AMD Zen 5 rumours

[JediFragger]

I'd not heard that before, so for anyone else here's an article about it - https://www.tomshardware.com/pc-com...heyll-be-close-improved-3d-v-cache-coming-too

11th June, check back a few pages and try to keep up

 

[Rob_B]

The pages appear too fast, no way I'm keeping up with this at my age

 

[Robert896r1]

It reminds to be seen, so far I've not seen tests that would show definite advantage of faster memory Vs 1:1 one with tighter timings.

Until fabric is widened, stick to 1:1 outside of vanity number for aida screenshots.

 

[LtMatt]

It reminds to be seen, so far I've not seen tests that would show definite advantage of faster memory Vs 1:1 one with tighter timings.

The difference is not huge on X3D parts, helps more on Zen 4 non X3D parts. For reference, I saw on average a 6% performance difference with 8000C34 vs 6000C30 tuned and running in 1:1 mode on my 7950X3D. Slightly higher in 1% lows too.

So, there is a performance advantage to using faster memory, despite not running in 1:1 mode. If you want every drop of performance on the table, 8000C34 is what you want. You might need a Gene to achieve it though.

I've seen a few lucky Gigabyte users get up to 8200Mhz on Zen 4, but for most it's 8000Mhz or lower. A lot of 2DPC boards will be stuck at 7600/7800Mhz.

Be interesting to see if there are improvements with Granite Ridge and X800 Chipset.

 

[wookiee87]

POWER!

AMD Ryzen 9 9950X gets tested at 253W and unlimited PPT settings - VideoCardz.com

AMD Ryzen 9 9950X tested in Blender again, now with unlimited power setting The 9950X ends up 35% faster than 14900K at the same power level. Courtesy: Lucasfilms Ltd./AMD AnandTech forum user “Igor_Kavinski” published another test featuring Blender and the Ryzen 9 9950X processor. It’s...

videocardz.com

Imagine what x3d will be like

 

[CaptainRAVE]

POWER!

AMD Ryzen 9 9950X gets tested at 253W and unlimited PPT settings - VideoCardz.com

AMD Ryzen 9 9950X tested in Blender again, now with unlimited power setting The 9950X ends up 35% faster than 14900K at the same power level. Courtesy: Lucasfilms Ltd./AMD AnandTech forum user “Igor_Kavinski” published another test featuring Blender and the Ryzen 9 9950X processor. It’s...

videocardz.com

I mean at this point all it needs is to have less than a 50% failure rate and it’ll be a success

 

[pete910]

Anyone know when the review NDA lifts on these? I guess reviewers have them by now.

 

[Tinek]

The difference is not huge on X3D parts, helps more on Zen 4 non X3D parts. For reference, I saw on average a 6% performance difference with 8000C34 vs 6000C30 tuned and running in 1:1 mode on my 7950X3D. Slightly higher in 1% lows too.

So, there is a performance advantage to using faster memory, despite not running in 1:1 mode. If you want every drop of performance on the table, 8000C34 is what you want. You might need a Gene to achieve it though.

I've seen a few lucky Gigabyte users get up to 8200Mhz on Zen 4, but for most it's 8000Mhz or lower. A lot of 2DPC boards will be stuck at 7600/7800Mhz.

Be interesting to see if there are improvements with Granite Ridge and X800 Chipset.

Interesting. The tests I've seen never went as high as 8000MT (likely they weren't lucky enough to push that high on their particular CPUs). Good to know it can go (with luck) that high and it gives proper perf. increase

BTW, related to x3D - upon updating BIOS to AGESA 1.2.0 I can now see in Windows it defaulting always to 3D cache cores, irrelevant of bios settings, drivers and even app that can manually choose between perf and 3D cache cores (through drivers). Problem is I also updated drivers and Windows to new build at the same time, so not sure now if it's that BIOS issue or the AGESA messed it up a bit or Windows... Not that it cause any issues in games but outside games I liked it to use perf. cores as priority, which doesn't happen now. Oddly, not seen any reports about it online, so it might be just MSI messing things up a bit here.

 

[LtMatt]

Interesting. The tests I've seen never went as high as 8000MT (likely they weren't lucky enough to push that high on their particular CPUs). Good to know it can go (with luck) that high and it gives proper perf. increase

BTW, related to x3D - upon updating BIOS to AGESA 1.2.0 I can now see in Windows it defaulting always to 3D cache cores, irrelevant of bios settings, drivers and even app that can manually choose between perf and 3D cache cores (through drivers). Problem is I also updated drivers and Windows to new build at the same time, so not sure now if it's that BIOS issue or the AGESA messed it up a bit or Windows... Not that it cause any issues in games but outside games I liked it to use perf. cores as priority, which doesn't happen now. Oddly, not seen any reports about it online, so it might be just MSI messing things up a bit here.

If you've ever looked at HUBs testing, they show a similar performance increase on Intel/AMD as you can from 6000Mhz and up. I'm not sure if HUB actually tune the other timings though outside of primaries.

 

[Nutty667]

Anyone know when the review NDA lifts on these? I guess reviewers have them by now.

Already in shops too!

AMD Ryzen 9000 Series Zen 5 Desktop CPUs Arrive at Retailers Already

New photos have emerged about AMD's Ryzen 9000 series using the Zen 5 architecture, including the Ryzen 7 9700X, an 8-core processor. Despite lacking official release dates, some units have reached consumers prematurely through leaks.

www.guru3d.com

"A Reddit user reported purchasing the Ryzen 7 9700X ahead of its official market release. This indicates that some units are circulating outside official channels. "

 

[LtMatt]

Interesting. The tests I've seen never went as high as 8000MT (likely they weren't lucky enough to push that high on their particular CPUs). Good to know it can go (with luck) that high and it gives proper perf. increase

BTW, related to x3D - upon updating BIOS to AGESA 1.2.0 I can now see in Windows it defaulting always to 3D cache cores, irrelevant of bios settings, drivers and even app that can manually choose between perf and 3D cache cores (through drivers). Problem is I also updated drivers and Windows to new build at the same time, so not sure now if it's that BIOS issue or the AGESA messed it up a bit or Windows... Not that it cause any issues in games but outside games I liked it to use perf. cores as priority, which doesn't happen now. Oddly, not seen any reports about it online, so it might be just MSI messing things up a bit here.

Try this.

Ryzen 7950X3D, 7900X3D, 7800X3D

with a 7800x3d and some gigabyte auros ****** b650 mobo, is it worth it bothering for the new chipset drivers?

forums.overclockers.co.uk

 

[humbug]

POWER!

AMD Ryzen 9 9950X gets tested at 253W and unlimited PPT settings - VideoCardz.com

AMD Ryzen 9 9950X tested in Blender again, now with unlimited power setting The 9950X ends up 35% faster than 14900K at the same power level. Courtesy: Lucasfilms Ltd./AMD AnandTech forum user “Igor_Kavinski” published another test featuring Blender and the Ryzen 9 9950X processor. It’s...

videocardz.com

Imagine what x3d will be like

There is almost no difference between unlimited PPT and 230 watts, the difference tho between the 14900K at 253 watts and the 7950X at 230 watts is 30%, or 36% at the same 253 watts

 

[wookiee87]

There is almost no difference between unlimited PPT and 230 watts, the difference tho between the 14900K at 253 watts and the 7950X at 230 watts is 30%, or 36% at the same 253 watts

i will admit the unlimited run is a ES chip but that being in the region of 35% faster than a 14900k, lets hope the retail chips continue this trend.
160-230w may be the sweet spot for the top 9950x and if the temps being reported stick, they will be monsters on any cooling (custom is the way to go XD)

 

[Noxia]

Custom water cooling can do one, so much faff for no benefit.

The last grasping at straws benefit I saw was: The components run cooler so they will last many more years.
Oh yes, because somebody who is enthusiast about PCS is keeping their GPU and CPU for years aren't they.

 

[StarShock]

Anyone know when the review NDA lifts on these? I guess reviewers have them by now.

Did see one reviewer say "ask me again tomorrow" yesterday. So if they were not just being facetious, there might be like an unboxing embargo today with release dates. But the only date I've seen thrown around is the 29th and not seen anything to back that up.

 

[Nutty667]

Did see one reviewer say "ask me again tomorrow" yesterday. So if they were not just being facetious, there might be like an unboxing embargo today with release dates. But the only date I've seen thrown around is the 29th and not seen anything to back that up.

If retailers already have units, I would expect reviews to be before then. Maybe 29th is official date of being able to buy them.

 

[Tinek]

Try this.

Ryzen 7950X3D, 7900X3D, 7800X3D

with a 7800x3d and some gigabyte auros ****** b650 mobo, is it worth it bothering for the new chipset drivers?

forums.overclockers.co.uk

Will do then I'm back home! It sounds a bit different from what I'm experiencing (as in, the opposite in a way) but maybe it is about 3dvc cores not getting parked properly outside games too. Worth checking! Otherwise I might go back 1 bios version to see if that changes it.

 

[LtMatt]

Will do then I'm back home! It sounds a bit different from what I'm experiencing (as in, the opposite in a way) but maybe it is about 3dvc cores not getting parked properly outside games too. Worth checking! Otherwise I might go back 1 bios version to see if that changes it.

Also check the BIOS and make sure a X3D preset has not been enabled, as it's possible to force apps to use just the cache cores primarily via the BIOS. The option is called CPPC, should be set to Auto.

 

[Tinek]

Also check the BIOS and make sure a X3D preset has not been enabled, as it's possible to force apps to use just the cache cores primarily via the BIOS. The option is called CPPC, should be set to Auto.

Yep, I've done that as the first thing (it's on auto). Then I even tried to manually change to performance and to driver mode - neither did anything. Then I used a software (forgot its name now) that uses AMD API of the driver to change which cores should be prioritised under Windows - no effect on that either. I suspect a bug in the new bios but will check the driver thing first.

 

[Vimes]

Is it still intended to release the new chipset motherboards later in the year when the 3d variant of the CPU's launch..? Thanks