• 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.

The (Ada) Lovelace architecture Thread.

Kopeite says both Lovelace and RDNA 3 are targeting roughly double performance over Ampere and RNDA 2.

However, both cards are also currently running into cost issues - that's not to say people won't accept higher prices anyway as we've seen, just that the new cards are more expensive to make than the current ones.

The big kicker is memory bandwidth, GDDR6 and GDDR6x just isn't cutting it - both Lovelace and RDNA 3 are taking large performance penalties when mated with G6 and G6x modules, to get the doubling of performance thats being estimated requires a large improvement in bandwidth which is what they are struggling with.

It will be interesting to see what Nvidia and AMD come up with to solve the bandwidth issue. Will AMD increase it's L3 game cache for RDNA 3 to try and fix the bandwidth limitation? What's Nvidia plan, expensive HBM2 modules? HBM2e supports up to 24GB capacity and up to 2500GB/s of bandwidth but it is several times more expensive than G6x
 
Last edited:
Kopeite says both Lovelace and RDNA 3 are targeting roughly double performance over Ampere and RNDA 2.
Nice. Will probably grab a 4070 and that would last me a long time. That said I will likely go RDNA 3 this time, as long as they are in stock and non ripoff prices. At least with Nvidia I can buy a founders edition at rrp like I did my 3080.
 
Bad @Cyber-Mav, bad! :D

I wonder, though, how do RT cores affect hash rates?
It's irrelevant. It's the memory bandwidth that matters, at least for the algos you'd actually run. Unless something takes Eth's place then mining is not something we need to worry about for RDNA 3 & Lovelace/Hopper because Eth is gonna be PoS by then.
 
People are seriously getting suckered into this? Think back in time and when has performance EVER doubled? EVER?

30-50% at the very most but this happens before every launch and its never what people claim early doors.
 
I keep watching regamingtech channel on youtube for rumors and news but not sure how well respected they are, seem to have a lot of subscribers for a UK channel
 
People are seriously getting suckered into this? Think back in time and when has performance EVER doubled? EVER?

30-50% at the very most but this happens before every launch and its never what people claim early doors.


Performance probably did double back in the early days of gpus at some points, if not they probably got close to it. Harder to do these days, whipping in a few extra billion transistors doesn't yield the bump you would expect for the most part.
 
Performance probably did double back in the early days of gpus at some points, if not they probably got close to it. Harder to do these days, whipping in a few extra billion transistors doesn't yield the bump you would expect for the most part.


It does for FP32. The 3090 is already double the performance of the 2080ti in Cuda workloads.

Games don't necessarily scale well with more cores but workloads that use Cuda cores have a nearly 1:1 scale - double the cores = double performance
 
It does for FP32. The 3090 is already double the performance of the 2080ti in Cuda workloads.

Games don't necessarily scale well with more cores but workloads that use Cuda cores have a nearly 1:1 scale - double the cores = double performance

But lets be honest here, that is a "technically".

We are talking about games mostly as that is what 90% of us use them for.
 
Will be very happy if we can get a 50% uplift in performance 3090 to 4090. Anything else is likely a pipe dream.

Would be happy with a 4080 with 16GB that has a 40% boost in performance over the 3080 at £649.

Ideally I want to see a bigger improvement in RT though. I expected bigger gains than we got with Ampere.
 
Back
Top Bottom