Yes Wendell know his stuff. He found out about the scheduling issues with Threadripper iirc.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
OcUK Ryzen 3000/Zen 2 review thread
- Thread starter CAT-THE-FIFTH
- Start date
Yes Wendell know his stuff. He found out about the scheduling issues with Threadripper iirc.
Looking across a whole range of reviews, the 1% and 0.1% lows seem to be all over the place for both AMD and Intel between different testers.
Gamers Nexus, for example, have 9700k/9900k processors showing terrible lows on some of their charts; no one else shows anything similar.
I think the best video I've seen so far has to be Richard's of Digital Foundry, as he really does highlight all the challenges of even attempting to come up with consistent results.
Then you've got stuff like this: https://www.youtube.com/watch?v=Cm9W6YHLa_s It shows dramatic improvement in the very games (Ubi) that Ryzen seems to lose most of it's performance compared to Intel. And yet Gamers Nexus didn't seem to show the same thing in their SMT on/off testing. People in the comments section of that video are also of the belief that there are still tweaks to the Windows scheduler from both Microsoft and AMD to come... No idea how true that is.
Soldato
Digital foundry used the stock cooler on the 3700X and an impressive AIO on the 9900k, no advantage to intel there then.
Doesn't change the thrust of his review, though: that getting consistent, repeatable test data from either Intel/AMD processors that includes stutter/dips/lows/etc. was still next to impossible. None of the simple bar graphs from other reviews show that. I still think his review was very well done. I mean, when the average frames rates are so high on all of these modern processors, it's only the lows/dips that really matter. Also, he was very positive on AMD and equally fair on Intel, so I can't see any room for complaint.
Associate
- Joined
- 21 Sep 2018
- Posts
- 895
Except the fact that he said he made things equal between the systems and just recently we found out that Ryzen 3000 performs optimally when cpu temp is maintained at or below 55c.
Yeah, best of luck with that on the stock cooler - which is what 99.9% of Ryzen buyers are going to be using..
To quote Gamers Nexus: "The chart starts at about 84 degrees Celsius, which is where you might be sitting with a 3900X with the stock cooler and with the average, modern case. ..a 55-degree load temperature is achievable primarily with high-end cooling solutions"
Any boost from such expensive cooling, in gaming, is pretty much unnoticeable in play. So his results are still valid, especially for the most common usage experience. And the "lows", which is all that really seems to matter with these modern processors, was still very well covered, IMHO.
Associate
- Joined
- 21 Sep 2018
- Posts
- 895
3900X. The vid you linked is an 8-core against an 8-core without HT in an AIO. Pretty even?
So, if the R7 was using custom and the 9700K was on air would that be fine with you?
https://imgur.com/mH8WcQ0
Last edited:
Soldato
Thing is, his results have the 3800x more or less equal to a 9900k in games.
No other reviewer has the two CPU's that close. 99% of them show still quite a big gap between the 9900k and the 3800x.
And we are calling this legit?
Associate
As for gaming, with my 3700x and knowing the benchmark results,
The 3700x or 3800x is around the intel 7700k level or a bit faster, between 7700k and 8700k > https://forums.overclockers.co.uk/posts/32883506 for gaming i think, if you disagree let me know.
I did a cpu z benchmark and according to that i beat a 9900k at multi thread benchmark > https://valid.x86.fr/bench/sstwfq/16
Or single thread same as a i5 9600k > https://valid.x86.fr/bench/sstwfq/1
Dan.
The 3700x or 3800x is around the intel 7700k level or a bit faster, between 7700k and 8700k > https://forums.overclockers.co.uk/posts/32883506 for gaming i think, if you disagree let me know.
I did a cpu z benchmark and according to that i beat a 9900k at multi thread benchmark > https://valid.x86.fr/bench/sstwfq/16
Or single thread same as a i5 9600k > https://valid.x86.fr/bench/sstwfq/1
Dan.
Soldato
Yes they are legit, for one he is using a 2070 super which will put less stress on the CPU than a 2080ti would. Goes to show how close performance is even with a $500 GPU.
Soldato
So @lltfdaniel and @RavenXXX2 comments seem to contradict each other. lltfdaniel is saying his 3700x is performing between a 7700k and 8700k whilst RavenXXX2 comments say that Level1Techs results that show the 3800x is on par with a 9900k are legit.
These comments are mutually exclusive!
These comments are mutually exclusive!
Associate
My comments should only be taken with a pinch of salt to be honest.
According to my cpu-z benchmarks, they agree with level1techs results, 9900k or 9600k.
Don't forget about games need to be optimised for amd processors as well.
edit,
You decide what is ****** but the benchmarks i did with cpu-z agrees with level1tech it beats the 9900k from what i have gathered or the same as a 9600k yeah.
Out of every benchmark in existent regarding the 3700x regarding games well.... i will still stand by it being between 7700k or 8700k with the statement i gave, but cpu z benchmarks tell a different story.
Yes between 7700k and 8700k that is not on 720p or 1080p more like 4k gaming.
Sometimes the 3700x is a little bit behind the 7700k though on gaming benchmarks at the end of the day you decide what is right or wrong, i only gave my estimate from researching results and knowing my own benchmarks so take it with a pinch of salt.
Ram speeds are also important some reviewers use slower ram, but i think these processors should be benched with no bottleneck instead of slow ram.
Dan.
Last edited:
Those fans look nice, what are they?Just done a quick 10min Eisbaer LT360 modAll controlled by Icue. Looks nice
These are Uphere PWM RGB T7SYC7 fans. Managed to connect them all to Icue
Wendell is pairing the £380 cpu with a £400 gpu and a £500 gpu not the £1000+ gpu which is the £2080 Ti, commonly used for benchmarks despite having a very small userbase.
A wide range of tests was not demonstrated.
You should expect Wendells results when using the hardware he is using and the benchmarks he is using.
I am in two minds.
One side of me thinks given the 3900X is a cpu for enthusiast users and not mainstream market, then its logical those users will buy high end cooling.
However the other side of me says if AMD are going to supply cooling with the chip then its fair and reasonable to expect that cooling to be tested and not just sidestepped. If they dont like that then supply a better cooler, simples.
So DF's review doing it no problem for me, it evens things out a bit as I expect the fanboy reviewers probably all used high end cooling anyway (with it also pegged at 100% static speed), so now that means consumers have stock cooling figures to digest. Which can only be a good thing.
AMD gives you a cooler because they want people to be able to buy the CPU, drop it in the socket, and use it. Its a more user friendly experience and saves money as well. If you want to use CPUs to their full potential then you need a high end cooler, specially with the 3900X. Intel on the other hand does not give you anything, which means more money to spend, and you cant use their K CPUs until you buy a 3rd party cooler.