Jump to content
Sign in to follow this  
Dasa

AMD Zen

Recommended Posts

Posted (edited)

Latency and bandwidth worse than 1700 in there tests which suggests something is wrong but despite that it is at least competitive with a stock 6700k although we have no idea what RAM was used in the other systems.

Edited by Dasa

Share this post


Link to post
Share on other sites
13 minutes ago, Dasa said:

Latency and bandwidth worse than 1700 in there tests which suggests something is wrong but despite that it is at least competitive with a stock 6700k although we have no idea what RAM was used in the other systems.

Yeah. We have to wait until it falls into the hands of the mainstream reviewers. I am not sure how extensive or skillful their tests are...

 

We need to know the ram, timings, etc that the others were using. We also need to see how much better running it on x570 with the 3733 "sweet spot" is. And overclocking the CPU

 

But it does at least show that the ryzen 3000 series will be a great gaming CPU I think.

Share this post


Link to post
Share on other sites

It was using the box cooler... So there's a few percent there as well.

 

The memory latency/performance and such is concerning though. This is why it has such a large cache. This is just the price of going chiplet, but being the first to work out the kinks might be a good long term strategy for AMD. 

Share this post


Link to post
Share on other sites

I do think we could have more latency as how it is designed...  More cache, like you said might help some... I wonder if it really matters if tests still show improvements and performance is there.  AMD mentioning 3733 sweet spot, so maybe it will improve slightly with those ram kits.

 

Someone did scaling for single core scores (9900k is 204):  If it scales to the 4.4ghz of the 3700x and 4.6ghz of the 3900x, we can guess those should score 205 and 214 respectively

Share this post


Link to post
Share on other sites

The earlier slide from AMD shows 69ns latency with 3200c14 which is close to 2600 and about the worst I would expect from 3600

There is definitely something up with that review at 80ns.

Share this post


Link to post
Share on other sites

I don't think El Chapuzas are very competent reviewers. Once we get the results from Anandtech, Tom's, Hardware Unboxed, Gamers Nexus, etc.. then we won't have any doubts.

Share this post


Link to post
Share on other sites
1 hour ago, SceptreCore said:

I don't think El Chapuzas are very competent reviewers.

Agreed

They have added X570 results now and the memory bandwidth anomaly remains but the latency is a little better

https://elchapuzasinformatico.com/2019/06/amd-ryzen-5-3600-x570-review/

Looking at the Intel results the bandwidth is about right for using the same 3200c14 kit but the latency for them is also ~10ns higher than it should be which makes me think they have a messy OS install.

Share this post


Link to post
Share on other sites

Mmmm... so it's hard to know if even these results can be trusted and that they fall into the acceptable margin of error. 3% uplift in TWW 2, also don't know if that's due to the board or not.

Share this post


Link to post
Share on other sites
Posted (edited)

Well if X570 really did drop latency by ~6% that could easily make a CPU limited game over 3% faster.

Or maybe it is just margin of error in there tests time will tell I guess.

Edited by Dasa
  • Like 2

Share this post


Link to post
Share on other sites
38 minutes ago, Dasa said:

Or maybe it is just margin of error in there tests time will tell I guess.

Yeah ... well.. if we can pick holes in some tests... then we can pick holes in all tests. Just gotta wait for the reviewers who know what they're doing. Just two more weeks.

Share this post


Link to post
Share on other sites
Posted (edited)

So just for kicks I thought I would see how much of a performance hit my 6700k takes if I try make the latency as bad as possible.

So here it is at 3200c21 any higher and it would fail to post.

This resulted in 59ns vs 42ns@3200c14 a 33% difference.

 

Cinbench 20 multi 2167 vs 2178 0.5% difference

Farcry5 83 vs 102FPS 20% difference

Arma III 53FPS vs 63FPS 17% difference <The reason the difference in Arma is lower than farcry is probably due to the lower FPS.

3200c14.jpg

3200c21.jpg

 

Edited by Dasa

Share this post


Link to post
Share on other sites

A huge difference in gaming performance.

 

Did you see those kits from G.Skill at computex that were ridiculously high speed and low latency?

Share this post


Link to post
Share on other sites

Yeah 3600c14 on X570 is nuts my kits may be able to do that but I would need to feed them over 1.5v

I know a lot of people have done ~4000c12 unstable with ~2v on b die for benchmarking

Share this post


Link to post
Share on other sites

Performing better at 3600 and boosting 150MHz higher.

65-70ns is where it should be.

Share this post


Link to post
Share on other sites

Final BIOS for X570 on the 7/7 will open overclocking functionality. So with the IMC OC'd we might squeeze some more out of it?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×