Why overclocking 40 series gives you to little advantage

[Gökyüzü]

I have Zotac 4070 ti Amp Extreme Airo

I test it at Kombustor benchmark while overclocking

Stock results : 8289 score average 138 fps
+50 core +1000 mem : 8381 score average 139 fps
.
.
.
at max i manage to see this results with out artifact
+175 core +2000 mem max power limit : 8628 score with average 143 fps
which that makes barely %4 fps advantage

If i some how manage to make that options to any 20 or 30 series card it would give me atleast %20 boost (i know its not possible toght)

Is there something im doing wrong ?
Or is it just this 40 series cards doesnt even worth to overclock because they made it this way?

[_I_]

newer cpus/gpus need better cooling to oc farther
they already boost to what the cooling allows, then throttle

thats probably the problem you are running into
or else the gpus vrm limits on how much power it can deliver to the gpu cores

[Crawl]

Modern hardware does a pretty good job of boosting on its own. The best way to get gains these days is to undervolt. Reduced power means less heat which gives you more headroom to boost higher and for longer.

[NEIL]

bro just buy the 6090 ti

[Illusion of Progress]

Some of the recent reviews for the RTX 4070 are mentioning it has next to no overclocking performance to be gained. I'm not sure if that also extends to the RTX 4070 Ti, but given they are almost the same GPU with the exception of CUDA core counts, it wouldn't be surprising if it was the same for it.

[Bad 💀 Motha]

All the NVIDIA GPUs since GTX 10 series do not OC well. Because they pretty much already come to you maxed out, or nearly maxed out.

[Gökyüzü]

Αναρτήθηκε αρχικά από NEIL:

bro just buy the 6090 ti

Lmao. Im not overclocking because i need more fps im overclocking because it was fun to overclock

[Bad 💀 Motha]

Well again it's been common knowledge that the GTX 10 and later GPUs are basically hard-capped when it comes to this.

[Komarimaru]

Αναρτήθηκε αρχικά από Bad 💀 Motha:

Well again it's been common knowledge that the GTX 10 and later GPUs are basically hard-capped when it comes to this.

I believe it depends on the AIB and model.

Before evga left the scene, their cards always Overclocked like crazy. My 1080ti, 2080ti, abd current 3080ti overclock really high.

For example, my 3080ti is setup as daily use with 2010Mhz core and 11,000Mhz on memory, slightly bypassing any stock 3090ti in performance while also being undervolted at the same time to not exceed 0.925mv.

[Bad 💀 Motha]

Well yes it depends on the branded model and how it is as "stock"

But many of then are already OCed near the max from the factory. Such as EVGA FTW3 series

[Komarimaru]

Αναρτήθηκε αρχικά από Bad 💀 Motha:

Well yes it depends on the branded model and how it is as "stock"

But many of then are already OCed near the max from the factory. Such as EVGA FTW3 series

Oh, you can always push evga more. Stock is 1920 and 9500 for core and men.

Downside is, my card is 450 watts, not 350 lol

[_I_]

if you really like overclocking, go back 20 years

[Soulreaver]

Αναρτήθηκε αρχικά από Bad 💀 Motha:

Well yes it depends on the branded model and how it is as "stock"

But many of then are already OCed near the max from the factory. Such as EVGA FTW3 series

Yeah but near the max meant for the 2080ti that the FTW3 one could get close to 3080 performance with a good OC compared to other AIB's that would sometimes be closer to 3060ti performance.

So yes they were pre overclocked to almost max and there was no big headroom left... but we are talking about massive improvements depending on what AIB you compare it to.

[Komarimaru]

Αυτό το σχόλιο αναμένει ανάλυση από το αυτόματο σύστημα ελέγχου περιεχομένου μας. Θα είναι προσωρινά κρυμμένο μέχρι να επιβεβαιώσουμε ότι δεν περιέχει βλαβερό περιεχόμενο (π.χ. συνδέσμους σε ιστοσελίδες που προσπαθούν να κλέψουν πληροφορίες).

[Pocahawtness]

I have heard that NVIDIA have really clamped down on overclocking this release.
NVIDIA are just getting way over the top mean with their products.

[Crawl]

Its not that they "clamped down" but they are pretty much pushing the cards to the limit out of the box with their boosting algorithms. For example a 4090 has a stock boost clock of 2520 MHz, my Strix 4090 OC has a boost clock of 2640 MHz according to Asus but my card actually boosts over 2700 MHz without me touching anything. It does that under full load while only pulling about 350w and never breaking 60C. Sure you could probably pick up a bit more with some manual tweaking but is it really worth bothering with it over what will ultimately result in such a small increase that it will make no real performance difference.