• 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.

AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

Status
Not open for further replies.
Joxeon said:
Just doubling a 5700XT die size would put it above a 3070 and that's before clock, IPC and power efficiency improvements which RDNA2 will bring.

I think AMD will have the performance but can they match nvidia's pricing?
Click to expand...

GPU's are not like lego, you can't just double the GPU and get double the performance. The whole architecture has to be designed to scale to a certain size. Hawaii (290X, 390X), Fiji (Fury) are prime examples of big GPU's that didn't have the architecture finesse to harness all that power. AMD are seemingly getting away from that approach.

If you just doubled the Navi10 (RX5700), you'd have a very poor performing GPU.
 
humbug said:
Also...

Something dodgy about Nvidia's marketing.

https://twitter.com/HardwareUnboxed/status/1301796954398101504

SvWSGTJ.png
Click to expand...

I called this one yesterday in the 10gb VRAM enough for the 3080 thread.

Joxeon said:
Maybe this was why the 3080 was so far ahead of the 2080 in the GF benchmarks at 4K as that card only has 8gb VRAM.
Click to expand...

Dave2150 said:
GPU's are not like lego, you can't just double the GPU and get double the performance. The whole architecture has to be designed to scale to a certain size. Hawaii (290X, 390X), Fiji (Fury) are prime examples of big GPU's that didn't have the architecture finesse to harness all that power. AMD are seemingly getting away from that approach.

If you just doubled the Navi10 (RX5700), you'd have a very poor performing GPU.
Click to expand...

Well the new cards will be designed that way.
 
Joxeon said:
I called this one yesterday in the 10gb VRAM enough for the 3080 thread.
Click to expand...


To be fair, even if you put the card up against the 1080ti which does have enough VRAM, the performance boost is very favourable for a generational leap in performance.

Its sneaky, no doubt about it. NVIDIA are trying to make the 2080ti out to be a redundant GPU.. without actually showing its redundancy... and I think those that are selling their 2080tis now will get a rude awakening when the benchmarks come out the 2080ti is still not as bad as NVIDIA have made it out to be.

Some people will be grabbing an absolute bargain with these knee jerk 2080ti sales; which for all intents and purposes are based off of 2080 vs 3080 % based hand-picked by NVIDIA benchmarks.
 
Oy Vey.......nvidia fanboys. Evil huang said during his kitchen table unveil of Ampere that he DOUBLED UP everything for Ampere. So the claimed cuda cores are -

RTX-3070 = 5888 cuda cores (but in reality is likely 2,944 cuda cores). RTX-2070 = 2304cc (2070 super = 2560cc).
RTX-3080 = 8704 cuda cores (but in reality is likely 4,352 cuda cores). RTX-2080 = 2944cc (2080 super = 3072cc and the 2080Ti = 4352cc).
RTX-3090 = 10496 cuda cores (but in reality is likely 5,248 cuda cores). For comparison, the RTX-Titan-X has 4608cc.

Bearing in mind top dog 'big Navi' has 80CU's (5120 Stream Processors), I think big Navi is not out of the race. Compare this with the RTX-Titan-X which has 4608cc.

I think evil huang is doing some major BS'ing and if he is lying through his teeth about something as basic as cuda core count on the 3070, 3080 and 3090, then I bet he's also greatly embellishing about their performance in gaming.
 
4K8KW10 said:
So cheap doesn't look right :( It means the performance is not there:
Click to expand...
That's not cheap, that's still too expensive in this horribly skewed market. Also, why do those price points automatically assume the performance isn't there? Maybe AMD's claims of being "disruptive" centre around significant price cuts? Hell, that statement also implies AMD wanted to charge an inflated price for the cards but were forced to lower it thanks to Ampere. Just like they did with the 5700 duo last year.

You did nothing but foam at the mouth about how Ryzen processors are far too expensive, but now you're saying their GPUs have to be expensive otherwise they won't perform?

Pick a story and stick to it, rather than flip-flopping your opinion with literally every post. Or troll off.
 
NZXT30 said:
So much hype around DLSS and RTX IO though I think people need assurances that AMD will have equivalent features.
Click to expand...

Well, let's look at all the other Nvidia 'inventions' shall we...

RTX - Actually Microsoft DXR
RTX IO - Actually Microsoft Direct Storage.
DLSS - Waiting to see what AMD use (possibly MS DirectML?)
 
Status
Not open for further replies.
Back
Top Bottom