AMD Radeon RX 7600

[Taisho]

• PCIe x8 interface

Why is this a negative?
If you run a 4090 at PCI 3.0x16\PCI 4.0x8 it only loses about 2 fps, so why should a very slow GPU need more lanes?
If it means they can save some cost and hopefully move said saving to customer, that makes it a plus.

I'm not saying it should be in the positive section but don't understand the negativity.

It hits 1% lows much more, and 0.1% lows much much more. Especially since you are talking about an average. If I average Cyberpunk RT and CS: GO FPS then I get a pretty playable experience in both.

A valid warning for PCI 3.0 users.

 

[AusWolf]

Maybe AMD wanted to only have one monolithic die in the Navi 33.

Yeah, but what about the defects? Yields can't be that good, surely.

 

[dgianstefani]

Yeah, but what about the defects?

I'm not an AMD engineer. Maybe defect rate is lower for older nodes.

 

[Rabit]

RX 7600 makes RX 6700 10GB and RX 6700 XT 12GB more attractive

 

[Chaitanya]

This one isn't highly recommended?

Which means you should actually get it if it fit your budget.

 

[btk2k2]

Why is it the spec 18gbps ram when it seems the MBA part and others are using 20gbps chips? Seems like a daft decision given more memory bandwidth would probably help quite a bit.

 

[Bomby569]

could this seriously not be a strategy from both nvidia and amd to keep moving stock of 3000 and 6000 gen cards that seem to never end? because it makes no sense to go with the new gen cards.
When stock finally ends they will drop prices of new gen like a rock... or just milk us idk

 

[TumbleGeorge]

Perf/watt is terrible. Literally equal to older rDNA2/1.

 

[rattlehead99]

Use more power than 4060ti, yet 30% slower. Yup, it's RDOA3 alright

It's on 6nm and not 5nm let alone 4nm, so it's normal to consume more power. The thing is that using 6nm, this card should be 210-220$ tops.

 

[rattlehead99]

Perf/watt is terrible. Literally equal to older rDNA2/1.

Because it's using 6nm and not 5nm or 4nm process node.

 

[Pumper]

Idle power - 2W.
Video playback - higher than a 4090.

Will AMD ever fix that shit?

 

[geniekid]

The cons list high power consumption for multi-monitor and video playback, but the benchmark data seems to suggest AMD fixed this. Am I missing something?

 

[Suspecto]

Terrible card for a terrible price.

 

[dirtyferret]

It's kind of weird for this generation of cards AMD has released its flagship and now entry level card and nothing in the mid tiers while Nvidia has already released the RTX 4060ti, 4070, 4070ti

 

[Denver]

Best performance per dollar so far, just needs 2-4Gb of extra memory to be perfect

 

[Mahboi]

Conclusion: "AMD's MSRP for the Radeon RX 7600 is $270, which is considerably lower than the $400 that NVIDIA wants for the RTX 4060 Ti, but there's also a 25% performance difference."
This is somewhat poorly worded. Is it 25% less performance than a 4060 Ti, or the 4060 Ti has 25% more performance?
It's the difference between:
100 - 125
and
100 - 133

Could you clarify this please?

 

[Pumper]

Conclusion: "AMD's MSRP for the Radeon RX 7600 is $270, which is considerably lower than the $400 that NVIDIA wants for the RTX 4060 Ti, but there's also a 25% performance difference."
This is somewhat poorly worded. Is it 25% less performance than a 4060 Ti, or the 4060 Ti has 25% more performance?
It's the difference between:
100 - 125
and
100 - 133

Could you clarify this please?

4060Ti is 25% faster at 1080p, 30% faster at 1440p, and 40% faster at 4K.

No matter what resolution, the 7600 is a turd, but is better value than the 4060Ti.

 

[dirtyferret]

Conclusion: "AMD's MSRP for the Radeon RX 7600 is $270, which is considerably lower than the $400 that NVIDIA wants for the RTX 4060 Ti, but there's also a 25% performance difference."
This is somewhat poorly worded. Is it 25% less performance than a 4060 Ti, or the 4060 Ti has 25% more performance?
It's the difference between:
100 - 125
and
100 - 133

Could you clarify this please?

based off of w1zzards and other game suites the 4060ti offers an additional 25% performance from the RX 7600 but the price difference is greater as the RTX4060ti is 48% more than the RX 7600; $270 * 1.48 = 400

 

[mb194dc]

It's trash.

The truth is that you could get a 5600xt in 2020 for 270. All the ones with Micron ram are unlockable and will get near a 5700xt at 1080p.

We've gone pretty much nowhere for 3 years in the sub 300 price range.

 

[catulitechup]

If intel play good can put in troubles amd and nvidia, only need add to A750 8gb 200us 256bit bus this:

A770 16gb 256bit bus to 300us (this could give rx 6700xt 12gb closer to 250us and rx 6700 10gb in someplace around 230us)

And after this, well begins the game!

 

[evolucion8]

What this proves for the first time since RDNA3 launched, is that RDNA3 is worthless.

We have a near perfect comparison with the 6650XT - the only significant difference being the architecture - and it achieves precisely nothing.

RDNA 2 was very powerful as its occupancy in resources was top notch, much higher than Ampere, but RDNA 3 just went back to what nVidia had been doing in the past which is trying to use software to increase resource utilization similar to what nVidia started to do back then with Kepler and then Ampere which relies too much on a stupid compiler to figure out the best way to dual issue instructions and it if fails, then it wont be any better than previous generations that didn't required that much of a hack. Instead of increasing the amount of cores linearly, they just partition the existing ones ala Ampere style. Reminds me of AMD's Bulldozer as well with the so called 8 core CPU lol

 

[W1zzard]

• PCIe x8 interface

Why is this a negative?
If you run a 4090 at PCI 3.0x16\PCI 4.0x8 it only loses about 2 fps, so why should a very slow GPU need more lanes?
If it means they can save some cost and hopefully move said saving to customer, that makes it a plus.

I'm not saying it should be in the positive section but don't understand the negativity.

You are absolutely right. It's a total non-issue for nearly all systems (will test to verify after Computex), but there's some people with older Gen 3 systems out there, and I rather have them scratch their head "what does this mean?", than buy the card and then scratch their head.

 

[W1zzard]

@W1zzard, due to the controversy over award designations, why not use the more objective data of your very nice X / Y analysis charts rather than the subjective choice/recommended awards? Here are some examples:

Highest performance award - #1 on relative performance per frame charts
High performance award - in the top 2-5
You could use a zig zag arrow going up and to the right as the award graphic.

Mother Nature award - #1 on efficiency charts
Good for the planet award - in the top 2-5
You could use a green leaf or a globe as the award graphic.

Best value award - #1 on performance per dollar charts
Great value award - in the top 2-5
You could use a coin as the award graphic.

Most innovative AIB award - #1 on noise, temp, cooler charts for a particular SKU (ie. 4090)
Innovative AIB award - in the top 2-5
You could use a fan as the award graphic.

I know it seems simple and a rehash of the data in another way but it would be harder for readers to argue the objective data found in your tests. The above would also limit the number of awards to the most standout products. Of course a single product could get multiple awards in this way. I also suggest placing the award icons at the top of the Value & Conclusions page. That would encourage readers to go back and focus on those awarded metrics and maybe read the conclusions rather than just stop reading at the Pros/Cons section.

Good ideas there, I think they need to be split by segment though

 

[mechtech]

Well I think I remember the 6600 MSRP was $330. So this has a new media engine, more performance 2 more years of inflation and MSRP is $270. I don’t see anything to complain about. That’s about $365 CAD which is what some 6600 are going for here. Seems like a win to consumers to me.

 

[W1zzard]

The cons list high power consumption for multi-monitor and video playback, but the benchmark data seems to suggest AMD fixed this. Am I missing something?

What makes you think it's fixed? 2 W -> 18 W ... 900% increase for connecting a 2nd monitor? 2 W -> 27 W for video?