Some people just love to hate AMD more than they love to love Nvidia/Intel. Unfortunately these people are never too bright in their criticisms or observations. The TUF A16 Advantage Edition presented in this review is but one of the configurations that will be provided with this chassis and cooling solution, further options are apparently available, including an RX7700S. It's safe to assume that the laptop was designed not just with a RX7600S in mind which doesn't reflect all that can be done with the additional efficiency improvements including a lighter, slimmer chassis if that was the case.

The TGP as stated in the review is a mere 95Ws, below the standard for most decent gaming laptops. Total average power consumption is 20W lower and total maximum power consumption is 40W lower than an Nvidia 3060 paired with an i7-12700 when both GPUs are limited to 95W, meaning the CPU/GPU combo delivers a similar performance while consuming less power, thus the battery life is excellent, much similar to that of a multimedia/office laptop.

RTX4000 Nvidia parts are better than the RTX3000 SKUs they seem to be replacing but in every segment, they are more expensive and the additional performance comes from frequency increases which eat back into the efficiency gain that comes with a more expensive manufacturing process. Even without the additional cost relayed on to the consumer, newer AMD CPU/GPUs deliver increased performance and efficiency.

Nice review as always. RX 7600S disappoints even more then expected... But, that's RDNA3, can't say I'm shocked.

For the guy thinking it's "impressive" - dude, read the specs. RX 7600S is the same number of shading units, same power consumption as 6700s. We can clearly see the generational gain from RDNA2 to RDNA3 - effectively ZERO (even worse for the multichip desktop design BTW)... If AMD just did die shrink, I would expect 10-15% maybe. And no, the price can't be "right", since they use more expensive process to achieve basically the same thing.

Next interesting thing will be to see how bad 780M is... The leaks suggest its better then what we see here, mostly because they clocked it higher. 680m is 2.2Ghz, while 780M will be up to 2.8Ghz. If they managed to keep the power consumption in check, it will be at least iGPU win for AMD. The rest of the RDNA3 lineup... oh boy, what a disaster...

Interestingly, it seems because of the pipeline optimization issues, the 7600S gains very little from the extra TFLOPS. But I do predict the gains in the 780M will be oversized in comparison because AMD improved lower level caches in RDNA3, which means a lot more when no VRAM. My bet is, even though the 680M needs DDR5 6400Mhz before it plateaus, the 780M will not need more, but will be 25% faster on average.

Quote from: NikoB on February 16, 2023, 13:39:58

Quote from: Russel on February 15, 2023, 23:12:202. Basic biometrics, a fingerprint scanner is a must.

By doing so, you intentionally lower the level of security and make it even easier for TNCs to collect your biometrics for the worldwide digital concentration camp. Although most idiots have already voluntarily handed over their fingerprints and facial biometrics on smartphones, so it's too late to rush about ...

Quote from: Russel on February 15, 2023, 23:12:20A high refresh rate

The problem is not the frame rate, but the terrible response time of most "60Hz" laptop panels. My monitor "60Hz" is more than 10 years old, at least 5 times faster than them by response time. And this is clearly seen in the frontal comparison.

Most of the population is above the roof of the screen with a response of 5-6ms, but the trouble is, even the fake "165Hz" panel from Asus has a monstrous response of 10ms+.

AMOLED is good for that (this is its key feature) - it has a response of 1-3ms even at "60Hz". But it has a bunch of other problems that immediately put an end to it as a screen for long-term work.

We continue to wait for microLED and hope that it will be very fast in response, no worse than AMOLED, flicker-free, with excellent viewing angles, matte (semi-matte) and with a resource of 15k+ hours up to 50% brightness drop, like IPS backlight. And of course from 300ppi+. They promise us a lot, but as with "super" batteries, the technology is still in the labs...

Yea. The response times are really abysmal in laptop screens especially..

Quote from: Russel on February 15, 2023, 23:12:202. Basic biometrics, a fingerprint scanner is a must.

By doing so, you intentionally lower the level of security and make it even easier for TNCs to collect your biometrics for the worldwide digital concentration camp. Although most idiots have already voluntarily handed over their fingerprints and facial biometrics on smartphones, so it's too late to rush about ...

Quote from: Russel on February 15, 2023, 23:12:20A high refresh rate

The problem is not the frame rate, but the terrible response time of most "60Hz" laptop panels. My monitor "60Hz" is more than 10 years old, at least 5 times faster than them by response time. And this is clearly seen in the frontal comparison.

Most of the population is above the roof of the screen with a response of 5-6ms, but the trouble is, even the fake "165Hz" panel from Asus has a monstrous response of 10ms+.

AMOLED is good for that (this is its key feature) - it has a response of 1-3ms even at "60Hz". But it has a bunch of other problems that immediately put an end to it as a screen for long-term work.

We continue to wait for microLED and hope that it will be very fast in response, no worse than AMOLED, flicker-free, with excellent viewing angles, matte (semi-matte) and with a resource of 15k+ hours up to 50% brightness drop, like IPS backlight. And of course from 300ppi+. They promise us a lot, but as with "super" batteries, the technology is still in the labs...

To me, a good 16" laptop should ideally have
1. A high refresh rate, pixel dense (near 300 ppi), ips, matte display. (16:10)
2. Basic biometrics, a fingerprint scanner is a must.
3. User replaceable RAM and an additional m2 slot.
4. At least 2 usb A ports and 2 usb c ports, 3 preferred.
5. Should be able to withstand some humid coastal conditions (laptop internals rusting has been a problem for me).
6. Good battery life and thermals.

But sadly only 3 and 5 are offered by A16.

Maybe the thermals would show improvement when you use universal x86 tuning utility?

Quote from: Florian Glaser on February 15, 2023, 09:44:19About power-measurements and battery life: We test the power draw with the power supply plugged in. Because of better / harder power savings in battery mode, the power draw without the power supply can be significantly lower, what explains the battery life.

Thanks- will know how to read the power-use data now.

I kind wished there was 10 to 20% more performance.  Maybe after drivers mature.  Still not horrible and the 7700s/7600m xt version should be in the ballpark as a 4060.  If AMD can push out a 36 cu sku they probably can compete with 4070s.  Hopefully navi 22 has something for mobile which can throw down with the 4080/90m or fill the giant gaping slot between 4070 and 4080.

Quote from: neblogai on February 15, 2023, 11:18:37He is mistaken (not without a fault of AMD too)-
(...)

So 6800S is a 11Bn transistor GPU, not 17Bn like he thought. AMD actually used 2Bn more transistors for Navi33, but the performance seems to be only a bit better. Hopefully it is more power efficient. And clocks higher on desktop..

Strange. Because 13.3 Bn is 20% more transistors than 11.1 Bn. I would have expected at least 20% performance increase then. With higher clocks maybe even 30%. But 5% just feels ridiculous.


***Oh wait... I noticed that in the "Port Royal Benchmark" it actually gained 20%. It's almost as if the card works better in Raytracing now. So maybe the majority of improvements only applies to Raytracing, not to other areas. That's maybe something worth to mention, and something that can easily be overlooked.

Quote from: neblogai on February 15, 2023, 11:18:37

Quote from: Räuber Hotz on February 15, 2023, 10:59:52

Quote from: Frans on February 15, 2023, 08:37:58Performance wise, assuming the spec on this site is accurate, 7600S managed match last gen higher tier part with much less transistor, 13.3 billion vs 17.2 billion!

So overall 7600S is actually impressive!

Ok, I get your point. *The progress was only made in efficiency*. That in itself is impressive.

The thing is, buyers don't really see that in performance. And buyers mainly buy because of more performance. So they would have preferred the same 17.2 billion transistors as before with that same efficiency. That would have been a good performance jump. So in the end, buyers will be disappointed.

He is mistaken (not without a fault of AMD too)- 'higher previous gen tier' that he mentions, the 6800S, is actually using the same die (Navi23) as 6600S and 6700S, just like 7600S-7700S are both Navi 33. So 6800S is a 11Bn transistor GPU, not 17Bn like he thought. AMD actually used 2Bn more transistors for Navi33, but the performance seems to be only a bit better (although some of that extra transistor budget was spent on an updated VCE engine). Hopefully it is more power efficient. And clocks higher on desktop..

The fault lies on this site. That is why I said assuming the spec on this site is accurate

QuoteThe AMD Radeon RX 6700S is a mobile high-end graphics card for gaming laptops. It uses the Navi 22 chip based on the new RDNA 2 architecture.

Performance wise, yes, it is not as impressive as I initially thought. Now I go straight to AMD site, apparently 6600S is basically just 4GB variant of 6700S (which is 8GB), thus yes, you can compare 7600S to 6700S directly if it is purely about comparing architecture. If they can sell 7600S at close to 6600S price, then it is a win for consumer, but I doubt they will do it.
Looking closely at the spec on AMD site, it seems current app or AMD driver have a hard time utilizing the dual issue capabilities of RDNA3, meaning for most apps, the amount of usable operation per second on RDNA3 is the same as on RDNA2 assuming CU count and clock speed are the same. Potentially there can be app that perform much better on 7600S vs 6700S, but right now it is mostly efficiency gain due to moving to 6nm. Other stuff potentially run much better on 7600S is something that uses raytracing since it is improved a lot on RDNA3.

Anyway, AMD mobile GPU naming is a mess. Apparently 6700M does use Navi22 so maybe that is where the mix up on notebookcheck site came from. 6700S Navi23, 6700M Navi22.
But ultimately it comes down to the pricing. If 7600S price is close to 6600S then it is great for consumer (You get slightly better performance for current application + extra 4GB and added features like AV1 encoding). 7700S does offer bigger performance boost vs 6700S, thus if it is comparing the same tier, then at the very least 7700S should perform 25% faster (around 25% higher TFLOPs and 28% higher memory bandwidth) vs 6700S.

Quote from: Räuber Hotz on February 15, 2023, 10:59:52

Quote from: Frans on February 15, 2023, 08:37:58Performance wise, assuming the spec on this site is accurate, 7600S managed match last gen higher tier part with much less transistor, 13.3 billion vs 17.2 billion!

So overall 7600S is actually impressive!

Ok, I get your point. *The progress was only made in efficiency*. That in itself is impressive.

The thing is, buyers don't really see that in performance. And buyers mainly buy because of more performance. So they would have preferred the same 17.2 billion transistors as before with that same efficiency. That would have been a good performance jump. So in the end, buyers will be disappointed.

He is mistaken (not without a fault of AMD too)- 'higher previous gen tier' that he mentions, the 6800S, is actually using the same die (Navi23) as 6600S and 6700S, just like 7600S-7700S are both Navi 33. So 6800S is a 11Bn transistor GPU, not 17Bn like he thought. AMD actually used 2Bn more transistors for Navi33, but the performance seems to be only a bit better (although some of that extra transistor budget was spent on an updated VCE engine). Hopefully it is more power efficient. And clocks higher on desktop..

Quote from: Russell on February 15, 2023, 03:34:11

Quote from: NikoB on February 14, 2023, 18:29:59Bad, slow (fake, as always, Asus in such series "165hz" with a monstrous response) screen with clearly visible pixels at 16", where only 4K@120-144Hz is suitable.

An obsolete Zen3+ generation processor with a shameful memory controller, which loses Raptor Lake in speed more than 1.5 times (by memory throughput and cpu).

I liked this ⬆⬆⬆ review more. LoL.

Well, part of it is cringy at best, especially the bit about memory controller. But overall- yeah, while i think the laptop looks good, and performs well enough, there seem to be many weak points- like poor cooling (=fan noise), low quality sound from speakers, etc.

I appreciate efficiency gains and they are worth something but RTX 4000 prices are a rip-off nevertheless. When price per efficiency stagnates, it feeds technological progress directly into Jensen's leather jacket.

Quote from: Frans on February 15, 2023, 08:37:58Performance wise, assuming the spec on this site is accurate, 7600S managed match last gen higher tier part with much less transistor, 13.3 billion vs 17.2 billion!

So overall 7600S is actually impressive!

Ok, I get your point. *The progress was only made in efficiency*. That in itself is impressive.

The thing is, buyers don't really see that in performance. And buyers mainly buy because of more performance. So they would have preferred the same 17.2 billion transistors as before with that same efficiency. That would have been a good performance jump. So in the end, buyers will be disappointed.

About power-measurements and battery life: We test the power draw with the power supply plugged in. Because of better / harder power savings in battery mode, the power draw without the power supply can be significantly lower, what explains the battery life.