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

[mmaenpaa]

Mitä se sinulle kuuluu?

Tuli vain sellainen tunne

 

[ondma]

That is because TPU are testing with GPU limited settings so the spread is reduced.

You mean not at 360p like Anand used?

 

[ondma]

As long as someone is happy using 8700K over 10700K or 10900K there no need to look at Zen 3 either.


The 8700k is 10% slower than 5800X in the 720p aggregate gaming bench.

Well, i dont game at 720p, so.......

Edit: Strange (not really) that low res gaming was so strongly denigrated (niche within a niche comes to mind from one poster) when Intel led more at low res, but now is somehow becoming the holy grail since that appears to be where AMD has the lead.

 

[Zucker2k]

We all know that but certain people were pushing very hard the super low res. testing as it *should* show which uarchitecture has more potential and brings less CPU bottleneck. Now the tables are turned and Zen3 pummels intel in the ground in 720p and lower - just check AT review and 384p numbers, it's sad

Those super low resolutions are basically a cache and memory benchmark, both of which were stacked in favor of Zen 3; the cache being acceptable because it's inherent to the uarch, but the memory not so much. Any consumer who buys Intel can overclock them and run the memory at 3600MHZ+ for instant and significant performance. That is what most real world gamers do. This attempt on your part to make a whole bunch of noise over resolutions no one plays at is hilarious. The shoe is on the other foot, so all of a sudden, 720p is the new 1440p.

 

[Veradun]

Those super low resolutions are basically a cache and memory benchmark, both of which were stacked in favor of Zen 2; the cache being acceptable because it's inherent to the uarch, but the memory not so much. Any consumer who buys Intel can overclock them and run the memory at 3600MHZ+ for instant and significant performance. That is what most real world gamers do. This attempt on your part to make a whole bunch of noise over resolutions no one plays at is hilarious. The shoe is on the other foot, so all of a sudden, 720p is the new 1440p.

Real world gamers buy stuff that runs at jedec specs, plug the thing to a 1080p monitor and play.

 

[lightmanek]

I just had a chance to run first few benches on my 5900X and in my old CB11.5 database I've found my fully tweaked 2500k at 4.73GHz and 1600MHz mem scoring ST of 1.91 pts where stock 5900X with 3666MHz mem scored 3.11 pts.
Not exactly clock for clock as Ryzen boosts to 4.9GHz but close enough to realise how much faster core has got in the last 9.5 years.

 

[inf64]

The butthurt level is enormous, it will be a tough year for some members on this forum

 

[jpiniero]

Real world gamers buy stuff that runs at jedec specs, plug the thing to a 1080p monitor and play.

I doubt that, given that XMP is Intel's own standard.

 

[Veradun]

I doubt that, given that XMP is Intel's own standard.

And RAM is shipped in systems that run at jedec specs for the ultra vast majority.

And RAM installed by Average Joe is run by the motherboard at jedec specs.

People here is not representative of the real world.

And even here a lot of people runs system at stock.

 

[GaiaHunter]

I want a 8 core zen3 but I'll wait for either a 5700X or a price drop.

 

[uzzi38]

Those super low resolutions are basically a cache and memory benchmark, both of which were stacked in favor of Zen 3; the cache being acceptable because it's inherent to the uarch, but the memory not so much. Any consumer who buys Intel can overclock them and run the memory at 3600MHZ+ for instant and significant performance. That is what most real world gamers do. This attempt on your part to make a whole bunch of noise over resolutions no one plays at is hilarious. The shoe is on the other foot, so all of a sudden, 720p is the new 1440p.

Core overclocking on Intel systems provided minimal benefits now as well. It only really helps with lower end dies - like the 10600K, and still doesn't allow them to beat the 10900K stock with the same memory.

You mention running the memory at 3600MHz+ on Intel, but you do realise that's also possible on Zen 3 systems? Furthermore, that is EXACTLY what Gamers Nexus and LTT have both done for both Intel and AMD systems - GN uses tuned memory at specific timings they standardise across all systems, and their the both of their Zen 3 results are amongst the most positive. Both also tested at 1080p high or ultra settings as far as I remember, so the "720p isn't real world" (which mind you, I agree with) don"t matter.

No matter how you may wish to cut it, Intel have lost the gaming crown for the time being. Not sure how you can argue otherwise at this point.

 

[jpiniero]

And RAM is shipped in systems that run at jedec specs for the ultra vast majority.

OEMs, yes. They will stick with JEDEC. But if you are talking DIY I would expect XMP to be turned on. It's not a big deal and pretty much guaranteed to work unlike overclocking.

 

[Veradun]

OEMs, yes. They will stick with JEDEC. But if you are talking DIY I would expect XMP to be turned on. It's not a big deal and pretty much guaranteed to work unlike overclocking.

As I said, Average Joe doesn't even know what a bios is.

 

[GaiaHunter]

Never understood the super low resolutions testing - bottlenecks just get shifted in a way that might not reflect future performance at higher resolutions and/or reflect how GPU architecture and game engine coding evolve.

I find much more interesting that zen3 perform generally better on newer titles.

 

[Zucker2k]

Core overclocking on Intel systems provided minimal benefits now as well. It only really helps with lower end dies - like the 10600K, and still doesn't allow them to beat the 10900K stock with the same memory.

You mention running the memory at 3600MHz+ on Intel, but you do realise that's also possible on Zen 3 systems? Furthermore, that is EXACTLY what Gamers Nexus and LTT have both done for both Intel and AMD systems - GN uses tuned memory at specific timings they standardise across all systems, and their the both of their Zen 3 results are amongst the most positive. Both also tested at 1080p high or ultra settings as far as I remember, so the "720p isn't real world" (which mind you, I agree with) don"t matter.

No matter how you may wish to cut it, Intel have lost the gaming crown for the time being. Not sure how you can argue otherwise at this point.

The 720p results he showed both have the Zen 3 systems running higher memory speeds. That's my point. Run all systems at the same speed and run the 720p tests again and watch the Intel systems do significantly better. 384p? Assuming we should even take that seriously, Zen 3 could probably fit a whole lot of the data in cache, and those that don't will go to a faster memory? TPU tested all systems with same memory speed in the 720p test, albeit with a 2080Ti, and the results were very different.

 

[jpiniero]

As I said, Average Joe doesn't even know what a bios is.

Those people aren't doing DIY. They are doing prebuilts.

 

[coercitiv]

Strange (not really) that low res gaming was so strongly denigrated (niche within a niche comes to mind from one poster) when Intel led more at low res, but now is somehow becoming the holy grail since that appears to be where AMD has the lead.

You're more than welcome to admit true gaming performance should be evaluated at higher resolutions, as long as that comes with the obvious admission that Zen 2 was within 3% of Skylake at gaming as well.

Also, mind your choice of benchmarks before RKL-S comes in, it may close some doors behind you.

 

[uzzi38]

The 720p results he showed both have the Zen 3 systems running higher memory speeds. That's my point. Run all systems at the same speed and run the 720p tests again and watch the Intel systems do significantly better. 384p? Assuming we should even take that seriously, Zen 3 could probably fit a whole lot of the data in cache, and those that don't will go to a faster memory? TPU tested all systems with same memory speed in the 720p test, albeit with a 2080Ti, and the results were very different.

1. Why are we still talking about TPU's results after TPU themselves said they recognised their results were far off base to what others saw and decided they needed to retest?

2. Resolution will have a minimal effect on what get stored in CPU cache. You'd have a point if you were talking about GPU cache, but you're not.

3. Like I said before, with tuned, equalised memory across both Intel and AMD platforms you have both LTT and GN showing that even at a higher resolution Zen 3 takes a very noticeable lead. (EDIT: and at 1080p) So your point about the cache is even more useless.

 

[lobz]

Well, i dont game at 720p, so.......

Edit: Strange (not really) that low res gaming was so strongly denigrated (niche within a niche comes to mind from one poster) when Intel led more at low res, but now is somehow becoming the holy grail since that appears to be where AMD has the lead.

I remember the opposite. It has always been used as a trouncing card against the gaming viability of Zen CPUs.

We feel your pain.

 

[jpiniero]

Also, mind your choice of benchmarks before RKL-S comes in

That reviewers mostly won't use the same memory makes Rocket Lake almost impossible to project at this point.

3. Like I said before, with tuned, equalised memory across both Intel and AMD platforms you have both LTT and GN showing that even at a higher resolution Zen 3 takes a very noticeable lead. So your point about the cache is even more useless.

I am looking at the Sweclockers review and outside of 720p it's pretty close between the 10900K and the 5900X/5950X. Maybe only a couple percent. Set to apparently 3600 the 10900K even wins a couple although it could be GPU limited. This includes games where the 5950X is 20%+ faster than the 3900X.

 

[Kenmitch]

Low res gaming results were used as both PUN and salt in the wounds by reviewers. I'm not sure if it was the intent, but that's the way I'm leaning.

I don't really see any reason to argue about the results as they speak for themselves.

Those arguing against low res gaming most likely were spewing it's importance in the past. Those defending it now most likely were downplaying it's importance in the past.

I wouldn't base my cpu purchasing on low resolution gaming, but it does serve a purpose in the end.

 

[lobz]

1. Why are we still talking about TPU's results after TPU themselves said they recognised their results were far off base to what others saw and decided they needed to retest?

Send nudes or didn't happen.

Joking aside, I don't think anybody took them seriously until Ondma came in and said triumphantly that the stock 8700K is 3% behind Zen 3.

 

[inf64]

Send nudes or didn't happen.

Joking aside, I don't think anybody took them seriously until Ondma came in and said triumphantly that the stock 8700K is 3% behind Zen 3.

Read the user's name backwards(<-) and you will figure it out

 

[scannall]

Low res gaming results were used as both PUN and salt in the wounds by reviewers. I'm not sure if it was the intent, but that's the way I'm leaning.

I don't really see any reason to argue about the results as they speak for themselves.

Those arguing against low res gaming most likely were spewing it's importance in the past. Those defending it now most likely were downplaying it's importance in the past.

I wouldn't base my cpu purchasing on low resolution gaming, but it does serve a purpose in the end.

I've always found the ultra low rez gaming benchmarks to be interesting, if not terribly relevant to current games and systems. More of an amusement than anything really useful.

 

[yeshua]

another Spec bench in zhihu by Edison Chen
compiler GCC 10.2 -march=znver2 for AMD, -march=native for Intel, -Ofast and other compatible flag，They are verified on the SPEC tools as a meaningful performance to upload.
third lib Jemalloc 5.21
SPEC CPU 2017 1.1 tune: base, 3 loops

author：Edison Chen
link：https://www.zhihu.com/question/428994199/answer/1561075014
Sourse : zhihu

A nice chart: shows just how great the Sky Lake uArch is as it's been able to compete successfully for five years in a row with almost zero underlying changes (aside from the number of cores/frequencies and microarchitectural tweaks to fix HW vulnerabilities like Meltdown). Thanks! Also puts to rest the claims that the Ryzen 5000 series is universally faster than ages old Sky Lake.

Is a Noctua NH-U12A adequate for cooling a Ryzen 5800x? I purchased this cooler about 2 weeks ago to be installed in my next rig.

More than.

The 5800X runs pretty hot in all-core workloads due to 142W PPT being spread across 8 cores. I'm not sure if the NH-U12A is going to be enough.

I don't know a single person who runs power viruses 7x24.

Quite happy with my decision to buy a 5900X. Out of the too three Ryzens, it definitely looks like it offers the best perf/value ratio if you plan to keep using your build for a long time.

It's because both 5600X and 5800X are overpriced junk in terms of performance per dollar - and this is where AMD has lost their mind. Many-cores CPUs should be a lot more expensive per core because they are used by professionals who actually can recuperate their cost.

This time around both 5600X and 5800X are absolute worst in this regard.

Eye watering MT performance as well. When a stock 16C/32T Zen3 chip on a dual channel mainstream platform utterly destroys a 4.8Ghz SKL-X 18C/36T with quad ch. memory, it's PG13 content : https://www.tomshardware.com/news/amd-ryzen-9-5950x-5900x-zen-3-review

What a win against a five years old uArch running on a much inferior node! Congrats! AMD fans should celebrate it until the rest of their lives. What next, should we compare Ryzen 5000 to Sandy Bridge CPUs for good measure? Alright! An amazing win!

Eye watering MT performance as well. When a stock 16C/32T Zen3 chip on a dual channel mainstream platform utterly destroys a 4.8Ghz SKL-X 18C/36T with quad ch. memory, it's PG13 content : https://www.tomshardware.com/news/amd-ryzen-9-5950x-5900x-zen-3-review

It must take a lot of effort to find tasks which are going to be seriously affected by running on two CCX'es simultaneously. 5800X defeats 5900X in certain scenarios but by a very small margin, however if you're going to run truly parallelized tasks like rendering, video encoding, compilation, AI - then obviously 5900X will be a much better choice not to mention it costs less than 5800x in terms of price per core.

Also 5900X runs quite cooler than 5800X:

https://tpucdn.com/review/amd-ryzen-9-5900x/images/cpu-temperature.png

https://tpucdn.com/review/amd-ryzen-7-5800x/images/cpu-temperature.png

So, if I had the money I'd buy the 5900X in an instant however given the new pricing policy I'm not touching Ryzen 5000 until prices go down and there's a sale. Besides I'm waiting for 5700X regardless - I've no desire to put a 142W CPU in my case.