Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,112
- Location
- Scotland
Yeah the post I quoted was talking about going to 80% perf for a power drop though. I don't know why anyone would buy a £1000 card to run it at 80% capability.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
NVIDIA ‘Ampere’ 8nm Graphics Cards
- Thread starter LoadsaMoney
- Start date
Yeah the post I quoted was talking about going to 80% perf for a power drop though. I don't know why anyone would buy a £1000 card to run it at 80% capability.
Soldato
- Joined
- 21 Apr 2007
- Posts
- 2,576
cos they like Minecraft?
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,112
- Location
- Scotland
cos they like Minecraft?
50 instances at the same time!
Not really sure why you think that Nvidia won't do it again with the 30 series cards, when they've clearly done it with the 20 series cards.
Fill every niche for every last $ seems to be the approach (from both manufacturers).
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,112
- Location
- Scotland
You think the Nvidia card will end up 10% slower than the high end RDNA2 card?
My question to both manufacturers this time is "How fast are you at given price point?"
RTX basically re-released the 1080 Ti and asked for more money because ray tracing. The only way to get more speed over the old generation was to spend more money.
I get the impression that Nvidia got this cadence going, from generation to generation, where each price point got meaningfully faster than the previous gen...then failed to maintain that cadence with RTX and just pretended that they were still on cadence.
I hope the market doesn't get fooled with names or status symbols this time. Make my money go meaningfully faster than the last gen.
RTX basically re-released the 1080 Ti and asked for more money because ray tracing. The only way to get more speed over the old generation was to spend more money.
I get the impression that Nvidia got this cadence going, from generation to generation, where each price point got meaningfully faster than the previous gen...then failed to maintain that cadence with RTX and just pretended that they were still on cadence.
I hope the market doesn't get fooled with names or status symbols this time. Make my money go meaningfully faster than the last gen.
Associate
- Joined
- 14 Aug 2017
- Posts
- 1,196
The RTX 2070 had the same launch price as the GTX 1080, and has somewhere around 10-15% better performance before you look at raytracing.
The boost at the same pricepoint was definitely not as much as previous generations, and the prices do seem to have got higher, but the performance per dollar also seems to have gone up.
RTX is obviously going to raise the price as It's a new useful feature and also there is additional hardware on the chip.
But I thought the reason we got such price hikes in the 2000 series was to beta test and pay for the R&D of RTX? At least that is what I recall a lot of people saying back when it came out
I'd be happy if they had of release gtx 1670 80 ti etc with the same perf as the rtx but without ray tracing and a price that reflected that.
I have the gtx 1070ti and even then I didnt think the performance was enough to warrant an upgrade.
The "usefulness" of RTX is questionable for this first gen.
Why? These RT Cores shouldn't just be used for RTX.
For an upgrade, I would like to be able to upgrade my HMD from a CV1 Oculus to a Reverb (or something similar)and maintain or improve my graphics settings. Basically, I want to move more pixels. The 2080 offered to move the same pixels as my 1080 Ti , for the same money. I don't care if RT cores could make me a tasty sammich. That's not what I'm in the market for.
Maybe 2020 will be the year when someone can make $600-$700 perform meaningfully better than it did 3 years ago. (It will actually be closer to 4 years by the time big Navi/30xx series drop)
What the hell is a Reverb ?
EDIT: Google answered this
https://www.anandtech.com/show/14104/hp-reverb-virtual-reality-headset
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
Full Fat Ampere GA100 detailed, this the maximum Ampere architecture allows
8192 Cuda Cores @ 2.0ghz
1024 Tensor Cores
130 RT Cores
48gb HBM2e memory @ 1.2ghz
300w TDP
Built on TSMC 7nm+
36 Teraflop peak output
Just beware this IS NOT a the specs for a gaming GPU. This is top of the line HPC Tesla, probably cost $10k USD.
Gaming chips max out at GA102, which will be cut down version from these specs and likely use GDDR6 memory not HBM2e
8192 Cuda Cores @ 2.0ghz
1024 Tensor Cores
130 RT Cores
48gb HBM2e memory @ 1.2ghz
300w TDP
Built on TSMC 7nm+
36 Teraflop peak output
Just beware this IS NOT a the specs for a gaming GPU. This is top of the line HPC Tesla, probably cost $10k USD.
Gaming chips max out at GA102, which will be cut down version from these specs and likely use GDDR6 memory not HBM2e
Last edited:
Associate
- Joined
- 14 Aug 2017
- Posts
- 1,196
Eh, more speculation and rumour...
It is always speculation and rumour though. I think we can all agree though that both NV and AMD and gearing up to launch products with big jumps in performance this time. The biggest unknown at the moment is launch date. Yes we know 2020 for RDNA2 but no real hint as to what part of 2020.