NV Super refresh reviews

[blckgrffn]

Jeff compares with the 3090, 2080ti. Titan XP, and 6900XT -

TL;DR man, that's just a tease for a video I won't watch.

I am glad you got my reference, despite me miswording it!

 

[SteveGrabowski]

Guess no one gives a crap about these either since just checked newegg and plenty are still available to order on launch day.

 

[Heartbreaker]

Guess no one gives a crap about these either since just checked newegg and plenty are still available to order on launch day.

Sure, just like no one cared about the 4070 for the exact same reason, so a similar "failure".

 

[beginner99]

The funny part is this would've been the launch segmentation if RDNA3 worked as intended.
Mind you, they had to do this refresh, it is yet another very well calculated plan. I'll let you lot speculate on why it is clever.

Not sure why it's clever. Anyone with a 4000 series (or 3000 series) will still not upgrade. All the chips were cut while yields clearly didn't have the need for that. NV could have launched like this easily and AMD would have been in trouble, 7900xt(x) would have been a hard sell against this line up so it would have lowered AMD revenue at little to no cost for NV because the chips are cut and I doubt yields made these cuts necessary.

 

[DavidC1]

The funny part is this would've been the launch segmentation if RDNA3 worked as intended.

How many times have AMD faltered vs Nvidia? Lot more.

If you look at the overall execution of the two companies, Nvidia is far better. Actually both x86 vendors have spotty records of execution. From multiple CEO changes, aiming for unsustainable goals, to pure batshit crazy ones. Both companies have fallen nearly completely flat on their faces.

Nvidia is in a perfect position of milking people when at top, and when the two competitors try to catch up, they can cut prices and still keep them at bay, until they get next generation.

Ah, but it makes sense. With the CEO still being the original Founder, stellar employee ratings with low attrition rate, and excellent engineering mindset.

 

[DAPUNISHER]

TL;DR man, that's just a tease for a video I won't watch.

$600 card on par with last gen $1500 card. Slightly slower at 4K, slightly faster at 1440.

 

[Aapje]

Not sure why it's clever. Anyone with a 4000 series (or 3000 series) will still not upgrade.

Not sure why you think that any sensible person would ever upgrade to a refresh of a card they already have, or why this matters.

The question is more if this will convince 10x0, 20x0 and 30x0 owners to upgrade.

 

[DAPUNISHER]

The question is more if this will convince 10x0, 20x0 and 30x0 owners to upgrade.

Right on.

The timing of this launch is well calculated. Here in the U.S. people will be getting their tax refund money shortly. Some may still be holding on to their Xmas money too.

 

[branch_suggestion]

Not sure why it's clever. Anyone with a 4000 series (or 3000 series) will still not upgrade.

Obviously Ada owners are set, not many of them though, sales are down from Ampere.
3070Ti and below buyers are the target and some will be convinced, maybe even the occasional 3080 owner.

All the chips were cut while yields clearly didn't have the need for that.

Top yields early on go to professional SKU's if they have room to do so, which they did.

NV could have launched like this easily and AMD would have been in trouble

NV would've made less money, demand wasn't there.

7900xt(x) would have been a hard sell against this line up so it would have lowered AMD revenue at little to no cost for NV because the chips are cut and I doubt yields made these cuts necessary.

NV would've made less money, a marginal market share slip is worth the higher margins.
AMD failed to execute, NV was therefore under zero pressure to bring their best lineup, so why would they do this?
Biggest reason is simple, RDNA4 is coming, and they will expect it to butcher their midrange, so this refresh softens the blow enough until client Blackwell, which is going to come out ~2Q afterwards.

But really, the goal isn't to hurt RDNA4 directly, but lessen the perception of dominance in the low/midrange. What they are really trying to do is force AMD to cut prices of RDNA3, this makes the value leap of RDNA4 lower.
N32/33 will stop being produced as soon as N48/44 go HVM, N31 might go on for some time longer but it is really hard to say.
RDNA4 is massively attractive to 3070 and lower users, which is the vast majority of the market today, NV therefore is trying to capture as many of them as possible before AMD launches without tanking margins.

Now BW will stop any bleeding, not sure how competitive both are but NV should remain dominant regardless, after all AMD have no high end until Q4 2025 at the earliest.
AMD will not make a dent without a peerless halo part, RDNA4 being cut down to midrange only gives them the best chance to make one in a timely manner, RDNA5 should come out before BW-Next after all.
Until then, expect high end BW to be awful value and a sandbag without any competition. If you care about client GPUs, you want AMD to execute properly from now, consumers win.
Also Intel exists, BM might be cute if it comes out before RDNA4, I guess.

 

[BFG10K]

That is a big issue. I wouldn’t get a GPU with that pin.

Same here. Actually 1x8-pin is my max, I won't buy a GPU with more than that.

 

[Aapje]

Same here. Actually 1x8-pin is my max, I won't buy a GPU with more than that.

Isn't that extremely limiting? Even my 6700 XT has two 8 pins.

 

[branch_suggestion]

Isn't that extremely limiting? Even my 6700 XT has two 8 pins.

Good enough for 225W including PCIe power. Most cards will move to 8+6 or dual 8 at that TDP though. 200W TDP is as high as you will see them usually.

 

[Heartbreaker]

Obviously Ada owners are set, not many of them though, sales are down from Ampere.

What's the evidence there aren't many of them?

Look at the Steam results. There are VERY many ADA cards in the hands of Steam Gamers.

 

[Aapje]

MLID claims that sales are really bad, especially for the over-MSRP cards. He suggests that Nvidia solved the wrong problem, by releasing a more powerful card with still just 12 GB, which makes sense to me.

In general I think that people are generally not very forward-looking or rational in general, but they tend to develop and hold on to stereotypes, good or bad. See the people still claiming that AMD drivers are bad.

I think that Nvidia messed up with 8 GB for the 3070 and 10 GB for the 3080. The idea has now been established in many people's minds that Nvidia is selling cards with too little VRAM. With 12 GB, it is at minimum questionable whether this will be sufficient for the lifespan of the GPU, especially in a UE5/RT world. With the stereotype being established, many people will not give Nvidia the benefit of the doubt.

If they had given 10 GB to the 3070 and 12 GB to the 3080, I think that people wouldn't have become primed to see the 12 GB as a huge issue.

 

[igor_kavinski]

Look at the Steam results. There are VERY many ADA cards in the hands of Steam Gamers.

9% for 4000 series whereas more than 25% for 3000 series.

Yeah. Very many

 

[igor_kavinski]

MLID claims that sales are really bad, especially for the over-MSRP cards. He suggests that Nvidia solved the wrong problem, by releasing a more powerful card with still just 12 GB, which makes sense to me.

For once, I have to agree with him.

 

[MrTeal]

9% for 4000 series whereas more than 25% for 3000 series.

Yeah. Very many

Not an entirely fair comparison, since the 4k series has been out for a bit over a year while the 30 series was the top pick for 2-2.5 years. I see 6.33% for Ada add in cards vs 19.53% for Ampere though. Laptops are always a little trickier because for most you just get what you get based on when you buy.
Still, the sell through on Ampere was probably much steadier over its lifespan due to how restricted supply was. Ada sales probably going to be more front loaded to the release of each card. Even the Supers haven't seen a crush of demand; here in Canada Best Buy dropped 170 of the FE at 8am CST, by 3pm there were still 40 left in stock. They added another 130 in the evening and the number has barely moved. We'll see at the end of this year by the time Blackwell rolls around if Ada can touch 20%, but I have my doubts.

 

[In2Photos]

MLID claims that sales are really bad, especially for the over-MSRP cards. He suggests that Nvidia solved the wrong problem, by releasing a more powerful card with still just 12 GB, which makes sense to me.

If they increased the amount of vRAM that would be "admitting" that they launched cards with too little. No way they would do that. I bet they don't even raise it for their next gen cards. People buying nVidia don't care. They see that nVidia's flagship card is the fastest card on the market so they buy the best nVidia card they can afford regardless of how it performs or what the specs are.

 

[igor_kavinski]

Not an entirely fair comparison, since the 4k series has been out for a bit over a year while the 30 series was the top pick for 2-2.5 years.

It does however highlight the fact that most gamers don't consider Ada a must have, especially when they have to pay considerably more than their existing Ampere card to get to a higher performance level. I guess they are waiting for either prices to come down (not very likely) or just use their savings to get a 5000 series card, whenever they become available. Seems like a repeat of the Pascal vs. Turing situation where most Pascal owners didn't feel there was much value to be had with Turing cards.

 

[blckgrffn]

I think we are maybe in the lull between console launches. Ampere, in general, from the 3060 (12GB) and up pretty much can play everything that's console optimized pretty convincingly. Which is basically everything.

DLSS and the like is a double edged sword, helping "low end" cards stay relevant, especially Ampere+.

What's really pushing people to refresh right now? VR is MIA, IMO. AAA titles again are pretty optimized at sub 4K settings. The last console upgrade was doubly painful, coming right when GPUs spiked. Those that upgraded anyway spent way too much for what they got and might be rationalizing keeping them longer due to the investment and huge depreciation they've experienced. (I am one of these. This $1k 6800 has a long life to live, hoping for the ~5+ years my 290x gave me.)

That and those willing to spend "real money" (in my mind over $500 for a GPU) have had plenty of opportunities now and these new cards are just lower power versions of what came before, not really pushing the needle on performance.

I guess this is true for the whole PC space and maybe why sales are not awesome across the board.

 

[Heartbreaker]

What's really pushing people to refresh right now?

Everyone has different age cards, and needs to upgrade at different times, so there are always people that need to upgrade now.

 

[Saylick]

Everyone has different age cards, and needs to upgrade at different times, so there are always people that need to upgrade now.

And the people that need to upgrade now seem like it’s a smaller group than the people who upgraded last year.

 

[Heartbreaker]

And the people that need to upgrade now seem like it’s a smaller group than the people who upgraded last year.

Go look at the 4070 launch, people here claimed it was a flop because it remained in stock. Same nonsense as this time. Same MLID drama claiming poor sales (MLID is basically a soap opera for nerds).

4070 sold extremely well, and I expect that will continue with the Super model.

So much "concern" here about "poor" Nvidia and it's "poor" sales.

 

[blckgrffn]

Everyone has different age cards, and needs to upgrade at different times, so there are always people that need to upgrade now.

I am talking aggregate PC sales, which slumped hard last year and is forecasted to maybe reach 2018 levels in 2024, a modest increase. However, they don't break out laptops and PCs in this, so its hard know if that is really the comeback of the tower or just more laptops. Growth in 2024 is all about W11 compliance it seems, but I don't know if many home users are going to be jumping up because of that reason in 2024.

I am not digging into more detailed nvidia or AMD quarterly recaps but all the growth is based on AI datacenter chips.

You posit that GPU purchases are linear. I am of the opinion they are more cyclical and needs/fomo based. I don't think there will be a definitive answer.

I am fine with folks buying whatever they want when they want, but we are still at the point where this refresh brings little beyond better efficiency except at the highest end. That's not the kind of headlines that Ampere ran with. The FOMO and urgency were real.

That's not only an nvidia issue, the AMD refresh was arguably worse. Again, outside of the top 2 SKUs.

 

[igor_kavinski]

So much "concern" here about "poor" Nvidia and it's "poor" sales.

Poor OEMs for the time being who bought into their lies. Which, if things don't improve, will lead to lesser GPU orders from OEMs in the next gen. So they won't get poor per se but they will definitely need to figure out some other snake oil to sell coz DLSS3 ain't it.