Ryzen 9000 officially announced for July 31st Release

TonyCarter

VALUED CONTRIBUTOR
Here come the reviews (R7 9700X)...but their R5 9600X was DOA (replacement was fine though)...and pricing is not looking particularly favourable to the 7000 series CPUs :unsure:





My summary of it would be "meh", and that was before I got to the end of the video where Tech Jesus (Tesus?) agreed with me:
Looks like the only way they're really better is in efficiency/temps, but whether you can add more power to give more performance hasn't been assessed yet. But there may also be improvements once the BIOS/AGESA code is optimised.

For now, I wouldn't recommend building anything with the 9000 series.
 
Last edited:

Scott

Behold The Ford Mondeo
Moderator
Looks poor but the headroom could be interesting. I think you could unlock a ton of performance given the TDP headroom there.
 

Paul1964

Silver Level Poster
The single core performance looked good but these are lacking cores compared to their Intel equivalents leading to disappointing benchmarks.

Lets see what the higher core counts of 9900X and 9950X look like. That could be where the real gains are ... we should find out next week.
 

SpyderTracks

We love you Ukraine
The single core performance looked good but these are lacking cores compared to their Intel equivalents leading to disappointing benchmarks.

Lets see what the higher core counts of 9900X and 9950X look like. That could be where the real gains are ... we should find out next week.
They’re not lacking cores, Intel are a Big / Little architecture, P cores are not the same as ecores, so an i9 is not a 24core cpu, it’s an 8+16 which is a VERY different thing. It’s part of Intel’s marketing to fool you into thinking it’s loads of cores

I do wonder if there will be following microcode and BIOS updates that unlock performance.

The efficiency is great, Single core is ok, multicore IMHO is really poor
 
Last edited:

Scott

Behold The Ford Mondeo
Moderator
Mutlicore looks to be deliberately power limited. It'll be the only saving grace for the chip IMO. If it can't be unlocked for serious gains then I think they definitely got the naming wrong and the 9700X should actually be called the 9600X to give a worthwhile jump in performance.

I honestly don't understand what AMD are doing with this release. I don't see their claims stacking up at all, outside of Photoshop. Are they trying to retire the 8 core and below? o_O
 

TonyCarter

VALUED CONTRIBUTOR
They're saving the 'K' 'suffix for their unlocked versions to grab the Intel fanboys who won't buy anything without a 'K' at the end!

They probably won't release the 9800X3D-KS until the very end ;)
 

Scott

Behold The Ford Mondeo
Moderator
I may be wrong here but I was positive that previous releases the top tier was out first followed by the lesser variants. This felt like a cash grab to get desperate early adopters buying the more expensive premium chips rather than waiting for the newer lower tier offerings.

Now we have the new chips releasing today but they're the lower tier models.

My guess is that the reason for this is they want the early adopters buying the trash release and not holding off for the actual decent chip(s) next week.

Your girl is disappointing me @SpyderTracks 😬
 

SpyderTracks

We love you Ukraine
I may be wrong here but I was positive that previous releases the top tier was out first followed by the lesser variants. This felt like a cash grab to get desperate early adopters buying the more expensive premium chips rather than waiting for the newer lower tier offerings.

Now we have the new chips releasing today but they're the lower tier models.

My guess is that the reason for this is they want the early adopters buying the trash release and not holding off for the actual decent chip(s) next week.

Your girl is disappointing me @SpyderTracks 😬
This is what confused me because leaked benchmarks of the 9950x look insanely good in multithreaded, of course may not be accurate, or perhaps that was with power limits removed but I suspect the 9900x and 9950x will be a rather different story.

But I’m very disheartened so far.
 

Scott

Behold The Ford Mondeo
Moderator
Just seen that HUB did the PBO testing that I was looking for..... uh oh!
 

TonyCarter

VALUED CONTRIBUTOR
Seems to perform slightly better on Linux benchmark suites, so wonder if there's a Windows issue in here somewhere as well?


2024-08-08_10-59-09.jpeg
 

SpyderTracks

We love you Ukraine
Seems to perform slightly better on Linux benchmark suites, so wonder if there's a Windows issue in here somewhere as well?


View attachment 41818
That’s very interesting, as you say could this be a windows scheduler issue for some reason? Don’t really see why as same core design and quantity

Perhaps windows chipset driver needs updating for 600 series boards?
 

Paul1964

Silver Level Poster
They’re not lacking cores, Intel are a Big / Little architecture, P cores are not the same as ecores, so an i9 is not a 24core cpu, it’s an 8+16 which is a VERY different thing. It’s part of Intel’s marketing to fool you into thinking it’s loads of cores
I understand what you are saying but benchmarks done by Jayz2Cents showed they were up there with the best on single thread but way behind on multithreaded benchmarks. Yes the e-cores are weaker, but more power efficient, than the p-cores and not hyperthreaded either but they will still make a meaningful contribution in the multithreaded synthetic benchmarks.
 

SpyderTracks

We love you Ukraine
I understand what you are saying but benchmarks done by Jayz2Cents showed they were up there with the best on single thread but way behind on multithreaded benchmarks. Yes the e-cores are weaker, but more power efficient, than the p-cores and not hyperthreaded either but they will still make a meaningful contribution in the multithreaded synthetic benchmarks.
That's what I mean though, I think there's more to this currently, I don't think we're seeing full performance from these chips as yet due to buggy release versions of drivers etc.

ecores aren't more power efficient at all, they are against the P cores, but that's no decent efficiency, they're still awfully inefficient. eCores are atom architecture which is god awful, they only reason they've done it (IMHO) is because they had buckets of wafers left over they needed to make some capital on.

But in an awful lot of workloads on Intel, it's still best advice to disable eCores fully, including gaming.
 

TonyCarter

VALUED CONTRIBUTOR
If my business depended on it, I'd rather have a AMD CPU that can run, flat-out, all day every day, and deliver 95% of the performance of the 'best' CPU, than one that will throttle after an hour and probably crash every few hours (or need to be artificially under-powered so that it's slower than the CPU I should have bought).
 
Top