AMD Planning September Launch for Radeon RX 7800 series and RX 7700 series

[btarunr]

AMD is planning to plug the Atlantic gap between its mainstream Radeon RX 7600 and enthusiast-class RX 7900 XT with the RX 7800 series and RX 7700 series, with either an announcement or teaser planned for 2023 Gamescom, which is scheduled for August. There could be up to four new graphics card SKUs announced, with their product launches spread across Q3 and Q4 2023. The "Navi 32" MCM is expected to power at least three of these SKUs, while it was recently rumored that AMD could design a new GPU that has the GCD of the "Navi 31" on the package of "Navi 32" with its four MCDs, to end up with a higher CU count than what the "Navi 32" can offer.

The "Navi 32" GPU is an MCM, just like the "Navi 31" powering the RX 7900 series. It is rumored to feature a 5 nm GCD (graphics compute die) with 60 RDNA3 compute units, which work out to 3,840 stream processors, 120 AI accelerators, 60 Ray Accelerators, 240 TMUs, and possibly 128 ROPs. The four 6 nm MCDs give it 64 MB of Infinity Cache, and a 256-bit wide GDDR6 memory interface. Assuming the RX 7800 XT uses the unnamed new MCM with the GCD of the "Navi 31" that has a CU count somewhere between 60 and 72, a maxed-out "Navi 32" could power the RX 7800, while its cut-down variants power the RX 7700 XT and RX 7700.



View at TechPowerUp Main Site | Source

 

[gmn17]

Nice AMD

 

[Firedrops]

up to four new graphics card SKUs

looking forward to the 7800 XT, 7800 XTX, 7800 XTXT, 7800 XTXTX

 

[vbq7qK68eyYAH4iR]

looking forward to the 7800 XT, 7800 XTX, 7800 XTXT, 7800 XTXTX

Come on. We all know you add the X on each side of the T. So it's the 7800 XXTXX.

 

[Daven]

There might also be a 7950XT and a 7950XTX by the end of the year.

 

[sethmatrix7]

Too late I already got a used 3080 for $400

 

[Chrispy_]

Finally, confirmation of Navi32
\o/

I was starting to worry that AMD had cold feet with GPU-based MCM, after the issues with Navi31 not clocking as intended - the unexpected artifacting that couldn't be fixed in software requiring them to dial it back by 10-15 from the originally-intended performance and clocks.

Do we know if this coming Navi32 is the cited respin with hardware fixes to tackle the clocking artifacts, or is this just the original silicon run that's limited to ~2.5GHz?

 

[ZoneDymo]

Sources: Moore's law is dead.... Deep sigh

 

[Chaitanya]

Hopefully they start with sensible prices instead of stupid pricing and then discounting them heavily few months of rotting on shevles at retailers.

 

[N/A]

back to school was it.

 

[TheDeeGee]

looking forward to the 7800 XT, 7800 XTX, 7800 XTXT, 7800 XTXTX

xXx_YahooGirl_xXx Edition

 

[Chrispy_]

Hopefully they start with sensible prices instead of stupid pricing and then discounting them heavily few months of rotting on shevles at retailers.

If the RX 7600 is anything to go by, AMD at least understands that stuff will rot on shelves if it's not price-competitive with other options in the same performance segment.

The official $279 MSRP of the RX 7600 wasn't awful against the selling price of the outgoing 6650XT and 6600XT, but the street pricing drop to $249 makes it a compelling buy among the 8GB competition and heavily-discounted Ampere stock that's still selling.

AMD don't need to compete with the 40-series on price, they know they need to compete with the entire open GPU market.

 

[ymdhis]

Let's hope the reason it took so long is because they took it for a respin to make sure all issues they had with the 7900 is fixed.

 

[Daven]

Hopefully they start with sensible prices instead of stupid pricing and then discounting them heavily few months of rotting on shevles at retailers.

They are constrained between the $269 of the 7600 and the $849 of the 7900XT for like 4-5 SKUs so we are talking every $80.

 

[AnotherReader]

Do we know if this coming Navi32 is the cited respin with hardware fixes to tackle the clocking artifacts, or is this just the original silicon run that's limited to ~2.5GHz?

While they have had enough time to do a respin for higher clocks, the rumoured use of a cut down Navi 31 for the 7800 XT suggests that clocks haven't improved. A 60 CU RDNA 3 with 30% higher clocks than the RX 6800 would be as fast as a 4070 Ti in rasterized workloads and then they wouldn't have needed to use the Navi 31 for the 7800 XT.

 

[Zareek]

Do we know if this coming Navi32 is the cited respin with hardware fixes to tackle the clocking artifacts, or is this just the original silicon run that's limited to ~2.5GHz?

I really hope so, if it is the original silicon design, then why the hell is it so late? Hopefully they've solved all the nagging issues as well, and it will have good drivers from day one. Well, maybe good launch drivers is asking too much, it would be a nice change. It's a bit tedious that it takes them a few months more to work out all the minor driver issues every single launch.

 

[rv8000]

Finally, confirmation of Navi32
\o/

I was starting to worry that AMD had cold feet with GPU-based MCM, after the issues with Navi31 not clocking as intended - the unexpected artifacting that couldn't be fixed in software requiring them to dial it back by 10-15 from the originally-intended performance and clocks.

Do we know if this coming Navi32 is the cited respin with hardware fixes to tackle the clocking artifacts, or is this just the original silicon run that's limited to ~2.5GHz?

What’s this imaginary 2.5ghz limit?

Average clock from reviews is already 300mhz over that? More unconfirmed bs about bad silicon.

 

[Chry]

Finally. a 7700+7700 PC will be possible. Now to clock the DDR5 @7700mhz...

 

[Dan.G]

It's nice and all, but... not really enthusiastic about new hardware.
If you already have a DX 12 Ultimate GPU, there aren't a lot of new features added to get excited about (not even DLSS 3 seems that interesting, to be fair).
It's possible we're looking at "diminishing returns" from the tech world: there's not much left to innovate.

 

[AnotherReader]

What’s this imaginary 2.5ghz limit?

View attachment 305190
Average clock from reviews is already 300mhz over that? More unconfirmed bs about bad silicon.

AMD claimed a 30% increase in clock speeds at the same power. The reference 6950 XT uses about 15 W less power than the reference 7900 XTX and is only 8% behind in average clock speeds in CyberPunk 2077. If we restrict ourselves to the 6900 XT due to the higher off chip power draw for the 7900 XTX, Computerbase reported a 14% increase in clock speeds. It seems rather obvious that clock speed targets were missed by about 15 percent.

 

[rv8000]

AMD claimed a 30% increase in clock speeds at the same power. The reference 6950 XT uses about 15 W less power than the reference 7900 XTX and is only 8% behind in average clock speeds in CyberPunk 2077. If we restrict ourselves to the 6900 XT due to the higher off chip power draw for the 7900 XTX, Computerbase reported a 14% increase in clock speeds. It seems rather obvious that clock speed targets were missed by about 15 percent.

View attachment 305191

I don’t know how closely anyone watches the owners threads over on OCN, but with the aqua bio these cards easily exceed 3ghz.

If you want argue they missed their power target sure. Chalk it up to semantics, but from what I’ve seen and personal experience 7900xt/XTX can regularly hit the 3ghz mark. There definitely isn’t a “2.5ghz limit” cuz of bad silicon.

 

[AnotherReader]

I don’t know how closely anyone watches the owners threads over on OCN, but with the aqua bio these cards easily exceed 3ghz.

If you want argue they missed their power target sure. Chalk it up to semantics, but from what I’ve seen and personal experience 7900xt/XTX can regularly hit the 3ghz mark. There definitely isn’t a “2.5ghz limit” cuz of bad silicon.

I am not saying that there's a 2.5 GHz limit. I'm just reporting the average clock speeds at the same power target. RDNA3, like RDNA2, before it, is an excellent overclocker. Recall that some RDNA2 SKUs also hit clock speeds of over 2800 MHz when overclocked. Even the maximum overclock figures aren't anywhere close to 30% above the figures for RDNA 2.

 

[Zunexxx]

While they have had enough time to do a respin for higher clocks, the rumoured use of a cut down Navi 31 for the 7800 XT suggests that clocks haven't improved. A 60 CU RDNA 3 with 30% higher clocks than the RX 6800 would be as fast as a 4070 Ti in rasterized workloads and then they wouldn't have needed to use the Navi 31 for the 7800 XT.

30% clock speed increase doesn't mean 30% performance increase, there's more to it, at most it will be 20-25%. There are tests already "emulating" the 7800xt out there, we are seeing a moderate boost of 8-10% over the 6800xt, which also seems to be about 20% faster than the 6800. Thou that was emulated using a wx7800 (navi31).

 

[rv8000]

I am not saying that there's a 2.5 GHz limit. I'm just reporting the average clock speeds at the same power target. RDNA3, like RDNA2, before it, is an excellent overclocker. Recall that some RDNA2 SKUs also hit clock speeds of over 2800 MHz when overclocked. Even the maximum overclock figures aren't anywhere close to 30% above the figures for RDNA 2.

You were not, nor did I originally reply to you. Chrispy stated it, as I quoted, and that’s just completely not true.

There’s air and water cooled cards between 3000-3100 mhz on UL leaderboards, which is just slightly under to over 30% in average raw clock speed vs the 6900XT. There are users showing ~3100 in cyberpunk 4K with RT at 411w. There are also other records pushing the card on exotic cooling above 3.3ghz+. Sure it uses a lot of power, new design, and worse node than ada, but there is certainly no 2.5ghz clock speed limit.

If you go through most of the reviews on TPU, on average for clock speeds the 7900 XTX has a 20% average lead over the 6900XT. AMD and Nvidia like to make loose claims all the time, never trust them.

 

[Zunexxx]

I don’t know how closely anyone watches the owners threads over on OCN, but with the aqua bio these cards easily exceed 3ghz.

If you want argue they missed their power target sure. Chalk it up to semantics, but from what I’ve seen and personal experience 7900xt/XTX can regularly hit the 3ghz mark. There definitely isn’t a “2.5ghz limit” cuz of bad silicon.

Doesn't hit that frequency in almost all games. Just for reference, I know my 4090 hits 3045-3060 under extreme low loads but it usually stays at 2865-2910 in most games. It should be same for AMD afaik. If I increase the PL, I am pretty sure I can hit higher frequency as I am still at stock now. 7900xtx hitting 3ghz seems totally plausible, but I don't think it stays there for most games if u know what I mean.