***OFFICIAL*** Ryzen 5000 / Zen 3 Launch Thread REVIEWS BEGIN PAGE 39

[amrnuke]

Maybe it's why buy the cow if you can get the milk for free.

However, consider another analogy:

AMD are putting cheap OEM tires on their shiny new sports car, and reviews are hammering it for subpar cornering, braking, and acceleration. Had they just put some Michelin Primacys on it, the reviews would have been much better.

This isn't about end-users tuning. This is about how AT and others test at stock, and there could be substantial benefit in mindshare from paying a dude 6 figures to do some fine-tuning.

 

[Hans de Vries]

"O processador Ryzen 5 5600X e melhor para jogos que o Intel i9-10900k"

"The Ryzen 5 5600X is beter for games as the Intel i9-10900k"

Parabéns AMD

Average over 16 games @ 1080p on an RTX 3070

 

[Panino Manino]

Strange results, when compared with that 3600XT.
Anyway, going to sleep now. Good dreams.

 

[Gideon]

A Polish site has also put up a review.

First full AMD Ryzen 5 5600X review published ahead of launch - VideoCardz.com

A Polish news site ITHardware published its AMD Ryzen 5 5600X review ahead of the official embargo. AMD Ryzen 5 5600X ITHardware are known for quality reviews, which often appear sooner than official embargoes lift. The site clearly has not signed the NDA with AMD, hence there was no reason to...

videocardz.com

The productivity benches are insane for a 6 core (just compare to 6/12 Comet Lake):

Gaming is a bit weaker than the other leak (probably down to the game choice). But comparing to where it's predecessor (3600X) the gains are ridiculous:

 

[Carfax83]

"O processador Ryzen 5 5600X e melhor para jogos que o Intel i9-10900k"

"The Ryzen 5 5600X is beter for games as the Intel i9-10900k"

Parabéns AMD

Average over 16 games @ 1080p on an RTX 3070
View attachment 33044

I remember that guy from the Zen 2 launch. I would view his "results" with extreme skepticism. I very much doubt he has the hardware he claims to, because some of his results are completely nonsensical.

 

[Carfax83]

A Polish site has also put up a review. The productivity benches are insane for a 6 core. Gaming is a bit weaker than the other leak (probably due to game choice):

First full AMD Ryzen 5 5600X review published ahead of launch - VideoCardz.com

A Polish news site ITHardware published its AMD Ryzen 5 5600X review ahead of the official embargo. AMD Ryzen 5 5600X ITHardware are known for quality reviews, which often appear sooner than official embargoes lift. The site clearly has not signed the NDA with AMD, hence there was no reason to...

videocardz.com

 

[Nereus77]

Man, I'm keen for that AnandTech review. And some forum users putting it through its paces....

 

[Kedas]

Based on this it seems that the 8 core 5800x may beat intels 10 core in video encoding.

 

[scineram]

Probably. I think they will want to launch something in 2021, if only to introduce AM5 and DDR5.

Looks like this wasn't just my theory around here, but this report seems to pour cold water on it.
https://www.trendforce.com/presscenter/news/20201103-10534.html

I really wonder what AMD plans to do next year for desktop. I doubt they want to go a calendar year without a release.

 

[Borealis7]

even with a mountain of salt, it seems the performance is there. whether its the 6C or the 8C parts, single threaded and gaming performance seems very good compared to Intel's top offerings according to the leaks.
what's going to decide it for me is the temps and power draw. if those line up just right, i might consider upgrading.
we will know later today.

 

[Carfax83]

Based on this it seems that the 8 core 5800x may beat intels 10 core in video encoding.

Any application that leverages SIMD is going to see large performance increases with Zen 3, courtesy of the extra AVX2 unit and the boost in general IPC.

You could see it in that 3ds Max benchmark I posted above. Zen 3 is 47% faster than Zen 2, and 18% faster than Intel Core i9 10850K, so 3ds Max is definitely heavily using AVX2 instructions.

Not surprising really. 3D rendering is the most parallel workload of them all I believe.

 

[Kedas]

Do we know if the die size increased? some things are removed some new added.

 

[Kedas]

I think AMD didn't want to claim/label the 5Ghz turbo boost (although it is doing it) to avoid getting to much attention to that clock instead of the results that matter.

 

[moinmoin]

Looks like this wasn't just my theory around here, but this report seems to pour cold water on it.
https://www.trendforce.com/presscenter/news/20201103-10534.html

I really wonder what AMD plans to do next year for desktop. I doubt they want to go a calendar year without a release.

AMD appears to have plenty of APU models it can launch to fill the time. Technically even consumer Renoir on desktop hasn't launched yet.

 

[JoeRambo]

You could see it in that 3ds Max benchmark I posted above. Zen 3 is 47% faster than Zen 2, and 18% faster than Intel Core i9 10850K, so 3ds Max is definitely heavily using AVX2 instructions.

I think it was Viewport testing, not rendering where AMD was already dominating due to ripping threads. And it bodes well for AMD as viewports are about strong CPU cache hierachy and IPC.

On the other hand the true test of SIMD - Linx is underforming. Makes very little sense given all other results.



97 GFlops is 4C Sandy Bridge zone, 8-10C Intel's have 500+ Gflops ( but Linx version is different, FMA enabled and latest Linpack library).

 

[Metalphase]

When can we see the official benchmarks? What precise hour today?

 

[PhonakV30]

https://twitter.com/x/status/1324315211034144768

PCMAG AMD Ryzen 9 5900X / Ryzen 7 5800X review leaked too


Page 1 / Page 2 / Conclusion / Brief

 

[Asterox]

This is very good or expected. Stock CPU setings R5 5600X vs R5 3600X much faster CPU but with lower power consumption.

 

[Rigg]

When can we see the official benchmarks? What precise hour today?

Anyone with that knowledge is probably under NDA. I believe reviews went up at 8am west coast on launch day of the Zen2 stuff. I could be mistaken though.

 

[Nereus77]

One... More... Hour...

 

[DrMrLordX]

Do we know if the die size increased? some things are removed some new added.

No clue. Hopefully some of the reviews will address that directly.

@Asterox

349W for the OCed 10850K, woof

 

[Asterox]

No clue. Hopefully some of the reviews will address that directly.

@Asterox

349W for the OCed 10850K, woof

Wait for Rocket lake 8/16, it will be even more power hog.

 

[DrMrLordX]

Wait for Rocket lake 8/16, it will be even more power hog.

Well like they used to say about the FX 9590: if you can cool it, you can clock it!

 

[tamz_msc]

On the other hand the true test of SIMD - Linx is underforming. Makes very little sense given all other results.

AFAIK LinX uses unmodified LINPACK libraries which aren't optimized for Zen. One needs to see whether Linpack Extreme works with Zen 3.

 

[Panino Manino]

This is very good or expected. Stock CPU setings R5 5600X vs R5 3600X much faster CPU but with lower power consumption.

HOW?!
AMD wasn't lying at all when they made their claims about power reduction on RDNA2 even with a much bigger and improved GPU?
What sorcery is this? Seriously, what they discovered to make such big improvement on the same nod?