Thanks for the info.
I can't get that low with stable UV I don't think, even with a negative frequency offset. Think I top put around -100mV in Steel Nomad if I remember correctly.
-
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 Radeon RX9070XT / RX9070 Owners Thread
- Thread starter CAT-THE-FIFTH
- Start date
I can't get that low with stable UV I don't think, even with a negative frequency offset. Think I top put around -100mV in Steel Nomad if I remember correctly.
LOL, I can tell exactly what website that is and I know a couple of people on Reddit ordered the Aorus Elite on there and received the Gaming OC. They must have got the SKU's mixed up! Little did they know they just had to order the cheaper oneOrdered a RX 9070 GAMING OC 16G as you can see below don't game as much as I used to and the 9000 series seems to be a good value for money.
and this turned up instead
Could not believe my eyes at first. Checked the website(tickled pink I am) and they have taken the item down. Currently in my PC now as I type this and it boosts upto 3.1ghz on "stock" values! (Time Spy was going over 3ghz average 2.9ghz)
Lucky you!!!
i bet only one side to the bad luck complained...
Base clock... Tell me more.
I'm tempted to jig up my AC unit used in summer when the house gets to hot for sleeping. Just pump it direct into the gaming room...
its very not advised... but you can overclock the 57/5800x3d but upping the base clock.
but my system maxed out at 106.. so nothing really to get excited about. but for benchmarking the 5% can matter.
there is no way i would run this day to day
stock 5700x3d 100 x 41.5 = 4150mhz
so 106 x 41.5 = 4399mhz
in the really world this make not differance, you should 100% not do is
Makes it even better.
5070: 51
5070 Ti: 65
5080: 82
You: 80
vs 5070 +57%
vs 5070 Ti 23%
vs 5080 -3%
Ah ok,.... need to run Port Royal for DXR.
Last edited:
this is a very interesting comparison my gpu maxed out and i mean maxed out is on pair with an average 5080.
obviously i can't game at my setting, but when i stripped my gpu looking at the die size i can see a 9080 or 9070xt been on the cards at some point
obviously i can't game at my setting, but when i stripped my gpu looking at the die size i can see a 9080 or 9070xt been on the cards at some point
Last edited:
Ah yes, I have read about this before but had heard it can break systems pretty quickly. I could probably gain that 5% by tightening timings on my RAM and doing a per core curve optimisation I would imagine.
Stock these things are about 74 FPS in Steel Nomad.
5070 Ti 65
5080 82
4090 93
Ok so the 4090 is 25% faster.
The 9070 XT die size is 357mm^2
The 4090 die Size is 609mm^2
They are both on the same TSMC N5 node, so on the same node RDNA 4 is near half the size for 80% the performance, at worst AMD would have to grow RDNA 4 by 25% to 446mm^2 to match the 4090's performance, that's not how it works, you don't double everything so AMD could probably do it with a 420mm^2 die.
I wonder if AMD underestimated the performance during the design stage? It wouldn't have taken much more to make it a Radeon RX 4090.
Also, don't you think that's impressive for what AMD have done here with such a small die?
5070 Ti 65
5080 82
4090 93
Ok so the 4090 is 25% faster.
The 9070 XT die size is 357mm^2
The 4090 die Size is 609mm^2
They are both on the same TSMC N5 node, so on the same node RDNA 4 is near half the size for 80% the performance, at worst AMD would have to grow RDNA 4 by 25% to 446mm^2 to match the 4090's performance, that's not how it works, you don't double everything so AMD could probably do it with a 420mm^2 die.
I wonder if AMD underestimated the performance during the design stage? It wouldn't have taken much more to make it a Radeon RX 4090.
Also, don't you think that's impressive for what AMD have done here with such a small die?
Last edited:
Unfortunately benchmarks are not indicative of actual gaming. Benchmarks are fun to tweak for best scores but the 5080 in general is going to be about 15% - 20% faster on average. The 9070 XT is doing great to get to around 85% of the 5080 for considerably less money.
Regardless of how disappointing Lackwell is, AMD and “could have/should have” scenarios mean nothing. AMD already had the 4090 performance to aim for and didn’t.
Regardless of how disappointing Lackwell is, AMD and “could have/should have” scenarios mean nothing. AMD already had the 4090 performance to aim for and didn’t.
Last edited:
Associate
- Joined
- 13 Apr 2025
- Posts
- 9
- Location
- Sweden
Yeah that is understandable.
The nitro+ TBP is ~10W less than the ASRock Taichi which also helps. I made some attempts lowering my PL to 107% in order to match the nitro+ at 110% to see if I could help out but I had crashes even at -90-95mV. That does not mean that it is impossible to luck out and do -105-110mV in a single run though and I feel like you should be able to get 8k+.
3DMark tests performs really poorly on Ultrawide resolutions, so in case you have an UW monitor I would suggest changing your desktop resolution to a resolution with a 16:9 aspect ratio.
In adrenaline I have pretty much the default graphics profile except that I have Texture Filtering Quality set to performance and Surface Format Optimization is on (the benefits of this seems to be within the margin of error so not sure if this actually helps or not).
Make sure other settings such as Anti-lag, Radeon boost and Image Sharpening is off if you are going for score. I would also make sure to turn off any sort of GPU scaling.
And just in case you don't know; Steel Nomad and Speed Way has VS mode which you can use to compare your run vs another in real time and it makes it easier to see how you are doing and it can really help you see if you have any major dips in performance which can be hard to spot otherwise. It can also help you see if your current settings has potential or are just garbage.
I also prefer to use MCT (MoreClockTool) over adrenaline for my OC's. Mainly because I found it so frustrating to wait forever for adrenaline to open whenever you had a driver crash and then redo all your settings if you forgot to save them. With MCT all your values are still filled in as long as you dont close it or hit reset and you can still save profiles.
Thanks for this. Lots of really good info.
The Texture Filter Quality setting I used to change on Nvidia to eek out a little extra score. Never knew I could do this in Adrenalin.
I didn't know about the Vs mode in Steel Nomad either. This sounds extremely useful.
Does MCT, or do you know if a tool exists, let you tweak the voltage frequency curve rather than set a blanket offset like Adrenalin?
Afterburner let me do this for my 3090 and I got a decent overclock this way, as there were certain frequencies that allowed much bigger UV. I basically mapped out a bespoke voltage offset across the frequency range of the GPU.
Associate
- Joined
- 13 Apr 2025
- Posts
- 9
- Location
- Sweden
It's not possible to do this with the newer cards to my knowledge unfortunately. I used to miss the ability to do that as well when I switched from nvidia to AMD about a year ago.
Lower frequency offset and higher voltage offsetgot into the top 100 but not sure where to go with it now
+300 Frequency Offset
-85mV Voltage Offset
2820 fast timing
+10% Board Power
You should be good upto about 105. My system would fail tests at 106
Last edited:
