Ampere Discussion Thread

[SteelCrysis]

https://www.tomshardware.com/news/new-nv...ere-series

Overclocking your brand new RTX 30 series GPU is a risky move but someone has already done this, hardwareluxx reports a ROG Strix RTX 3080 BIOS has been flashed to use the BIOS for an RTX 3080 TUF Gaming using a new version of NVFlash that includes support for Nvidia's new Ampere graphics cards.
...
Now, all we're left with is flashing the graphics card BIOS, since we aren't modifying the actual BIOS, this is still something we can do to hopefully get higher clock speeds. Technically as long as you grab a BIOS from another card with the same GPU (say RTX 3080 Founders Edition to RTX 3080 TUF Gaming) that can work.

But should you do it? Probably not, flashing a BIOS from one card to a different card is obviously very risky and not something the cards were designed to do. In a worst-case scenario, your graphics card will be completely bricked, and the warranty rejected because you tampered with your card. Hardcore overclocking is the only real reason to do it, and if you don't mind risking your GPUs life in the process the performance gains could be incredible.

https://www.extremetech.com/gaming/31570...ity-issues

We’ve covered the reports of instability across the RTX 3080 and RTX 3090 product families, even cards from different manufacturers. An early theory, coined by Igor of Igor’s Lab, is that the issue could be caused by sub-par power circuitry, especially since there seemed to be evidence that the problems were concentrated in certain specific GPU families much more so than others.
...
Here’s the truth of the situation as we know it today:

Nvidia launched Ampere. A lot of customers had problems with Ampere. A logical theory based on the quality of the power rail circuitry was advanced and endorsed as plausible.

Meanwhile, people discovered that lowering your GPU boost clock by ~100MHz or locking your GPU clock to a high, fixed frequency both produced more stability. This was not in tension with the first theory — a circuit that can’t quite keep up with the demand places on it at 2.1GHz might be perfectly at ease at 1.9GHz. Similarly, repeatedly switching a GPUs clock speed up and down requires much more complex power distribution compared with running things at a static clock.

After several days of speculation, Nvidia has released a new driver, 456.55, that seems to have dramatically improved the situation for most players. Games that were previously unstable at low clocks now run rock-solid at higher frequencies. There are some people still having problems after the update, but it appears to have worked for the majority of people.

We’re still going to keep an eye on this solution as it evolves, but Nvidia’s latest driver appears to resolve the issue quite effectively.

https://www.extremetech.com/gaming/31571...nstability

One of the frustrating things about trying to sort out the RTX instability issues from last week’s launch has been the relative paucity of comments from vendors. Now, however, they’ve collectively broken their silence — and they’re all saying pretty much the same thing: Update your video drivers.
...
Overclocker der8auer, otherwise known as “Person who does things I don’t have the guts to try,” decided to replace two of Gigabyte’s stock 470u CP-CAP capacitors with twenty 47u MLCC capacitors (this works out to the same power capacity for both setups). His maximum stable overclock went up 2 percent as a result, or about 30MHz.

Der8auer’s results do show that power rail hardware can make a small difference, but it’s not enough to really move the needle one way or the other. Instead, the problem really does appear to have been driver-related.
...
The example above is hypothetical; we don’t know what Nvidia adjusted in its driver to improve stability, and while there have been reports of clock drops, there have also been reports of clock improvements.

I’ve been following this story since it broke and I’ve written a number of updates to illustrate how quickly something can evolve — and how early reports, even when they accurately identify a problem, can incorrectly identify the cause. Until and unless new evidence emerges showing the problem is still somehow linked to the POSCAP / MLCC question, the Nvidia 455.56 driver appears to resolve the extant problems. If they stay resolved, they’ll be remembered as a hiccup on the way to a successful overall launch.