Soldato
Who is Patrick Schur?
-
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
- Thread starter Gregster
- Start date
- Status
Soldato
Surprised nvidia haven't got a proprietary closed box brand for that..
Associate
- Joined
- 19 Jun 2017
- Posts
- 1,029
Soldato
Someone who got that data from an AIB
Yes but who is he, some random, or someone with history?
Associate
- Joined
- 19 Jun 2017
- Posts
- 1,029
Reverse engineer.. check his Twitter acct
Soldato
Is the guy entrusted to deliver another hint this weekend!
Haha, nice one
I've seen his name with other leaks, but I'm not sure if he's truly connected
Soldato
Hang on so he's got his hands on an AIB board already, or hes speaking to someone that is part of the AIB and sharing some testing?
Or, he's making a credible guess based on other rumours. Very hard to say.
Soldato
Well a while ago it was shared the lower sku (navi22?) would be clocked high, just not as powerful as the navi21 which will boost lower than that mhz.
Even TSMC's refined 7nm with EUV is only a 15% frequency up lift over their first version and there are smaller and simpler parts than a typical high performance GPU that struggle to get 2.3GHz so it would be pretty incredible engineering if they've pulled it off on anything bigger than a 5700XT die.
The parts getting those kind of speeds and upwards are like 70mm2.
Soldato
- Joined
- 8 Jun 2018
- Posts
- 2,827
2.4Ghz sounds tasty
Soldato
Insane clocks. I am thinking the reveal will be a success for once, if they have a lineup muscling in on nvidia I do hope the stock materialises so we can bag one before scalpers and bulk buyers ruin it.
Some performance estimates based on conservative values for 2.4Ghz 80 CU and 72CU parts.
80% CU scaling and 50% clockspeed scaling with 10% IPC uplift. 5700XT as baseline.
80CUs = 1.8 x 1.125 x 1.1 = 2.23x 5700XT. This is neck and neck with the 3090FE @4k based on TPU numbers.
72CUs = 1.64 x 1.125 x 1.1 = 2.03x 5700XT. This is neck and neck with the 3080FE @4k based on TPU numbers.
I think AMD are bringing some genuine top tier competition this go around.
80% CU scaling and 50% clockspeed scaling with 10% IPC uplift. 5700XT as baseline.
80CUs = 1.8 x 1.125 x 1.1 = 2.23x 5700XT. This is neck and neck with the 3090FE @4k based on TPU numbers.
72CUs = 1.64 x 1.125 x 1.1 = 2.03x 5700XT. This is neck and neck with the 3080FE @4k based on TPU numbers.
I think AMD are bringing some genuine top tier competition this go around.
Soldato
Its almost too close like they planned it. What makes me believe its not that is the 'cheap' and rushed nvidia release with the massive regression in power consumption + card sizes. This story all points to nvidia backtracking as they got tipped off how fast big navi was predicted.
What will be the icing on the cake is (for AMD sympathisers) should the gap have been nullified within one generation yet coming out with a better power draw, the last piece of the puzzle is at what price?
What will be the icing on the cake is (for AMD sympathisers) should the gap have been nullified within one generation yet coming out with a better power draw, the last piece of the puzzle is at what price?
Soldato
Lets hope so.
So do you think the 3080 was going to be slower, then Nvidia got tipped off and had to push the 3080 to its limits? Or, do you think that AMD are now pushing their card to the limits knowing what the 3080 and 3090 can do? Either way round is feasible.
- Status