-
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.
AMD Zen 5 rumours
- Thread starter Grim5
- Start date
I know it's probably my not on the cards but I'm kinda of hoping we get at least a further two releases on AM5.
Because even a '10,000' series CPU will be a bit of a lacklustre upgrade from a 7800x3d for most use cases and one of the major reasons a lot of people bought into AM5vwas thr prospect of a significant uplift over its lifetime (maybe even not if as big as the jump from say a 1800x to 5800 x3d where it seems anything from 40% to well over 100% gains were seen in games)
Are we really looking at DDR6 in the consumer sector that soon?
DDR5 hasn't seen that much improvement in its speeds compared to previous generations of ddr memory yet...
And in AMD's case the 9000 series doesn't seem to have advanced workable memory speeds at all (unless this can be improved later on)
Caporegime
For me that just reinforces my theory about the Windows Scheduler.
Whatever the case is, it looks like AMD has changed something, its just whether its AMDs change thats the issue, or whether Windows is having a hard time dealing with the change and needs an update, or maybe a combination of drivers/bios and windows updates is needed
It would be great if AMD can quickly clarify what has happened as the media is just running around trying to guess whats wrong
Last edited:
To be fair something doesn't seem to have gone right somewhere - I don't think the older AMD slides and developer comments were entirely made up. The latency is an odd one but given some of the issues with a minority of reviewer sample CPUs I suspect volume production quality isn't up to expectation but too late in the day to fix for this release - sometimes the 9950X is behind the 7950X even.
When you predict a million things, your bound to get a few right. He gets more wrong than right. Yet you still watch his content
Actually it looks like whatever AMD did its not just CCX to CCX latency affected, though that is the worst. A thread jumping between the CCXs has a latency penalty of 250% over Zen4, however the latency of a thread jumping from one core to another within the same CCX also has a latency increase of 40% compared to Zen4.
Meaning even single CCX Zen5 CPUs like the 9700x are suffering from higher latency between the cores. Can Windows fix that?
And both of these scenarios where the latency is increased is occurring on both the Ryzen 9000 desktop parts and also the Strix point AI HX mobile Zen5 parts.
Meaning even single CCX Zen5 CPUs like the 9700x are suffering from higher latency between the cores. Can Windows fix that?
And both of these scenarios where the latency is increased is occurring on both the Ryzen 9000 desktop parts and also the Strix point AI HX mobile Zen5 parts.
Last edited:
even talking to SMT thread on same core increased latency from 6-7ns to 15-16ns
If it is an issue with Windows they can't really say so, last thing they want to do is pee off MS.
Soldato
The thing with all this latency talk, AMD clearly can't have missed that and then they have come out and told us that performance seen in reviews is what they expected so it all seems like AMD claiming nothing wrong and it is what it is.
People on forums claiming must be issues and defending probably the worst launch of a CPU for a decade!
People on forums claiming must be issues and defending probably the worst launch of a CPU for a decade!
Latency doesn’t explain the hit and mis performance between tasks. Is it a memory flushing problem?
I’m tempted to buy a single and dual CCD chip and test. It could make me a YouTube tech journalist billionaire sensation, over night.
I’m tempted to buy a single and dual CCD chip and test. It could make me a YouTube tech journalist billionaire sensation, over night.
Last edited:
Man of Honour
That's too much bother. You should film yourself flushing a Zen 5 processor down the toilet and be guaranteed to become a YouTube hit.
After going through all the reviews last night, this is concerning. Luckily I can wait until X870 is out, hopefully they will have worked out what the issue is by then and patched it. If not it looks a choice between the 7950X and 9950X on an X870 board which is a crazy situation considering we have just had a refresh.
I don't care about gaming performance but the performance variation on the productivity apps is just crazy
I don't care about gaming performance but the performance variation on the productivity apps is just crazy
I actually think latency can explain all of it
In single thread Zen 5 is great, core-to-core latency doesn't affect it
In multithreaded tasks of embarassingly parallel nature (rendering etc), threads don't need to talk to each other, latency doesn't matter only thoughput does. So Zen 5 is good but sometimes held back by IOD bandwidth.
But there are a few benchmarks and modern games where its not fully loading all threads and they do need to talk. And thats the hit and miss area.
The few linux games benchmarked are mostly single thread and don't expose latency. Linux vs Windows is a red herring. Linux tests in phoronix are just dominated by embarassingly parallel server optimised benchmarks
This now looks like a repeat of first gen Threadripper. Chiplets communicated to each other via main memory and caused all kinds of weirdness
has any reviewer published results WITHOUT core parking properly done? I bet that would be all over the place too
Last edited: