AMD RX 460 BIOS unlocked for 1024 Stream Processors

[HEXUS]

BIOS mod has been verified by a couple of tech sites this weekend.

Read more.

[Jace007]

I'm glad you mentioned the new AMD drivers picking up on these BIOS Mods, Do it at your own risk!

[CAT-THE-FIFTH]

Wow,it actually hardly changes power consumption either:

https://www.techpowerup.com/228601/a...eam-processors

The Asus Strix gains nearly 10% overall in performance just from the unlocked shaders.



So basically AMD just made the GTX1050 look better for no reason.

[scaryjim]

... So basically AMD just made the GTX1050 look better for no reason.

Probably the early yields weren't good enough to produce enough fully-enabled cards. They'd have looked a lot worse if 1 in every 10 cards was borked straight out of the box, and we've only had, what, 7 confirmed successes with this technique so far? Plus they were obviously reserving the top bin dies for the professional cards which have a much higher margin.

Yields should have improved over time, so I'd expect to see a fully enabled RX 465 down the line...

[DanceswithUnix]

Yields should have improved over time, so I'd expect to see a fully enabled RX 465 down the line...

It still doesn't beat the 1050 on performance, so if it hurt the £90 starting price of the RX 460 I think it would be a backwards step.

An extra crippled 470 might be a better move.

[CAT-THE-FIFTH]

It still doesn't beat the 1050 on performance, so if it hurt the £90 starting price of the RX 460 I think it would be a backwards step.

An extra crippled 470 might be a better move.

The Witcher 3 generally runs better on Nvidia cards anyway.

If you look at the TPU review of the GXT1050,its considered to be around 10% faster on average(TPU also tends to be harder on AMD cards,too if you look at the GTX1050TI vs RX470 performance difference),so a fully enabled Polaris 11 would probably be around GTX1050 level performance.

This is why the RX460 4GB should have at least launched as a fully enabled chip IMHO OFC.

[kompukare]

This is quite common from AMD though: they don't bin their products very aggressively.
I guess lower volumes explains some of it, but it seems Nvidia are willing to release higher bins first and then keep binning until they have set aside enough to release a lower binned part.
Unless the initial Polaris 11 yields are very, very poor indeed that would have been a better strategy for AMD too.
From the TPU power figures, downclocking the 1024 shader part to the 896 performance might make for a very efficient part.

[scaryjim]

... downclocking the 1024 shader part to the 896 performance might make for a very efficient part.

This is essentially what they are already doing for the professional grade cards, and the profit margins on the professional cards are much higher - which is why I suspect most of the top-bin dies went into the professional cards in the first place, leaving the consumer product with insufficient fully-functioning dies to risk a fully-enabled consumer part.

I can only see them releasing a fully-enabled RX 465 if the yields have improved sufficiently for them to do so at no extra cost: I can easily see a big trumpets-blaring Vega launch being followed closely by a Polaris refresh, with xx5 cards coming out to replace the incumbents at the same price (and the existing cards getting a price cut to shift the supply whilst they're still available).

[jigger]

Nice. I'm getting a single slot one in the new year.