AMD Radeon RX 7600 Early Sample Offers RX 6750 XT Performance at 175W: Rumor

[btarunr]

AMD is expected to debut its performance-segment Radeon RX 7600 RDNA3 graphics card in May-June 2023, with board partners expected to show off their custom-design cards in the 2023 Computex (June). Moore's Law is Dead reports that they've spoken to a source with access to an early graphics card sample running the 5 nm "Navi 33" silicon that powers the RX 7600. This card, with development drivers (which are sure to be riddled with performance limiters); offers a 11% performance uplift over the Radeon RX 6650 XT, and a gaming power draw of 175 W (the RX 6650 XT pulls around 185-190 W).

This is still an early sample running development drivers, but a 11% performance boost puts it in the league of the Radeon RX 6700 XT. Should a production RX 7600 with launch-day drivers put on another 5-7% performance over this, the RX 7600 could end up with performance roughly matching the RX 6750 XT (a slim performance lead over the RTX 3070 in 1080p gaming). Should its power draw also hold, one can expect custom-design graphics cards to ship with single 8-pin PCIe power connectors. A couple of nifty specs of the RX 7600 also leaked out in the MLID report: Firstly, that 8 GB will remain the standard memory size for the RX 7600, as it is for the current RX 6650 XT. Secondly, the RX 7600 engine clock is reported to boost "above" 2.60 GHz.



View at TechPowerUp Main Site | Source

 

[Chaitanya]

At what price though, anything above $300 not worth it.

Edit: also I hope there are more 2 slot cards and not the nonsense 2.5 slot cards seen currently.

 

[N/A]

apples to apples, stop comparin it to team green, this is a RX 6800 with 8GB what a shameful display,

 

[Ruru]

apples to apples, stop comparin it to team green, this is a RX 6800 with 8GB what a shameful display,

16GB would be too much and I don't know would 12GB be possible with a 128-bit bus. Nvidia did it with mixed density chips 550 Ti, 650 Ti Boost, 660 & 660 Ti as they had 1 or 2GB & 192-bit bus.

 

[wolf]

I am prepared to be underwhelmed, but the real win here will be price and availability, if they're abundant and cheap, it could be a big success.

Unfortunately if the *crappy source* rumour is to be believed, that efficiency still leaves a lot to be desired.

 

[AusWolf]

Why do we still have articles reporting on rumours spread by MLID? That channel has been confirmed to be a hoax and fake on multiple occasions.

 

[tvshacker]

Why do we still have articles reporting on rumours spread by MLID? That channel has been confirmed to be a hoax and fake on multiple occasions.

The most recent I can think of was that the 4070 would be 750$ and performed within 10% of the 4070TI.

 

[Broken Processor]

Yeah need's more ram. As usual AMD fails to read the room and stomp's on the opportunity as they walk on by.

 

[konga]

The article headline here is a joke. The leak only indicates an 11% improvement over the 6650 XT, and any additional theoretical gains from driver improvements are a hallucination on the article author's part. It's okay to theorize, but to present that theory as a fact in the headline is extremely bad practice. Haven't you learned your lesson the first time you overhyped RDNA3?

 

[ToTTenTranz]

apples to apples, stop comparin it to team green, this is a RX 6800 with 8GB what a shameful display,

This is completely wrong.
The RX7600 should be a RX6600 that they managed to make 16% smaller on the cost-focused N6 process that provides no density or performance advantage over the original.

The fact that the cut-down Navi 33 is performing well above the fully enabled Navi 23 at a lower TDP is pretty great on its own. This is a low-cost dGPU first and foremost.

 

[Vayra86]

8GB at 6750XT perf?

I'd run to the store now to get RDNA2 then instead.
But... rumor be rumor. Let's see where it truly lands and where it truly prices itself

However I don't quite get anyone who's pissing over general RDNA3 efficiency, its damn close to Ada really. And it does that while having much more VRAM, which does cost power.

I mean look... its single digit at most and you'd do well to disregard the 4080 because its quite honestly the shittiest offer in the Ada stack, plus 4080 and up are hitting some small CPU hit/bottleneck.

 

[ZoneDymo]

Why do we still have articles reporting on rumours spread by MLID? That channel has been confirmed to be a hoax and fake on multiple occasions.

This its absolutely embarrasing...... if I could vote down an article I would.

 

[Denver]

Well, the source of information is not the best, but we already know more or less what to expect since this chip is already being used in laptops.

Being produced in 6nm, the chances of having a price at least equal to the previous generation seem viable to me. My bet: About 10-15% more performance, more vram and better efficiency at $250.

 

[napata]

8GB at 6750XT perf?

I'd run to the store now to get RDNA2 then instead.
But... rumor be rumor. Let's see where it truly lands and where it truly prices itself

However I don't quite get anyone who's pissing over general RDNA3 efficiency, its damn close to Ada really. And it does that while having much more VRAM, which does cost power.

I mean look... its single digit at most and you'd do well to disregard the 4080 because its quite honestly the shittiest offer in the Ada stack, plus 4080 and up are hitting some small CPU hit/bottleneck.

View attachment 293187

Effiency is a complicated topic and can't be reduced to a single game. If you use Doom Eternal for example then it's suddenly a 20-30% difference. 4080 only uses 280-290W in it while being faster than a 7900XTX IIRC so it's very much an example in favor of ADA, but it's a good example to show how much a game can swing the results in favor of either Nvidia or AMD.

Although you're that guy who mistakenly believes GPUs always hit max power consumption, right? That reminds me that I should deliver you some screenshots of synthetic GPU workloads, like TSE or Port Royal, where my 4090 runs below 400W.

Also once you cap games with a framelimit RDNA3 loses a ton of efficiency. That might not matter to you if you never run a cap but a lot of people run games with frame caps.

 

[Mahboi]

Why do we still have articles reporting on rumours spread by MLID? That channel has been confirmed to be a hoax and fake on multiple occasions.

Because many occasions he's wrong, and many he's right?

This is completely wrong.
The RX7600 should be a RX6600 that they managed to make 16% smaller on the cost-focused N6 process that provides no density or performance advantage over the original.

The fact that the cut-down Navi 33 is performing well above the fully enabled Navi 23 at a lower TDP is pretty great on its own. This is a low-cost dGPU first and foremost.

I agree, but the fact is also that it comes with a limited amount of VRAM at a time where VRAM is extremely sought after.
It's going to be a cheaper and more powerful, but ultimately pointless, addition to the pool of GPUs on offer.

The only thing I can hope from it is that it's priced equal to an rx 6600. If you put a really low price with insufficient VRAM amounts, you get a GPU that makes little sense, but isn't outright terrible. Not that I expect AMD to shit out something as egregious as a 3060Ti/3070s at any point.

 

[sLowEnd]

That's...not very impressive, both performance-wise and power consumption-wise.

 

[Mahboi]

Also once you cap games with a framelimit RDNA3 loses a ton of efficiency. That might not matter to you if you never run a cap but a lot of people run games with frame caps.

Rocking an rx 7900 xt right now, I can say that the biggest sign if immaturity of the card is its power draw. It constantly stays above what's necessary, or jitters in power draw and has a very hard time properly downpowering.

For example I can run Overwatch 2 on Ultra (also Epic, but frankly I saw no visual difference so why bother) with Radeon Chill on, and comfortably get 140fps: card eats 150W.
I then start a basic Youtube video/Twitch and just idle away: card eats 70W.
Downpowering back to the best possible idle power draw (29W) is a test of patience because the card will literally dance around 40-55W for several minutes before entirely downpowering to its real minimum. Even when all you do is idle away or just type on these forums:

All of that to say that the card clearly has the chops for a good efficiency, but is woefully immature in its power usage.
I'm eagerly waiting for the next driver update to see what actually gets done, and I'm expecting to gain a few watts on each new driver or so in the next ~1 year.

That's...not very impressive, both performance-wise and power consumption-wise.

To be fair, between the huge upcoming APUs and the value of cards that have less than 12Go of VRAM, I'm really expecting low end enthusiast to be squeezed away in the next years. No need to put big efforts when you can just shove 10-40 RDNA3 CUs on your CPU...

 

[tfdsaf]

If the performance numbers are true than this is another DOA card! Hey I'm saying this for both gpu makers, what a shock, we can be objective and not hold ANY company in our hearth!

For this to actually be good it would need to be at least 6-7% faster as the report suggested and not cost a penny over $300 and also be available with 16GB of vram for $40 more!

This needs to be on par with the RX 6800, draw less power and cost $300 or less to be actually good value! If AMD are smart they will go with this strategy and offer a 16GB model as well for $40 or $50 more!

 

[Chris34]

What about FSR 3? Or did that feature became a meme already?

 

[Nostras]

If the performance numbers are true than this is another DOA card! Hey I'm saying this for both gpu makers, what a shock, we can be objective and not hold ANY company in our hearth!

For this to actually be good it would need to be at least 6-7% faster as the report suggested and not cost a penny over $300 and also be available with 16GB of vram for $40 more!

This needs to be on par with the RX 6800, draw less power and cost $300 or less to be actually good value! If AMD are smart they will go with this strategy and offer a 16GB model as well for $40 or $50 more!

No such thing as a bad product, only a bad price.
Well, yes, obviously it's a bit more nuanced than that, but if the card is like 300$ it's offering a serious discount over an 6750XT.
The reduction in VRAM is a bit disappointing, but for a 300$ card I think it's an acceptable compromise.

 

[antuk15]

The most recent I can think of was that the 4070 would be 750$ and performed within 10% of the 4070TI.

10%? Techpowerups review has it at 20%+

 

[Nostras]

10%? Techpowerups review has it at 20%+

Quite sure shacker is commenting that ML was wrong. They stated price and performance would be very close where in reality price and performance were significantly lower.

 

[AusWolf]

Because many occasions he's wrong, and many he's right?

You can't afford to be wrong when you consider yourself newsworthy. Anything coming from him/them is hot air, imo.

 

[Denver]

If the performance numbers are true than this is another DOA card! Hey I'm saying this for both gpu makers, what a shock, we can be objective and not hold ANY company in our hearth!

For this to actually be good it would need to be at least 6-7% faster as the report suggested and not cost a penny over $300 and also be available with 16GB of vram for $40 more!

This needs to be on par with the RX 6800, draw less power and cost $300 or less to be actually good value! If AMD are smart they will go with this strategy and offer a 16GB model as well for $40 or $50 more!

You dream too far..
In the real world, 10-15% performance above 6650x at a cost of $250 will sell very well.

 

[docnorth]

At what price though, anything above $300 not worth it.

Edit: also I hope there are more 2 slot cards and not the nonsense 2.5 slot cards seen currently.

2 slot and dual fan, not triple.