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

[scannall]

Excuse me if it has been mentioned before, because I am sure it has, but when on earth are the actual reviews going to hit and when do they hit the stores? hmmm

Stores November 5th. Reviews likely a few days before that.

 

[misuspita]

I don't think there is another NDA date for reviews except 5 November. So, most probably, except additional info given at the Radeon launch event, we should expect reviews coming online in 05.11

 

[Kenmitch]

I don't think there is another NDA date for reviews except 5 November. So, most probably, except additional info given at the Radeon launch event, we should expect reviews coming online in 05.11

Zen 3 will most likely make a appearance during the Big Navi announcement event. It won't show a lot, but it'll probably at least show it flexing it's gaming muscles.

 

[leoneazzurro]

Wrong thread

 

[yeshua]

I'm curious why all the "leaked" Ryzen 5000 GB5 results in multi-core mode have been quite ... abysmal, i.e. there's zero increase over the Ryzen 3000 series:

5600X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5800X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5900X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5950X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

One explanation is that the multi-core test is memory speed/timings sensitive and these results are from motherboards vendors which often use the cheapest RAM available.

 

[Thunder 57]

I'm curious why all the "leaked" Ryzen 5000 GB5 results in multi-core mode have been quite ... abysmal, i.e. there's zero increase over the Ryzen 3000 series:

5600X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5800X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5900X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

5950X - Geekbench 5 CPU Search - Geekbench Browser

browser.geekbench.com

One explanation is that the multi-core test is memory speed/timings sensitive and these results are from motherboards vendors which often use the cheapest RAM available.

It's GeekBench? I'm not worried about it. We'll see the real numbers soon enough.

 

[yeshua]

It's GeekBench? I'm not worried about it. We'll see the real numbers soon enough.

There's a direct correlation between GB5 and SPEC results and GB5 is a good indicator of overall performance.

 

[inf64]

There's a direct correlation between GB5 and SPEC results and GB5 is a good indicator of overall performance.

It's just 10 days until NDA is lifted, no stress.

 

[Thunder 57]

There's a direct correlation between GB5 and SPEC results and GB5 is a good indicator of overall performance.

Oh yay Nuvia. The golden crew that has shown jack so far. I think things will look good come Nov 5th.

 

[Tuna-Fish]

I got a feeling from the presentation that a 5000 series CPUs are for enthusiasts and gamers, other people can live with the last year CPUs.

AMD still wants to sell them CPUs, there is not a substantial amount of stock of last year's CPUs, and the manufacturing resources used to make last year's CPUs are exactly the same as the modern ones, so AMD will just make lower-end models of 5000 series.

More than that, they mostly want fully functional 8-core chiplets for the server and other higher-ASP markets. Meaning, they will have large amounts of 6-core chiplets available, and based on their pricing, they really want to sell those to consumers.

 

[Mopetar]

Zen 3 will most likely make a appearance during the Big Navi announcement event. It won't show a lot, but it'll probably at least show it flexing it's gaming muscles.

It had better be there. If you're going to show off a top of the line GPU then you absolutely want the best CPU so that there's nothing holding it back.

If I were AMD I'd sent some to NVidia to use at any of their events or for doing benchmarks for any unreleased products they have coming up.

AMD still wants to sell them CPUs, there is not a substantial amount of stock of last year's CPUs, and the manufacturing resources used to make last year's CPUs are exactly the same as the modern ones, so AMD will just make lower-end models of 5000 series.

I'm just glad we're getting anything at all this year. With the limited wafers between all of the products that they're launching right now it would have made a lot of financial sense for AMD to target the server or professional market first since as good as the margins are for these high end consumer models, it's even better for Threadripper and Epyc.

But they probably want some extra time to work out any bugs for those product segments. Hopefully they have enough stock on hand so we don't wind up in a similar situation to the recent Ampere launch where there's a lot of scalpers buying them up to flip.

 

[Gideon]

Here's what optimized memory (and good cooling) can do on a 3950X:

Geekbench ST: 1700 MT: 14018

System manufacturer System Product Name - Geekbench Browser

Benchmark results for a System manufacturer System Product Name with an AMD Ryzen 9 5950X processor.

browser.geekbench.com

I don't think it's overclocked as MT score would be much better then (this is below average).

 

[Hitman928]

Here's what optimized memory (and good cooling) can do on a 3950X:

Geekbench ST: 1700 MT: 14018

System manufacturer System Product Name - Geekbench Browser

Benchmark results for a System manufacturer System Product Name with an AMD Ryzen 9 5950X processor.

browser.geekbench.com

I don't think it's overclocked as MT score would be much better then (this is below average).

It shows it reaching over 5 GHz single core. Maybe PBO is on and works better now with Zen3?

 

[Hans de Vries]

PassMark Results for the Ryzen 9 5950X

Single Threaded:
Ryzen 9 5950x: 3693 ---> 134%
Ryzen 9 3950x: 2747 ---> 100%

Multi Threaded
Ryzen 9 5950x: 45564 ---> 116%
Ryzen 9 3950x: 39279 ---> 100%

PassMark - AMD Ryzen 9 5950X - Price performance comparison

www.cpubenchmark.net

 

[Gideon]

It shows it reaching over 5 GHz single core. Maybe PBO is on and works better now with Zen3?

It might me, but unless they've significantly changed the way PBO works, it should also have much higher MT scores (but these are equal to others or even lower)

IMO it's more likely it just boosts over spec (just as Renoir does as well as 3000 series XT chips). If that's the case AMDs confidence really surprises me. They possibly have a chip that does 5 Ghz about as reliably as 3950x did 4.7 (for very brief moments in some workloads) but they chose no to market it as a 5 Ghz chip (due to 3xxx series frequency backlash).

 

[scannall]

It might me, but unless they've significantly changed the way PBO works, it should also have much higher MT scores (but these are equal to others or even lower)

IMO it's more likely it just boosts over spec (just as Renoir does as well as 3000 series XT chips). If that's the case AMDs confidence really surprises me. They possibly have a chip that does 5 Ghz about as reliably as 3950x did 4.7 (for very brief moments in some workloads) but they chose no to market it as a 5 Ghz chip (due to 3xxx series frequency backlash).

Better to under promise and over deliver.

 

[Hitman928]

It might me, but unless they've significantly changed the way PBO works, it should also have much higher MT scores (but these are equal to others or even lower)

IMO it's more likely it just boosts over spec (just as Renoir does as well as 3000 series XT chips). If that's the case AMDs confidence really surprises me. They possibly have a chip that does 5 Ghz about as reliably as 3950x did 4.7 (for very brief moments in some workloads) but they chose no to market it as a 5 Ghz chip (due to 3xxx series frequency backlash).

I didn't realize the XT and Renoir chips boosted over spec without PBO. Did any reviews document this?

 

[itsmydamnation]

I didn't realize the XT and Renoir chips boosted over spec without PBO. Did any reviews document this?

its well documented on Renoir

 

[inf64]

PassMark Results for the Ryzen 9 5950X

Single Threaded:
Ryzen 9 5950x: 3693 ---> 134%
Ryzen 9 3950x: 2747 ---> 100%

Multi Threaded
Ryzen 9 5950x: 45564 ---> 116%
Ryzen 9 3950x: 39279 ---> 100%

Great find Hans!

Looking at ST scores, 5950X @ 4.9Ghz Vs 10900K @ 5.3Ghz, it verifies that average IPC advantage Zen3 has over Skylake is ~26%.

 

[nicalandia]

Great find Hans!

Looking at ST scores, 5950X @ 4.9Ghz Vs 10900K @ 5.3Ghz, it verifies that average IPC advantage Zen3 has over Skylake is ~26%.

Yean, Zen 2 had an IPC advantage over Skylake refresh of about 6%, there is also a leaked bench where it shows that the 5600X is within 6%(short) of the 10900K and people were saying that it's Impossible, because the 10900K has a 40% core advantage. But it is very possible if we take into account the 26% IPC advantage + 8-10% SMT(when compared to Hyperthreading) and most likely was a Stock vs Stock comparison

 

[Hitman928]

its well documented on Renoir

I did a quick search for documentation, I didn't find much. I did find this though:

While our Ryzen 7 3800XT showed around 50MHz higher boost behaviour than our 3900XT sample in our Cinebench 1T test run, we spoke to AMD and there should be no particular reason why the eight-core chip is pushing higher than the twelve core. Both CPUs have the same maximum frequency rating, so this could be related to individual processor variance (which was also highlighted as unlikely by AMD), operating temperatures, or the way in which each CPU interacts with our monitoring software.

AMD Ryzen 9 3900XT and Ryzen 7 3800XT CPU Review - KitGuru

AMD is launching new, higher-frequency Ryzen 3000XT processors codenamed ‘Matisse 2’. Core to the up

www.kitguru.net

So Kitguru's 3800xt was pushing slightly above its rated max boost which AMD even seems to be surprised by. Maybe motherboards pulling some shenanigans again? I don't know, will be interesting to see what happens with the Zen3 chips and if boosting past spec is an officially supported behavior or not.

 

[Hans de Vries]

Thanks to TUM_APISAK:

https://twitter.com/i/web/status/1320922391950708736

CPU-Z benchmark:

Single Threaded:
Ryzen 9 5800X: 650 ---> 125%
Ryzen 9 3800X: 519 ---> 100%

MultiThreaded:
Ryzen 9 5800X: 6593 ---> 116%
Ryzen 9 3800X: 5665 ---> 100%



..................


Of course: remember how the CPU-Z score was adapted when Ryzen came out:

BONUS via HXL from Chiphell: https://twitter.com/9550pro/status/1320949122728296449?s=20



Single Threaded:
Ryzen 9 5950x: 690.2 ---> 132%
Ryzen 9 3950x: 524.0 ---> 100%

Multi Threaded
Ryzen 9 5950x: 13306.5 ---> 122%
Ryzen 9 3950x: 10867.0 ---> 100%

 

[Carfax83]

How does PassMark compare to Geekbench?

 

[Makaveli]

Thanks to TUM_APISAK:

https://twitter.com/i/web/status/1320922391950708736

CPU-Z benchmark:

Single Threaded:
Ryzen 9 5800X: 650 ---> 125%
Ryzen 9 3800X: 519 ---> 100%

MultiThreaded:
Ryzen 9 5800X: 6593 ---> 116%
Ryzen 9 3800X: 5665 ---> 100%

View attachment 32421

View attachment 32418 .........View attachment 32419.........View attachment 32420


Of course: remember how the CPU-Z score was adapted when Ryzen came out:

BONUS via HXL from Chiphell: https://twitter.com/9550pro/status/1320949122728296449?s=20

View attachment 32424

Single Threaded:
Ryzen 9 5950x: 690.2 ---> 132%
Ryzen 9 3950x: 524.0 ---> 100%

Multi Threaded
Ryzen 9 5950x: 13306.5 ---> 122%
Ryzen 9 3950x: 10867.0 ---> 100%

There bench was also on DDR 2400 memory with pretty crap timings.

There 3800X does 519

My 3800X with PBO Fmax on with the newest Bios at DDR4 3200 with a Ryzen Dram fast profile applied.

Single (5800X) 650 vs (3800X) 541.5 = 16%

Multi (5800X) 6593 vs (3800X) 5599.4 = 15%

Looking forward to this upgrade.



The last bios update has provided a nice boost in scores for me and I believe its because I'm hitting higher boost clocks now. Prior to this bios upgrade max boost for me was 4490

 

[Panino Manino]

Tomorrow will be a "busy" day.
Not only there's RDNA 2 event, looks like we'll see some early reviews of Zen 3 from youtubers.