Well, I had to shoe-horn the BIOS in there, by using the flashback feature. This meant putting the BIOS ROM on a USB stick, renaming the ROM file and putting it in a dedicated USB port and press a BIOS flash button on the Mobo itself.
I tried it via the flash tool, but it couldn’t find the file. I should’ve taken the hint right there, but…
A filename is a poor choice for a flashing tool to identify a firmware, because you guessed it, it can be changed by the user (or an attacker).
Ideally cryptographic methods are used to sign a firmware so the tool can ensure:
a) that It actually comes from the manufacturer and
b) that it is the right firmware for the hardware it’s being applied to.
Every appstore should do this nowadays even for apps than run in heavily sandboxed environments. It really shouldn’t be optional for something that has the privilege level of a firmware.
But in AsRocks defense, the flashback method is a last ditch effort, when all else fails.
When I tried the default method, I got an error message saying the file was not found. This made me look for alternate flashing solutions. I guess it would’ve been better if it found the file but told me it was incompatible…
The fact that the flashback functionality seemingly doesn’t include checks is both a security issue as well as a liability for the user (or the manufacturer as well if they accept that as RMA) in case someone bricks their mobo. I’d think about reporting that fact to ASRock to see if they are aware of that (perhaps they can/want to do something about it, even if unlikely).
“Reloaded BIOS, could not replicate fault”
Got fed up with the market and went for plan B. I installed my new 7800 XT yesterday and will look for an upgrade when we actually see good deals for buyers.
Now for the fun of making all the controller profiles work on my new machine. I wrote I guide somewhere here for exactly this use case…
I could use that too…
Here you go
Today one shop had the ROG Astral gaming OC 5090 available for “a not completely insane but still way too bloody high price” available so I pulled the trigger. I might have a complete rig this weekend
Mein Beileid
Ich weiß, es ist ziemlich bescheuert… naja, EGAL
YOLO
This thing is nuts:
The GPU support thingy supplied with the 5090 is too tall for my case so Lego to the rescue until I can craft a more elegant solution
Are you referring to the hole it made in your savings account?
Maybe
With my old graphics settings and PD at 1.0 it’s solid at 90fps. After I cranked some settings and put PD to 1.2 I get some drops but I basically cranked everything up with DLSS on quality. It should be fairly easy to dial it back slightly to get absolutely smooth 90fps everywhere
Closed the case as I am happy with the setup for now. I’ll look into a wooden GPU support to replace the Legos but I want wood that fits the case. Put the baby 5090 in as well
Max current when running the card at 600W (power limit) with furmark was 8.8A at the pin with the highest current, 7.4A at pin with the lowest current so it looks like the connector does make a decent enough connection. No danger of melting anything
I’m not convinced running Furmark is such a good idea. I feel like it unreasonably stresses your card beyond what it would do even under intense gaming scenarios which could weaken the cooling ability in the future by messing with the heatsink adhesive.
All you need is two things:
Thick cables
Plenty of airflow
I am not constantly running furmark. I ran it for a few minutes until temps stabilized and checked that no pin exceeded its rated power. It’s an easy way to get the card to run at its power limit. No game I tried, even Cyberpunk 2077 at max settings with all the ray tracing, path tracing and other stuff enabled got even close to the 600W limit. DCS in VR draws about 400-450W max, Cyberpunk 2077 is a little higher at around 475W.
Interestingly, DCS in VR can actually use close to 32GB of VRAM and I don’t just mean allocated VRAM. Just place a ton of assets around an airfield and bam 30-31GB VRAM usage, the F-16C instant action from Bagram AFB is a good candidate to get it all used up.
Edit to elaborate:
That’s the problem and the reason why I chose the way too expensive Asus top tier card (well there’s still the AIO version but that’s 250€ more expensive than the already ridiculously priced ROG Astral lol) and then ran furmark before closing the case and once again after I closed it because the cable does touch the front glass slightly.
The issue is not the cables or the cooling, it’s the flimsy 12VHPWR connector that nvidia insists on using. The ROG Astral is the only 5090 that can monitor the current on each pin of the connector.
You can have all the cooling and the thickest cables in the world but if the connector doesn’t make proper contact it will inevitably start to melt since the design of the 12VHPWR connector does not feature a sufficient safety margin and has no failsafe against a bad connection. It’s rated for up to 9.5A per pin and the cable is designed as a 600W cable so you end up with a headroom of just 1.17A per pin, which is ridiculous considering the 8-Pin PCIe connector is rated for 150W and can easily achieve double that without melting the connector, it’s got a more than 2x safety margin.
The cooling of the 5090 Astral is bonkers btw. At 600W, GPU temps topped out at 73°C and then stabilized at 71° and the memory stayed at 80°C after reaching 82°C. This is with the case closed and the midle fan and rear fan limited to 50% RPM and the outer fans on auto.
That’s such a rocket sled of a GPU, lol.
I’m side-eying at RX 9070 XT prices, but they still feel unjustified coming from a RX 6800 XT. Maybe later.
You misunderstand.
I would never run Furmark for even one second. I see it as akin to driving your car around at redline RPM in first gear to make sure the oil and carb cooling works. I think the entire concept of overstressing the card is flawed. I do not believe the information gained is worth the cost to the hardware.
If your connector wasn’t working great, you would have just melted it down. How does that help?