1000 points by taking it outside ? That's not too bad !
I only got +50 :D, but was happy already I could actually complete multiple +24000 runs again ... but needing to put this outside in below 0°C to get there, and the hotspot getting too easily to 110°C, feels like something is not right ... so I drained the loop, cleaned out the water blocks and re-pasted, and will soon do some tests again (on new AM5 platform)

Also, I just saw my RAM was running at default 2133 instead of 3200CL14 (probably because of a crash earlier, and I forgot to set it back to DOCP), I wonder how that would have influenced the score ?


if +1000 is any indication, makes us wonder what score you could get when moving your PC outside :D
Click to expand...

Ram is what gave me 700 pts... going from 6800 XMP to 7800 tuned
 
Last edited:
ZeroBarrier said:
Okay, it doesn't matter to you. Got it. 👍
Click to expand...

Catsonar said:
Doesn't matter to me either and if you aren't a successful professional Streamer living off your income from Streams, then those are features that don't carry much weight. Sure, lots of people use those but are not Pro Streamers or make their whole income off streams. Besides the Streaming trendy craze..........

When you look at just performance and playing games, it does become a toss up and the hardware is overkill anyways for the games people play.
Click to expand...

Same here.

I have a 6900xt right now, but prior to that I used only Nvidia GPU's since ~2011, and prior to a brief stint in 2010-2011 when I went from a single Radeon HD 6970, to two Radeon HD 6970's to a single Radeon HD 7970, I used nothing but Nvidia GPU's since I got my GeForce 2 GTS in 2001.

By my count, I've had 11 Nvidia GPU's in my main rig over the last ~20 years and I've never once used any of the mentioned features.

  • Shadowplay: I don't know what this is, but it seems streaming/capture related, and I have zero interest in anything streaming related as either a viewer or a poster.
  • Gamestream: I've never had any interest in playing games away from my desktop, so this feature has no value to me. I don't want gaming in my living room, I don't want it on any of my mobile devices. I play games in one place and one place only, alone at my desk in my office. Not to mention, Nvidia just ended support for this feature on Shield devices... Apparently they want you to pay them for a streaming service rather than streaming from your Desktop :p Turns out Sunshine is a good option though, and it works on all GPU's.
  • Freestyle: I don't know what this is. Seems to be some sort of modding tool? I'm fine without it.
  • Ansel: I don't know what this is. Appears to have something to do with streaming and/or capturing 3d environments, which is ultimately of no interst to me.
  • RTX Voice: Don't really care about RTX voice. While I used to be really into them, I've barely played any multiplayer titles in a decade, so I don't really use voice at all anymore. I don't even have a Discord account. Last time I spent any time playing multiplayer titles, everyone was using private TeamSpeak servers.
  • NVENC: I have no need for real-time video compression because I don't stream, I never have streamed, and I never will stream, and even if I did, I wouldn't want my GPU to be slowed down by having to encode that shit. On the rare occurence I need to transcode something I do it on the CPU using my 24C/48T Threadripper, which gives me more control over the algorithm and settings.

There are certainly going to be people for whom these features are of value, and maybe even important, but don't assume everyone plays games the same way you do :p

It's a common source of cognitive bias, inferring your own interests and usage cases on other people, but there are A LOT of people who just play single player games casually on their own, and aren't into all the modern multiplayer streaming nonsense. Just because they aren't in your circle doesn't mean they don't exist and aren't numerous.

It's the same form of cognitive bias that results in statement such as "how could the other party win the election, they couldn't possibly have that many voters, I've never even met a person who would vote for them" that people from both sides of the political debate not infrequently utter.

We tend to be blinded by our own personal experiences and our own circles.
 
Last edited:
bl4C3y3 said:
and the hotspot getting too easily to 110°C, feels like something is not right ... so I drained the loop, cleaned out the water blocks and re-pasted, and will soon do some tests again (on new AM5 platform)
Click to expand...
yes, something was not right ... now hotspot reaches around 80-85°C max, and I can get the same frequencies and a slightly higher score of 24467 vs 24438, but with a lower average temperature of 51°C vs 53°C ... and that is inside at ambient of around 19° vs outside at freezing temperatures :D

24467: 1287mV, MPT 517W, 2850core, 2140mem
https://www.3dmark.com/3dm/85513441?

Below Ambient said:
I took mine outside also and got a Timespy gpu score of 25695 using mpt at 1275 but with my mv at 1255 2940 highest clock
https://www.3dmark.com/3dm/85243423
Click to expand...
How do you get your CPU score so high ?
I'm just starting with a 7950X and I get mostly around 17000 CPU score (with SMT off = 16c/16t ... SMT on = 16c/32t seemed to be worse, but didn't try much yet on the CPU side).
Your 12900K seemed to have SMT on = 16c/24t ?
 
bl4C3y3 said:
yes, something was not right ... now hotspot reaches around 80-85°C max, and I can get the same frequencies and a slightly higher score of 24467 vs 24438, but with a lower average temperature of 51°C vs 53°C ... and that is inside at ambient of around 19° vs outside at freezing temperatures :D

24467: 1287mV, MPT 517W, 2850core, 2140mem
https://www.3dmark.com/3dm/85513441?


How do you get your CPU score so high ?
I'm just starting with a 7950X and I get mostly around 17000 CPU score (with SMT off = 16c/16t ... SMT on = 16c/32t seemed to be worse, but didn't try much yet on the CPU side).
Your 12900K seemed to have SMT on = 16c/24t ?
Click to expand...
Am5 is a dumpster fire is why...5950x at all core actually scores higher lol...

As for mine all core helps, as does 7800 ram speed
 
Finally got my PC setup again after moving this year (sorry I disappeared for a bit there). After updating the BIOS on my X470 and the newest AMD drivers I discovered that somewhere along the way the issue I was having with hard system reboots when approaching the 375W MPT limit has been resolved. (I think the issue was previously that the card was trying to pull too much over pci-e, but really either could have fixed it).

That said, I'm now able to run consistently at 2825mhz set (2760mhz) in benchmark. This was good for 22,756 graphics while at 8x pci-e 3.0. I'll have to pop out the pcie ssd and see what it scores with a full 16 lanes.

Still, not bad considering the circumstances!
https://www.3dmark.com/3dm/85996500?
 
R-Type said:
Finally got my PC setup again after moving this year (sorry I disappeared for a bit there). After updating the BIOS on my X470 and the newest AMD drivers I discovered that somewhere along the way the issue I was having with hard system reboots when approaching the 375W MPT limit has been resolved. (I think the issue was previously that the card was trying to pull too much over pci-e, but really either could have fixed it).

That said, I'm now able to run consistently at 2825mhz set (2760mhz) in benchmark. This was good for 22,756 graphics while at 8x pci-e 3.0. I'll have to pop out the pcie ssd and see what it scores with a full 16 lanes.

Still, not bad considering the circumstances!
https://www.3dmark.com/3dm/85996500?
Click to expand...
You'll get better results ccx multiplier or all core overclock... You can also use mpt and tune to 1250mv to get a bit more... Nice out of the box set up though...pretty much at the 5% more for 5 more fps stage lol
 
Below Ambient said:
You'll get better results ccx multiplier or all core overclock... You can also use mpt and tune to 1250mv to get a bit more... Nice out of the box set up though...pretty much at the 5% more for 5 more fps stage lol
Click to expand...
My issue is that I can saturate my 375W increased power budget for 80% of the run with 1063mV... more voltage wouldn't do me any good if I didn't also increase the power budget. What are people running with the reference boards these days?
 
Graphics Score of 23,452 at 16x vs 8x pcie 3.0 link width.
https://www.3dmark.com/3dm/86001914?

Given I daily drive these settings, I'm pretty content. I think that puts me in 4th place for the Watercooled rankings as well. :)
 
Last edited:
R-Type said:
I think that puts me in 4th place for the Watercooled rankings as well. :)
Click to expand...
I think 5th :p

Updated this quote from post #373 with latest additions:
Catsonar said:
here we can have some friendly competition for achieving the highest Time Spy GPU score. Overclock all you want. Time Spy is available for free here.


We are focused on graphics score, not overall! This allows anyone with a 6900XT and or Cracked out 6800XTs to be competitive since CPU does not play a major role in the GPU score.

The rules are:
1. In your post, please include the score and a link to your validated result, and tell us if it is air cooling, water cooling, or extreme (AC, chiller, dry ice, LN2, hard mods, shunt mods, etc).
2. Leaderboard will consist of top 10 scores per cooling type only.


Have fun!
Click to expand...

Extreme:
1.
2.
3.

Water Cooling:
1. 25695 BelowAmbient https://www.3dmark.com/3dm/85243423 (or should this be under extreme ? because of 'chiller')
2. 25627 Nordskov https://www.3dmark.com/spy/26614660
3. 24467 bl4C3y3 https://www.3dmark.com/3dm/85513441?
4. 24412 Catsonar https://www.3dmark.com/spy/26633741
5. 23452 R-Type https://www.3dmark.com/3dm/86001914?
6. 23411 Zarathustra[H] https://www.3dmark.com/3dm/68596722?
7. 23029 Gideon https://www.3dmark.com/3dm/71914412?
8. 22850 NightReaver https://www.3dmark.com/3dm/75867005?
9. 22062 Chassepot (6800XT)
10. 22048 Killroy67

Air Cooling:
1. 23079 Mysterfix https://www.3dmark.com/spy/28585113
2. 22735 bl4C3y3 https://www.3dmark.com/3dm/72273323?
3. 22663 John1780 https://www.3dmark.com/3dm/72045575?
4. 22397 Starfalcon https://www.3dmark.com/spy/26262182
5. 22049 alceryes3 https://www.3dmark.com/spy/26674135
6. 21364 Thecold
7. 21269 Gideon
8. 21261 AVATARAT (6800XT) https://www.3dmark.com/3dm/72606418
9. 20994 kwikgta
10. 20978 newls1
 
Last edited:
bl4C3y3 said:
I think 5th :p

Updated this quote from post #373 with latest additions:


Extreme:
1.
2.
3.

Water Cooling:
1. 25695 BelowAmbient https://www.3dmark.com/3dm/85243423 (or should this be under extreme ? because of 'chiller')
2. 25627 Nordskov https://www.3dmark.com/spy/26614660
3. 24467 bl4C3y3 https://www.3dmark.com/3dm/85513441?
4. 24412 Catsonar https://www.3dmark.com/spy/26633741
5. 23452 R-Type https://www.3dmark.com/3dm/86001914?
6. 23411 Zarathustra[H] https://www.3dmark.com/3dm/68596722?
7. 23029 Gideon https://www.3dmark.com/3dm/71914412?
8. 22850 NightReaver https://www.3dmark.com/3dm/75867005?
9. 22062 Chassepot (6800XT)
10. 22048 Killroy67

Air Cooling:
1. 23079 Mysterfix https://www.3dmark.com/spy/28585113
2. 22735 bl4C3y3 https://www.3dmark.com/3dm/72273323?
3. 22663 John1780 https://www.3dmark.com/3dm/72045575?
4. 22397 Starfalcon https://www.3dmark.com/spy/26262182
5. 22049 alceryes3 https://www.3dmark.com/spy/26674135
6. 21364 Thecold
7. 21269 Gideon
8. 21261 AVATARAT (6800XT) https://www.3dmark.com/3dm/72606418
9. 20994 kwikgta
10. 20978 newls1
Click to expand...
Damn, tough competition around here.

Faster than any aircooled cards at least. ;)
 
R-Type said:
Your score is really good for the average being below 2600mhz. Are you running MPT?
Click to expand...
Yeah. It's set at 345W & 365A. This card likes wattages. I had it at 325W / 345A & 2600 / 2700 and only managed 21538. Setting it to 345W + 15% gave me those numbers. Memory sucks though, can't get over 2020.

I"m not sure what AMD did to the reference card bios but I'm not able to go above 1.175v. Increasing to 1.20v in MPT and it'll just freeze the desktop. Same thing happens if I try to go above 2700 MHz.
 
nxcess said:
Yeah. It's set at 345W & 365A. This card likes wattages. I had it at 325W / 345A & 2600 / 2700 and only managed 21538. Setting it to 345W + 15% gave me those numbers. Memory sucks though, can't get over 2020.

I"m not sure what AMD did to the reference card bios but I'm not able to go above 1.175v. Increasing to 1.20v in MPT and it'll just freeze the desktop. Same thing happens if I try to go above 2700 MHz.
Click to expand...

Here's some wattage for you lol

 
nxcess said:
Yeah. It's set at 345W & 365A. This card likes wattages. I had it at 325W / 345A & 2600 / 2700 and only managed 21538. Setting it to 345W + 15% gave me those numbers. Memory sucks though, can't get over 2020.

I"m not sure what AMD did to the reference card bios but I'm not able to go above 1.175v. Increasing to 1.20v in MPT and it'll just freeze the desktop. Same thing happens if I try to go above 2700 MHz.
Click to expand...

I had also issues with the voltage in the beginning, as soon as I set it above 1.175, the frequency locked at 500 MHz.
Doing this trick below in MPT solved that for me:

Nordskov said:
bl4c3y3

You go to features and make sure temp dependent vmin is marked on. Then go to power table and at right side second from bottom it says something with 800-800. Then change to Maybe 1200-1200. Then your Max vcore Will be 1.2v vs Stock 1.175v. And therefor Maybe some higher boost. You should Daily be able to do 1.250v. But wouldnt go higher Then that for Daily. At that voltage with Max 380w+15%pl im doing like 24500graphic score for Daily gaming and Can run games at 2900mhz boost clock.
Click to expand...
 
You must log in or register to reply here.
Back
Top