Soldato
Its all about the power efficiency now! Oh and mid range has tried to be pulled up to £500+ will see how long intel take to join that conga..
-
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.
It looks like the 'real' /affordable RDNA3 + next gen NV desktop launch won't launch until September. Thoughts?
- Thread starter g67575
- Start date
Its all about the power efficiency now! Oh and mid range has tried to be pulled up to £500+ will see how long intel take to join that conga..
Soldato
Navi 33 is more entry level/low end now.
Well, we can directly compare a RX 6800 RDNA2 GPU to an RDNA3 GPU with 60 CUs.
You should get performance that is 15-20% faster than the RX 6800, so it should be a bit faster than the 72 CU RX 6800 XT.
I'd imagine they'd call it a RX 7800, considering the top Navi32 die has the same CU count as the last gen 60 CU part.
It should compete with the RTX 3080 and RTX 4070.
That would mean that they could release the RX 7800 XT with 70 CUs (on Navi31) at a later point.
That would be congruent with the naming scheme of the workstation W7800 (70 CUs).
In term of the RDNA3 GPUs still to be released, maybe the clock speed fixes that have been talked about, are simply a way to achieve good clock speeds even on the lower quality + lower tier silicon. So, perhaps Navi33 and Navi32 will have similar maximum clocks to Navi31.
You should get performance that is 15-20% faster than the RX 6800, so it should be a bit faster than the 72 CU RX 6800 XT.
I'd imagine they'd call it a RX 7800, considering the top Navi32 die has the same CU count as the last gen 60 CU part.
It should compete with the RTX 3080 and RTX 4070.
That would mean that they could release the RX 7800 XT with 70 CUs (on Navi31) at a later point.
That would be congruent with the naming scheme of the workstation W7800 (70 CUs).
In term of the RDNA3 GPUs still to be released, maybe the clock speed fixes that have been talked about, are simply a way to achieve good clock speeds even on the lower quality + lower tier silicon. So, perhaps Navi33 and Navi32 will have similar maximum clocks to Navi31.
Last edited:
Associate
- Joined
- 21 Jun 2018
- Posts
- 154
To borrow a term from F1, is it fair to say AMD are going for the overcut relative to Nvidia - deliberately releasing later than the equivalent competitive card? Rather than releasing later because they couldn't release in April or earlier.
Obviously ignoring the 4090 as that comparison would be more like a formula-E car against an F1 car.
Obviously ignoring the 4090 as that comparison would be more like a formula-E car against an F1 car.
Will the FIA charge Jensen for overspend/overcharging?
Soldato
If they go first they have to have a lower price. This way they can wait for Jen to shaft his customers, then shaft theirs. That is why they have stopped going first.
Well, even if dgpus don't shift, AMD is still shifting consoles.
AMD in this logic doesn't need to react at all, they can in theory hold high prices and Nvidia won't budge until they can't shift any reasonable volume
Its funny people have deluded themselves into thinking AMD are making much money on consoles. AMD also supplied the chips for the last gen consoles. Sony and Microsoft basically bought the chips at barely above cost and AMD didn't make much money.
When Sony and Microsoft stocked up in chips (driving higher revenue even post crypto bubble), AMD's operating margin dropped significantly in 2022Q3 and their actual operating income dropped a lot.
This continued into 2022Q4.
They finally got some respite in 2023Q1 with new cards presumably having an effect. Operating income still dropped, although by not as much.
The money isn't in selling commoditised chips to companies which will have squeezed AMD into barely profitable agreements.
When Sony and Microsoft stocked up in chips (driving higher revenue even post crypto bubble), AMD's operating margin dropped significantly in 2022Q3 and their actual operating income dropped a lot.
This continued into 2022Q4.
They finally got some respite in 2023Q1 with new cards presumably having an effect. Operating income still dropped, although by not as much.
The money isn't in selling commoditised chips to companies which will have squeezed AMD into barely profitable agreements.
Last edited:
Soldato
I don't recall where...maybe it was a MLID guest but I recall hearing AMD get about $80 per PS5 sold and they don't supply the chips. They designed the APU for Sony so Sony are effectively paying $80 in Royalties per PS5 for the AMD design.
Caporegime
Its funny people have deluded themselves into thinking AMD are making much money on consoles. AMD also supplied the chips for the last gen consoles. Sony and Microsoft basically bought the chips at barely above cost and AMD didn't make much money.
When Sony and Microsoft stocked up in chips (driving higher revenue even post crypto bubble), AMD's operating margin dropped significantly in 2022Q3 and their actual operating income dropped a lot.
This continued into 2022Q4.
They finally got some respite in 2023Q1 with new cards presumably having an effect. Operating income still dropped, although by not as much.
The money isn't in selling commoditised chips to companies which will have squeezed AMD into barely profitable agreements.
AMD don't make these chips, Sony and MS do, its purely a licencing agreement.
Your analysis also doesn't match up with your own slides, it clearly says SoC revenue has increased by double digits but more than off set by gaming graphics, in other words dGPU's are eating in to gaming segment margins.
Last edited:
Associate
- Joined
- 21 Jun 2018
- Posts
- 154
That sounds like a lot IF its the same throughout the life of the console rather than starting at that for at least the first couple of years, and decline later in the consoles lifecycle.
Caporegime
They aren't getting the same level of "gaming" revenue as Nvidia from GPU's sales.
Lol, AMD don't even make their own chips. Nor do Sony or Microsoft. Such a nothing comment.
These are custom chips designed by AMD, built by TSMC.
I see again someone focussing on revenue. AMD's operating income is way down because these sales make hardly any money.
Nothing is eating into anything. AMD haven't been able to sell many GPUs for the past year. Possibly changed a bit last quarter.
Don't run a business. You'll end up selling loads of something with no margin. It's obvious the 'graphics revenue' brings home the bacon.
AMD aren't deliberately prioritising a low profit segment over a high one. They have no choice as they don't have any products.
Last edited:
It's never gonna be $80 (on top of what TSMC charge to actually make it and the cost of GDDR6?). That's a made up number. When will people stop believing anything MLID says.
Soldato
Last quarter there was barely $200 million between them:
Nvidia and AMD Gaming Revenues Are Closer Than You Think
And Nvidia shares are up 10% afterhours, due to better than expected results and a booming AI business.
It just shows how much OTT the margins Nvidia are making,and why PCMR needs to stop defending dGPU pricing.
Last edited:
Caporegime
Caporegime
Last quarter there was barely $200 million between them:
Nvidia and AMD Gaming Revenues Are Closer Than You Think
And Nvidia shares are up 10% afterhours, due to better than expected results and a booming AI business.www.tomshardware.com
It just shows how much OTT the margins Nvidia are making,and why PCMR needs to stop defending dGPU pricing.
About 65% Nvidia vs 44% AMD.
Soldato
Which proves the point about Nvidia pricing being excessive. This is why all the cretins defending dGPU pricing need to stop.
Last edited:
Caporegime
yeah.....
Caporegime
If the 3070 was the same sort of £350 as the 6700XT Nvidia would have the same sort of margins.