Recent content by midnightoil

They changed nomenclature shortly before launch, and decided to call them the 5900X and 5950X instead. If Zen 4 stays on target for Q4 next year, I doubt they'll bother with any XT parts.

RIP Intel. Pretty sure 5700X or 5800 non-X and a 5600 will launch early next year, once stock of the 3xxx has tapered.

Of course it's constrained. It's not infinite. But it vastly exceeds one vendor's leading edge products.

What would Apple produce? Why would they want all of 5nm? What are they going to sell? TSMC's capacity far exceeds one vendor.

HBM2e is unlikely to be more expensive than GDDR6X, from what I've heard, especially when you account for more expensive, complicated power delivery as the latter is so power hungry. HBM2, sure. 2e ... highly unlikely. But I doubt AMD are using 2e. I suspect they're waiting for RDNA3 to do...

They're going to be virtually unobtainum. Huge die, ultra low yields, limited wafers initially. Tbh if they didn't charge it this high, the extra margin would just go to gougers. The only reason you're seeing the 3090 is because NVIDIA are very worried about RDNA2. The 3080 die would have...

I think you can forget it being a 3090 die going down. There are already a load of disabled shaders. You'd likely be dealing with REALLY REALLY shitty dies if they were worse than those meeting the minimum requirements for the 3090, of which there will be few anyway. They'll be hoping yields...

Every single indicator I've seen is that there'll be virtually no stock, and yields are awful. All the analysis I've seen agrees. There's no way they'd have such a high % of disabled cores otherwise. 20% would seem to indicate it's at the limits of the process. Power consumption is another...

No chance in the near term. Volume is gonna be tiny, and likely there will be a high number of RMAs which will eat stock. TSMC -> Samsung; new modified 10nm process (8nm) not really designed for this kind of thing; large dies; high TDPs; 20% of CUs disabled (that's a massive number -...

Based on all the performance data from both next gen consoles, technical document releases, and that they clearly will have evaluated PS5 and XB dev kits, and likely had rough performance of RDNA2 desktop leaked to them. The pricing and gigantic, inefficient 3090 reflect this. Why else would...

They're expecting RDNA2 to be faster and more efficient than Ampere ... whilst hoping the huge, brute force 3090 will retain them the absolute crown. Given how inefficient this is looking as an architecture, and that they're on a slightly shrunk 10nm process (that's what Samsung 8nm is), that...

To me, this suggests that NVIDIA *think* they are likely to be in serious trouble with RDNA2. This card screams massive die pushed way past a reasonable point on the clock / voltage (efficiency) curve. This appears to be a much larger die in relative terms than the already huge 2080Ti die...

That valuation is ludicrous. But all these tech behemoths need to be broken up. If the US won't do it, then other localities need to make sure the regional operations are broken up. It's worse than the Standard Oil days.

Haha. Johnson and his cronies wouldn't give a shit, and none of them are likely to be remotely informed enough (or bothered to read briefings) even if they did care about stuff like this. They're not going to block the sale of a UK company to a US one anyway. Would imperil the trade deal they...

This is incredibly unlikely to happen. 1) Competition. This will never get past competition authorities in the EU. It shouldn't in the US either ... but typically they will do what ever they think is best for 'brand USA'. 2) Because of the profound competition implications, it's very...