Gskill ripjaws 3600mhz cl16 2x8gb
-
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
- Thread starter Kaapstad
- Start date
CPU score 8,250. Overall score around 14,650. 3d mark says average for 5600x + 3080 should be around 15k.
Regarding temps, on stock 3080 targets 74c. So if it exceeds 74c then it will start to throttle. In non-demanding games with a very good case airflow its not surprising to get 65c, many people are quite surprised at how quiet and cool these card are. Try playing some heavy RT games such as Metro Exodus or Control, the cards will start to sweat.
I also had a 3090 which was about 2-3% below average scores in 3dmark and at the time I had 3700x (8 cores) so I don't think this is something to do with the CPU.
I suspect something is off with the motherboard or mobo+ram combo. But in my case (which has lower score than yours) I am 3% under average in 3dmark, this is not significant enough for me to lose any sleep over it. I tried to find the issue but after days of googling I could not "fix" this.
Also just in case you have no heard, a lot of people were reporting underperformance on gigabyte amd motherboards due to some USB issues. If you have gigabyte mobo, look into this.
Tbh though your score looks fine, a lot of people running these benchmarks will do so with there cards heavily overclocked and power modded while running the fans at 100%.
Your cpu score is 8250? Stock right? Mine was 8286 max with pbo auto(which is default). Pretty similar. I have an msi b550 tomahawk. But the ram im using is not in the qvl list of the website but in gskill’s website it was mentioned that it is compatible.
And I’ve metro too and it doesn’t cross 65 lol
What’s the average graphics score for the 3080fe? I guess I shouldn’t sweat too much given I get 17300.
I also have dropped about 1000 graphics score since my 1st run on 20th September. I pair Gigabyte Gaming OC with a 7700K @5.1, so not the best but still shows the drop.
22/9 Release driver, 17263 GPU score:
4/2 461.40 driver, 16144 GPU score
I haven't got above 17000 since the 3rd driver on from the card release date
22/9 Release driver, 17263 GPU score:
4/2 461.40 driver, 16144 GPU score
I haven't got above 17000 since the 3rd driver on from the card release date
Which drivers though I went back to 460.79 clean drivers because everything after that was bad for me. The latest drivers provide sound popping and high latency.
Also I tried HWInfo last night on Hitman 3 and my GPU mem maxed at 96c. Also my FE was going mad over 80c then I realised I had gsync and frame limiter off. I usually have mine set to gsync on, vsync on in Nvidia CP and frame limit to 75fps (monitor is 75mhz) so this is probably why and I was too tired to try it again. It was getting over 200fps before so no wonder it was hot.
Also I tried HWInfo last night on Hitman 3 and my GPU mem maxed at 96c. Also my FE was going mad over 80c then I realised I had gsync and frame limiter off. I usually have mine set to gsync on, vsync on in Nvidia CP and frame limit to 75fps (monitor is 75mhz) so this is probably why and I was too tired to try it again. It was getting over 200fps before so no wonder it was hot.
CPU stock yes. As far as I understand PBO on auto doesnt actually do anything as turning it off does not affect the score.
65c in metro exodus with RT, hairworks and physx on? That's crazy good cooling you have there or perhaps you're gaming at 1080p and the card doesnt have to work as hard? I'm gaming at 3440x1440 and my fans get quite loud when playing metro exodus as they try to keep card temps at 74c (i do have minimal noise case fans at lowest rpm though).
As mentioned previously average 5600x + 3080 combo has 17,500 graphics score for timespy and this score includes all 3080, including aib, which have better cooling and hence better sustained boost clocks / higher stock clocks / higher stock power limit. From what I've seen 17,300 is about all right.
Also as was previously mentioned there was definitely driver nerf to the card performance. On old nvidia drivers my score increases by about 300 points. Original guru3d had 3080fe timespy graphics score at 17,800 but since then at the very least it was nerfed to 17,500. Perhaps there were subsequent boost clock reduction through drivers.
But definitely look into undervolting, you seem to have crazy good cooling so with reduced voltage you will not run into the power limit as much, which will increase your boost clocks and boost sustainability, just make sure to test stability in metro exodus on all max including RT. (to give you an example my 1845 @ 812mv undervolt was stable in everything, timespy extreme stress test pass, control game stable for several hours but metro was the one game that actually not stable (crash after couple of hours of gaming) and had to bump it up to 818mv to make it absolutely stable).
Last edited:
Wow this is very interesting. Also have you noticed your CPU score went down by almost 10%.
Also your 3080 have the same boost average clock frequency between the two runs so the card is performing exactly as it was before (my understanding is that nvidia drivers specifically reduced boost clocks to address the crashing issue).
Looks to me that something is off with your CPU or some software issue? 16.1k graphics score for 3080 definitely looks very wrong, 3d mark reporting average graphics score for your cpu+gpu combo at 17,559.
my CPU score increased from 5213 to 5885, but that isn't surprising as I closed stuff like iCUE this time
Ah you're right. I got them mixed up. Make this performance drop more peculiar. CPU was faster and gpu average clocks the same yet performance dropped. How very strange!
Must be something with the drivers, also I tried looking up my same cpu and gpu clock speeds including VRAM and there aren't many results with that combination. Mostly the results have additional overclocks on the memory VRAM at least, and then quite a lot with gpu clock increases too. I haven't bothered changing mine for the factory OC yet, as I haven't put a waterblock on it so far, so don't want the fan noise too loud. Also it's bouncing off the power limit most of the time anyway
That’s horrible, what is nvidia doing? But is this the case only for a handful of us?
Sorry couldn’t reply sooner. It wouldn’t let me do since I’m a newbie here and can only do a max of 15posts per day.
Anyways, I’m on 1440p. And yeah haha my temps are pretty cool for my 32C ambient. Idk whether to suffer from success or if it’s an actual problem lmao my airflow config:
Three front intake, two bottom intake, two top exhaust and one rear exhaust. Arctic p12 pwm fans running at the default bios settings. Case is a Lian Li lancool mesh 2.
Coming back to the topic at hand, might your resolution be the reason why you get lower scores than me? Since I’m at 2560x1440p. And I’m at the latest driver 146. Many people had issues with this. Which driver you on? Maybe we need to dial back with ddu?
In timespy no, but generally in games I heard that lower resolution in games achieves higher boost clocks. I'm on the latest nvidia drivers as well and I wouldn't roll back nvidia drivers, they reduced boost clocks for a reason - games were crashing.
I remember when I was googling the issue I saw one guy get his score improved significantly after changing mobo but im not doing that for at least a year or two so at this point im quite happy with what I got.
By the sounds of it you have great cooling, if you are worried the card is not 100% utilised then just run a benchmark with gpu-z running to make sure its in 98+ gpu load and see whether you are running into the power limit. Perhaps some FE cards have higher stock voltage hence we get power limited quicker and 1-3% behind averages benches. Not a big deal as an undervolt resolves this issue. I can get 18,000 timesply score with an undervolt at 1920 @ 868mv but prefer 1845 @ 818mv because its quieter and cooler. As you have great cooling you will likely get really great undervolt performance.
So three of my friends have fe’s. I asked all of them to do timespy and none crossed 17,300. Hovered in the 100s and 200s. So yeah I guess we have nothing to worry about.
You can run a unigine heaven bench to be sure. There was a person in this forum going by the name of Guest2. Don’t know how to tag him, amazing person, he was very patient and helped me out a ton. He has an fe and 5800x and we both got similar results in heaven 4.0. Thats when I knew that there was nothing wrong with mine, I guess a lot of factors go into synthetic scores. Here is mine for reference, try out with the same settings and see if yours holds up.
Make sure that the res is 2560x1440.
Last edited:
So three of my friends have fe’s. I asked all of them to do timespy and none crossed 17,300. Hovered in the 100s and 200s. So yeah I guess we have nothing to worry about.
You can run a unigine heaven bench to be sure. There was a person in this forum going by the name of Guest2. Don’t know how to tag him, amazing person, he was very patient and helped me out a ton. He has an fe and 5800x and we both got similar results in heaven 4.0. Thats when I knew that there was nothing wrong with mine, I guess a lot of factors go into synthetic scores. Here is mine for reference, try out with the same settings and see if yours holds up. Make sure that the res is 2560x1440.
Yeah got around the same score.
So going by your friend's feedback and more people getting FE in this forum it appears that yes 17-17.5k is an absolutely normal score.
Enjoy your card now!
Absolutely. Cheers.
Hey guys, got a question about thermal throttling (again)
So I've replaced and reapplied new thermal pads and wow I've gone from 106c junction temps to mid 70s Celsius, so that's worked great! It posts and works,
However it still says thermal throttling, I myself think it's the thermal paste, I only had noctua nt-h2 lying around that I got about 6 months ago. My core temps seem to be around the 70s Celsius during stress testing, I applied it by blob dead center, do you think that's the issue?
Should I of spread it instead? I don't mind taking it apart again just thought I'd ask here first.
Thanks in advance.
So I've replaced and reapplied new thermal pads and wow I've gone from 106c junction temps to mid 70s Celsius, so that's worked great! It posts and works,
However it still says thermal throttling, I myself think it's the thermal paste, I only had noctua nt-h2 lying around that I got about 6 months ago. My core temps seem to be around the 70s Celsius during stress testing, I applied it by blob dead center, do you think that's the issue?
Should I of spread it instead? I don't mind taking it apart again just thought I'd ask here first.
Thanks in advance.
Msi suprim x 3080 in stock elsewhere....£1100.
These at least are sold out, if its these 'VGA5916'
Indeed.