Oh, on the topic of power, known bugs are power draw issues that they have said they are prioritising for the next driver.
I want to undervolt the thing, but I’m going to wait for better drivers as stability could change quite soon if they are messing with power stuff.
Obviously an important metric!
Comes with the kernel. You can run the in development branch of a kernel or at least pull the firmware from that kernel and hope for the best.
I loved that GN included that in the benchmarks.
I just read TPU’s review on Sapphire RX7900XTX Nitro+. And saw the idle power consumption of the 7000 series:
Single monitor
AMD RX 7900XT 11W
AMD RX 7900XTX 13W
Sapphire RX 7900XTX Nitro+ 16W
Fair enough. But…
Multi-monitor
AMD RX 7900XT 85W
AMD RX 7900XTX 103W
Sapphire RX 7900XTX Nitro+ 104W
If you’re a multi-monitor user, and can’t decide GREEN or RED, you can always delay your decision for another six months. Give AMD driver time to mature a bit.
yeah i know there firmware thats loaded by kernel… buts only useful when you’ve got an updated OS…
but ive seen jayz2cents talk about updating firmware on nvidia before and didnt know if there was a way to do the same for amd gpus…
since what if you wanted to try older hardware&OS with a rx7000 gpu… you’d want the firmware on the ROM updated since the older OS couldnt have something for it.
you need to tweet this to steve so he can update his chart
Twitter is still going? 
Has anyone undervolted these?
I got my 7900xt today in the post.
The swap over was a simple reboot into safe mode, run DDU and shutdown. (Download the AMD GPU driver software also)
Swap card over and unplug ethernet
Boot up and install GPU drivers.
Reboot and plug in ethernet again as you go…
…profit?
Software picked up my monitor which is freesync premium (didn’t even know it was this i thought it was just VESA’s sync not AMD offering as well)
First game i have played is Forza Horizon 5, which while isn’t an AMD 7000 strong suit, for me has been AWESOME over the 1080 it replaces.
I get if your on a 5000 series or a 6000 series, the 7000 is probably a waste of time, but as someone who didn’t get the chance to upgrade till now, I am one happy customer… and now all AMD for this build atm
I have ordered an XTX, will be moving up from a 1080 as well! Can’t wait for the uplift in performance. 
Another year and I might have got Nvidia, but they are so anti-consumer I just can’t abide it. So I used my wallet to make my statement.
Have all the parts for a wicked XP/XP-era Linux box sitting on a test bench in the garage along with a dozen plus GPUs I bought in a frenzy this fall for just such an occasion.
As usual, life got in the way time-wise and I haven’t fired her up yet.
As far as the 7 series GPUs, there’s still so much conjecture and speculation around performance that I just don’t know. That $1,000 price point feels so steep, yet affordable based on the new baselines set by so many market forces. Wait and see, I guess.
As far as the 7 series GPUs, there’s still so much conjecture and speculation around performance that I just don’t know. That $1,000 price point feels so steep, yet affordable based on the new baselines set by so many market forces. Wait and see, I guess.
My advice is just to look at the performance we have now and base ones decision on that. We may get some nice driver updates that fix up some things in the future, but it’s best not to bet on that.
On the other hand, they seem fine in the market we have at the price they are.
The ATOM BIOS on the AMD setups don’t work that way. If you update to new firmware and then want to use it on an old system, the old system would not know how to talk to the ATOM BIOS and you would have to do custom patching to make that work, hence why AMD just upstreams it all to the kernel. nVidia is closed source and proprietary and that is why you have to jump through the hoops you have to for that side of the house. they are just two different philosophies of how to support a product.
READ: AMD is pretty much plug and play once the initial support has been upstreamed to the official Linux kernel. You can run the proprietary driver if you need day one specific feature support but within a kernel release cycle, the open source driver will surpass the official driver and firmware combo.
That’s not true.
If your card has a newer ATOM BIOS, older kernel/system simply won’t upload the older firmware to the card.
Regardless what version of ATOM BIOS your card has, it should work in old and new system as long as the AMD driver comes with the system said with support of your card.
That’s how a sane engineer would design firmware and drivers…
Have you worked on the AMD Open Source drivers? In a sane world, yes, but you are dealing with an open source community that has different aims and a for profit company with different aims. They goals and deadlines are completely different. Unfortunately, they cannot all agree on the same things except for nVidia sucks to work with.
True, thus my point. If you are newer firmware, then in order for it to work on a kernel that has not been back ported with the new hotness, it will not work correctly. See the whole GCN family, especially the RX 290X series. A total dumpster fire on *nix systems because of all of the hardware and firmware revisions.
That is not how that works. basic VESA support will always be there but we have to qualify “work”. Can it light up pixels, yes. Can it drive 3D, more than likely. Will is support the features of the card that did not exist in that kernel version + mesa combo, no. No only that, some cards refuse to work (again, see the above).
Uploading ATOM BIOS is handled by AMDGPU driver. It’s as simple as that. The driver knows when to upload and when not required to upload. ‘Upload’ means by mapping the firmware into RAM accessible by the card.
Yeah, I know that. I am not saying that is not how it works But you cannot have a new card with an old firmware from before the card existed. You also have cut offs of what old cards get as far as new firmware; AMD makes that decision. Availability is really what determines that.
I am going to disengage at this point. The asker of the question can decide which information is useful to them.
Well, did you ever get one, locally?
